Search

CN-121978949-A - Model prediction-based self-adaptive control method for quick reflection mirror

CN121978949ACN 121978949 ACN121978949 ACN 121978949ACN-121978949-A

Abstract

The invention discloses a fast reflecting mirror self-adaptive control method based on model prediction, which relates to the technical field of fast reflecting mirror control, and comprises the steps of constructing a self-adaptive control model based on parameter information of a fast reflecting mirror, splitting the self-adaptive control model into a plurality of sub-control areas, and executing model simulation on each sub-control area in a preset simulation period; and the whole working procedure of the local optimal control is packaged into a control assembly and stored into a shared data pool for the self-adaptive control call of the fast reflecting mirror under other periods. The invention improves the control precision and response efficiency of the fast reflection mirror, reduces the computing and compiling resource expenditure and is suitable for fast reflection mirror control scenes with high precision requirements through partition control, state prediction and cross-period component multiplexing.

Inventors

  • ZHENG MINGCHUN
  • ZHAO SHUBIN

Assignees

  • 北京迅来光电技术有限公司

Dates

Publication Date
20260505
Application Date
20260206

Claims (9)

  1. 1. The model prediction-based adaptive control method for the quick reflection mirror is characterized by comprising the following steps of: step S1, constructing a corresponding self-adaptive control model based on parameter information of a quick reflection mirror, splitting the self-adaptive control model to obtain a plurality of sub-control areas, and executing model simulation on each sub-control area in a preset simulation period; s2, predicting the fast reflection mirror deflection states under different sub-control areas based on the result of model simulation, taking the fast reflection mirror deflection states as training parameters of the self-adaptive control model, and executing local optimal control in one period on each sub-control area by the self-adaptive control model; And step S3, packaging all working procedures of local optimal control in one period into a control component, storing the control component into a shared data pool, and selecting the control component in the shared data pool for calling by the adaptive control of the fast reflection mirror in other periods.
  2. 2. The adaptive control method of a fast mirror based on model prediction according to claim 1, wherein the process of constructing the corresponding adaptive control model based on the parameter information of the fast mirror comprises: the parameter information of the fast reflecting mirror comprises body structure parameters, frequency characteristic parameters, environment state parameters and calibration feedback information, the parameter information is processed through an engineering modeling method, an example model of the fast reflecting mirror is constructed, and the example model is subjected to twin processing based on a digital twin technology to obtain a twin model of the fast reflecting mirror; Selecting a plurality of control points on the twin model, and sequentially executing sweep frequency test, step test and ray tracing test on the fast reflecting mirror at each control point to obtain a space feature vector and a dynamic response parameter corresponding to each control point of the fast reflecting mirror; The twin model of the fast mirror is built as an adaptive control model based on a plurality of control points.
  3. 3. The adaptive control method for a fast reflecting mirror based on model prediction according to claim 2, wherein the process of constructing an adaptive control model based on control points and splitting the adaptive control model to obtain a plurality of sub-control areas comprises: the space feature vector is used for representing the unique space coordinate of each control point on the quick reflection mirror; the dynamic response parameters are used for representing the working condition characteristics of the quick reflection mirror at each control point; Clustering the spatial feature vectors of a plurality of control points based on respective dynamic response parameters to obtain a plurality of clustering set clusters, and setting virtual points corresponding to the mean value vectors of all the spatial feature vectors as logical clustering centers of the corresponding clustering set clusters for the spatial feature vectors of the dynamic response parameters of all the control points in each clustering set cluster; Mapping all the space feature vectors to a twin model, converting the twin model into an adaptive control model, sequentially connecting other control points except for a clustering center in each clustering cluster to the respective clustering centers, and constructing a sub-control area of the adaptive control model on the quick reflection mirror.
  4. 4. A model prediction based adaptive control method for a fast mirror according to claim 3, wherein the process of performing model simulation for each sub-control region within a preset simulation period comprises: Presetting a simulation period, and synchronously starting a self-adaptive control model to simulate the model of a plurality of sub-control areas where the fast reflection mirror is positioned in the simulation period, wherein the model simulation comprises task configuration, environment configuration, simulation operation and simulation evaluation; The task configuration is used for creating a primary simulation task of the sub-control area, and the simulation task comprises a plurality of instruction signals for controlling the quick reflection mirror to finish correct procedure operation and disturbance conditions for causing disturbance to the quick reflection mirror; The environment configuration is used for building a simulation environment for executing the simulation task; The simulation operation is that the quick reflection mirror sequentially carries out a simulation task in a simulation environment based on a plurality of instruction signals, the positions of disturbance conditions inserted into the instruction signals are set, and the response state, time domain information and frequency domain information of the simulation task after each disturbance condition is added are recorded; and after the simulation operation is executed, performing simulation evaluation.
  5. 5. The model prediction-based adaptive control method for a fast mirror according to claim 4, wherein the process of predicting the fast mirror shift state under different sub-control areas based on the result of the model simulation comprises: constructing a standard model of the quick reflection mirror, dividing the quick reflection mirror into a plurality of objects to be analyzed, setting model parameters of each object to be analyzed, and constructing an instability model of the whole quick reflection mirror composed of all the objects to be analyzed based on the model parameters; And determining the mirror surface offset of each state analysis point of the quick reflection mirror based on the instability model and the standard model, and integrating the mirror surface offset of a plurality of state analysis points to serve as the quick reflection mirror offset state under the corresponding sub-control area.
  6. 6. The model prediction based adaptive control method for a fast mirror according to claim 5, wherein the process of performing the locally optimal control for each sub-control region by the adaptive control model using the fast mirror offset state as a training parameter of the adaptive control model comprises: Integrating the fast reflection mirror offset states of all sub-control areas as a training parameter set; counting the control request frequency of each sub-control area in a preset time window, determining a common area and a common area based on the control request frequency, creating a container queue for the common area, and creating a data bin for the common area; Connecting a resource operation node with a container queue, constructing a first communication channel, processing all instruction signals of the self-adaptive control model into a control script, and transmitting the control script carrying analysis resources at the resource operation node to the container queue through the first communication channel; performing operation through the container to finish local optimal control of the common area in one period; And connecting the resource operation node with the data bin, constructing a second communication channel, transmitting the analysis resource and the control script carried by the resource operation node into the data bin through the second communication channel, and processing the control script based on the analysis resource to finish the local optimal control of the common area in one period.
  7. 7. The model prediction based adaptive control method of a fast mirror according to claim 6, wherein the process of the container queue performing an operation on a container of the control script in the common area based on the parsing resource comprises: the container queue is composed of a plurality of container nodes, each container node is used for processing a sub-control area of the common area, the first container node of the container queue is used as a source node, and the rest other container nodes are used as additional nodes; compiling a control script at a source node based on analysis resources, transmitting a compiling result to an additional node, selecting an adaptive part of compiling content by the additional node based on respective control requirements of the additional node on the fast-reflection mirror, and creating an adaptive part of compiling content and a complement content part of own control requirements as an auxiliary control script for executing respective partial control operations of all container nodes on the fast-reflection mirror; Each container node integrates and creates the adapted compiled content and the auxiliary control script into respective history call templates, and the respective history call templates are cached in the respective container nodes, and when the subsequent corresponding sub-control area executes the local optimal control, the adaptation is preferentially performed from the history call templates.
  8. 8. The model predictive based adaptive control method as recited in claim 7, wherein the process of encapsulating all of the locally optimal control procedures within a cycle as control components comprises: The local optimal control comprises a plurality of working procedures for executing mirror control operation, and the working procedure time length and the front and back linking conditions of each working procedure are obtained, wherein the front and back linking conditions of the working procedure comprise parallel execution conditions and sequential execution conditions; Enabling a plurality of procedures of the parallel execution conditions to be executed concurrently, setting a maximum operation time sequence axis and a residual operation time sequence axis, screening candidate procedures based on the procedure duration of each procedure of the sequential execution conditions, judging whether the candidate procedures meet the requirements of primary packaging with the procedures under the parallel execution conditions, processing to generate a corresponding sub-control assembly when the candidate procedures meet the requirements, and directly packaging the corresponding procedures to generate the sub-control assembly when the candidate procedures do not meet the requirements; And based on the condition that the accumulated value of the process duration is minimum, integrating the sub-control assemblies of all the processes of the local optimal control in one period, and obtaining the control assembly of the sub-control area corresponding to the local optimal control.
  9. 9. The model predictive based adaptive control method as recited in claim 8, wherein storing the control components in the shared data pool, and selecting the control components in the shared data pool for invocation by the adaptive control of the fast mirror during other periods comprises: Setting a database, setting a plurality of data interfaces of a communication connection database, setting a management port for establishing communication paths from all the data interfaces to the management port, creating a data sharing window at the management port, and taking the database under the data sharing window and the plurality of data interfaces of the connection database as a shared data pool; Storing the control components into a shared data pool, taking the component information of each control component as a retrieval information certificate, selecting a data interface as an information uploading interface when the operation of the self-adaptive control of the quick reflection mirror under other periods needs to be executed, and inputting the operation information of the self-adaptive control of the quick reflection mirror under other periods into the shared data pool; and calling a control component which accords with the adaptive control corresponding to the execution process of the fast mirror under the current period based on the matched retrieval information credential in the operation information matched shared data pool.

Description

Model prediction-based self-adaptive control method for quick reflection mirror Technical Field The invention relates to the technical field of fast reflection mirror control, in particular to a fast reflection mirror self-adaptive control method based on model prediction. Background The quick reflection mirror is used as a core execution component in a high-precision optical system, is widely applied to the fields of astronomical observation, laser communication, self-adaptive optics and the like, and the control precision and response speed of the quick reflection mirror directly determine the upper limit of the performance of the whole optical system. The existing fast reflection mirror control method mostly adopts a global unified control strategy, namely a single control model is built based on the overall parameters of the fast reflection mirror, however, the body structure of the fast reflection mirror has spatial heterogeneity, and the frequency domain characteristics and dynamic responses of different areas are different, so that the single control model is difficult to adapt to the personalized control requirements of each area. Meanwhile, the traditional control method does not establish a multiplexing mechanism of cross-period control experience, model training and control parameter debugging are required to be carried out again in each control period, so that control delay is increased, calculation resources are wasted, the prediction of the offset state of each area of the quick-reflection mirror is lacked, the establishment of a control strategy often depends on experience parameters, local optimal control for different areas is difficult to realize, the overall control precision of the quick-reflection mirror is insufficient, and the requirement of a high-precision optical system on mirror stability cannot be met. Disclosure of Invention The invention aims to provide a model prediction-based adaptive control method for a quick reflection mirror, which aims to solve the problem of the deficiency in the background technology. In order to achieve the above purpose, the invention provides a fast reflection mirror self-adaptive control method based on model prediction, which comprises the following steps: step S1, constructing a corresponding self-adaptive control model based on parameter information of a quick reflection mirror, splitting the self-adaptive control model to obtain a plurality of sub-control areas, and executing model simulation on each sub-control area in a preset simulation period; s2, predicting the fast reflection mirror deflection states under different sub-control areas based on the result of model simulation, taking the fast reflection mirror deflection states as training parameters of the self-adaptive control model, and executing local optimal control in one period on each sub-control area by the self-adaptive control model; And step S3, packaging all working procedures of local optimal control in one period into a control component, storing the control component into a shared data pool, and selecting the control component in the shared data pool for calling by the adaptive control of the fast reflection mirror in other periods. In a preferred embodiment, the process of constructing the corresponding adaptive control model based on the parameter information of the fast mirror includes: the parameter information of the fast reflecting mirror comprises body structure parameters, frequency characteristic parameters, environment state parameters and calibration feedback information, the parameter information is processed through an engineering modeling method, an example model of the fast reflecting mirror is constructed, and the example model is subjected to twin processing based on a digital twin technology to obtain a twin model of the fast reflecting mirror; Selecting a plurality of control points on the twin model, and sequentially executing sweep frequency test, step test and ray tracing test on the fast reflecting mirror at each control point to obtain a space feature vector and a dynamic response parameter corresponding to each control point of the fast reflecting mirror; The twin model of the fast mirror is built as an adaptive control model based on a plurality of control points. In a preferred embodiment, the process of constructing an adaptive control model based on control points, and splitting the adaptive control model to obtain a plurality of sub-control areas includes: the space feature vector is used for representing the unique space coordinate of each control point on the quick reflection mirror; the dynamic response parameters are used for representing the working condition characteristics of the quick reflection mirror at each control point; Clustering the spatial feature vectors of a plurality of control points based on respective dynamic response parameters to obtain a plurality of clustering set clusters, and setting virtual points corresponding to the