基于BFOA-PSO-GMM的轨道电路故障诊断研究

Research on Fault Diagnosis of Track Circuit Based on BFOA-PSO-GMM

针对轨道电路系统庞大、故障种类繁多等问题,提出一种融合细菌觅食优化算法和粒子群优化算法的高斯混合模型,对轨道电路的多种故障类型进行诊断。该模型通过融合细菌觅食优化算法与粒子群优化算法,找寻适合EM算法的初始值,利用合适的初始值有效避免EM算法陷入局部最优,提高模型的故障诊断能力。通过对实测数据的训练和测试实验表明,本模型比传统高斯混合模型的故障诊断准确率提高了31.85%,比采用粒子群优化算法改进模型的故障诊断准确率提高了9.4%,即本模型对轨道电路的故障诊断更加有效。

In view of the huge track circuit system and various types of faults,this paper proposed a Gaussian mixture model that combines bacterial foraging optimization algorithm and particle swarm optimization algorithm to diagnose multiple failure types of the track circuit.By fusing bacterial foraging optimization algorithm and particle swarm optimization algorithm to find the initial value suitable for the EM algorithm,the model effectively avoided the EM algorithm from falling into local optima,resulting in the improvement of its fault diagnosis ability.Through the training and testing experiments on the measured data,it is shown that the fault diagnosis accuracy of the model is 31.85%higher than that of the original Gaussian mixed model,and 9.4%higher than the fault diagnosis accuracy of the improved model using the particle swarm optimization algorithm,proving that the model is more effective in fault diagnosis of track circuits.