Research On Performance Assessment Algorithm Of Process Industry Control System

In the process of using industrial control system, with the passage of time, theequipment aging and many other reasons to cause a decline in controller performance,resulting in lower production efficiency, and higher production cost. The report showsthat about60%of the industrial controllers have control performance problems in theoperation. On the other hand, Along with the widely used DCS control system inprocess industries, the acquisition and storage of operation data is easier.Corresponding controller performance indicators can be obtained by using controlperformance assessment theory with these data, providing more information supportfor the process industry, so this theory is of great importance in the process industry.Based on the research of domestic and international numerous controlperformance assessment algorithm, aiming at the shortcomings of the current controlperformance assessment theory, using the assessment method based on the data ofsingle variable and makes an analysis of the multivariable control system, the mainresearch content is as follows:In view of the single variable control system performance assessment method, theminimum variance control algorithm is studied, and is verified on the single watertank model, the results can reflect the performance of the controller. Based on thetraditional algorithm, using the linear regression algorithm is used to estimate systemminimum variance. An optimize linear regression algorithm with forgetting factor isproposed to make the estimation result fits the actual performance, and is verified onthe mathematical model, the result is closer to the original minimum variancebenchmark. MATLAB/GUI is used for developing related software, the performanceassessment on the data from the fractionating tower, the performance indicators of thecontrol system can be obtained, and the control system performance can bemonitored.For the performance assessment method of multivariable control system,multivariable minimum variance algorithm is studied firstly, this algorithm needs touse system identification model, priori information and the calculation of relevantcorrelation matrix, so the recognition have identification problems when it is appliedto the process industry control system. Subspace identification method is proposed toapply in controller performance assessment, sandwich matrix can be obtained byusing subspace transformation, the minimum variance benchmark can be estimated byusing the subspace matrix, the identification of system model can be avoided. On thisbasis, the estimation method of LQG benchmark scheme using the subspace matrixalgorithm is proposed, it can be applied to more control systems. The assessmentresults of algorithm is analyzed, and the effectiveness of the algorithm is verified bythe results of corresponding digital simulation.