| Omethoate (C5H12NO4PS), a kind of pesticide, is widely used in agriculture. The process of its synthesis is very complex. The temperature of the process is required to be under -12癈, and the reaction should be finished as quickly as possible. The model of the reaction is characterized with multiple parameters, non-linear features and complex disturbances. It is difficult to be controlled automatically. Exact mathematic model is required in classic and modern control theories. But the model of the omethoate synthesis process is very complex and its parameters continuously change during the course of reaction. It is difficult to acquire sound control effect using traditional control methods. It is necessary to study new methods to control the process. Having analyzed its running and controlling features, we decided to adopt an intelligent control method.In this thesis, we firstly introduced the process, its features, and the required adding speed of NH2CH3. According to the characteristics of the process and the request of our client, we developed the scheme of the control system. The six kettles were divided into two groups. Each includes three kettles and was controlled with one industrial control computer.Then, we studied the basic control theories of omethoate synthesis. Expert system and neural network theories were specially discussed. We decided to adopt "IF-THEN" phrase to express expert knowledge. The back propagation neural networks were particularly constructed.According to the scheme of the omethoate synthesis intelligent control system, we designed and chose the hardware of the system. It included the industry control computer, the analog input card, the analog output card, the temperature and level transducer, etc. The algorithms, which are about transferring parameters from analog to digital, from digital to analog, and from voltage to project value, were also discussed.There are various disturbances in the omethoate synthesis plant. We adopted some measures to obtain high stability and to enhance the system's anti-jamming ability. The system's lines are distributed separately from other power lines. The frame was soundly connected to the ground. Light-power separators were used inthe analog input and output cards. Regulators were adopted to confine the voltage value of input signals. Capacitors were used to filter noises. We also studied filter algorithm to dispose every input signals.Because the devices of the control system have some limitation in their function, we used the virtual instrument technique to expand them and to full fill some special system requirements. A virtual signal source v/as developed for the control system's simulated running. The analog input card was amended with an algorithm of our software. Soft springs were designed to prevent the valve's electromotor from running when it was closed. Virtual flow meters were designed to gain the flow value of NH2CH3.After the designing of the control system, we discussed the intelligent algorithm of the control system. First, we analyzed factors that affect the temperature of the kettle. They are the flow value of NH2CH3, the flow value and the temperature of the cooling water, the cooling system's opening or closed status of other kettles, etc. The features of the process were also discussed. We considered the process as a three inputs, one output, time variant system. We constructed a model characterized with simple inertia and delay processes. The parameters of the model change continuously during the reaction. The model, with multiple inputs, disturbances, and highly non-linear features, is difficult to be controlled using traditional methods. Considering the characteristics of the process, we divided it into four parts, which are the beginning, the ascending, the balancing and the ending periods. Expert control techniques were adopted to control the first and the last periods. In the ascending and balancing parts, which are crucial to the process, we used BP artificial neural networks to control the synthesis... |
PDF Full Text Request