The Resistance Furnace Temperature Control System Based On Model-free Learning Adaptive Control Method

Resistance furnace is widely used in metallurgy, chemical industry,machinery and other kinds of industrial control process. For a resistance furnace temperature control system, it's important to design a control strategy for temperature control. Temperature control strategy is the key to both temperature measurement and temperature control. Temperature measurement is based on temperature control, this technology is quite mature. But the resistance furnace is a kind of large inertia and time delay systems, door switches, heating materials, ambient temperature and power voltage affect the control process, so how to improve the resistance furnace temperature control accuracy has been an important research topic.In resistance furnace temperature control PID algorithm is the most widely used, however, its performance may degrade when applied to highly nonlinear processes. When the plant's dynamics are hard to characterize precisely or are subject to environmental uncertainties, one may encounter difficulties in applying the conventional controller design methodologies. It is difficult to guarantee accuracy when using conventional PID control algorithm for plant's time-varying, disturbance and parameter uncertainty on the resistance furnace.In recent years, a new theory and technology is raised, model-free learning adaptive control(MFLAC).The basic idea is to use a newly introduced concept of pseudo gradient vector, replace the general discrete time nonlinear system on one point in abstract model of the controlled object with the linear dynamic, and only use the data of the object to compute the pseudo gradient vector in order to achieve adaptive control of nonlinear systems. It is designed only with the controlled system input and output data controller, the controller does not contain any information of mathematical model and control theory in the control system.This paper introduces the primary theory of PID control algorithm, model-free adaptive control, BP neural network, then design a model-free adaptive control method using BP neural network compensator, which is applied in the resistance furnace control system based on detailed research. In this paper, a resistance furnace system is designed with labview as a development environment, which has system temperature monitoring, temperature control, data analysis and other functions. The system can select PID control, model free adaptive control or model free adaptive control with BP neural network compensation to carry on experimental study. A large number of experimental studies showed that the system has good robustness and adaptability.