The Research Of Multivariable System Control Performance Assessment Based On Generalized Minimum Variance

There are many control loops in the modern process industry. Thechanging of actual working conditions, nolinear process or equipmentaging could reduce control performance greatly. So, evaluating systemperformance accurately has a great significance for equipment real-timemonitoring, reduction of energy consumption, production efficiencyimprovement and so on. Recently, many scholars focus on the field ofcontrol performance assessment, but a serious of difficulties such asmodeling limit its wide application in the industry process. A method ofapplying diagonal interactor to obtain system performance index wasproposed in this paper, which based on several traditional assessmentmethods. The main research results and innovation are described asfollows:1. The history and current research status about the development ofmultivariate performance assessment is analyzed and compared in thispaper, and then summarized industrial production, analyzed existing andpotential problems in theory and practical application. 2. Selection of method plays an important role to the complexity ofthe calculation and accuracy of performance index when one assessesmultivariable system. Unfortunately there is no effective way to obtainaccurate performance index using simple method. This paper proposed amethod of diagonal interactors which is applied to generalized minimumvariance system to obtain performance index. The proposed approachcould obtain accurate assessment variance and optimal control modelthrough a little priori knowledge such as disturbance model, time delaymatrix and so on; it also avoided system errors generated in the process ofidentifying interactors.3. Multivariable system performance with linear time varyingdisturbance has been researched in this paper. A new method ofmultivariable performance assessment is proposed through dividing timevarying disturbance into three types of constant disturbance. A weightedperformance index is constructed and the weight matrix corresponding todifferent disturbance types and priority. Also, this method allows usersspecify reasonable output variance and it could obtain generalizedmultivariable system closed-loop output variance upper and lower boundthrough the calculation of diagonal interactors. Minimum variancecontrol parameters are ensured and achieved optical control model isproposed according to ITAE optimizing. The simulation results prove theavailability of the proposed approach. Through these researches, calculations are reduced in the field ofcontrol performance assessment technology which is more convenient inindustrial applications. What's more, this technology expanded to linertime-varying multivariable system and the proposed output variance limitmethod is more practical.