摘要
非线性系统广泛存在于工业过程,目前非线性系统的描述方法较多、较分散,其待辨识的模型特性获取也较复杂。针对单输入单输出Hammerstein动态模型,研究了一种可以统一描述死区、饱和、迟滞、加载等4种典型不连续非线性特性的结构,通过选择不同的结构参数可以得到9种非线性特性,结合关键变量分离原则对这些特性进行处理。应用基本粒子群优化算法和一种改进的粒子群优化算法对含有统一非线性特性的Hammerstein ARMA模型进行辨识,通过仿真得到包含9种不同非线性特性的Hammerstein模型的参数估计,其结果表明统一非线性结构的描述及所应用的辨识方法具有良好的效果,说明了其有效性。
Nonlinear systems widely exist in industry process. However, the descriptions are various and decentralized. The characteristics for identification are difficult to obtain. For a single-input and single-output nonlinear dynamic Hammerstein model, a uniform description was introduced, which could describe four typical discontinuous nonlinear characteristics such as dead-zone, saturation, hysteresis and preload. Through different parameters choice, nine different nonlinearities could be obtained. Key-term separation principle was used to process the characteristics. The basic Particle Swarm Optimization （PSO） and a modified PSO method were applied to identify the Hammerstein ARMA model containing the unified nonlinear characteristics. Though simulation, parameters of this Hammerstein model with 9 different nonlinear characteristics were estimated. The simulation results show the unified nonlinear characteristics description and the identification method applied are effective.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2014年第12期2887-2891,共5页
Journal of System Simulation
基金
国家自然科学基金(61273132)
