摘要
线性直流电源是电子设备中最易发生故障的薄弱环节之一,其不完善的故障模型和容差等问题使故障诊断变得复杂。神经网络的自组织学习能力为故障诊断问题提供了一种新的解决途径,反向传播算法是神经网络中应用广泛的一种多层前馈神经网络模型。但算法有求解精度低、搜索速度慢、易于陷入局部极小值的缺点。蚁群算法是一种新型的模拟进化算法,有正反馈、分布式计算、启发性收敛等特性。本文将基于蚁群算法的神经网络的方法应用于线性直流电源的故障诊断中,仿真实验表明:此方法提高了网络的训练效率和故障定位的准确性。
Linear DC electric source circuit is one of the weak parts that breaks down most easily in electronic equipment. Its incomplete fault model and tolerance problem make the fault diagnosis more complicated. The self-organized learning ability of the neural network provides a new way to solve the problem. Back propagation algorithm is a model of feedback neural networks widely used in many areas, but it has some shortcomings, such as low-precision solution, slow search speed and easy convergence to the local minimum points. Ant colony system is a novel simulated evolutionary algorithm. Ant system has the advantages such as positive feedback, distributed computation and using a constructive greedy heuristic. A method for fault diagnosis of linear DC electric source circuits based on ant colony algorithm and neural technology is introduced. Simulation experiment shows that the new method reduces the dimension of input to the neural network, raises the training efficiency and improves the fault classification accuracy.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2009年第3期515-520,共6页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(50677069)资助项目
关键词
蚁群算法
神经网络
线性直流电源
故障诊断
ant colony algorithm
neural network
linear DC electric source
fault diagnosis