Research And Optimization Design Based On BP Network About Tetra Pak Euipment Peventive Mintenance System

In recent years ,the domestic liquid food industry competitive is becoming high and high,and the cost control in Enterprises is increasingly stringent. The Tetra Pak equipment mainly used in packaging dairy products, has become the competing target of compression in enterprises and failed to play the effectiveness of preventive maintenance, because it not only uses TPMS (Tetra Pak Maintenance System) with high cost but also been tested that the replacement of spare parts is too late or too early in the course of operation, which. In order to adapt domestic demands in TPMS operation and optimize the original preventive maintenance system, it is necessary to develop intelligentized software system as the auxiliary system for TPMS.This paper in-depth studed the TPMS system and, first of all, we established the way that one week is a unit for maintainance, carried out the concept of residual life of spare parts, and as well used the operation and maintenance records of Tetra Pak equipment enterprises over the past years to set up the Relational DataBase Management System. Secondly, we utilized BP artificial neural network which developed more mature and prediction function to achieve the goal of predicting residual life of spare parts of equipment maintenance. Finally, we established the ratio of maintenance security residual life, a comprehensive reliabile parameter, and used the reliabile indication of equipment operation as selection criteria for prediction, which fully embodies the thought that reliability is the core for equipment maintainance.The development of our software systems is based on Windows XP environment, and it uses Visual Basic6.0 as the main development language, VB-embedded commercialization database management system; Access2003 provides data support for system prediction and caculating of reliability parameters; the interface between database and systems is achieved through ADO control; use Matlab as a operation platform for the BP neural network to realize the BP network training and maintain the content of prediction; uses ActiveX control on interface between Matlab and systems. All these methods supplied the soft flat for the clarity of equipment system's maintenance and management.Through the compare between the prediction results and corresponding original datas, we find that our system which make use of the combination design of nerve center and networks structure can accurately reflects the intinsic relationship between operating environment and equipment spare parts, and then guarantee the accuracy results of the system forecast. It can be concluded that our study is the effective method to resolve the contradiction between reliability and economy of Tetra Pak maintenance system.