Technique Of PLC Users Dynamic Evaluation Based On Rough Set And SVM

Programmable Logic Controller (PLC) is the kernel product of autocontrol field. With the development of economy and the elevation of automatization level, the global market share is improved continuously. For the sake of holding the market trend adequately and improving the competitive power of enterprise, the dynamic evaluation of customer value and the implementation of classification management have been the important market strategy. Some customer evaluation methods in common use classify customer simply according to the attributes of customer without sufficient learning and generalization. However, it will result in the incorrect classification of customer if the variation attribute of customer value influenced by uncertain factors is ignored.According to the characteristics of PLC users, a customer dynamic evaluation method based on rough set and support vector machine is advanced. Customer evaluation indicators are established from three dimensions of current value, potential value and odditional value. The relative weights of indicators are confirmed by the improved analytic hierarchy process. The dynamic evaluation of customer value is realized based on the annual amplitude rate of indicators. The periodic observed customer value is discretized by way of rough set and boolean inference arithmetic for keeping the indiscernibility relation of information system. It gets much better majorization results and convergence speed to reduce redundant attributes by particle swarm optimization than genetic algorithm. The customer classification model of scaling dataset is established and the optimal model parameters are obtained by10-cross validation and grid-search for preventing unbalanced train by some oversize or undersize attributes and improving the classification accuracy of model. For validating the validity of this method, the original data of sample customers from2006to2011can be found by systematic sampling taking some automatization company as example. The data is pretreated by rough set and the train and test of svm one against one classifier are accomplished. The result indicates that the evaluation method can not only implement the periodic evaluation and classification of customer value, but also have a better generalization.The customer relationship dynamic management system is designed based on the foregoing theory. The building process of database and the design of UML class diagram are dissertated detailedly. This system is used by the market department of some automatization company and the work flow is viewable. It is more significant to classify customer based on customer value.