Research On Data Rectification And Its Application In Manufacturing Execution System

The keen competition environment puts forward higher requirement for fine management, manufacturing execution system(MES) emerges as the time require. As one of the key techniques of MES, data rectification provides accurate datas for business modules of MES, as well as other related application modules. The dissertation, taken the process industry, discusses some solutions and an application in logistics balance of MES.The main contents in this dissertation are outlined as follows:1) Data rectification and its application are reviewed, the research status of manufacturing execution system are also mentioned.2) As it comes to data rectification, the most common technique is the method of least squares(LS), however, its fitting degree of reconciled value and true value is not always satisfactory. Mixed integer linear programming(MILP) performs better than LS, but it needs great amount of calculation due to the introduction of too many binary variables. In the dissertation, a simultaneous data reconciliation and measurement bias detection algorithm based on the mixed integer linear programming framework (SE-MILP) is presented. At the first stage of the algorithm, a set of measurements suspect is obtained. Next, the mixed integer optimization approach is used, where binary variables are only added to process variables that belong to the suspect set. By reducing the number of binary variables, the algorithm decreases computational complexity in some extent.3) As we all know, there are two kinds of gross errors, namely measurement related errors and process related errors. The latter errors such as process leaks are seldom considered in the existing approaches. But process related errors can also deteriorate the calibration result badly. In view of this, process related errors are taken into account of SE-MILP algorithm. A set of measurements and units suspect of being biased or having leaks is also obtained at the first stage of the extended algorithm. The simulation result shows that the extended algorithm is of high reliability and low computational complexity.4) The complex logistics characteristics and management weakness of petrochemical enterprises bring many problems to logistics balance of manufacturing execution system. The problem of process data modeling is discussed. Finally, the extended SE-MILP is used for the resolution of logistics balance problems of petrochemical enterprises, which also shows the good performance of the algorithm for practical problems.The conclusion and perspective are given at the end of the dissertation.