Decision-making Expert System In Production Cell Changeover Based On Interactive Simulation And Neural Network

With the continuous strengthening of the trend of economic globalization, and theaccelerating of the pace of industrial restructuring, the market environment ofmanufacturing enterprise has been facing to changing dramatically, especially in theseaspects: individual needs of customers for products, timeliness of product delivery andtransient lifecycle. The production cell can shorten significantly the manufacturingcycle, save the production area, avoid the backlog of inventory, improve utilization ofequipment because it could coordinate the production process to efficient logistics, so ithas been widely using in discrete manufacturing, metallurgy, ship building industry andso on.In practice, the production cell changeover frequently, which is caused by regularorder inserting, numerous items of parts, complex produce structure, and flexibleprocess route and device configuration. At the same time, the decision-making ofchangeover is characterized by complex, dynamic randomness and multi-objective, sothe traditional method of changeover which mainly depends on the experiencedengineer could not meet the requirements of increasingly complex productionenvironment. But the merits and defects of a decision of changeover directly impact onproduct quality, production cycle and production cells flexibility. Therefore, how toobjectively and accurately reflect the above-mentioned characteristics of the productionchangeover decision-making and rationally make the decision of the production cellchangeover, has became a major issue for academia and the business community.At present, due to the production cell autonomy, evolution and complexity, and theuncertain and dynamic work method, task process and behavior of people in theproduction cell, the changeover decision-making method that is simply based onmathematical optimization model exists the following questions. First, the simplemathematical model is difficult to meet the global optimum and high feasibility of theproduction requirement because the production cell has increasingly shown a dynamicand multi-objective characteristics. Second, the production cell has been tending to be acomplex giant system because of the improve of requirement to flexible manufacturingsystem, which is effected by diversification of product demand, the temporary insertingorder, emergency order and product delivery deadline, so mathematical modeling isdifficult to reflect fully and accurately the practical producing situation. Third, it is difficult to quantify the factors effecting person making a decision, and simplemathematical modeling methods can not reflect the role of these factors. In the view ofthe limitations of traditional changeover decision-making method, and the shortage ofcurrent expert system based on neural networks such as the lack of mean of knowledgecollection, the absence of the rule extracting of productionchangeover expert systemand narrow application mode, this paper puts forward the production cell changeoverdecision-making expert system based on interactive simulation and the hybrid neuralnetwork. The content includes the following:Firstly, on the basis of the research in general expert systems and the expert systemstructure and principle based on neural network, this paper comes up with theimplement principle and the key route and technology of the production cell changeoverdecision-making expert system based on interactive simulation and the hybrid neuralnetwork. In addition, taking into account the knowledge expansion of the changeoverexpert, the paper proposes the working process and application model based on WebService.Secondly, to solve the problem of the lack of the current expert system knowledgeacquisition, the paper uses interactive simulation to access knowledge acquisition,elaborates the building process of simulation models of manufacturing systems andinteractive simulation model, puts forward the decision algorithm and the expertdecision-making selection algorithm, and introduces the process of interactivesimulation by an actual examples.Thirdly, the paper provides specifically the implementation technology of theproduction changeover decision-making expert system. For the complexity anddiversity of the influence factor of production changeover decision-making, usesprincipal component analysis to reduce the dimensionality of the production state vector;deals with multi-mode of variable in the production control strategy by using integrationFisher classifier and regression neural network computing technology; puts forward theextractive technique of the rules of inference engine based on the IER-Trepan algorithmand preset text technology, which solves the shortage of the extraction mean oftraditional expert system rule. For the narrow nature of the application mode oftraditional production changeover decision-making expert system, proposes theapplication model of productioncell changeover decision-making expert system basedon Web Service, which is achieved by using the package of two-tier COM objectstructure and expert system program. Finally, taking the production changeover decision-making of a motorcycle engineproduction cell as a research example, the paper verifies the feasibility of theimplementation principles and techniques. The results show that the production cellchangeover decision-making expert system based on interactive simulation and thehybrid neural network is feasible and operable, and it provides a feasible method for theproduction cell changeover decision-making intelligently.