Quadratic polynomial-based nonlinear compound PID (proportional-integral-differential) neural network control method

Abstract

The invention discloses a quadratic polynomial-based nonlinear compound PID (proportional-integral-differential) neural network control method. The method comprises the following steps of: firstly analyzing roles of three gain parameters of PID during the control process, obtaining rough curves of the three gain parameters along with error variation, constructing three quadratic polynomial-based nonlinear compound gain functions according to the three gain curves to further obtain nonlinear compound PID control law, taking the nonlinear compound PID control law as a neural network model, respectively fusing seven operation units including a linear proportional operation unit, a nonlinear proportional operation unit, a linear integral operation unit, a nonlinear integral operation unit, a linear differential operation unit and two nonlinear differential operation units into hidden layer neurons, and constructing a quadratic polynomial-based nonlinear compound PID neural network controller. Nonlinear compound PID control signals can be generated by on-line real-time training through a neural network, thereby implementing dynamic control on a nonlinear controlled object. According to the invention, the nonlinear object can be quickly and accurately controlled, the calculation quality is small, the real-time property is good, and the robustness is strong.

Description

technical field [0001] The invention belongs to the field of automatic control, and relates to an intelligent control method for online self-stabilization of parameters of hidden layer neurons by integrating a linear and nonlinear compound PID operation unit. Background technique [0002] Proportional, integral, and differential (P, I, D) control according to the deviation is the control method with the longest history and the strongest vitality. Although advanced control strategies are gradually being promoted today, more than 90% of the control loops currently in operation are still PID controllers. However, with the increase of the complexity of the system and the increase of the uncertain factors of the object, the traditional linear PID control is no longer applicable, but the nonlinear PID control can truly reflect the nonlinear relationship between the control amount and the deviation signal. To a certain extent, it overcomes the shortcomings of the linear PID contro...

AI Technical Summary

Problems solved by technology

However, on the one hand, this method involves two-layer neural network weight adjustment, and there is a phenomenon of weight coupling, so the algorithm converges slowly and requires a large amount of calculation, which is not conducive to the real-time control of the fast sampling system; on the other hand, this method only uses linear PID The computing unit is integrated into the hidden layer neurons of the neural network after clipping processing, and its nonlinear control ability is not strong

Images