采用多目标粒子群算法的模拟电路故障诊断研究

Analog Circuit Fault Diagnosis with Multi-Objective Particle Swarm Optimization

提出了一种容差条件下基于多目标粒子群(MOPSO)算法的模拟电路软故障诊断方法.通过灵敏度分析,建立模拟电路故障诊断的约束线性规划方程组,以元件参数变化量与标称值的百分比作为故障判据.针对MOPSO中目标空间增加时种群选择压力影响算法性能的问题,采用阶有效优化准则代替传统的Pareto优化准则,引入最优折中解作为全局最优解,从而提出基于阶有效的平衡全局搜索策略多目标粒子群(ESEO-MOPSO)算法,并将其用于模拟电路故障诊断的约束线性规划方程组的求解中.仿真结果表明,该方法兼顾故障元件的定位和故障元件参数变化量的估计,可以有效地实现模拟电路在容差条件下的软故障定量诊断.

Taking tolerance as the parameter, a method based on multi-objective particle swarm optimization （MOPSO） for soft fault diagnosis of analog circuit is proposed. The constraint linear programming equation is constructed according to the sensitivity analysis of node-voltage. The percentage of the parameter deviation against the nominal value is considered as the diagnosis criterion. Aiming at the selective pressure in the MOPSO caused by the number of the increased objectives, the preference order is chosen instead of the traditional Pareto optimum. Then ＂the best compromise＂ is introduced as a global best to update the particle velocity. Thus an equilibrium selection of global search following the effective ordering （ESEO-MOPSO） is employed to divide the node-voltage incremental equations. The simulation illustrates that the proposed method enables to locate the faulty element and estimate the parameter deviation effectively.