Hierarchical stochastic gradient identification for non-uniformly sampling hammerstein systems with colored noise

摘要

The system inputs and/or outputs are interference with various noise, such as the data sampler errors, environment changing, some manipulating and so on. Many conventional algorithms can estimate system parameters, most of them assume that the system with single-input single-output, but in process industries, multiple-input multiple-output (MIMO) systems exist widely. A hierarchical multi-innovation stochastic gradient identification algorithm is proposed for non-uniformly sampling Hammerstein systems with colored noise. The corresponding state space models of Hammerstein systems are derived by using the lifting technique. Based on the hierarchical identification principle, the Hammerstein system with colored noise is decomposed into two subsystems firstly. Then the parameters are identified by using the multi-innovation based stochastic gradient algorithm with forgetting factors. Simulation results demonstrate the effectiveness of proposed algorithm.