Application Of Multivariable Non-linear Decoupling Control Based On Neural Network In DCS

Along with the development of artificial neural network, a neuron network technology with its unique characteristics, has penetrated into every branch of control theory. It has the ability to learn, generalization ability, tolerance and nonlinear mapping capability. Being used for decoupling, it can solve problems that the traditional decoupling method can't do, it provide a new idea. Today, traditional analog control system has been replaced by digital intelligent control system, intelligent control system offers an open software development platform, and the convenient for user is the secondary development. But the platform own control algorithm function is very lack, often only realizes independent single-loop PID regulation and switch quantity of process control. the requirements of many industrial fields cannot be satisfied by the present on-the-spot control requirements. How to use the existing mature control algorithm in intelligent control system? This paper introduced an idea with an experimental feasible example.In this thesis, based on Matlab mathematical software and the BP neural network control technology we solve the loop coupling problem of the PTC heater ceramic in DCS. PTC heater is typical nonlinear ceramic object, we analyse the object properties, and establish the system simulation model in the Matlab simulation software, then train the neural network decouple controller, at last use the result of study, combined with the ABB Freelance 800F distribution control system, we design the BP neural network decoupling controller, now the neural network control technology is applied in the DCS system. By the way we can take advantage of the configuration software's rapidly developing application, and also use the neuron network technology in the decoupling control. the value of the design is that the control algorithm is no longer limited to simple PID control in DCS.