Research On Data Reconciliation Theory And Applications In Manufacturing Execution System

As the bridge of process control system and enterprise resource planning, manufacturing execution system plays an important role in the production management. Data reconciliation, as one of the key techniques of MES, makes full use of redundancy information to reduce process measurement error. More accurate and reliable data is provided to other application modules of MES to optimize the production management.This thesis concerns on multi-resolution data rectification modeling and gross error detection based on evidence theory. The main contents of this thesis are outlined as follows,(1) On the basis of two kinds of general equipment models in petrochemical enterprise, the concept of material flow adjacency matrix is proposed to describe the global material flow directions and mass balance equations. According to the information of material-flow adjacency matrix as well as the material balance of equipment, data rectification model can be built accurately and conveniently. By converting the material flow adjacency matrix based on practical demand, is converted, multi-resolution data rectification models with different time and spatial scales can be obtained.(2) Based on D-S evidence theory, a novel method combined two traditional gross error detection algorithms is proposed. The evidences of gross error are obtained via Measurement Test (MT) and Node Test (NT). Then these evidences are simplified in accordance with some rules. After evidence synthesis, gross error can be found. The simplification rules presented in the paper make the amount of evidence synthesis computation reduce sharply. When there are several gross errors, sequential compensation strategy is used to avoid matrix becoming singular during the process of iterative calculation.(3) The characteristics and weakness of logistics management of petrochemical enterprises are discussed. Then, a brief introduction of Sinopec Manufacturing Execution System, as well as its material and energy balance modules, is also given. Finally, the approach of multi-resolution data rectification modeling and gross error detection method based on evidence theory, which are proposed in this thesis, are applied for the resolution of material balance problem in a petrochemical enterprise. The result demonstrates the correctness and practicability of the proposed methods.At the end of the thesis, the conclusion and perspective are given.