Research On The Method Of Inverse Control Of Industrial Air-Conditioning System

With the development of Chinese Textile Industry, the demand of productivity effect and textile quality is increasing, which needs higher environment of textile mill. Preference, stationary temperature and relative humidity are the key factors of increasing productivity and insuring the product quality. Therefore, the control performance of industrial air-conditioning system in textile mill should fit increasing. Based on application of Inverse System Theory and requirement of Neural Network method, the Neural Network inverse system is adopted to meet the control demand of air-condition system.The dissertation introduces the background and development of nonlinear control system firstly. As the difficulties of nonlinear control system, the concept of Neural Network inverse system is proposed. Also its merits of resolving nonlinear problems are involved. Secondly, the investigation is taken on industrial air-conditioning structure and operation principle. Considering the influences of time-delay and nonlinear sections, a nonlinear industrial air-conditioning model, including 2 inputs, 2 outputs and strong coupling performance, is built. At last, air-conditioning model is verified according to the actual air-conditioning.Textile mill temperature and relative humidity system is a multi-variable,nonlinear and strongly coupling controlled plant. The dissertation, applying Inverse System Theory, takes a research on the mathematical model of nonlinear plant, testifying system invertibility. Cascading the inverse system which consists of a static Neural Network and integrators with the air-conditioner, controlled plant is decoupled into two independent first-order linear subsystems, with the characters of independent of the air-conditioning model and parameters.Because of many influences, the pseudo-linearization system can not realize exact linear decoupling. So, additive controllers are needed, which adopts PID control strategy. The combinational controller, including Neural Network and addictive controllers, is founded. As a result, a Neural Network inverse combinational control system is built.The simulation system was based on MATLAB environment. The temperature and relative humidity of textile mill are effectively controlled by collecting data, training network and adjusting PID parameters. The result shows that Neural Network inverse control method has a better performance.