Improved Renyi Entropy Benchmark for Performance Assessment of Common Cascade Control System

Abstract

To deal with the inconsistency of the minimum variance (MV) benchmark in evaluating non-Gaussian disturbance systems, this paper proposed a new benchmark, which combined entropy with output mean value. For a cascade control system, the new benchmark was constructed by analyzing the weakness of the MV benchmark and the pure Renyi entropy benchmark. In order to estimate the more accurate performance of the unknown system, an improved estimation of distribution algorithm based on entropy criterion is given. It can identify the disturbance distribution and calculate the new index evaluation value. Finally, different disturbance distributions were used to verify the consistency of the new index. The experimental results show that the proposed index and algorithms are consistent and effective in evaluating the performance of the unknown systems with non-Gaussian disturbance.