故障树法和改进PSO-PNN网络的电梯故障诊断模型

针对电梯故障问题，提出一种将故障树分析法、改进的粒子群优化算法和概率神经网络相结合的方法用于电梯的故障诊断。以电梯的安全回路系统为例，用故障树法对回路进行分析，获得训练样本与故障类型；使用粒子群算法对概率神经网络的平滑因子进行优化，在优化过程中，针对粒子群算法存在易陷入局部最优的缺陷，提出对惯性权重的改进策略；采用相对误差对诊断效果做出评估，并与传统的概率神经网络和基本粒子群算法优化的概率神经网络在各种故障类型输出和最大相对误差等方面进行比较，结果表明:该模型能够有效诊断电梯故障。

Diagnosis model of elevator fault based on fault tree analysis and improved PSO-PNN network

Aiming at the problem that the elevator fault occur frequently, a new method combining fault tree analysis, improved particle swarm optimization (PSO) and probabilistic neural network (PNN) was proposed to diagnose the elevator fault. Taking the safety circuit system of elevator as example, the circuit was analyzed by the fault tree analysis, and the training samples and the fault types were obtained. The PSO algorithm was used to optimize the smoothing factor of PNN. In the optimization process, the improvement strategy of inertia weight was put forward according to the weakness of PSO which is easy to fall into the local optimum. The relative error was applied to evaluate the diagnosis effect, then the output of various fault types and the maximum relative error of this method were compared with the traditional PNN and the PNN optimized by the basic PSO. The results showed that this model can diagnose the elevator fault effectively.