JP2021071060A - Control condition setting device, control condition setting method, control device and control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a control condition setting device capable of reducing an amount of data required for determining a control condition for appropriately controlling an internal combustion engine mounted on a vehicle. A control condition setting device is estimated from a model representing a distribution of states of an internal combustion engine 10 in a beta distribution when the internal combustion engine 10 is controlled according to the control conditions among a plurality of control conditions under a predetermined operating condition. The control condition selection unit 31 that selects the control condition that maximizes the dispersion of the state of the internal combustion engine 10 and the measured distribution of the state of the internal combustion engine 10 when the internal combustion engine 10 is operated a predetermined number of times according to the selected control condition. In addition, the model update unit 33 that updates the model so that the distribution of the state of the internal combustion engine 10 represented by the model corresponding to the selected control condition approaches, and the internal combustion engine 10 estimated according to the updated model are abnormal. It has a control condition setting unit 34 which sets a control condition in which the probability of becoming a state is equal to or less than a predetermined threshold value as a control condition that can be used in a predetermined operation condition. [Selection diagram] Fig. 4

Description

The present invention relates to a control condition setting device and a control condition setting method for setting control conditions used for controlling an internal combustion engine mounted on a vehicle, and a control device and control for controlling an internal combustion engine using such control conditions. Regarding the method.

The internal combustion engine of a vehicle is required to control the internal combustion engine so that an abnormality such as knocking does not occur. Therefore, a technique for controlling an internal combustion engine by using an output based on a model using a Gaussian process has been proposed (see, for example, Patent Document 1).

For example, the control device described in Patent Document 1 calculates the probability distribution of the probability of occurrence of an output parameter with respect to the value of the control parameter based on the current value of the operation parameter using a model. Then, the control device sets the target value of the control parameter based on the calculated probability distribution of the occurrence probability of the output parameter so that the probability that the value of the output parameter becomes equal to or higher than the target value is closest to the target probability. The model used to calculate the probability distribution of the probability of occurrence of the output parameter is a model using the Gaussian process that outputs the probability distribution of the probability of occurrence of the output parameter when the value of the operation parameter and the value of the control parameter are input. It is composed.

JP-A-2019-157652

However, in the above technique, the amount of data required to train the model is large, and therefore it takes time to be able to appropriately determine the control parameters.

Therefore, an object of the present invention is to provide a control condition setting device capable of reducing the amount of data required for determining control conditions for appropriately controlling an internal combustion engine mounted on a vehicle.

According to one embodiment of the present invention, there is provided a control condition setting device that sets control conditions that can be used to control an internal combustion engine of a vehicle. This control condition setting device sets an abnormal state of the internal combustion engine and a normal state of the internal combustion engine when the internal combustion engine is controlled according to the control conditions for each of a plurality of control conditions under a predetermined operating condition of the internal combustion engine. The distribution of the state of the internal combustion engine estimated from the model is maximized among the storage unit that stores the model that represents the distribution of the state of the internal combustion engine including the beta distribution and the plurality of control conditions under the predetermined operating conditions of the internal combustion engine. The internal combustion engine represented by the model corresponding to the selected control condition is shown in the control condition selection unit that selects the control condition and the measured distribution of the state of the internal combustion engine when the internal combustion engine is operated a predetermined number of times according to the selected control condition. The model update unit that updates the model corresponding to the selected control condition so that the distribution of the engine state approaches, and the internal combustion engine that is estimated according to the updated model among multiple control conditions becomes an abnormal state. It has a control condition setting unit that sets a control condition whose probability is equal to or less than a predetermined threshold as a control condition that can be used in a predetermined operation condition.

In this control condition setting device, when the control condition selection unit assumes that the frequency of abnormal conditions of the internal combustion engine follows a model for the control conditions for each of a plurality of control conditions under a predetermined operating condition of the internal combustion engine. Regarding the control condition when it is assumed that the probability that the internal combustion engine is in an abnormal state is a predetermined probability with respect to the model likelihood representing the plausibility of the detection result of the state of the internal combustion engine with respect to the control condition. The ratio of the observation state likelihood, which indicates the plausibility of the detection result of the state of the internal combustion engine, is calculated, and among a plurality of control conditions, the control conditions in which the ratio is equal to or higher than a predetermined threshold, and the internal combustion engine estimated from the model It is preferable to select the control condition that maximizes the dispersion of the state.

Further, in this control condition setting device, the model update unit sets the measured distribution of the state of the internal combustion engine when the internal combustion engine is operated a predetermined number of times according to the selected control condition, and the control conditions around the selected control condition. It is preferable to further update the model corresponding to the control conditions around the selected control conditions so that the distribution of the state of the internal combustion engine represented by the corresponding model approaches.

According to another embodiment of the present invention, a control device for controlling an internal combustion engine of a vehicle is provided. This control device can be used under each of a plurality of operating conditions of the internal combustion engine, and the distribution of the state of the internal combustion engine including the abnormal state of the internal combustion engine and the normal state of the internal combustion engine. Is measured by a storage unit that stores one or more control conditions in which the probability that the internal combustion engine will be in an abnormal state is equal to or less than a predetermined threshold, and an operating condition detection unit that detects the operating conditions of the internal combustion engine, according to a model representing the beta distribution. It has a control unit that specifies the operating conditions of the internal combustion engine based on the measured signal and controls the internal combustion engine according to any one or more of the control conditions in the specified operating conditions among the plurality of operating conditions.

This control device includes a control condition selection unit that selects a control condition that maximizes the dispersion of the state of the internal combustion engine estimated from the model among a plurality of control conditions under a predetermined operating condition of the internal combustion engine, and a selected control condition. According to the selected control conditions, the distribution of the internal combustion engine state represented by the model corresponding to the selected control condition approaches the measured distribution of the internal combustion engine state when the internal combustion engine is operated a predetermined number of times according to the above. The model update unit that updates the corresponding model and the probability that the internal combustion engine will be in an abnormal state, which is estimated according to the updated model among a plurality of control conditions under the predetermined operating conditions of the internal combustion engine, is equal to or less than a predetermined threshold. It is preferable to further have a control condition setting unit that sets the control condition as a control condition that can be used in a predetermined operation condition.

According to still another embodiment of the present invention, there is provided a control condition setting method for setting control conditions that can be used to control an internal combustion engine of a vehicle. This control condition setting method includes a state in which the internal combustion engine is abnormal and a state in which the internal combustion engine is normal when the internal combustion engine is controlled according to the control conditions among a plurality of control conditions in a predetermined operating condition of the internal combustion engine. A control condition that maximizes the dispersion of the state of the internal combustion engine estimated from a model that expresses the distribution of the state of the internal combustion engine as a beta distribution is selected, and the internal combustion engine is operated a predetermined number of times according to the selected control condition. The model corresponding to the selected control condition is updated so that the distribution of the state of the internal combustion engine represented by the model corresponding to the selected control condition approaches the measured distribution of the state. It includes setting a control condition that is estimated according to the updated model and that the probability that the internal combustion engine is in an abnormal state is equal to or less than a predetermined threshold value as a control condition that can be used under a predetermined operating condition.

According to still another embodiment of the present invention, there is provided a control method for controlling an internal combustion engine mounted on a vehicle. This control method specifies the operating conditions of the internal combustion engine based on the measurement signal measured by the operating condition detection unit that detects the operating conditions of the internal combustion engine, can be used under the specified operating conditions, and is an internal combustion engine. According to a model that represents the distribution of the state of the internal combustion engine including the abnormal state and the normal state of the internal combustion engine as a beta distribution, the probability that the internal combustion engine will be in the abnormal state is one or more control conditions that are equal to or less than a predetermined threshold. Includes controlling the internal combustion engine according to either.

The control condition setting device according to the present invention has an effect that the amount of data required for determining the control conditions for appropriately controlling the internal combustion engine mounted on the vehicle can be reduced.

It is a hardware block diagram of the electronic control device which is one Embodiment of a control device and a control condition setting device. It is a figure which shows an example of the model which shows the distribution of the state of an engine. It is a figure which shows the outline of the parameter which defines the model which represents the distribution of the state of an engine. It is a functional block diagram of a processor of an electronic control device which is related to an engine control process including a control condition setting process. It is an operation flowchart of the control condition setting process. It is an operation flowchart of an engine control process.

Hereinafter, with reference to the figure, a control condition setting device and a control condition setting method for setting control conditions used for controlling an internal combustion engine mounted on a vehicle, and an internal combustion engine are controlled using such control conditions. The control device and the control method will be described. This control condition setting device is a distribution of two states that are discrete and do not depend on the previous state for each combination of operating conditions and control conditions that can be applied to the internal combustion engine in the internal combustion engine mounted on the vehicle. For example, the distribution of the abnormal state in which knocking occurs and the normal state in which knocking does not occur is modeled by approximating the beta distribution, particularly the first-class beta distribution. At that time, this control condition setting device operates the internal combustion engine under the control condition that maximizes the dispersion of the state estimated from the model for each operating condition, measures the state of the internal combustion engine, and measures the state of the internal combustion engine according to the measurement result. Repeat updating the model. Then, in this control condition setting device, the state of the internal combustion engine is abnormal (for example, knocking) according to the model for each control condition with respect to the operating condition in which the reliability of the model for each control condition is estimated to have increased to a predetermined degree. A control condition in which the probability of the occurrence of the above-mentioned condition is equal to or less than a predetermined threshold value is set as a control condition that can be used for controlling the internal combustion engine. The operating conditions are defined by the current operating conditions of the internal combustion engine and external control signals, such as the internal combustion engine speed, load factor, air-fuel ratio, and a combination of two or more of them. Yes, it is a condition that is not the control target of the control device of the internal combustion engine. On the other hand, the control conditions are conditions to be controlled by the control device of the internal combustion engine, such as the ignition timing of the spark plug, the opening / closing timing of the valve, or a combination thereof.

In the embodiment shown below, it is assumed that the distribution of the abnormal state in which knocking occurs and the normal state in which knocking does not occur is modeled.

FIG. 1 is a hardware configuration diagram of an electronic control device which is one embodiment of a control device and a control condition setting device. In the present embodiment, the electronic control unit (ECU) 1 for controlling the engine 10, which is an example of an internal combustion engine mounted on a vehicle, has a communication interface 21, a memory 22, and a processor 23. Further, the engine 10 to be controlled by the ECU 1 has a plurality of cylinders, spark plugs 11 provided in each cylinder, and a variable valve timing mechanism 12 for controlling the valve timing of the intake valve or the exhaust valve of each cylinder. Note that in FIG. 1, for simplification, only one spark plug 11 and one variable valve timing mechanism 12 for one cylinder are typically shown.

The communication interface 21 is an example of a communication unit, and has an interface circuit for connecting the ECU 1 to an in-vehicle network (not shown). Then, the communication interface 21 determines various sensors mounted on the vehicle, for example, a knock sensor 13 for measuring the knock strength of the engine 10, an in-cylinder pressure sensor 14 for measuring the pressure in each cylinder of the engine 10, and the rotation speed of the engine 10. The sensor signal from the rotation sensor 15 to be measured or the air flow meter 16 for measuring the amount of air sucked into each cylinder of the engine 10 is received, and the received sensor signal is passed to the processor 23. The knock sensor 13 and the in-cylinder pressure sensor 14 are examples of a state detection unit, and the rotation sensor 15 and an air flow meter 16 are examples of an operating condition detection unit, respectively.

The memory 22 is an example of a storage unit, and includes, for example, a volatile semiconductor memory and a non-volatile semiconductor memory. Then, the memory 22 stores various data used in various processes executed by the processor 23. As such data, the memory 22 contains, for example, a plurality of control conditions prepared in advance for each operating condition of the engine 10, a measurement result of the state of the engine 10, and the engine 10 for each combination of the operating condition and the control condition. The parameters that define the model representing the distribution of the states of, and the initial values of such parameters are stored.

ここで、パラメータαは、エンジン１０が異常な状態となる頻度を表すパラメータであり、パラメータβは、エンジン１０が正常な状態となる頻度を表すパラメータである。また、関数B(α,β)は、ベータ関数であり、xは、エンジン１０の状態を表す。 As described above, in the present embodiment, the model representing the distribution of the state of the engine 10 is represented by the type 1 beta distribution. That is, the model Beta (x | α, β) is expressed by the following equation.

Here, the parameter α is a parameter representing the frequency with which the engine 10 is in an abnormal state, and the parameter β is a parameter representing the frequency with which the engine 10 is in a normal state. Further, the function B (α, β) is a beta function, and x represents the state of the engine 10.

FIG. 2 is a diagram showing an example of a model showing the distribution of the state of the engine 10 including the normal state and the abnormal state of the engine 10. In FIG. 2, the horizontal axis represents the state of the engine 10 (abnormal state in which knocking occurs: 1, normal state in which knocking does not occur: 0), and the vertical axis represents the probability density for each state. The model 201 represents the distribution of the state when the state of the engine 10 has been measured 40 times in the past and the engine 10 is in an abnormal state in 20 of them. Further, the model 202 represents the distribution of the state when the state of the engine 10 is measured 60 times in the past and the state of the engine 10 is in an abnormal state in 30 times among them. Further, the model 203 represents the distribution of the state when the state of the engine 10 is measured 60 times in the past and the engine 10 is in an abnormal state in 10 times. Further, the model 204 represents the distribution of the state when the state of the engine 10 has been measured 70 times in the past and the state of the engine 10 is in an abnormal state in 60 of them. As described above, in the present embodiment, each model is defined according to the beta distribution, so that the number of times the engine 10 has been in an abnormal state and the number of times that the engine 10 has been in a normal state are close to the higher state. Has a peak. Moreover, the higher the number of measurements, the higher the peak. Further, by using the beta distribution as a model, it can be seen that the distribution of the state of the engine 10 is accurately represented even if the state of the engine 10 is measured several tens of times.

FIG. 3 is a diagram showing an outline of parameters that define a model representing the distribution of states of the engine 10 stored in the memory 22. Tables 300 and 310 shown in FIG. 3 represent parameters α and β for each operating condition for one control condition. Each column of the table 300 and the table 310 corresponds to the individual operating conditions of the engine 10. In this example, the operating condition is represented by a combination of the engine speed and the load factor. Therefore, the parameters α and the parameters β under the operating conditions corresponding to the columns are stored in each column of the table 300 and each column of the table 310. Then, when the parameters α and β are updated for any of the operating conditions and the control conditions in the control condition setting process described later, the columns corresponding to the operating conditions in the table 300 and the table 310 corresponding to the control conditions are displayed. The value of the parameter is rewritten to its updated value.

The processor 23 has one or a plurality of CPUs (Central Processing Units) and peripheral circuits thereof. The processor 23 may further include other arithmetic circuits such as a logical operation unit or a numerical operation unit. Then, the processor 23 controls the engine 10. In the present embodiment, the processor 23 sets applicable control conditions for each operating condition of the engine 10. Further, the processor 23 determines the ignition timing of the spark plug 11 and the valve timing of the variable valve timing mechanism 12 provided in each cylinder of the engine 10 in each combustion cycle according to the applicable control conditions set for each operating condition. Control at least one. In the following, the case where the processor 23 controls the ignition timing will be described, but the processor 23 may also control the valve timing in accordance with the engine control process described below.

FIG. 4 is a functional block diagram of the processor 23 related to the engine control process including the control condition setting process. The processor 23 includes a control condition selection unit 31, a control condition change unit 32, a model update unit 33, a control condition setting unit 34, and a control unit 35. Each of these parts of the processor 23 is realized by a computer program running on the processor 23. Alternatively, each of these parts included in the processor 23 may be a dedicated arithmetic circuit provided in the processor 23. Further, among these units of the processor 23, the control condition selection unit 31, the control condition change unit 32, the model update unit 33, and the control condition setting unit 34 execute the control condition setting process. In the present embodiment, the processor 23 separately executes the control condition setting process for each of the plurality of cylinders of the engine 10. In this case, usable control conditions are set for each cylinder. However, the processor 23 may execute one control condition setting process for two or more cylinders among the plurality of cylinders included in the engine 10. In this case, the set usable control conditions are commonly used for the two or more cylinders.

The control condition selection unit 31 selects a control condition to be updated as a model for each operation condition. In the present embodiment, the control condition selection unit 31 selects the control condition that maximizes the variance of the state estimated from the model in the operating conditions currently applied to the engine 10.

Therefore, the control condition selection unit 31 specifies the operating condition of the engine 10 at the present time based on the sensor signal representing the operating condition received by the processor 23 from the rotation sensor 15 or the air flow meter 16 via the communication interface 21. To do. This sensor signal is an example of a measurement signal representing an operating condition. At that time, the control condition selection unit 31 may specify the operating conditions from these sensor signals according to any of various methods for specifying the operating conditions of the engine 10.

ここで、μ_iは、制御条件z_iについて、これまでに観測された、ノッキングが発生した状態の回数α_i及びノッキングが発生していない状態の回数β_iから算出される、ノッキングの発生割合（すなわち、μ_i=α_i/(α_i+β_i)）を表す。 The control condition selection unit 31 has so far determined the number of knocking states when it is assumed that the knocking occurrence frequency follows the model for each _{control condition z i} under the operating conditions currently applied to the engine 10. The number of times α _i was observed and the number of times in the state where knocking did not occur is the number of times β _i observed so far. (Hereinafter, for convenience of explanation, it is referred to as model likelihood). Calculate L (μ _i ). The number α _i and the number β _i are parameters and engines 10 that represent the frequency of abnormalities in the engine 10 in the beta distribution that defines the model for the _{control condition z i,} which are updated by the model update unit 33, respectively. It can be a parameter indicating the frequency with which no abnormality has occurred. If the model for the control condition z _i has never been updated, the engine 10 has an abnormality in the beta distribution that defines the model representing the state distribution of the engine 10 as the _{number α i} and the number β _i. The initial value of the parameter representing the frequency of occurrence and the initial value of the parameter representing the frequency of occurrence of no abnormality in the engine 10 may be used. Further, the control condition selection unit 31 assumes that the knocking occurrence probability is a predetermined probability Pbd (for example, 0.1) for each _{control condition z i} under the operating conditions currently applied to the engine 10. In this case, the _{probability of the number of knocking states α i} and the number of knocking-free states β _i observed so far (hereinafter referred to as the observation state likelihood for convenience of explanation) L Calculate (Pbd _i ). The predetermined probability Pbd is, for example, an allowable upper limit value of the probability that knocking occurs in the engine 10 (hereinafter, may be referred to as an allowable upper limit probability), which is used for setting usable control conditions, which will be described later. ) Is set. However, the predetermined probability Pbd may be different from the permissible upper limit probability. Then, the control condition selection unit 31 sets a set of control conditions in which _{the ratio λ (z i} ) = L (Pbd _i ) / L (μ _i ) of the observed state likelihood to the model likelihood becomes larger than the predetermined threshold value Λ _T. Ask. L (μ _i ) and L (P bd _i ) are expressed by the following equations.

Here, μ _i is the knocking occurrence rate calculated from _{the number of knocking states α i} and the number of knocking non-knocking states β _i observed so far for the control condition z _i. (That is, μ _i = α _i / (α _i + β _i )).

制御条件選択部３１は、選択した制御条件を制御条件変更部３２へ通知する。 The model likelihood L (μ _i ) becomes smaller as the reliability of the model in the state of the engine 10 becomes lower. Therefore, the above ratio λ (z _i ) becomes larger as the reliability of the model is lower. Therefore, the set of selected control conditions represents a set of control conditions with low model reliability. Therefore, the control condition selection unit 31 selects the control condition that maximizes the variance of the state estimated from the model from the set. As a result, the control condition selection unit 31 can appropriately select the control conditions corresponding to the model to be updated. According to the modification, the control condition selection unit 31 may select the control condition that maximizes the variance of the state estimated from the model from all the control conditions regarding the operating conditions at the present time. In this embodiment, the model is represented by a type 1 beta distribution, so the variance V (x) of the state estimated from the model is the following equation based on the beta distribution parameters α and β representing the model. Is calculated by.

The control condition selection unit 31 notifies the control condition change unit 32 of the selected control condition.

When the set of the above control conditions is an empty set, there is no control condition with low model reliability in the operating conditions applied to the engine 10 at present. Therefore, the control condition selection unit 31 finishes updating the model with respect to the operating condition, and notifies the control condition setting unit 34 that the update of the model is completed.

Alternatively, the control condition selection unit 31 ends the model update with respect to the operating condition even when the number of trials for updating the model reaches a preset upper limit with respect to the operating condition applied to the engine 10 at the present time. Then, the control condition setting unit 34 may be notified that the model update has been completed.

The control condition changing unit 32 sets the control condition notified from the control condition selection unit 31 as the target control condition under the current operating conditions. Then, the control condition changing unit 32 changes the control condition so that the control condition becomes the target control condition. At that time, the control condition changing unit 32 may change the control condition linearly from the current control condition to the target control condition, or change the control condition according to another method for changing the control condition. May be good. Then, when the control condition reaches the target control condition, the control condition changing unit 32 notifies the model updating unit 33 to that effect.

Regarding the current operating conditions, the model update unit 33 determines the state of the engine 10 from the knock sensor 13 or the in-cylinder pressure sensor 14 received by the processor 23 via the communication interface 21 when the control conditions reach the target control conditions. Based on the represented sensor signal, the state of the engine 10 is detected for each of a predetermined number of combustion cycles (for example, 10 to 50 times). This sensor signal is an example of a measurement signal representing the state of the internal combustion engine. In the present embodiment, the model update unit 33 determines whether or not knocking has occurred for each of the predetermined number of combustion cycles, and stores the number of knocking occurrences in the memory 22 among the predetermined number of times. The model update unit 33 may determine whether or not knocking has occurred according to any of various methods for determining whether or not knocking has occurred from these sensor signals.

ここで、パラメータαj、βjは、それぞれ、モデルの更新対象となる制御条件z_j（目標制御条件z_targetを含む）における、モデルを表すベータ分布を規定するパラメータである。また、tricube(z_j,z_target,k)は、tricube関数を表し、定数kは、tricube関数の拡がり度合いを規定する。なお、tricube関数の代わりに、ガウス関数が用いられてもよい。そしてMは、上記の所定回数を表し、mは、所定回数のうち、ノッキングが発生した回数を表す。 Based on the detection result of the state of the engine 10 a predetermined number of times stored in the memory 22, the model update unit 33 states the state of each engine 10 under the target control condition and the surrounding control conditions under the current operating conditions. Update the model. For example, the model update unit 33 represents the distribution of the state of the engine 10 represented by the detection result of the state of the engine 10 a predetermined number of times according to the following equation, with each model of the target control condition and the control conditions around it. The model is updated so that the distribution of the states of the engine 10 is approached.

Here, the parameters αj and βj are parameters that define the beta distribution representing the model under the _{control condition z j} (including the target control condition z _{target) to be updated, respectively.} Further, tricube (z _j , z _target , k) represents the tricube function, and the constant k defines the degree of spread of the tricube function. A Gaussian function may be used instead of the tricube function. Then, M represents the above-mentioned predetermined number of times, and m represents the number of times knocking has occurred among the predetermined number of times.

As shown in the equation (4), the model update unit 33 uses the detection result of the state of the engine 10 for the _{target control condition z target} a predetermined number of times, and uses the detection result of the engine 10 for the _{target control condition z target.} Not only the model of the state distribution, but also the model of the state distribution of the engine 10 with respect to the control condition z _{j around the target control condition z target can be updated.} Therefore, even for the model of the control condition that is not set in the target control condition, the model update unit 33 updates each control condition with a smaller number of measurement data because the distribution of the state of the engine 10 is updated to be able to be expressed to some extent accurately. The accuracy of the model can be improved.

According to the modification, the model update unit 33 may use the likelihood ratio λ (z _{target ) calculated for the target control condition z target} instead of the constant k in the equation (4). In this case, the lower the reliability of the model for the target control condition, the lower the likelihood ratio λ (z _target ). Therefore, the detection result of the state of the engine 10 can be applied to the model of the control condition farther from the target control condition. It will be reflected. Therefore, the learning speed of the model of each control condition becomes high.

_{The model update unit 33 stores the parameters α j} and β _j of the model of each updated control condition in the memory 22.

The control condition setting unit 34 can use the control condition for controlling the engine 10 with respect to the operation condition (hereinafter, may be referred to as the operation condition of interest) notified by the control condition selection unit 31 that the model update is completed. To set. In the present embodiment, the control condition setting unit 34 has an abnormality in which knocking has occurred, which is estimated from a model of the state of the engine 10 corresponding to the control conditions among the control conditions prepared in advance in the operation condition of interest. Select one or more control conditions in which the probability of being in a state is equal to or less than a predetermined allowable upper limit probability (for example, 0.1). The probability of knocking occurring in an abnormal state is, for example, an integral of the probability of becoming an individual state closer to the abnormal state than the value of a predetermined state (for example, 0.5) in a model representing the state of the engine 10. Calculated as a value. Then, the control condition setting unit 34 sets each selected control condition as a control condition that can be used for controlling the engine 10. The control condition setting unit 34 stores in the memory 22 a flag indicating that the selection has been completed for the operating condition for which the usable control condition has been selected.

FIG. 5 is an operation flowchart of the control condition setting process. The processor 23 executes the control condition setting process according to the following operation flowchart.

The control condition selection unit 31 of the processor 23 determines the operating conditions of the engine 10 at the present time based on the sensor signal representing the operating conditions received by the processor 23 from the rotation sensor 15 or the air flow meter 16 via the communication interface 21. Identify (step S101). Then, the control condition selection unit 31 reads from the memory 22 a model representing the distribution of the state of the engine 10 corresponding to the control condition for each control condition in the current operating condition (step S102).

_{The control condition selection unit 31 calculates the model likelihood L (μ i} ) for each control condition under the current operating conditions (step S103). Further, the control condition selection unit 31 _{calculates the observation state likelihood L (Pbd i} ) for each control condition under the current operating conditions (step S104). The control condition selection unit 31 may execute either the process of step S103 or the process of step S104 first. The control condition selection unit 31, for each control conditions in the operating conditions of the current, calculates the ratio of the observed state likelihood for the model likelihood λ a (z _i) (step S105).

Control condition selecting section 31, from among the control conditions in the operating conditions of the current, the ratio lambda (z _i) selects a larger control conditions than a predetermined threshold value lambda _T (step S106). Then, the control condition selection unit 31 determines whether or not the set of selected control conditions is an empty set (step S107).

If the set of selected control conditions is not an empty set (step S107-No), the control condition selection unit 31 maximizes the variance of the state of the engine 10 estimated from the model from the set of control conditions. The control condition is selected as the target control condition (step S108).

The control condition changing unit 32 of the processor 23 changes the control condition so that the control condition becomes the target control condition (step S109).

When the control condition reaches the target control condition, the model update unit 33 of the processor 23 measures the state of the engine 10 based on the sensor signal from the knock sensor 13 or the in-cylinder pressure sensor 14 for each of a predetermined number of combustion cycles. (Step S110). Then, the model update unit 33 updates each model so that the model of the state of the engine 10 with respect to the target control condition and the control conditions around the target control condition approaches the measured distribution of the state of the engine 10 a predetermined number of times (step). S111). Then, the processor 23 repeats the processing after step S101.

On the other hand, if the set of selected control conditions in step S107 is an empty set (step S107-Yes), the reliability of the model for each control condition under the current operating conditions is high. Therefore, the control condition setting unit 34 of the processor 23 has a probability of becoming an abnormal state in which knocking has occurred, which is estimated from a model of the state of the engine 10 corresponding to the control conditions among the control conditions in the current operating conditions. Selects a control condition in which is equal to or less than a predetermined allowable upper limit probability as a usable control condition (step S112). Then, the processor 23 ends the control condition setting process. As described above, regarding the operating conditions applied to the engine 10 at the present time, even when the number of trials for updating the model reaches the preset upper limit number, the control condition setting unit 34 determines the operating conditions. The process of step S112 may be executed.

The control unit 35 selects one of the available control conditions from the available control conditions for the operating condition for which the selection of the control conditions that can be used for controlling the engine 10 (hereinafter referred to as the usable control conditions) has been completed. The engine 10 is controlled according to the selected control conditions.

For example, the control unit 35 may specify the current operating condition based on the sensor signal from the rotation sensor 15 or the air flow meter 16, similarly to the control condition selection unit 31. Then, when the control unit 35 stores in the memory 22 a flag indicating that the selection of the usable control condition has been completed with respect to the specified operating condition, the current time for each cylinder of the engine 10 is set at the present time for each combustion cycle. The timing of igniting the spark plug 11 is controlled according to the control conditions selected from the available control conditions for the operating conditions. Therefore, the control unit 35 reads one or more usable control conditions under the current operating conditions from the memory 22. For example, the control unit 35 sets the ignition timing to MBT when the usable control condition includes a control condition that defines the ignition timing (Minimum advance for the Best Torque, MBT) at which the torque is maximized. Set. On the other hand, when the control condition that defines the MBT as the ignition timing is not included in the usable control condition, the control unit 35 sets the ignition timing closest to the MBT among the usable control conditions in which the ignition timing is later than the MBT. The ignition timing may be set according to the specified control conditions.

The control unit 35 outputs a control signal for igniting the spark plug 11 to the spark plug 11 at the determined ignition timing.

FIG. 6 is an operation flowchart of the engine control process. The control unit 35 of the processor 23 executes the engine control process according to the following operation flowchart for each combustion cycle.

The control unit 35 reads the usable control conditions and the MBT under the current operating conditions of the engine 10 from the memory 22 (step S201). Then, the control unit 35 determines whether or not the usable control condition includes a control condition that defines the MBT as the ignition timing (step S202).

When the usable control condition includes a control condition that defines the MBT as the ignition timing (step S202-Yes), the control unit 35 sets the ignition timing to the MBT (step S203). On the other hand, when the control condition that defines the MBT is not included as the ignition timing (step S202-No), the control unit 35 sets the ignition timing closest to the MBT among the usable control conditions in which the ignition timing is later than the MBT. The ignition timing is set according to the specified control conditions (step S204).

After step S203 or S204, the control unit 35 outputs a control signal for igniting the spark plug 11 of the engine 10 to the spark plug 11 at the set ignition timing (step S205). Then, the control unit 35 ends the engine control process.

As described above, the control condition setting device and the control device are in the state of the internal combustion engine, including the normal state and the abnormal state of the internal combustion engine for each combination of the operating conditions and the control conditions of the internal combustion engine. The distribution is represented by a model that approximates the beta distribution. Then, according to the model, this control device selects a control condition in which the probability that the internal combustion engine becomes abnormal is equal to or less than a predetermined allowable upper limit probability from a plurality of control conditions prepared in advance for each operating condition. , Control the internal combustion engine according to the selected control conditions. Therefore, this control device can appropriately control the internal combustion engine so as to prevent the internal combustion engine from being in an abnormal state. Furthermore, since this control device represents the distribution of the state of the internal combustion engine with a model that approximates it with a beta distribution, it is discrete and pre-stated, such as a state in which knocking occurs and a state in which knocking does not occur. An accurate model can be obtained by using a relatively small number of measurement data for the distribution of two independent states. Therefore, this control device can reduce the amount of data required to determine the control conditions for appropriately controlling the internal combustion engine mounted on the vehicle.

Further, in each of the above embodiments or modifications, it is configured to be communicable with a device provided separately from the ECU 1, for example, an inspection device for the engine 10 installed in a factory, or a vehicle on which the ECU 1 is mounted. The server may execute the control condition setting process. That is, the control device of the engine 10 and the control condition setting device may be provided separately. In this case, if the usable control conditions for each operating condition set by the separately provided control condition setting device (for example, the above-mentioned inspection device or server) are stored in the memory 22 of the ECU 1 in advance. Good. Further, in this case, since the processor 23 of the ECU 1 does not have to execute the control condition setting process, in the processor 23, the control condition selection unit 31, the control condition change unit 32, the model update unit 33, and the control condition setting unit 34 It may be omitted.

According to this modification, the ECU 1 does not have to execute the control condition setting process, so that the calculation load of the ECU 1 is reduced.

The distribution of the states of the internal combustion engine to be modeled is not limited to the above example, and for example, the distribution of an abnormal state in which a misfire has occurred and a normal state in which a misfire has not occurred may be modeled. In this case as well, the control condition setting device and the internal combustion engine control device may execute the same processing as the control condition setting process and the internal combustion engine control process according to the above-described embodiment or modification.

As described above, those skilled in the art can make various changes within the scope of the present invention according to the embodiment.

1 Electronic control unit (ECU)
10 Engine 11 Spark plug 12 Variable valve timing mechanism 13 Knock sensor 14 In-cylinder pressure sensor 15 Rotation sensor 16 Air flow meter 21 Communication interface 22 Memory 23 Processor 31 Control condition selection unit 32 Control condition change unit 33 Model update unit 34 Control condition setting unit 35 Control unit

Claims (7)

A control condition setting device that sets control conditions that can be used to control the internal combustion engine of a vehicle.
The internal combustion engine includes an abnormal state of the internal combustion engine and a normal state of the internal combustion engine when the internal combustion engine is controlled according to the control conditions for each of a plurality of control conditions under a predetermined operating condition of the internal combustion engine. A storage unit that stores a model that represents the distribution of the state of the engine in beta distribution,
A control condition selection unit that selects a control condition that maximizes the dispersion of the state of the internal combustion engine estimated from the model among the plurality of control conditions under the predetermined operating conditions of the internal combustion engine.
Distribution of the state of the internal combustion engine represented by the model corresponding to the selected control condition to the measured distribution of the state of the internal combustion engine when the internal combustion engine is operated a predetermined number of times according to the selected control condition. A model update unit that updates the model corresponding to the selected control condition so that
Among the plurality of control conditions, a control condition estimated according to the updated model in which the probability that the internal combustion engine becomes abnormal is equal to or less than a predetermined threshold value is used as a control condition that can be used in the predetermined operation condition. Control condition setting unit to be set and
Control condition setting device having. When the control condition selection unit assumes that the frequency with which the internal combustion engine becomes abnormal for each of the plurality of control conditions under the predetermined operating conditions of the internal combustion engine follows the model for the control conditions. , Regarding the control condition when it is assumed that the probability that the internal combustion engine is in an abnormal state is a predetermined probability with respect to the model likelihood representing the plausibility of the detection result of the state of the internal combustion engine with respect to the control condition. The ratio of the observation state likelihood representing the plausibility of the detection result of the state of the internal combustion engine is calculated, and the ratio is estimated from the model among the control conditions in which the ratio is equal to or higher than a predetermined threshold among the plurality of control conditions. The control condition setting device according to claim 1, wherein the control condition that maximizes the dispersion of the state of the internal combustion engine is selected. The model update unit corresponds to the control conditions around the selected control conditions in the measured distribution of the state of the internal combustion engine when the internal combustion engine is operated a predetermined number of times according to the selected control conditions. The control condition setting device according to claim 1 or 2, further updating the model corresponding to the peripheral control conditions of the selected control conditions so that the distribution of the states of the internal combustion engine represented by the above is closer. A control device that controls the internal combustion engine of a vehicle.
For each of the plurality of operating conditions of the internal combustion engine, the distribution of the states of the internal combustion engine including the abnormal state of the internal combustion engine and the normal state of the internal combustion engine, which can be used under the operating conditions. A storage unit that stores one or more control conditions in which the probability that the internal combustion engine will be in an abnormal state is equal to or less than a predetermined threshold according to a model represented by a beta distribution.
The operating conditions of the internal combustion engine are specified based on the measurement signal measured by the operating condition detection unit that detects the operating conditions of the internal combustion engine, and one or more of the above-mentioned controls under the specified operating conditions among the plurality of operating conditions. A control unit that controls the internal combustion engine according to any of the conditions,
Control device with. A control condition selection unit that selects a control condition that maximizes the dispersion of the state of the internal combustion engine estimated from the model among a plurality of control conditions under a predetermined operating condition of the internal combustion engine.
Distribution of the state of the internal combustion engine represented by the model corresponding to the selected control condition to the measured distribution of the state of the internal combustion engine when the internal combustion engine is operated a predetermined number of times according to the selected control condition. A model update unit that updates the model corresponding to the selected control condition so that
Among the plurality of control conditions, a control condition estimated according to an updated model in which the probability that the internal combustion engine will be in an abnormal state is equal to or less than the predetermined threshold value can be used in the predetermined operation condition. Control condition setting unit to be set as
The control device according to claim 4, further comprising. A control condition setting method for setting control conditions that can be used to control an internal combustion engine of a vehicle.
The internal combustion engine including an abnormal state of the internal combustion engine and a normal state of the internal combustion engine when the internal combustion engine is controlled according to the control conditions among a plurality of control conditions under a predetermined operating condition of the internal combustion engine. Select the control condition that maximizes the dispersion of the state of the internal combustion engine estimated from the model that expresses the distribution of the state of
Distribution of the state of the internal combustion engine represented by the model corresponding to the selected control condition to the measured distribution of the state of the internal combustion engine when the internal combustion engine is operated a predetermined number of times according to the selected control condition. Update the model corresponding to the selected control condition so that
Among the plurality of control conditions, a control condition estimated according to the updated model in which the probability that the internal combustion engine becomes abnormal is equal to or less than a predetermined threshold value is used as a control condition that can be used in the predetermined operation condition. Set,
Control condition setting method. A control method that controls the internal combustion engine of a vehicle.
The operating conditions of the internal combustion engine are specified based on the measurement signal measured by the operating condition detection unit that detects the operating conditions of the internal combustion engine.
According to a model in which the internal combustion engine can be used under the specified operating conditions and the state of the internal combustion engine including the abnormal state of the internal combustion engine and the normal state of the internal combustion engine is represented by a beta distribution, the internal combustion engine Controls the internal combustion engine according to any one or more control conditions in which the probability that the internal combustion engine becomes abnormal is equal to or less than a predetermined threshold value.
Control method.

Images