Biomimetic Approach to the Development of Nonlinear Filters

将进化计算技术（遗传算法（GA）、进化策略（EC）等）和群体智能（粒子群优化（PSO）、蚁群优化（ACO）、细菌觅食优化（BFO）算法等）等仿生方法与蒙特卡洛滤波、粒子滤波等计算机密集型非线性滤波方法相结合，提出了基于ES的无气味粒子滤波器（ESUP）和具有自适应增益的随机逼近型非线性滤波器。提出了一种有效的多目标粒子群导引选择方法，以获得Pareto最优前沿，可用于多准则下的非线性滤波器设计。利用模糊逻辑和蚁群算法，提出了静态P-V曲线估计的非线性辨识方法。

Integrating the biomimetic approaches such as evolutionary computation techniques (genetic algorithm (GA), evolutionary strategies (EC), etc.) and swarm intelligence (particle swarm optimization (PSO), ant colony optimization (ACO), bacteria foraging optimization (BFO) algorithms, etc.) with the computer-intensive nonlinear filtering methods such as Monte Carlo filters and particle filters, new particle filters including the ES based unscented particle filter (ESUP) and stochastic approximation type nonlinear filter with an adaptive gain were developed.An efficient guide selection method in the multi-objective PSO was proposed to obtain the Pareto optimal front, and it may be applicable to nonlinear filter design under the multi-criteria. Moreover, nonlinear identification methodology was developed for the static P-V curve estimation using Fuzzy logic and ACO.