| In the business community of China,for the manufacturing execution system(MES)understanding and application is only in its infancy.The enterprise has not rich experience about how to build the use of MES,and the enterprise is relatively unfamiliar for the core concept of MES,key technology and the way of the fusion of the latest technology.After modern tobacco industry has experienced reform for the "automatic" and "intelligent",all kinds of system specification through the whole production process in tobacco enterprises,but problems exist is still in the initial stage of the finished product inspection,control and feedback the results,on the level of quality management and process control ability is relatively low,set quality goals and objectives is not high.Most of the production quality problem by quality management professionals to deal with,the quality problems that arise cause serious losses,even brings irreparable situation.In this paper,add on sub-module about quality process management system and quality traceability system in MES original modules.This article describes the quality of the process management system,is the change about the results of control and feedback in the past,and can be done in the process of production for product quality and timely feedback and control,and improves the level of process control ability.We analyze and clean a financial transaction data set containing 30 features and normal and abnormal tags,on this basis,study and construct an auto-encoder neural network.The neural network is used to study the reconstruction error of the normal behavior data in the data set,and then we predict a small amount of abnormal behavior data and get the reconstruction error.In the optimization process of the former,it is found that the size of the auto-encoder reconstruction error does not affect the detection ability of the classifier,and the training time of the auto-encoder neural network is shortened to the original 86% by the optimization of the latter. |
PDF Full Text Request