Heating Furnace Temperature Control Method Based On Fuzzy Neural Network Research

Crude oil as an important material resources, have a crucial impact on national economic development, considering the characteristics of crude oil in China, high wax content, high freezing point and high viscosity, need to constantly to heat crude oil, preventing the precipitation occurred in the process of transportation of crude oil or set. The heating furnace is one of the main equipment in the process, how to design the control system of heating furnace, to improve the level of automation control system improve the efficiency of system control and shortening the time of system control has a very important role.Heating furnace is a nonlinear, strong coupling, multivariable and interference of complex control system, it mainly contains three control direction:adjust the ventilation and air volume, ensure the pressure in the chamber of a stove or furnace, insures the safe running of the system; Maintain within the scope of the best air-fuel ratio, and has reached the higher combustion efficiency, guarantee the economy of system control process; Ensure the output of crude oil temperature within the normal range, to ensure smooth delivery of crude oil. And these requirements by using traditional control methods can not meet, combined with modern advanced control methods, fuzzy control and fuzzy neural network combined with the advantage of BP neural network, fuzzy system not only don’t need to establish an accurate mathematical model of control system, and the use of BP neural network can be between actual output and desired output error back propagation, to fix the system of membership functions and weights, the system output is stable in a shorter time. This paper introduces the fuzzy system with the characteristics of BP neural network, and combining site and characteristics of two control methods. Considering the shortcomings of BP neural network, is easy to make the system into the defects of slow convergence speed and easy to fall into local minimum, the initial parameters and network Settings has a great influence on the performance of the BP algorithm, the GA is a kind of group type iterative adaptive heuristic global search algorithm, the optimization of neural network weights and parameters has global convergence, fuzzy neural network with GA training can avoid falling into local minimum, so choose GA- BP network, the genetic algorithm combined with BP neural network to modify the initial data of the system.Finally user MATLAB to fuzzy neural network control system simulation, compared with the traditional PID control methods, analysis of this kind of control method in a variety of Angle control the superiority of the system.