| The Contemporary manufacturing system is developing towards discentralization, intelligentization and integration. Based on these new characteristics of agile manufacturing system, this dissertation studies systematically and roundly on the theory and methodology for New Control System (NCS), that lays a good foundation for realizing on NCS.Based on the structure design of network computing, function, software of NCS, with using B/S model, O-O(Object-Oriented), COM/DCOM, OLE, fieldbus, JAVA, Web, software bus, Plug-and-play technologies, the architecture and integrated model of NCS are built. The method of process model making is also proposed.To realize and enhance the field process ability, the conception of IP (Intelligent Pro-node) is proposed. Based on IP, the field information model is built. After introducing intelligent cell, the function realization of IP is proposed. On the basis of fuzzy diagram theory, the information-processing model of IP is also proposed. The task-allocation and reference model of MIPS(Multi Intelligent Pro-node System) based on knowledge cooperation are proposed in detail. Based on standard on smart networked transducers IEEE 1451, neuron-chip, Lonworks technology, the structure model and prototype of IP are designed.A self-adaptive PID controller based on a neural network is presented. It possesses not only the self-learning ability and adaptability, but also the function of self-adjusting factors. Based on fuzzy set, neural network theory, the fuzzy control model and fuzzy neural network control model of multi variable system are presented. Based on the Automatic Core-Welding Line of Shop Floor Control System in YangZhou Radiator Plant Computer Integrated Manufacturing System (YS-CIMS/SFCS), the Fuzzy Logic theory was applied to the controlling device and established the main heating room fuzzy temperature controller and finally was put into practical use. After a period of running, it's proved that the system is reliable, efficient with improved production quality and lower energy consuming.Based on the characteristics of diagnostic knowledge of equipments' faults, the hierarchy tree is adopted to represent both deep and shallow diagnostic knowledge. Then extended Petri Nets are used to realize the knowledge representation and fault diagnosis. Based on DAI (Distributed Artificial Intelligent), the structure of the distributed multi-agent monitoring and diagnostic system is firstly created. Then the fuzzy relationship model of the distributed multi-agent monitoring and diagnostic system is researched. The feasibility of system is improved. The method of multi-agent integrated monitoring and diagnostic system based on knowledge influence diagram is put forward. Then the integrated decision-making structure including the local decision-making and unitary decision-making is presented. For the purpose of the realization of the integrated decision-making, the knowledge -based influence diagrams and their reasoning methods are designed. Real-timeillperformance of system is improved.The operation optimization strategy in process industry is proposed, including off-line optimization, non-linearity solving strategy, on-line optimization and self-learning model. Based on polymerization reaction of the Nylon-6 rubberized cord fabric production of distributed control system in Yangzhou organic chemical plant Computer Integrated Manufacturing System (YH-CIMS/DCS), the multiple stepwise regression method was used to build the statistic mathematical models of the molecule weight and the monomer quantum of casting slice belt. Then the optimization model of polymerization reaction was presented, which was solved by using Simulation Annealing algorithm to obtain the best techniques parameters. The improved hybrid genetic algorithm and back propagation algorithm are combined to train neural network, brought out the neural network prediction model of casting slice belt's average molecule weight to guide the technologist on-line.
|
PDF Full Text Request