Dynamic economic performance optimization and control integrated design method for non-Gaussian system

Abstract

The invention discloses a dynamic economic performance optimization and control integrated design method for a non-Gaussian system, and belongs to the field of random system optimization control. Themethod comprises the steps: 1, building a nonlinear random system model under the non-Gaussian disturbance; 2, building a two-layer dynamic optimization control framework structure; 3, carrying out the performance index selection based on the statistical information; 4, carrying out the dynamic economic performance optimization and the design of a model prediction controller. Compared with a conventional two-layer RTO-MPC employing a steady-state model at an upper layer, the method gives full consideration to the dynamic characteristics of a process, highlights the reachability of the obtainedoptimal set track, and has greater practical engineering significance.

Description

technical field [0001] The invention specifically relates to an integrated design method of economic performance optimization and control of a system under the influence of non-Gaussian random noise, and belongs to the field of random system optimization control. Background technique [0002] In practical systems, random disturbances exist widely, especially in industrial processes, such as the random factors of temperature concentration of reactors in chemical processes. Therefore, the optimal control of stochastic systems has always been a hot research topic in the field of system control. Based on the assumption of Gaussianity of system variables, a lot of research results have been achieved by optimizing and controlling the mean and variance, and have been widely used in engineering practice. However, the actual industrial process variables are often non-Gaussian, which makes the study of non-Gaussian distributed control systems more necessary. Inspired by the classic ...

AI Technical Summary

Problems solved by technology

But in fact, the set value / reference trajectory needs to be set according to the actual economic performance index, which is a movement trajectory that changes at any time, so this solution needs further consideration;

[0006] (2) Similarly, in terms of non-Gaussian stochastic control, the existing literature mainly constructs instantaneous performance indicators based on statistical information to obtain a single-step control law, which cannot predict the future control action trend, and has a certain degree of conservatism;

[0007] (3) In the traditional two-level economic process optimization of real-time optimization and model predictive control, real-time optimization does not consider the dynamic characteristics of the process, so the accessibility of the obtained set values ​​cannot be guaranteed

Images