JP5771893B2 - Turbulence suppression control method - Google Patents

Description

  The present invention relates to a technique for suppressing vertical acceleration caused by turbulence when an aircraft capable of acquiring the state of motion of the aircraft can also acquire wind speed and direction data of turbulence in front of the aircraft.

  Most recent large-scale aircraft accidents in Japan are caused by turbulence, and in order to realize safer aircraft operation, control technology that prevents sudden fluctuations due to turbulence, that is, prior information on turbulence, which is a disturbance, is somehow used. Control that suppresses aircraft shake due to turbulence by acquiring and effectively using the information (hereinafter referred to as turbulence suppression control) is required. There have been some proposals for the control algorithm. For example, Patent Document 1 proposes “a method and system for generating a rudder correction command for reducing undesirable lateral movement”. It is an object of the present invention to reduce undesirable lateral movement of an airplane by reducing side loads and upsets caused by turbulence and gusts, and therefore, for airplanes caused by turbulence and gusts. The configuration is such that a rudder correction command is generated that changes the rudder command of the airplane so as to reduce the effective force on the vertical stabilizer. More specifically, the differential pressure is measured on both sides of the vertical stabilizer and used to produce a rudder deflection value that is compensated for roll rate and yaw rate. The compensated deflection value is high pass filtered with a corner frequency that is 25% of the airplane's Dutch roll frequency. The result is a first rudder deflection value that is subtractively synthesized with the second rudder deflection value. The second rudder deflection value is derived by gain adjusting and low-pass filtering the inertia skid angle ratio value generated by the aircraft yaw damper. Further, an object of the invention of “a system for predicting and measuring turbulence upstream of an aircraft”, which is Patent Document 2, is an airflow working surface flight control system that uses a rider and is in a forward phase with respect to the system. The purpose is to measure the wind speed in front of the aircraft at a sufficient distance to allow operation by. The present invention relates to a system that pre-measures turbulence upstream of an aircraft and is mounted on the aircraft, the system transmitting a light beam toward the front of the aircraft, and a scattered light beam A direct detection device associated with the control means, a first processing element using a first internal correction algorithm, and at least one aircraft control surface actuator And a second processing element that uses a second external correction algorithm to transmit instructions that can be executed.

In US Pat. No. 6,057,049, a method for reducing the influence of vertical turbulence on an aircraft is presented, which presents a method that can particularly effectively reduce the influence of at least one vertical turbulence encountered in flight on an aircraft, in particular a transport aircraft. An apparatus "is disclosed. The method presented by the invention comprises the following series of sequential steps during flight:
a) The vertical component of the wind outside the aircraft is determined at the current location of the aircraft,
b) This wind vertical component determines the intensity level for vertical turbulence outside the aircraft at the current position above,
c) With the wind vertical component described above, at least one control command is calculated for at least one controllable movable member capable of acting on the lift of the aircraft, and this control command is a load factor caused to the aircraft by vertical turbulence. That can reduce the size of
d) A check is made to see if an operating condition that depends at least on the level of severity is achieved,
e) If the above operating state is realized, the control command is transmitted to at least one actuator of the controllable movable member;
Is performed automatically and repeatedly.

  The purpose of “minimizing the dynamic structural load of the aircraft” disclosed in US Pat. No. 6,057,086 is to provide effective minimization of the dynamic structural load of the aircraft as the name suggests. The present invention provides a method for minimizing the dynamic structural load of an aircraft introduced to an aircraft by external excitation, the method of generating a signal indicative of the external excitation, and introduced to the aircraft Deriving a preceding control signal for driving an aircraft control element from the signal indicating the excitation according to a preceding control rule so as to reduce a dynamic structural load, and an error signal indicating a result of the preceding control And optimizing a predecessor control rule with the error signal and / or a signal indicative of the excitation so as to minimize the dynamic structural load.

  However, these conventional control algorithms for suppressing the swaying of the aircraft due to turbulent air flow are, at best, control algorithms for problems that only consider the uncertainty of the system, and many are control algorithms for problems that do not even take into account. A control algorithm that takes into account the observation error for the advanced information data of the advanced disturbance and the observation error for the state data of the system has not been proposed so far. However, these observation values always include observation errors, and it is impossible to perform observations that do not include observation errors. In addition, if the control considering these observation errors is not performed, there is a possibility that the influence of the turbulence will be exacerbated, that is, the control for amplifying the turbulence fluctuation may be added. There is a strong need for algorithms that take into account errors and system uncertainties.

  The object of the present invention is to perform two turbulence suppression controls in consideration of the two observation errors pointed out in the background art, that is, the observation error of the prior turbulence and the state observation error of the own aircraft, and the uncertainty of the system. The goal is to develop and provide an algorithm that takes into account all the effects of errors or uncertainties and provides control that minimizes the effects of turbulence encountered even when they affect each other worst.

The turbulence suppression control method according to the present invention controls a system that discretizes a linear motion equation of a continuous-time aircraft, and combines them as a discrete-time system with dynamic characteristics such as an actuator that moves a control surface and an engine that causes a change in thrust. Preliminary turbulence information observation using a delay system with uncertain dead time shown in [Equation 1] at the control input to ensure robustness against system modeling errors (robustness) In order to ensure robustness against errors, the preliminary turbulence information using the presumed error width shown in [Equation 2] is used, and further, robustness against observation errors in the state of motion of the own aircraft is ensured [ Using the aircraft motion state using the presumed error range shown in Equation 3], the state and input related to the motion of the aircraft In the next constraint, the evaluation function is optimized, and the first input data is added as the actual input in the obtained optimal input time series. In the next step, the newly obtained turbulence information and information on the body motion are obtained. Is used to solve the optimization problem again, add the first data of the obtained optimum input time series as an actual input, and repeat this step every step.
Where T _d is a set of delay steps composed of delay steps corresponding to uncertain delay times, d, d + 1,... Are the delay times of each element, and d underbar is a time step. The smallest delay step value expressed as a number, the d upper bar encounters the largest delay step value expressed as a number of time steps, and w _{k + j | k} encounters j steps ahead in k steps. observations of turbulence, w k _{+ j} is the true value of the turbulence encountered j step ahead in k steps, X _j delta _w represents the observation error of turbulence.

Further, by adding an observation error of a state quantity related to mechanical motion, the turbulence suppression control method of the present invention performs a formulation for obtaining a common input time series for all combinations of the above-described observation error and uncertainty. .

The effects of using the present invention are as follows in comparison with the prior art.
1) Even if there is an observation error of prior turbulence, if it is within the assumed range, the influence of the motion when receiving actual turbulence is below the evaluation function value obtained as a result of optimization.
2) Even if there is an observation error in the state of motion of the aircraft, the effect of the motion when receiving actual turbulence is within the expected function value if it is within the assumed range.
3) Even if there is an uncertain delay in the on-board system, the effect of motion when subjected to actual turbulence is within the assumed range and is below the evaluation function value obtained as a result of optimization.
4) Since a control input that does not exceed the drive input range and input rate limit of the control input such as an elevator can be obtained, the body motion does not become unstable due to the saturation of the control input and the rate limit.

It is a conceptual diagram of the turbulence suppression control which concerns on this invention. It is a figure which shows the influence of the uncertain dead time in a control input terminal. It is a figure which shows the prior disturbance information containing an observation error. It is the graph which put together the result of having simulated the gust suppression performance of this invention. It is a graph showing an example of the result of having simulated the gust suppression performance of this invention.

FIG. 1 shows a conceptual diagram of the present invention. According to the control method of the present invention, a flying aircraft can acquire turbulent wind speed / direction data (expressed in Vg) by any method, and can also acquire information related to aircraft motion (expressed in V, θ, etc.). In such a case, if the control is not performed, a motion that is affected by the airflow will occur, but a control input command that suppresses the motion will be sent to the turbulence information observation error, the aircraft The basic concept of the present invention is a control method that is generated so as to be robust against motion observation errors and airframe motion model uncertainty.

The first feature of the present invention is to use a delay system having an uncertain dead time at the control input end in order to ensure robustness (robustness) with respect to a modeling error of the system, and to be robust against an observation error of prior turbulence information. Aircraft motion state using pre-estimated error width to ensure robustness against observation error of the state of motion of own aircraft, using prior turbulence information using error width assumed in advance to ensure It is in the point using.
The second feature point is that a formulation for obtaining a common input time series for all combinations of the above-described observation error and uncertainty is performed.

Now, let us consider micromotion from the steady flight state of an aircraft. It is assumed that a body motion model representing the minute motion is given by a discrete time system. However, this is usually given a continuous time system and the period T of the computer _s It may be a discretized system with zero-order hold by [s]. Similarly, an actuator system such as an elevator (elevator) is provided as a discrete time system. This is also usually given a continuous time system, the period T of the computer _s It may be a discretized system with zero-order hold by [s]. At this time, it is assumed that a system combining them is given by the following.

Where x _k  Is the state of the coupled system in k steps, w _k  Is the turbulence input in k steps, u _k  Is the control input of the elevator etc. in k steps, z _k  Represents the fluctuation from the steady value of the vertical acceleration in k steps. U _k  Input one step before u _k-1  And the difference Δu _k  Decomposed into u _k = U _k-1 + Δu _k  It expresses.
Furthermore, at the control input end to which input by a control input device such as an elevator is applied, T _d  ∈ [T _dmin , T _dmax And a delay time system having an indefinite delay time. Note that this indefinite delay is a virtual system for describing a modeling error that occurs during modeling of an actuator or the like, and there is no need for a delay in the actual system. At this time, the indefinite delay time T _d  Is the following T for the above discrete time system P: _d  This corresponds to an uncertain delay step due to.
Therefore, T at the control input terminal _d A discrete-time system P with an uncertain delay of T _d  It is expressed by a system set composed of a finite number of elements obtained by combining a finite number of delays expressed by ## EQU1 ## and the system of [Equation 1]. The system uncertainty expression using the uncertain delay is now possible.

Next, consider the observation error of prior information on turbulence. Suppose that turbulence information from the current time k step to N-1 step is observed.
Where w _{k + j | k}  Is the observed value of turbulence encountered at j steps ahead in k steps, w _{k + j}  Is the true value of the turbulence encountered at j steps ahead in k steps, X _j Δ _w  Represents the observation error of turbulence.
Next, let us consider the state observation error related to the motion of the aircraft. k-step true state x _k  X _{k | k}  And the true state and the observed value are x _k = X _{k | k} + Δ _x The relationship is established. However, the observation error Δ _x Is predetermined.
At this time, an input time series common to all combinations of the uncertain delay step [Equation 2], the turbulence observation error [Equation 3], and the state observation error is obtained under the constraints on the state and input related to the motion of the aircraft. Sought to optimize a predetermined evaluation function. This problem is formulated as a Second Order Cone Programming (SOCP) problem and can be obtained using commercially available tools.
Of the obtained optimum input time series, the first data is added as an actual input.
In the next step, the optimization problem is solved again using the newly obtained turbulence information and information on the airframe motion, and the first data of the obtained optimum input time series is added as an actual input. This is repeated every step.

Hereinafter, embodiments of the present invention will be described in detail.
The symbols used in this specification are defined below. 0 _n the n-dimensional zero matrix, 0 _{n, m} are n × m-dimensional zero matrix, I _n is the n-dimensional unit matrix, 1 _n is n-dimensional vector where all elements are 1, 0, appropriately sized zero matrix Or a zero vector, R ⁿ is a set of n-dimensional real vectors, R ^{n × m} is a set of n × m-dimensional execution sequences,
Throughout this specification, for variable p 1 in a discrete time system, pk represents the variable p ₁ at k steps, and p _{k + j | k} represents the predicted value at k + j steps predicted at k steps.

In this specification, a linear longitudinal micromotion of an aircraft is a control target. In addition, it is assumed that the time change of turbulence does not affect the airframe motion as in the widely used literature (see Non-Patent Document 1). Considering that most of the current airborne computers are digital computers, the system is a system that discretizes the linear motion equation of a continuous-time aircraft, and the dynamic characteristics of the actuator that moves the control surface and the engine that changes the thrust, etc. Is also given as a discrete-time system. At this time, it is assumed that a system obtained by combining these discrete time systems is given by equation (1).
^{_{Here, x k ∈ R n, w}} k ∈ R nw, u k ∈R nu, z k ∈R nz , respectively, the state variables (specifically the speed, angular, elevator steering angle, etc.) of the system P, disturbance This is a vector that represents input (specifically, turbulence), control input (specifically, aerodynamic control surface displacement command, etc.), and control output (variables that characterize the motion that is caused by turbulence such as vertical acceleration, etc. and is to be reduced). . Note that it is assumed that the state x _k includes all observation errors but is observable. The sampling period is T _s [s].

In general, it is impossible to obtain a system without a modeling error, and design of a control system for an actual system is performed in consideration of uncertainties such as modeling errors. There are parametric and frequency domain representation methods for the uncertainty of this system. One of these methods is uncertainty using a time-invariant dead time system using bounded dead time at the control input. There is an expression. This method shows effectiveness as an uncertainty expression even in different design specifications, such as the tracking problem in Non-Patent Document 2, which is the prior art, and the model matching problem in Non-Patent Document 3. Therefore, also in this specification, an uncertainty expression using dead time at the control input end is used. Specifically, a system is considered in which a time-invariant dead time system in which the _dead time T _d [s] is T _d ∈ [T _dmin , T _dmax ] is added to the control input terminal. However, it is assumed that T _s ≦ T _dmin is satisfied in consideration of a one-step delay caused by the on-board computer. Now, even if a dead time system with an uncertain dead time _Td is inserted at the control input end, since the target is a discrete time system, it can be seen that the number of steps to which a control input is applied is limited to a finite number. (See FIG. 2).

Further, as a control input u _k of the equation (2), expressed by decomposing the value of one step before and to the difference.
u _k = u _k−1 + Δu _k (2)
This is to impose not only constraints on input saturation but also constraints on input rate.
At this time, the system P _u in which an uncertain time-invariant dead time system satisfying T _d ∈ [T _dmin , T _dmax ] is added to the control input terminal in the system of the formula (1) is the following system using Δu _k as an input: Is described as

In this specification, it is assumed that the state quantity x _k of the system (1) is observed including an observation error. Assume that the current time is k steps, and the state quantity x _{k | k} observed at time k satisfies the following equation using a known time-invariant vector Δ _x representing an observation error.
x _k = x _{k | k} + Δ _x (5)
However, the existence range of Δ _x is given by a known convex polyhedron, and a set composed of the end points is defined as Φ _x . That is, the following formula is established.
Δ _x ∈ coΦ _x (6)
In the above equation, co is a convex set, that is, coΦ _x represents the whole and the inside of the convex polyhedron.
Here, the existence range Ω _x of x _k is defined.

Further, it is assumed that turbulent airflow that is a disturbance input is obtained as information including an observation error from the present to N-1 steps ahead. In other words, it is assumed that the following equation holds.
Here, w _{k + j | k} is an observed value of the turbulent air flow encountered in j steps ahead in k steps, w _{k + j} is a true value of the turbulent air flow encountered in j steps ahead in k steps, and X _j ∈ R ^{nw × nw} (j = 0, 1, ··· N-1) is given a constant matrix for representing the observation error of the pre-disturbance information, delta _w is an uncertain constant vector that satisfies the following expression (Fig. 3 reference).

Since prior disturbance information is obtained up to N-1 steps ahead, it is assumed that state prediction is performed up to N-1 steps ahead.
Under the above assumption, (3) and (4) the system P _u with uncertainty defined in formula, optimization problem such as the following consider (problem 1). That is, when the state quantity related to the airframe motion including the observation error can be observed as in (5) and turbulence information including the observation error is obtained as in (9), (13), (14), Under the constraints of Eq. (15), the maximum value of the evaluation function J _i (x ⁱ , Δ _{u, z} ⁱ ) (specifically, the observation error of the preturbulence and the observation error of the state quantity of machine motion) The common input time series Δ _u that minimizes the maximized J _i (the maximum value for i of x ⁱ , Δ _{u, z} ⁱ ) is obtained.
What is important here is that a common input difference time series that does not depend on the index i of the system P _u ⁱ , the observation error set Ω _w of the prior disturbance information, and the state observation error set Ω _x is required. As a result, the uncertainty described by the system set P _u, it is possible to obtain a robust input time series relative to measurement error for observed values of the observation error and aircraft motion state quantity of the pre-disturbance information.

The method of Non-Patent Document 5, which is the conventional turbulence suppression control, “Unlike equation (5), there is no observation error and the state quantity related to the body motion can be observed, and unlike equation (9), the observation error is Common to minimize the maximum value of the evaluation function J _i (x ⁱ , Δ _{u, z} ⁱ ) under the constraints of the equations (13), (14), (15) seek input time series Δ _u. "I have proposed a method to solve the problem. Here, the index i is due to the uncertainty of the system, and handles the “maximum value” considering only this one type of uncertainty. The turbulence suppression control algorithm that takes into account only the turbulence observation error, which is one step ahead of this, `` Unlike (5), it can observe state quantities related to aircraft motion without any observation error, but as in (9), When turbulence information including errors is obtained, the maximum value of the evaluation function J _i (x ⁱ , Δ _{u, z} ⁱ ) is minimized under the constraints of equations (13), (14), and (15) seek common input time series Δ _u to. "I have proposed a method to solve the problem. That is the control method of solving Problem 2 every step.
That is, Non-Patent Document 5 formulates the case where there is no observation error in the observation values of the prior disturbance information and the body motion state quantity. In the turbulence suppression control algorithm considering only the turbulence observation error, Non-Patent Document 5 The above results have been developed to formulate the case where there is an observation error only in the prior disturbance information. Therefore, only the results of that study are shown here. In the following, since there is no observation error in the airframe motion state quantity, it is assumed that x _k = x _{k | k} holds for Equation (5).

That is, in Non-Patent Document 5, Problem 1 becomes Problem 2 based on the above preparation.
To summarize the above, the control method for Problem 1 when there is no state quantity observation error related to machine motion is a problem using the expanded state of Equation (8) and the turbulence information of Equation (9) that are updated every step. 2 is a control method using the input according to the equation (2) in which step 2 is solved and Δu _k = Δu _{k | k} .

Problem 3 is a problem obtained by adding an observation error of a state quantity related to airframe motion to Problem 2. Therefore, problem 1 and problem 3 are the same and only the expression is changed.
At this time, since the problem 3 is evaluated for all combinations of end points of (Δ _w , Δ _x ), the calculation complexity is larger than the problem 2. However, the uncertainty of the plant is described by an uncertain time-invariant dead time at the control input, and the observation errors for the prior disturbance information and the observed values of the airframe motion state quantities are expressed by equations (9) and (5), respectively. Therefore, it should be noted that no conservativeness such as approximation is introduced when the problem 3 is formulated. Problem 3 is to create an optimal input time series from among the control input time series that satisfies the constraints (19) and (20) such as the state. From the above, it is possible to perform performance limit evaluation of gust mitigation control by performing a simulation using Problem 3.

In this section, as in Non-Patent Document 5, an example of examining the possibility of suppressing vertical acceleration due to turbulence is shown.
Similar to Non-Patent Document 5, a small vertical vertical approximation was performed from the horizontal steady flight state of the experimental aircraft MuPAL-α of the Japan Aerospace Exploration Agency at 1520 [m] and true air speed of 66.5 [m / s]. Consider an aircraft motion that adds an elevator actuator model modeled by a 1 / (0.1s + 1) first order lag model to the motion. The sampling period T _s was set to 0.1 [s]. The state variable x _k is [u _i w _i q θ δ _e ] ^T , the disturbance input x _k is w _g , the control input u _k is δ _ec , and the control output z _k is Δa _z .
Here, u _i , w _i , q, θ, δ _e , w _g , δ _ec , and Δ a _z are the aircraft coordinate system x-direction inertia velocity [m / s] and the aircraft coordinate system z-direction inertia velocity [m / s], respectively. ], Pitch angular velocity [rad / s], pitch angle [rad], elevator rudder angular displacement [rad], inertial coordinate system z direction gust [m / s], elevator rudder angular displacement command [rad], inertial coordinate system z direction It represents the variation of acceleration [m / s ² ]. Note that w _g is a positive wind that blows upward from the bottom of the aircraft.

The indeterminate dead time T _d [s] is T _d ∈ [0.1, 0.4]. At this time,
The endpoint set Φ _x that defines the existence range of the vector Δ _x that represents the observation error for the observed value of the airframe motion state quantity is
Both the weight matrices Q and S in the evaluation function J _i are determined to be zero matrices.
Note that the control output constraint of Equation (15) is not imposed, and R, which is a weight for the control input rate, and the time length T _s × N of the prior turbulence information are simulation parameters.
The turbulent airflow encountered is set to be given by the following equation.
w _g (t) = sin (ωt) (21)
Here, t [s] represents the time from the start of simulation, and ω [rad / s], which is the frequency of turbulence, was used as a simulation parameter.
The matrix X _j representing the observation error of the prior disturbance information was determined as follows.
X _j = 0.2 + 0.1 × (66.5 / 100 × T _s ) j (22)
Each set of simulation parameters R, Ta, and ω is defined as follows.
R ∈ {10 ¹ , 10 ² , 10 ³ , 10 ⁴ , 10 ⁵ }
T _a ∈ {1.0, 1.2, 1.4,1.6, 1.8, 2.0, 2.2, 2.4}
ω ∈ {0.1, 0.5, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0,7.0}
A simulation of 20 [s] was performed using an airframe motion linear time-invariant system and an elevator actuator model. The solution to the SOCP problem was obtained by SeDuMi in Non-Patent Document 6 via YALMIP in Non-Patent Document 4. Moreover, the case where the turbulence suppression control by the problem 3 on the conditions which cannot obtain prior turbulence information as a comparison object was also performed in the same environment. For descriptive purposes, case names are defined below.

Although this FIG. 4 is a graph made with color display, since color display cannot be performed with a patent drawing, a color graph will be submitted separately as an attachment. The reason why the number of simulations is 2 × 2 is to take the combination of the maximum and minimum for turbulence information error and the maximum and minimum case for airframe state quantity error. FIG. 4 shows the results for disturbance input ω (turbulence frequency) of 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, and 7.0. In this graph, for performance comparison, the number N of steps is taken on the y-axis, the weight R for the control input rate is taken as the simulation parameter on the x-axis, and J _A / J _B for judging the possibility of suppressing turbulence, and prior turbulence information J _A / J _C representing the usefulness of is plotted in the z-axis direction. In addition, mesh-like surfaces representing J _A / J _B = 1.0 and J _A / J _C = 1.0 representing equivalent performance were also plotted. In this case, the combination of R and T _a being plotted below the mesh-like surfaces, more of J _A means a better performance. In the case of ω = 0.1, 0.5, since J _A / J _B and J _A / J _C both exceeded 2 in most cases, it was omitted. From this result, in the problem setting assumed in this paper, it is considered possible to reduce the effect of turbulence with a frequency of 2.0 to 6.0 [rad / s]. It is considered that the influence of turbulence smaller than s] and turbulence larger than 7.0 [rads /] can hardly be suppressed even if turbulence information is obtained in advance.

As an example of a simple simulation result, FIG. 5 shows a simulation result when the turbulence frequency is ω = 3.0, R = 10, and Ta = 2.4 (N = 24). In the simulation, there are cases in which the number of uncertain delay steps is 1 or 2 or 3 or 4 when the observation error of the prior turbulence is the maximum or minimum, the observation error of the state quantity of the airframe motion is the maximum or minimum. , All these combinations are implemented. The values JA, JB, and JC of Case A, B, and C are obtained from these simulation results , and the points where JA / JB and JA / JC are plotted are X and Y at ω = 3.0 in FIG. 4, respectively. In FIG. 5, the solid line indicates the above Case A (data for turbulence suppression control of the present invention) , the broken line indicates the above Case B (data without control) , and the dotted line indicates the above Case C (preliminary turbulence information pear data) . . It can be seen that for all 16 pairs shown in the graph, Case A has a remarkably small control width compared to Case B in which control is not performed. In addition, the control width of Case C indicated by the dotted line is slightly smaller than that of Case B, but the control width of Case A is much smaller than this. It can be confirmed at a glance how Case A, which has been subjected to turbulence suppression control according to the present invention, suppresses the movement of the aircraft by comparing the solid line graph, the broken line graph, and the dotted line graph .

The turbulence suppression control method according to the present invention can observe disturbances in advance, and can be used in general for systems that have been given optimization specifications such as minimizing the effects of disturbances. Suppose that the wave information to be encountered is obtained in advance. In such a case, it is possible to obtain a steering input that most suppresses shaking based on wave information obtained in advance.
As another application, it is also possible to perform optimum blade pitch angle control of a wind turbine that performs wind power generation. This is because wind turbines that generate wind power have their rotational speed controlled by the pitch angle of the blades. However, it is important to rotate the wind turbine at a certain rotational speed, and it is desirable to maintain the rotational speed without being affected by gusts. It is rare. In such a case, by adding a device that measures the turbulence in the distance to the windmill legs or the center of the windmill, it becomes possible to observe the disturbance of the wind speed, which is a disturbance. It is possible to calculate the optimum pitch angle for power generation in consideration.

Japanese Patent Laid-Open No. 8-310495 “Method and System for Generating Rudder Correction Command to Reduce Undesirable Lateral Movement” Published on Nov. 26, 1996 Special table 2008-500525 gazette "System for predicting and measuring turbulence upstream of aircraft" Published January 10, 2008 JP 2009-511339 Gazette “Method and apparatus for reducing the influence of vertical turbulence on aircraft” Published on March 19, 2009 JP 2009-524547 A “Minimization of the Dynamic Structural Load of an Aircraft” Released on July 2, 2009

Stengel, R.E .: “Flight Dynamics” Princeton University Press, 2004, pp.274-297. Masahiro Ohno, Yasuhiro Yamaguchi, Takeshi Hata, Morio Takahama, Yoshikazu Miyazawa, Tatsuji Izumi, Robust Flight Control Law Design of ALFLEX, Transactions of the Society of Instrument and Control Engineers, Vol. 34, (1998), pp. 1905-1912. Masayuki Sato, Satoshi Sato “Flight Controller Design for Simulating Gust Response and Steering Response: Design and Experimental Results for Lateral / Directional Motion of MuPAL-α” Japan Aerospace Society Proceedings, Vol.54, (2006) , Pp. 71-81. Lofberg, J.:YALMIP: A Toolbox for Modeling and Optimization in MATLAB, Proc. The CACSD Conference, Taipei, Taiwan, 2004, Masayuki Sato, Nobuhiro Yokoyama, and Satoshi Sato “Gust Alleviation Control by Robust Model Predictive Control Using Prior Information of Turbulence” Japan Aerospace Society Proceedings, Vol.57 (September 28, 2009), pp. 345-353. Sturm, J.S .: Using SeDuMi1.02, a MATLAB Toolbox for Optimization Over Symmetric Cones, Optimization Methods and Software, Vols. 11-12, (1999), pp. 625-653.

Claims (2)

A system that discretizes the linear motion equation of a continuous-time aircraft is to be controlled, and these coupled systems are given as a discrete-time system with dynamic characteristics such as an actuator that moves the control surface and an engine that causes a change in thrust. To ensure robustness against observation errors of prior turbulence information using a delay system with an uncertain dead time shown in [Equation 1] at the control input to secure robustness (robustness) against modeling errors In order to secure robustness against the observation error of the state of motion of the own aircraft, using the pre-turbulent airflow information using the presumed error range shown in [Equation 2], it is assumed in advance. The aircraft motion state using the measured error width is evaluated under the constraints on the state and input related to the motion of the aircraft. Optimize the function, add the first data of the obtained optimal input time series as the actual input, and in the next step, use the newly obtained turbulence information and information on the aircraft motion to optimize again The turbulence suppression control method is characterized by adding the first data of the optimum input time series obtained as an actual input and repeating this step every step.
Where T _d is a set of delay steps composed of delay steps corresponding to uncertain delay times, d, d + 1,... Are the delay times of each element, and d underbar is a time step. The smallest delay step value expressed as a number, the d upper bar encounters the largest delay step value expressed as a number of time steps, and w _{k + j | k} encounters j steps ahead in k steps. observations of turbulence, w k _{+ j} is the true value of the turbulence encountered j step ahead in k steps, X _j delta _w represents the observation error of turbulence. The turbulence suppression control method according to claim 1, wherein a formulation for obtaining a common input time series for all combinations of the observation error and the uncertainty is added by adding an observation error of a state quantity related to mechanical motion.