A performance evaluation method for load control system of thermal power units was put forward

on the basis of multivariable data-driven technology.By adopting the historical data based covariance benchmark indexes and taking covariance matrix of the process output data as the evaluation functions

this method calculates the evaluation function's ratio of the benchmark data and monitoring data

thus the performance evaluation indexes of the control system were obtained.At the same time

through eigenvalue analysis

the calculation of covariance indexes is converted into product of the generalized eigenvalue of the output matrix.Using the statistical inference theory can make the index value calculated by the part of the sample data closer to the general characteristics of the control system.With application of the MATLAB graphical user interface tool-GUI

the simulation interface of the performance assessment of load control system of thermal power-generation units was designed

based on the covariance indexes.Finally

the model simulation data and the industrial field data are used to evaluate the load control system.The results verified the effectiveness and rationality of this method.