基于AGA优化RBF神经网络的矿井通风机故障诊断  

Fault diagnosis of mine ventilator based on AGA optimized RBF neural network

作　　者：余发山[1] 高勇[1] Yu Fashan Gao Yong(School of Electrical Engineering and Automation, Henan Polytechnic University, Jiaozuo 454000,Chin)

机构地区：[1]河南理工大学电气工程与自动化学院,焦作454000

出　　处：《电子测量技术》2017年第9期241-245,共5页Electronic Measurement Technology

摘　　要：针对径向基神经网络(RBF)用于故障诊断时存在收敛速度慢、诊断结果准确率低等问题,提出了一种基于自适应遗传算法(AGA)优化RBF神经网络的矿井通风机故障诊断方法。采用AGA对RBF神经网络的隐含层节点数、隐层基函数的中心和宽度进行优化,以此提高RBF网络的泛化能力。通过大量收集和整理工作形成样本集,使用训练样本训练RBF网络,根据网络输出结果对通风机故障进行诊断。仿真结果表明,相较于RBF神经网络,AGA优化的RBF神经网络收敛速度更快,迭代次数更少,能够有效识别通风机故障类型,诊断结果准确率更高。Aimingat the problem that RBF neural network have a slow convergence speed and the diagnostic accuracy is low when applied to fault diagnosis, a fault diagnosis method of mine ventilator based on Adaptive genetic algorithm optimized RBF neural network is proposed. Using adaptive genetic algorithm to optimize the number of hidden layer nodes, the center and width of hidden layer function, and the generalization ability of network is improved. Through a large number of collection and finishing work to form a sample set,using the training sample set to train the network, make/ault diagnosis of mine ventilator according to the network output results. The simulation reveals that compared with RBF neural network, the RBF neural network optimized by adaptive genetic algorithm has a faster convergence speed and less number of iterations. It can effectively identify the type of failure, and it has a higher accuracy of fault diagnosis.

关 键 词：矿井通风机 故障诊断 自适应遗传算法 神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TN707[自动化与计算机技术—控制科学与工程]