Steering engine control method, device and equipment based on neural network and disturbance observer

Abstract

The invention relates to the technical field of steering engine automatic control, in particular to a steering engine control method, device and equipment based on a neural network and a disturbance observer, and the method comprises the following steps: obtaining model parameters of a digital steering engine model; establishing a steering engine control model based on a sliding mode variable structure control theory and a steering engine kinetic equation; performing disturbance compensation on the equivalent friction torque of the motor output end to obtain an equivalent friction torque estimated value of the motor output end; inputting the tracking error of the steering engine deflection angle and the derivative thereof into the trained neural network to obtain a model parameter and a disturbance comprehensive correction value; and according to the equivalent friction torque estimation value of the output end of the motor, the model parameters, the disturbance comprehensive correction value and the steering engine control model, determining the controlled input of the steering engine. The technical problems that in the prior art, the coupling relation of an aircraft digital steering engine system is high in nonlinearity, external disturbance is unknown, the required system response speed is high, and a traditional PID controller and a traditional intelligent algorithm are difficult to meet the requirements are solved.

Description

Technical Field

[0001] The present application relates to the technical field of servo control, and particularly relates to a servo control method, device and equipment based on a neural network and a disturbance observer. Background Art

[0002] During the operation of an aircraft, multi-dimensional variables such as angle of attack, Mach number, and high-terminal velocity fall angle need to be comprehensively analyzed and considered. Moreover, under high-overload and high-dynamic pressure trajectory conditions, the moment of the air rudder surface increases significantly, posing a certain challenge to the stability of the control system and putting forward higher requirements for the tracking accuracy, response speed, etc. of the driving servo.

[0003] Currently, electric servos are increasingly being applied to aircraft and are developing towards miniaturization and high power density. This article mainly conducts research on the control method with a high-performance miniaturized digital servo as the object.

[0004] Traditional digital servos usually adopt a PID controller. The PID controller does not depend on the specific model of the system and can achieve good control effects by adjusting parameters. However, the digital servo system of an aircraft has a coupling relationship, strong nonlinearity, complex application scenarios, unknown external disturbances, requires a fast system response speed, high tracking accuracy, and strong robustness. Traditional PID controllers and traditional intelligent algorithms are difficult to meet the above requirements. Summary of the Invention

[0005] The present application provides a servo control method, device and equipment based on a neural network and a disturbance observer, which can solve the technical problems existing in the prior art that the digital servo system of an aircraft has a coupling relationship, strong nonlinearity, complex application scenarios, unknown external disturbances, requires a fast system response speed, high tracking accuracy, and strong robustness, and traditional PID controllers are difficult to meet the above requirements.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] In a first aspect, the present invention provides a servo control method based on a neural network and a disturbance observer, including the following steps:

[0008] Obtain the model parameters of the digital servo model, where the model parameters include the equivalent frictional torque at the motor output end, the servo deflection angle, the moment of inertia of the rudder shaft, and the theoretical equivalent load coefficient;

[0009] Based on the sliding mode variable structure control theory and the servo dynamics equation, establish a servo control model;

[0010] Use a disturbance observer to perform disturbance compensation on the equivalent frictional torque at the motor output end to obtain an estimated value of the equivalent frictional torque at the motor output end;

[0011] Input the tracking error of the steering gear deflection angle and its derivative into the trained neural network to obtain the model parameters of the theoretical steering axis moment of inertia and the theoretical equivalent load coefficient and the comprehensive correction value of the disturbance.

[0012] Determine the controlled input of the steering gear according to the estimated value of the equivalent frictional torque at the motor output end, the model parameters, the comprehensive correction value of the disturbance, and the steering gear control model.

[0013] In some alternative solutions, the disturbance compensation for the equivalent frictional torque at the motor output end by using a disturbance observer includes:

[0014] Define the auxiliary variable of the disturbance observer based on the disturbance variable of the equivalent frictional torque at the motor output end and the nonlinear function to be designed in the disturbance observer.

[0015] Obtain the state estimate value of the auxiliary variable based on the observation gain and the auxiliary variable in the disturbance observer.

[0016] Determine the estimated value of the equivalent frictional torque at the motor output end based on the state estimate value of the auxiliary variable and the nonlinear function to be designed in the disturbance observer.

[0017] In some alternative solutions, the auxiliary variable is: z = τ f -p(x);

[0018] where Z is the auxiliary variable, τ f is the disturbance variable of the equivalent frictional torque at the motor output end, and p(x) is the nonlinear function to be designed;

[0019] The state estimate value of the auxiliary variable is determined according to the formula ;

[0020] where is the derivative of the state estimate value of the auxiliary variable; is the state estimate value of the auxiliary variable; l(x) is the observation gain; K is the theoretical equivalent load coefficient; is the derivative of the steering gear deflection angle; M is the input torque of the digital steering gear; p(x) is the nonlinear function to be designed;

[0021] The estimated value of the equivalent frictional torque at the motor output end is:

[0022] where is the estimated value of the equivalent frictional torque at the motor output end.

[0023] In some alternative solutions, the neural network adopts an RBF neural network. Input the tracking error of the steering gear deflection angle into the RBF neural network, and use the output of the RBF neural network as the model parameters of the theoretical steering axis moment of inertia and the theoretical equivalent load coefficient and the comprehensive correction value of the disturbance.

[0024] In some alternative solutions, establishing a servo control model based on the sliding mode variable structure control theory and the servo dynamics equation includes:

[0025] Establishing a servo dynamics equation for the servo deflection angle, the theoretical inertia of the steering axis, the theoretical equivalent load coefficient, and the equivalent friction torque at the motor output end in the model parameters;

[0026] Based on the errors of the theoretical inertia of the steering axis and the theoretical equivalent load coefficient, rewriting the servo dynamics equation to obtain a first rewritten servo dynamics equation;

[0027] Based on the sliding mode variable structure control theory, rewriting the servo dynamics equation to obtain a second rewritten servo dynamics equation;

[0028] Combining the first rewritten servo dynamics equation and the second rewritten servo dynamics equation to establish a servo control model.

[0029] In some alternative solutions, rewriting the servo dynamics equation based on the errors of the theoretical inertia of the steering axis and the theoretical equivalent load coefficient to obtain a first rewritten servo dynamics equation includes:

[0030] Decomposing the theoretical inertia of the steering axis into the actual inertia of the steering axis and the inertia error value of the steering axis, and decomposing the theoretical equivalent load coefficient into the actual equivalent load coefficient and the equivalent load coefficient error value;

[0031] Based on the actual inertia of the steering axis and the inertia error value of the steering axis, and the actual equivalent load coefficient and the equivalent load coefficient error value, rewriting the servo dynamics equation to obtain a first rewritten servo dynamics equation.

[0032] In some alternative solutions

[0033] The first rewritten servo dynamics equation is:

[0034] where J0 is the actual inertia of the steering axis, is the second derivative of the servo deflection angle, K0 is the actual equivalent load coefficient, φ is the servo deflection angle, M is the controlled input of the servo, τ f is the equivalent friction torque at the motor output end, f is the model parameter error of the theoretical inertia of the steering axis and the theoretical equivalent load coefficient and the unknown external nonlinear disturbance, J′ is the inertia error value of the steering axis, K′ is the equivalent load coefficient error value, and d is the unknown external nonlinear disturbance;

[0035] The second rewritten servo dynamics equation is:

[0036]

[0037] Wherein, J is the theoretical rudder shaft moment of inertia, K1 is the coefficient of the switching function proportional term, and the rudder angle tracking error e = φ - φ d , φ d is the desired trajectory of the rudder angle, is the derivative of the rudder angle tracking error, is the sliding mode switching surface, K a is the convergence speed coefficient of s, K r is the convergence speed coefficient of sgn(s), and sgn() is the sign function, is the second derivative of the desired trajectory of the rudder angle, and K is the theoretical equivalent load coefficient.

[0038] In some alternative solutions, the controlled input of the servo is:

[0039]

[0040] Wherein, M is the controlled input of the servo, is the comprehensive correction value of the model parameters and disturbances, is the estimated value of the equivalent friction torque at the motor output end.

[0041] In a second aspect, the present invention further provides a servo control device based on a neural network and a disturbance observer, including:

[0042] A parameter acquisition module, which is used to acquire the model parameters of the digital servo model, and the model parameters include the equivalent friction torque at the motor output end, the servo deflection angle, the rudder shaft moment of inertia, and the theoretical equivalent load coefficient;

[0043] A model establishment module, which is used to establish a servo control model based on the sliding mode variable structure control theory and the servo dynamics equation;

[0044] A friction torque compensation module, which is used to use a disturbance observer to perform disturbance compensation on the equivalent friction torque at the motor output end to obtain an estimated value of the equivalent friction torque at the motor output end;

[0045] A model parameter and disturbance correction module, which is used to input the tracking error of the servo deflection angle and its derivative into the trained neural network to obtain a comprehensive correction value of the model parameters and disturbances regarding the theoretical rudder shaft moment of inertia and the theoretical equivalent load coefficient;

[0046] A servo controlled input determination module, which is used to determine the servo controlled input according to the estimated value of the equivalent friction torque at the motor output end, the comprehensive correction value of the model parameters and disturbances, and the servo control model.

[0047] Thirdly, the present invention also provides a servo control device based on a neural network and a disturbance observer. The servo control device based on a neural network and a disturbance observer includes a processor, a memory, and a servo control program based on a neural network and a disturbance observer that is stored on the memory and can be executed by the processor. When the servo control program based on a neural network and a disturbance observer is executed by the processor, the steps of the servo control method based on a neural network and a disturbance observer described in any one of the above are implemented.

[0048] Compared with the prior art, the advantages of the present invention are as follows: In this solution, a friction torque generated by a digital servo is compensated by a disturbance observer, an uncertain term of a dynamic model and an unknown external disturbance are approximated by a neural network, and trajectory tracking control is completed by combining an inverse sliding mode controller. It is applied to the motion tracking control of a small digital servo of an aircraft, and the effectiveness of the composite controller proposed in this paper is verified by simulation. Compared with traditional PID controllers and sliding mode controllers, the solution proposed by the present invention has good disturbance compensation and approximation capabilities, high tracking accuracy, fast response speed, and strong robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0050] Figure 1 It is a flowchart of the servo control method based on a neural network and a disturbance observer in an embodiment of the present invention;

[0051] Figure 2 It is a structural schematic block diagram of the servo control device based on a neural network and a disturbance observer in an embodiment of the present invention;

[0052] Figure 3 It is a schematic diagram of the RBF neural network structure in an embodiment of the present invention;

[0053] Figure 4 It is a control structure block diagram of the RBF neural network in an embodiment of the present invention;

[0054] Figure 5 It is an expected trajectory tracking curve of the rudder deflection angle in an embodiment of the present invention;

[0055] Figure 6 It is a tracking error curve of the rudder deflection angle in an embodiment of the present invention;

[0056] Figure 7It is the disturbance variable estimation curve of the equivalent friction torque at the motor output end in the embodiment of the present invention;

[0057] Figure 8 It is a schematic diagram of the servo control device based on neural network and disturbance observer in the embodiment of the present invention. Specific implementation manners

[0058] In order to enable those skilled in the art of the present technology to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present application.

[0059] As Figure 1 shown, in the first aspect, the present invention provides a servo control method based on neural network and disturbance observer, including the following steps:

[0060] S1: Obtain the model parameters of the digital servo model.

[0061] In this example, the model parameters of the digital servo model include: servo deflection angle, theoretical inertia of the servo shaft, theoretical equivalent load coefficient, and equivalent friction torque at the motor output end.

[0062] S2: Based on the sliding mode variable structure control theory and the servo dynamics equation, establish a servo control model.

[0063] Step S4 includes:

[0064] S21: Establish a servo dynamics equation for the servo deflection angle, theoretical inertia of the servo shaft, theoretical equivalent load coefficient, and equivalent friction torque at the motor output end in the model parameters.

[0065] The dynamics equation of the miniaturized digital servo is established as: where is the servo deflection angle, M is the input torque of the digital servo, J is the inertia of the servo shaft, τ f is the equivalent friction torque at the motor output end, K is the equivalent load coefficient, is the second derivative of the servo deflection angle.

[0066] where M max is the maximum load torque, φ max is the maximum servo deflection angle.

[0067] S22: Rewrite the servo dynamics equation based on the errors of the theoretical rudder shaft moment of inertia and the theoretical equivalent load coefficient to obtain the first rewritten servo dynamics equation.

[0068] Step S42 specifically includes:

[0069] A: Decompose the theoretical rudder shaft moment of inertia into the actual rudder shaft moment of inertia and the rudder shaft moment of inertia error value, and decompose the theoretical equivalent load coefficient into the actual equivalent load coefficient and the equivalent load coefficient error value.

[0070] Specific decomposition results in: where J is the theoretical rudder shaft moment of inertia, J0 is the actual rudder shaft moment of inertia, J′ is the rudder shaft moment of inertia error value; K is the theoretical equivalent load coefficient, K0 is the actual equivalent load coefficient, and K′ is the actual equivalent load coefficient error value.

[0071] B: Rewrite the servo dynamics equation based on the actual rudder shaft moment of inertia and the rudder shaft moment of inertia error value, as well as the actual equivalent load coefficient and the equivalent load coefficient error value, to obtain the first rewritten servo dynamics equation.

[0072] The first rewritten servo dynamics equation is:

[0073] where J0 is the actual rudder shaft moment of inertia, is the second derivative of the servo deflection angle, K0 is the actual equivalent load coefficient, φ is the servo deflection angle, M is the controlled input of the servo, τ f is the equivalent frictional torque at the motor output end, f is the model parameter error of the theoretical rudder shaft moment of inertia and the theoretical equivalent load coefficient and the external unknown non - linear disturbance, J′ is the rudder shaft moment of inertia error value, K′ is the equivalent load coefficient error value, and d is the external unknown non - linear disturbance.

[0074] S23: Rewrite the servo dynamics equation based on the sliding mode variable structure control theory to obtain the second rewritten servo dynamics equation.

[0075] First, define the rudder angle tracking error: e = φ - φ d , where φ is the servo deflection angle and φ d is the desired trajectory of the rudder angle.

[0076] According to the sliding mode variable structure control theory, a sliding mode switching surface can be designed: where K1 represents the proportional term coefficient of the switching function, and K1 > 0.

[0077] Then the dynamics equation can be rewritten as:

[0078]

[0079] Therefore, to ensure the stability of the controlled system, observing the above equation, the desired controlled input of the digital servo can be defined, that is, the second rewritten servo dynamics equation is:

[0080]

[0081] Where J is the theoretical moment of inertia of the steering axis, K1 is the coefficient of the proportional term of the switching function, the steering angle tracking error e = φ - φ d , φ d is the desired trajectory of the steering angle, is the derivative of the steering angle tracking error, is the sliding mode switching surface, K a is the convergence speed coefficient of s, K r is the convergence speed coefficient of sgn(s), sgn() is the sign function, is the second derivative of the desired trajectory of the steering angle, and K is the theoretical equivalent load coefficient.

[0082] If the controlled input torque satisfies the above relationship, the steering angle tracking error can approach 0.

[0083] Next, verify the rationality of the desired controlled input of the servo:

[0084] Construct the Lyapunov function V:

[0085] Take the derivative of the Lyapunov function V: Where, is the derivative of V, is the derivative of s. Substitute Equation Equation into Equation After arrangement, we get:

[0086] Equation satisfies the Lyapunov stability criterion, so the system is stable, verifying the rationality of the design of the controlled input.

[0087] S24: Combine the first rewritten servo dynamics equation and the second rewritten servo dynamics equation to establish a servo control model.

[0088] The servo control model is:

[0089] Where M is the controlled input of the servo, f is the model parameter error of the theoretical moment of inertia of the steering axis and the theoretical equivalent load coefficient and the unknown external nonlinear disturbance, τ f is the equivalent frictional torque at the output end of the motor.

[0090] S3: Use a disturbance observer to perform disturbance compensation on the equivalent frictional torque at the motor output end to obtain an estimated value of the equivalent frictional torque at the motor output end.

[0091] For the non-linear frictional torque generated during the output of the steering gear, a non-linear disturbance observer can be used to perform disturbance compensation on the frictional torque, and it does not need to rely on a specific friction model, improving the versatility of the controller.

[0092] Step S2 specifically includes:

[0093] S31: Based on the disturbance variable regarding the equivalent frictional torque at the motor output end and the non-linear function to be designed in the disturbance observer, define the auxiliary variable of the disturbance observer.

[0094] Define the auxiliary variable of the disturbance observer as: z = τ f -p(x).

[0095] where, Z is the auxiliary variable, τ f is the equivalent frictional torque at the motor output end, and p(x) is the non-linear function to be designed.

[0096]

[0097] In the formula, l(x) is the observation gain. To simplify the derivation, it is designed as a linear function vector. t is the time variable, x is the observer input value, J is the theoretical moment of inertia of the steering axis, is the second derivative of the steering gear deflection angle.

[0098] S32: Based on the observation gain and the auxiliary variable in the disturbance observer, obtain the state estimated value of the auxiliary variable.

[0099] The state estimated value of the auxiliary variable z can be expressed as:

[0100] where, is the derivative of the state estimated value of the auxiliary variable; is the state estimated value of the auxiliary variable; l(x) is the observation gain; K is the theoretical equivalent load coefficient; is the derivative of the steering gear deflection angle; M is the digital steering gear input torque; p(x) is the non-linear function to be designed.

[0101] S33: Based on the state estimated value of the auxiliary variable and the non-linear function to be designed in the disturbance observer, determine the estimated value of the equivalent frictional torque at the motor output end.

[0102] The estimated value of the equivalent frictional torque at the motor output end is: where, is the estimated value of the equivalent frictional torque at the motor output end.

[0103] Next, it is proved whether the designed disturbance observer is stable.

[0104] Define the estimation error of the disturbance observer:

[0105] For equation Taking the derivative gives: Here, is the derivative of and f is the derivative of τ Here, is the derivative of

[0106] For equation Solve the differential equation:

[0107] When t→∞, where ε N is an infinitesimal positive constant. Therefore, it can be shown that the designed disturbance observer is stable and bounded.

[0108] S4: Input the tracking error of the steering gear deflection angle and its derivative into the trained neural network to obtain the model parameters of the theoretical steering shaft moment of inertia and the theoretical equivalent load coefficient and the disturbance comprehensive correction value.

[0109] In this example, the neural network uses an RBF (Radial basis function network) neural network. The tracking error of the steering gear deflection angle is input into the RBF neural network, and the output of the RBF neural network is used as the model parameters of the theoretical steering shaft moment of inertia and the theoretical equivalent load coefficient and the disturbance comprehensive correction value.

[0110] The RBF neural network has good generalization performance, a simple network structure, avoiding unnecessary complex calculations. Theoretically, it can approximate any nonlinear function with arbitrary precision in a compact set. Its network structure is as Figure 3 shown. The RBF neural network consists of three layers: the input layer, the hidden layer, and the output layer.

[0111] Input layer: x = [x1, x2, …, x n T represents the input of the network, where n represents the input dimension of the network.

[0112] Hidden layer: The network output of the hidden layer h = [h1, h2, …, h m T , m represents the output dimension of the hidden layer. Select the Gaussian basis function as the membership function of the input layer. The output of the jth neuron is:

[0113] ​​

[0114] In the formula, is the coordinate vector of the center point of the Gaussian function of the j-th neuron in the hidden layer, and b = [b1, b2, …, b m T is the width of the Gaussian function of the j-th neuron.

[0115] Output layer: The output of the RBF neural network is:

[0116] y(t) = W T h = W1h1 + W2h2 + … + W m h m

[0117] When the system has large uncertainties, the RBF neural network can approximate any nonlinear function and improve the performance of the controller; the adaptive control method has the advantage of not requiring prior knowledge of unknown parameters, and its adaptive rate can be derived by constructing a Lyapunov function in neural network control, and the adjustment of the adaptive weights is used to ensure the convergence and stability of the entire closed-loop system. Based on the RBF neural network, this paper designs an adaptive RBF-NN controller to approximate the model error and external unknown disturbances, and improves the stability of the system through online estimation. The controller structure is as Figure 4 shown.

[0118] Select the input of the neural network The theoretical output of the neural network is: f * = W *T h(x) + ε. The actual output is That is, the comprehensive correction value of the model parameters and disturbances regarding the theoretical moment of inertia of the rudder shaft and the theoretical equivalent load coefficient.

[0119] In the formula, e is the rudder angle tracking error, is the derivative of the rudder angle tracking error, W * is the ideal weight of the neural network, ε is the approximation error of the neural network, and h(x) is the network output of the hidden layer.

[0120] Define the error between the ideal weight and the actual weight * as the weight error, W is the ideal weight,

[0121] S5: Determine the controlled input of the steering gear according to the estimated value of the equivalent friction torque at the output end of the motor, the comprehensive correction value of the model parameters and disturbances, and the steering gear control model.

[0122] ​Substitute the estimated value of the equivalent friction torque at the motor output end, the model parameters, and the comprehensive correction value of the disturbance into the servo control model, and the controlled input of the servo is obtained as follows:

[0123]

[0124] where M is the controlled input of the servo, is the comprehensive correction value of the model parameters and the disturbance, is the estimated value of the equivalent friction torque at the motor output end.

[0125] To verify the stability of the system corresponding to the control algorithm proposed in this paper, select the Lyapunov function:

[0126] Derive the formula to obtain:

[0127] Substitute the formula e = φ - φ d and into the formula Finally, we can obtain:

[0128]

[0129] Take the RBF neural network adaptation rate: Substitute it into the above formula, and finally we can get:

[0130] Since when and only when s = c = 0, that is, when s = c = 0. According to the LaSalle invariance principle, the closed-loop system is asymptotically stable, that is, when t → ∞, c → ∞ and s → ∞, so q → q d .

[0131] To verify the effectiveness of the proposed adaptive composite control algorithm, write the S function of the relevant controller in Matlab, and build a simulation environment through Simulink to conduct simulation analysis and verification on the traditional PID controller, the sliding mode variable structure controller (SMC), and the designed adaptive composite controller. Select the system parameters as: J = 0.158, K = 2.5; the model uncertainty is set as: The desired output deflection angle of the servo is set as Assume the friction disturbance is

[0132] See Figure 5 and Figure 6 and Figure 7 , the simulation curves of the PID, SMC, and the controller in this paper are shown. Among them Figure 5is the desired trajectory tracking curve of the rudder deflection angle, Figure 6 is the tracking error curve of the rudder deflection angle, Figure 7 is the estimated curve of the disturbance variable regarding the equivalent friction torque at the motor output end.

[0133] It can be seen that although all three controllers can achieve the tracking of the desired trajectory, the controller proposed in this paper has the fastest convergence speed, smaller tracking error, the highest accuracy, can better suppress the influence brought by the nonlinear disturbance, and has the strongest robustness. From the simulation results, it can be known that the RBF neural network adaptive controller, the disturbance observer and the backstepping sliding mode controller can be effectively combined to suppress the nonlinear disturbance and uncertainty of the system, and can further improve the response speed, tracking accuracy and stability of the system.

[0134] In summary, the model parameters of the digital servo model are obtained; the equivalent friction torque at the motor output end is compensated by using a disturbance observer to obtain the estimated value of the equivalent friction torque at the motor output end; the tracking error of the servo deflection angle is input into the trained neural network to obtain the model parameters and the disturbance comprehensive correction value regarding the theoretical rudder shaft moment of inertia and the theoretical equivalent load coefficient; based on the sliding mode variable structure control theory and the servo dynamics equation, according to the estimated value of the equivalent friction torque at the motor output end, the model parameters and the disturbance comprehensive correction value, the controlled input of the servo is determined. The friction torque generated by the digital servo is compensated by the disturbance observer, the uncertain terms of the dynamic model and the external unknown disturbances are approximated by the neural network, and the trajectory tracking control is completed by combining the backstepping sliding mode controller. It is applied to the motion tracking control of the small digital servo of the aircraft, and the effectiveness of the composite controller proposed in this paper is verified by simulation. Compared with the traditional PID controller and the sliding mode controller, the proposed scheme of the present invention has good disturbance compensation and approximation capabilities, and has high tracking accuracy, fast response speed and strong robustness.

[0135] As Figure 2 shown, on the second aspect, the present invention also provides a servo control device based on a neural network and a disturbance observer, including: a parameter acquisition module, a friction torque compensation module, a model parameter and disturbance correction module, and a servo controlled input determination module.

[0136] Among them, the parameter acquisition module is used to acquire the model parameters of the digital servo model; the friction torque compensation module is used to compensate the equivalent friction torque at the motor output end by using a disturbance observer to obtain the estimated value of the equivalent friction torque at the motor output end; the model parameter and disturbance correction module is used to input the tracking error and its derivative of the servo deflection angle into the trained neural network to obtain the model parameters and the disturbance comprehensive correction value regarding the theoretical rudder shaft moment of inertia and the theoretical equivalent load coefficient; the servo controlled input determination module is used to determine the servo controlled input based on the sliding mode variable structure control theory and the servo dynamics equation, according to the estimated value of the equivalent friction torque at the motor output end, the model parameters and the disturbance comprehensive correction value.

[0137] Among them, the function implementation of each module in the above servo control device based on a neural network and a disturbance observer corresponds to each step in the above embodiment of the servo control method based on a neural network and a disturbance observer, and its function and implementation process will not be elaborated here one by one.

[0138] In a third aspect, an embodiment of the present application provides a servo control device based on a neural network and a disturbance observer. The servo control device based on a neural network and a disturbance observer can be a device with data processing functions such as a personal computer (PC), a laptop computer, a server, etc.

[0139] Refer to Figure 8 , Figure 8 which is a schematic diagram of the hardware structure of the servo control device based on a neural network and a disturbance observer involved in the solution of the embodiment of the present application. In the embodiment of the present application, the servo control device based on a neural network and a disturbance observer may include a processor, a memory, a communication interface, and a communication bus.

[0140] Among them, the communication bus can be of any type and is used to interconnect the processor, the memory, and the communication interface.

[0141] The communication interface includes input / output (I / O) interfaces, physical interfaces, and logical interfaces, etc., which are used to implement the interconnection of components inside the servo control device based on a neural network and a disturbance observer, and interfaces for implementing the interconnection of the servo control device based on a neural network and a disturbance observer with other devices (such as other computing devices or user devices). The physical interface can be an Ethernet interface, a fiber optic interface, an ATM interface, etc.; the user device can be a display screen, a keyboard, etc.

[0142] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical memory, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0143] The processor can be a general - purpose processor, which can call the servo control program based on neural network and disturbance observer stored in the memory and execute the servo control method based on neural network and disturbance observer provided in the embodiments of the present application. For example, the general - purpose processor can be a central processing unit (CPU). Among them, the method executed when the servo control program based on neural network and disturbance observer is called can refer to the various embodiments of the servo control method based on neural network and disturbance observer in the present application, which will not be elaborated here.

[0144] Those skilled in the art can understand that Figure 8 the hardware structure shown in does not constitute a limitation to the present application, and it may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.

[0145] In the fourth aspect, the embodiments of the present application further provide a computer - readable storage medium.

[0146] The computer - readable storage medium of the present application stores a servo control program based on neural network and disturbance observer. When the servo control program based on neural network and disturbance observer is executed by a processor, it realizes the steps of the servo control method based on neural network and disturbance observer as described above.

[0147] Among them, the method realized when the servo control program based on neural network and disturbance observer is executed can refer to the various embodiments of the servo control method based on neural network and disturbance observer in the present application, which will not be elaborated here.

[0148] It should be noted that the serial numbers of the above - mentioned embodiments of the present application are only for description and do not represent the superiority or inferiority of the embodiments.

[0149] The terms "including" and "having" and any variations thereof in the specification, claims and drawings of the present application are intended to cover non - exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices. The descriptions of terms such as "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit that "first", "second" and "third" are of different types.

[0150] In the description of the embodiments of this application, words such as "exemplary", "for example", or "for illustration purposes" are used to indicate examples, illustrations, or explanations. Any embodiment or design solution described as "exemplary", "for example", or "for illustration purposes" in the embodiments of this application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "exemplary", "for example", or "for illustration purposes" is intended to present the relevant concepts in a specific manner.

[0151] In the description of the embodiments of this application, unless otherwise specified, " / " means "or". For example, A / B can mean A or B; "and / or" in the text is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "a plurality of" means two or more than two.

[0152] In some processes described in the embodiments of this application, a plurality of operations or steps appear in a specific order. However, it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of this application or may be executed in parallel. The serial numbers of the operations are only used to distinguish different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in order or in parallel, and these operations or steps may be combined.

[0153] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium as described above (such as ROM / RAM, magnetic disk, optical disc) and includes several instructions for causing a terminal device to execute the methods described in the various embodiments of this application.

[0154] The above are only the preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or equivalent process transformation made by using the content of the specification and drawings of this application, or directly or indirectly applied in other related technical fields, is equally included in the patent protection scope of this application.

Claims

1. A servo control method based on a neural network and a disturbance observer, characterized in that, It includes the following steps: Obtain the model parameters of the digital servo model, where the model parameters include the equivalent frictional torque at the motor output end, the servo deflection angle, the moment of inertia of the steering shaft, and the theoretical equivalent load coefficient; Based on the sliding mode variable structure control theory and the servo dynamics equation, establish a servo control model; Use a disturbance observer to perform disturbance compensation on the equivalent frictional torque at the motor output end to obtain an estimated value of the equivalent frictional torque at the motor output end; Input the tracking error of the servo deflection angle and its derivative into the trained neural network to obtain the model parameters and disturbance comprehensive correction values for the theoretical moment of inertia of the steering shaft and the theoretical equivalent load coefficient; Determine the controlled input of the servo according to the estimated value of the equivalent frictional torque at the motor output end, the model parameters, the disturbance comprehensive correction values, and the servo control model.

2. The servo control method based on neural network and disturbance observer according to claim 1, characterized in that The use of the disturbance observer to perform disturbance compensation on the equivalent frictional torque at the motor output end includes: Based on the disturbance variable regarding the equivalent frictional torque at the motor output end and the nonlinear function to be designed in the disturbance observer, define the auxiliary variable of the disturbance observer; Based on the observation gain in the disturbance observer and the auxiliary variable, obtain the state estimated value of the auxiliary variable; Based on the state estimated value of the auxiliary variable and the nonlinear function to be designed in the disturbance observer, determine the estimated value of the equivalent frictional torque at the motor output end.

3. The servo control method based on a neural network and a disturbance observer according to claim 2, characterized in that: The auxiliary variable is: z = τ f -p(x); where Z is an auxiliary variable, τ f is a disturbance variable regarding the equivalent frictional torque at the motor output end, and p(x) is a non-linear function to be designed; The state estimation value of the auxiliary variable is determined according to the formula ; wherein, is the derivative of the state estimation value of the auxiliary variable; is the state estimation value of the auxiliary variable; l(x) is the observation gain; K is the theoretical equivalent load coefficient; is the derivative of the steering gear deflection angle; M is the digital steering gear input torque; p(x) is the nonlinear function to be designed; The estimated value of the equivalent frictional torque at the motor output end is: Among them, is the estimated value of the equivalent friction torque at the motor output end.

4. The servo control method based on a neural network and a disturbance observer according to claim 1, wherein The neural network uses an RBF neural network. Input the tracking error of the servo deflection angle into the RBF neural network, and use the output of the RBF neural network as the model parameters and disturbance comprehensive correction values for the theoretical moment of inertia of the steering shaft and the theoretical equivalent load coefficient.

5. The servo control method based on a neural network and a disturbance observer according to claim 1, wherein The establishment of the servo control model based on the sliding mode variable structure control theory and the servo dynamics equation includes: Establish a servo dynamics equation regarding the servo deflection angle, the theoretical moment of inertia of the steering shaft, the theoretical equivalent load coefficient, and the equivalent frictional torque at the motor output end in the model parameters; Based on the errors of the theoretical moment of inertia of the steering shaft and the theoretical equivalent load coefficient, rewrite the servo dynamics equation to obtain the first rewritten servo dynamics equation; Based on the sliding mode variable structure control theory, rewrite the servo dynamics equation to obtain the second rewritten servo dynamics equation; Combine the first rewritten servo dynamics equation and the second rewritten servo dynamics equation to establish a servo control model.

6. The servo control method based on neural network and disturbance observer according to claim 5, characterized in that The rewriting of the servo dynamics equation based on the errors of the theoretical moment of inertia of the steering shaft and the theoretical equivalent load coefficient to obtain the first rewritten servo dynamics equation includes: Decompose the theoretical moment of inertia of the steering shaft into the actual moment of inertia of the steering shaft and the moment of inertia error value of the steering shaft, and decompose the theoretical equivalent load coefficient into the actual equivalent load coefficient and the equivalent load coefficient error value; Based on the actual moment of inertia of the steering shaft and the moment of inertia error value of the steering shaft, as well as the actual equivalent load coefficient and the equivalent load coefficient error value, rewrite the servo dynamics equation to obtain the first rewritten servo dynamics equation.

7. The servo control method based on a neural network and a disturbance observer according to claim 1, characterized in that: The first rewritten servo dynamics equation is as follows: Among them, J0 is the actual inertia of the rudder shaft, is the second derivative of the servo deflector angle, K0 is the actual equivalent load coefficient, φ is the servo deflector angle, M is the controlled input of the servo, τ f is the equivalent frictional torque at the motor output end, f is the model parameter error of the theoretical inertia of the rudder shaft and the theoretical equivalent load coefficient and the unknown external nonlinear disturbance, J′ is the error value of the inertia of the rudder shaft, K′ is the error value of the equivalent load coefficient, and d is the unknown external nonlinear disturbance; The second rewritten servo dynamics equation is: Among them, J is the theoretical moment of inertia of the rudder shaft, K1 is the coefficient of the proportional term of the switching function, and the rudder angle tracking error e = φ - φ d , φ d is the desired trajectory of the rudder angle, is the derivative of the rudder angle tracking error, is the sliding mode switching surface, and K a is the convergence speed coefficient of s, and K r is the convergence speed coefficient of sgn(s), where sgn() is the sign function, is the second derivative of the desired trajectory of the rudder angle, and K is the theoretical equivalent load coefficient.

8. The servo control method based on neural network and disturbance observer according to any one of claims 1-7, characterized in that The controlled input of the servo is: where M is the controlled input of the steering gear, is the comprehensive correction value of the model parameters and disturbances, is the estimated value of the equivalent friction torque at the motor output end.

9. A servo control device based on a neural network and a disturbance observer, characterized in that, It includes: A parameter acquisition module, which is used to acquire the model parameters of the digital servo model, and the model parameters include the equivalent friction torque at the motor output end, the servo deflection angle, the moment of inertia of the servo shaft, and the theoretical equivalent load coefficient; A model establishment module, which is used to establish a servo control model based on the sliding mode variable structure control theory and the servo dynamics equation; A friction torque compensation module, which is used to use a disturbance observer to perform disturbance compensation on the equivalent friction torque at the motor output end to obtain an estimated value of the equivalent friction torque at the motor output end; A model parameter and disturbance correction module, which is used to input the tracking error of the servo deflection angle into the trained neural network to obtain the model parameter and disturbance comprehensive correction values of the theoretical moment of inertia of the servo shaft and the theoretical equivalent load coefficient; A servo controlled input determination module, which is used to determine the servo controlled input according to the estimated value of the equivalent friction torque at the motor output end, the model parameter and disturbance comprehensive correction values, and the servo control model.

10. A servo control device based on a neural network and a disturbance observer, characterized in that, The servo control device based on the neural network and the disturbance observer includes a processor, a memory, and a servo control program based on the neural network and the disturbance observer that is stored on the memory and can be executed by the processor. When the servo control program based on the neural network and the disturbance observer is executed by the processor, the steps of the servo control method based on the neural network and the disturbance observer as described in any one of claims 1 to 8 are implemented.

Images