Universal identification method for unknown nonlinear system

Abstract

The invention discloses a universal identification method for an unknown nonlinear system. The method comprises the following steps: S1, collecting data; S2, inputting the collected data into a Pade sparse framework formed by a Pade transformation module and an identification module, and processing to obtain a predicted system control equation P (x); and S3, performing feedback control on data collection according to the deviation degree of the system control equation P (x) obtained in the step S2 so as to automatically optimize the obtained system control equation P (x). According to the invention, the method does not aim at a certain type of specific system, and has universality and universality; the Pade function base adopted in the method is obtained through Pade approximate transformation and has good convergence, so that convergence can be well achieved for a nonlinear system, and the characteristic of high precision is achieved; and according to the method, the complete control equation of the system can be explicitly identified, and compared with neural network type machine learning, the method has white box performance.

Description

technical field [0001] The invention relates to the fields of mathematics and intelligent control, in particular to a general identification method for unknown nonlinear systems. Background technique [0002] As human knowledge advances, our ability to simulate and predict the phenomena we observe around us has grown, from the understanding of astronomical phenomena to the engineering of advanced technologies. But there are still countless phenomena that we cannot effectively model and understand, such as the generation of brain consciousness, the internal laws of some complex physical systems, etc. For these complex systems, we lack an effective systematic framework for modeling prediction and interpretation, so as to understand the behavior of these complex systems. [0003] With the improvement of computing power and the development of sensor technology, for many systems of interest, we can obtain a large amount of measurement data, even beyond our ability to parse and p...

AI Technical Summary

Problems solved by technology

[0023] 1. There are few existing identification methods for explicit identification of completely unknown nonlinear systems;

[0024] 2. The existing identification methods are black-box;

[0025] 3. Existing identification methods require prior knowledge of the system and are not universal

Images