Gaussian process model-based predictive control method for multi-variable nonlinear dynamic system model

Abstract

The invention discloses a Gaussian process model-based predictive control method for a multi-variable nonlinear dynamic system model, belongs to the technical field of predictive control of the multi-variable nonlinear dynamic system model and aims at solving the technical problem of providing improvement of the Gaussian process model-based predictive control method for the multi-variable nonlinear dynamic system model. The technical scheme adopted for solving the technical problem is as follows: the predictive control method comprises the following steps of (1) building an external dynamic PLS framework; (2) predicting output data and obtaining a plurality of single-input and single-output systems in hidden space through decoupling of a dynamic GP-PLS model; (3) carrying out control by using the dynamic GP-PLS model and designing a model predictive controller in each single-input and single-output system; (4) obtaining an optimum control action through minimizing an objective function; and (5) reconstructing the model predictive control result in the hidden space back to original space and controlling the original space. The Gaussian process model-based predictive control method is applied to the multi-variable nonlinear dynamic system model.

Description

technical field [0001] The invention discloses a multivariable nonlinear dynamic system model predictive control method based on a Gaussian process model, and belongs to the technical field of predictive control of multivariable nonlinear dynamic system models. Background technique [0002] With the rapid development of industry and information science and technology, the scale of industrial production is getting larger and larger, and the production process and production process are becoming more and more complex, which poses a major challenge to traditional mechanism modeling and control strategies, especially in Petroleum, chemical, metallurgy, machinery and other industries for application. Model Predictive Control (MPC), as an advanced computer control algorithm, estimates and calculates the system's future state optimization sequence based on the current and past operating states of the system. The first input value of the optimization sequence is used on the system. ...

AI Technical Summary

Problems solved by technology

[0005] 1. At present, in the field of control, the RBF neural network is mainly used as the nonlinear prediction model in predictive control. However, the structure design of the neural network itself still mainly relies on experience, and there are many optimization parameters and the training process is complicated, which makes the neural network has many limitations;

[0006] Second, before implementing MPC, a well-described model must be developed for the process, but accurate first-principles models are not only difficult to obtain, but also because complete process knowledge is often scarce, when used for MPC design, it is usually preferred to start from Data-driven models identified in input / output data, however, when implementing MPC using data-driven models, especially when building well-described models from highly correlated data, the process is cumbersome and complex;

[0007] Third, for large-scale multiple-input multiple-output (MIMO) systems, in MIMO processing, the solution of control actions becomes expensive and time-consuming, and the cross-coupling of process variables leads to difficult controller design, which has great impact on the real-time application of MPC in industry. The application is very limited;

[0008] Fourth, for systems with higher dimensions, the use of traditional MPC algorithms will not only lead to complex design of the controller, but also increase the amount of calculation of the controller, so it is necessary to improve the control algorithm to overcome the shortcomings of traditional model predictive control

Images