| With the constant progress of industrial production technology,the product quality and economic profit is becoming more and more necessary in industrial production process.At present,in order to solve the operation state monitor of complex industrial progress,relevant scholars have done a lot of research work.Processing monitoring is for monitoring whether production process is running "normally" or "faults" are occurring.However,in order to get high quality products and a higher economic efficiency,only distinguishing"normal" and "fault" is not enough.In the condition of normal operation,it is important that industrial production process is running in the best state.So,on the basis of process monitoring,in order to provide reasonable advice for the adjustment of production process optimization operation,it is necessary to evaluate the advantages and disadvantages of production process running states.The online evaluation of the production process running status is used for the real-time evaluation of the operation of the production process through mode identification,feature extraction and feature matching based on the running data gained online,which is able to monitor the running conditions timely,accurately and comprehensively and has an important practical significance in improving production efficiency and economic benefit.For realizing operation status evaluation in multimodal complex industrial production process,this thesis proposes an online operation status evaluation method based on Gaussian Mixture model.Under the condition of knowing data is "optimal" or "non-optimal",in order to build offline models by using multimodal data,this thesis researches an offline data modal division method based on the complex Mahalanobis distance and ISODATA clustering algorithm,which realizes modal division of offline data.For the online evaluation,the complex Mahalanobis distance corresponding to each GMM mode is computed via online data and compared with its corresponding statistic control limit contrast in order to realize the online judgement of the running state of the production process and its corresponding mode.Finally,this thesis takes advantage of Tennessee Eastman simulation to verify the proposed method,of which the result shows the effectiveness of the method. |
