蝙蝠算法优化模糊神经网络的25Hz相敏轨道电路故障诊断研究

Fault Diagnosis Research for 25 Hz Phase Sensitive Track Circuit Based on Bat Algorithm to Optimize Fuzzy Neural Network

针对传统25Hz相敏轨道电路故障诊断网络求解时收敛速度慢、诊断精度不高的问题,提出智能蝙蝠算法(BA)与模糊逻辑理论(FS)、神经网络(NN)相融合的BA-FNN模型,优化网络参数,对25Hz相敏轨道电路进行故障诊断。考虑到轨道电路特征参数的不确定性、模糊性,运用模糊逻辑理论对轨道电路特征参数进行模糊化预处理;为了克服传统的BP算法存在易陷入局部极小点和速度收敛慢等问题,引入蝙蝠算法,模拟蝙蝠的飞行过程对模糊BP网络的相关参数进行优化;建立最优BA-FNN模型用于网络诊断。仿真结果表明,相对于GA-FNN、PSO-FNN比较算法模型,BA-FNN模型不仅学习训练次数明显减少,训练时间缩短,且寻优精度较高,可有效提高25Hz相敏轨道电路故障诊断的精度。

Aiming at slow convergence speed and low diagnostic accuracy in the traditional fault diagnosis of the 25 Hz phase sensitive track circuit,this paper proposed a BA-FNN model which combines intelligent bat algorithm(BA)with fuzzy system(FS)and neural network(NN)to optimize fuzzy neural network(FNN)parameters and diagnose the fault of 25 Hz phase sensitive track circuit.First of all,considering the uncertainty and fuzziness of the characteristic parameters of the track circuit,the fuzzy logic theory was used to conduct fuzzy preprocessing of the characteristic parameters of the track circuit.In order to overcome the slow convergence speed and local minimum point of faults of the traditional BP algorithm,the bat algorithm was introduced to simulate the flight process of bats,to optimize the parameters of the fuzzy BP network,and to establish the optimal BA-FNN model for network diagnosis.The simulation results show that compared with GA-FNN and PSO-FNN comparison algorithm models,the BA-FNN not only reduces the number of,training times,shortens the training time,but also has higher optimization accuracy.It can effectively improve the 25 Hz phase sensitive diagnostic accuracy of track circuit fault diagnosis.