粒子群优化BP网络在本安参数评定中的应用

Design Application of BP Network in Parameter Assessment of Intrinsic Safety

为了有效地评定本质安全电气参数,建立了火花试验是否点燃与影响火花点燃能力的主要因素(电压U、电流I、电容C、电感L和电阻R)的BP神经网络预测模型。采用粒子群算法优化了BP神经网络的权值和阈值,构建了基于粒子群神经网络的本质安全电气参数的预测模型。所提出的PSO-BP算法解决了一般BP算法迭代速度慢,且易出现局部最优的问题,并以实际电路设计参数为例,进行算法实现。仿真结果表明,该方法可有效地预测本质安全参数。

In order to effectively assess intrinsically safe electrical parameters,established the spark test whether lighting and the main factors influencing the ability to spark ignited（voltage U,current I,capacitance C and inductance L and R）of the BP neural network prediction model. Adopting the particle swarm algorithm to optimize the BP neural network weights and threshold,constructed based on particle swarm of intrinsically safe electrical parameters of neural network prediction model. The proposed PSO and BP algorithm to solve the general BP algorithm iterative speed is slow,and problems of local optimum easily,and in the actual circuit design parameters,for example,algorithm implementation. The simulation results show that this method can effectively predict intrinsically safe parameters.