Search

CN-121995760-A - Intelligent optimal control method of networked system

CN121995760ACN 121995760 ACN121995760 ACN 121995760ACN-121995760-A

Abstract

The invention discloses an intelligent optimal control method of a networked system, which comprises the steps of S1, establishing a dynamic model of an uncertain strict feedback nonlinear system, defining an event trigger mechanism based on a relative threshold value, S2, adopting a radial basis function neural network to perform online approximation on an unknown nonlinear function in the system, S3, defining a coordinate transformation error variable based on a back-stepping method, designing a virtual control law and a parameter self-adaptive law for an intermediate subsystem, S4, designing an actual control law and an inverse optimal control law based on the event trigger mechanism and a Lyapunov stability theory, wherein the inverse optimal control law has a relation with the actual control law, S5, executing event trigger control according to the actual control law and the event trigger mechanism, and applying a control signal meeting the inverse optimality to a controlled object to realize closed-loop stable control. The invention can solve the problems of signal discontinuity, system uncertainty and network resource limitation.

Inventors

  • WU JIE
  • LIU ZHI

Assignees

  • 广东工业大学
  • 人工智能与数字经济广东省实验室(广州)

Dates

Publication Date
20260508
Application Date
20260210

Claims (10)

  1. 1. An intelligent optimal control method for a networked system is characterized by comprising the following steps: s1, establishing a dynamic model of an uncertain strict feedback nonlinear system, and defining an event trigger mechanism based on a relative threshold value; S2, adopting a radial basis function neural network to perform online approximation on an unknown nonlinear function in the system; S3, defining a coordinate transformation error variable based on a back-stepping method, and designing a virtual control law and a parameter self-adaptive law for an intermediate subsystem; S4, designing an actual control law and an inverse optimal control law based on an event triggering mechanism and a Lyapunov stability theory, wherein the inverse optimal control law has a relation with the actual control law; S5, executing event trigger control according to the actual control law and the event trigger mechanism, and applying a control signal meeting inverse optimality to the controlled object to realize closed-loop stable control.
  2. 2. The intelligent optimal control method of a networked system according to claim 1, wherein the dynamic model of the uncertain strict feedback nonlinear system is set up as follows: ; Wherein, the As a system state vector of the system, , In order to control the input of the device, For the output of the system, Is an unknown smooth nonlinear function; Represent the first A subsystem; Is the total number of subsystems.
  3. 3. The intelligent optimal control method of a networked system according to claim 2, wherein the event trigger mechanism is defined as: controlling update time By triggering conditions Determining, wherein In order to measure the error of the measurement, For the output of the ideal controller, In order to actually control the input of the device, And Is a design parameter; 。
  4. 4. A method for intelligent optimal control of a networked system according to claim 3, wherein the event trigger mechanism satisfies: at intervals of time In, actual control input Expressed as: ; Wherein, the And Is a time-varying parameter and meets And 。
  5. 5. The intelligent optimal control method of a networked system according to claim 1, wherein the online approximation of the unknown nonlinear function in the system using the radial basis function neural network comprises: approximating an unknown function using a radial basis function neural network Wherein As an ideal weight vector for the model of the model, As a vector of the gaussian basis function, Is an approximation error; the following relationship is satisfied: ; Wherein, the As a center point of the lens, the lens is, Is the width.
  6. 6. The intelligent optimal control method of a networked system according to claim 1, wherein step S3 includes: Definition of tracking error Wherein Is a reference track; Definition of the first embodiment Error variable of subsystem Wherein Is the first Virtual control law of step design; And constructing a model and other unknown parameters of the adaptive law online estimated neural network weight.
  7. 7. The intelligent optimal control method of a networked system according to claim 6, wherein the auxiliary function is defined to handle non-smooth points in the back-stepping method, including soft sign functions Switching function ; ; ; Wherein, the method comprises the following steps of.
  8. 8. A method for intelligent optimal control of a networked system according to claim 3, wherein designing the actual control law and the inverse optimal control law comprises: Design of ideal controller output signals To compensate for errors caused by system nonlinearities and event triggers; Actual control signal applied to system At the moment of triggering Updated to Remains unchanged during the triggering interval; constructing inverse optimal control law Wherein To design gain parameters.
  9. 9. The intelligent optimal control method of a networked system according to claim 8, wherein the ideal controller outputs signals The design is as follows: ; Wherein, the , For estimating ; ; 。
  10. 10. The intelligent optimal control method of a networked system according to claim 8, wherein step S5 includes: the actual control signal determined in step S4 Applied to the controlled object as a final instruction; Actual control signal Output of signals by an ideal controller Obtained by sampling and holding by an event trigger mechanism and the actual control signal Satisfy the following requirements The inverse optimal relation of the system is ensured, and the optimality and stability of the system under a preset performance index are ensured.

Description

Intelligent optimal control method of networked system Technical Field The invention relates to the technical field of automatic control, in particular to an intelligent optimal control method of a networked system, which is particularly suitable for a remote network control scene of a strict feedback nonlinear system with uncertainty. Background With the rapid development of modern networked control systems, optimal control design of nonlinear systems has become an important research direction in the field of automatic control. The core goal of optimal control is to find a control strategy, and minimize a preset cost function while guaranteeing the stability of the system, so as to realize global optimization of the system performance. However, for nonlinear systems with uncertainty, classical optimal control theory generally requires solving the Hamilton-Jacobi-Bellman (HJB) equation. Since the HJB equation is essentially a nonlinear partial differential equation, it is difficult to obtain an analytical solution in most cases, which makes classical optimal control a great computational challenge in practical engineering applications. To overcome the difficulty of directly solving the HJB equation, an inverse optimal control (InverseOptimalControl) method is proposed. According to the method, a cost functional function which enables the known calm controller to be an optimal controller is constructed, so that a complicated HJB equation is prevented from being directly solved. On the other hand, in modern network control systems, communication bandwidth and computing resources are often limited. To increase resource utilization, event-triggered control (Event-TriggeredControl, ETC) mechanisms have evolved. Unlike conventional periodic sampling control, event-triggered control updates the control signal only when the system state meets certain trigger conditions, effectively reducing unnecessary communication transmissions and actuator actions. However, combining event triggering mechanisms with inverse optimal control faces serious theoretical challenges. The main problem is that conventional inverse optimal control designs are typically based on lyapunov control function (CLF), requiring a smooth resolution of the controller, whereas the control input generated by the event trigger mechanism is essentially a piecewise constant and discontinuous signal. The discontinuity of the signal breaks the requirement of the traditional inverse optimal control theory on smoothness, so that an event-triggered inverse optimal nerve control method of an uncertain nonlinear system is realized. The existing theoretical framework is difficult to be directly applied. Furthermore, for a strictly feedback nonlinear system common in practical engineering, the interior of the system often contains an unknown nonlinear dynamic function, which further increases the complexity of the controller design. The existing reinforcement learning or self-adaptive dynamic programming method can solve the problem of partial optimal control, but is often heavy in calculation load and difficult to provide strict stability theoretical guarantee. Therefore, a control scheme capable of solving the problems of signal discontinuity, system uncertainty and network resource limitation is needed. Disclosure of Invention The invention aims to overcome the defects of the prior art and provide an intelligent optimal control method of a networked system, which can solve the problems of signal discontinuity, system uncertainty and network resource limitation. In order to achieve the above purpose, the technical scheme provided by the invention is as follows: An intelligent optimal control method of a networked system, comprising the following steps: s1, establishing a dynamic model of an uncertain strict feedback nonlinear system, and defining an event trigger mechanism based on a relative threshold value; S2, adopting a radial basis function neural network to perform online approximation on an unknown nonlinear function in the system; S3, defining a coordinate transformation error variable based on a back-stepping method, and designing a virtual control law and a parameter self-adaptive law for an intermediate subsystem; S4, designing an actual control law and an inverse optimal control law based on an event triggering mechanism and a Lyapunov stability theory, wherein the inverse optimal control law has a relation with the actual control law; S5, executing event trigger control according to the actual control law and the event trigger mechanism, and applying a control signal meeting inverse optimality to the controlled object to realize closed-loop stable control. Further, the dynamic model of the uncertain strict feedback nonlinear system is established as follows: ; Wherein, the As a system state vector of the system,,In order to control the input of the device,For the output of the system,Is an unknown smooth nonlinear function; Represent the first A