The Application Of Neural Networks On The Supervise And Control System Of Centeral Air Condition And The Research Of System Integration

This thesis was carried out basing on the project that was central air condition supervise and control of some cigarette factory, whose area is 1200 square meter and very great. Because the request of control sensitivity for temperature and humidity was very high, the area of air condition was carved up, and every subarea had its independent the central air condition system. The system of central air condition is most important part of power, but the main control model of PID modulate applied on the system of central air condition had some problems.which made each system and equipment free connect and communication and realized more far, seamless, real open and free communication between control layer and management layer of system and equipments. As a result, a satisfying result was achieved. In addition, in the control system of central air condition, in order to make the system integration better, and realize the communication and integration of different systems, the author brought out a kind of software technique. It was simply introduced of composing and work characteristic of supervise and control system of the central air condition. We analyzed the main reasons affecting the temperature and humidity of central air condition, and affirmed the mission and characteristic of central air condition. And analyzed many problems of present central air condition supervise and control system in the pattern of classic PID control, especial great area air condition system. We summarized the basic theory and ways of Artificial Neural Networks. According to existing problems of present great area central air condition supervise and control system in the pattern of classic PID control, the feasibility was analyzed, which is the techniques of Artificial Neural Networks applying in the air condition supervise and control air condition. We deeply analyzed of the classic arithmetic of Back-Propagation (BP) networks, and worked out the reasons that caused the Error Shock and brought out the theoretic solving project. At the same time, the processing method to the sample data was improved.We applied the Artificial Neural Networks theory on the temperature and humidity supervise and control system of central air condition, and selected input parameters, output nodes, the reasonable the structure and active function, the best learning velocity and momentum coefficient through the study of sample data, which improved the learning velocity and bettered the project of control basing on the convergence of system, and reach the anticipative purpose of control. According to the existing problems in the integration between control network and information network, and through the research that OPC techniques applied on the data exchange of supervise and control system of the central air condition, we deeply analyzed the structure and function of all kinds of rules in the techniques on OPC. Basing on it, we researched and empolder the server and client of OPC DA, and given the program code of many interfaces. The research showed that the Artificial Neural Network applied on the temperature and humidity control system of central air condition, which made up the shortage of the classic PID control in the face of unsteadily, badly fit the changes of work condition and big inertia of control. As a result, the outlets are satisfied. In addition, OPC not only improved the interoperability of supervise and control system, but also easily extended functions, which made the system more opening and brought us a new way of system integration.