高斯和声粗糙集BNN光纤管道泄漏监测

Gauss improved harmony search based rough-set block-neural-network for optical fiber pipeline leakage monitoring

为解决长距离管道泄漏监测中干扰多,传统识别算法精度不高的问题,提出一种基于高斯改进和声搜索算法的粗糙集块神经网络。提出一种高斯改进和声搜索算法,利用高斯分布特性对其即兴创作过程进行改进,给出理论分析,保证改进算法的收敛性;利用粗糙集进行采集信号的特征提取预处理,使之适合神经网络处理,简化数据计算量;对粗糙集参数和块神经网络(block neural network,BNN)参数进行编码,利用改进和声搜索算法和实测数据对监控模型进行处理。仿真结果表明,该监控模型可有效对泄漏信号进行识别,精度高,满足实际需要。

To solve the problem of leakage interference monitoring in long distance pipeline,and the low accuracy problem of traditional recognition algorithm,an rough blocked neural network based on Gauss improved harmony search algorithm was proposed.A kind of Gauss improvement harmony search algorithm was designed,in which the Gauss distribution characteristics were used on the improvisation process for improvement,and its theoretical analysis was given,the convergence of the improved algorithm was then ensured.The signal was pre-processed using rough set,making it suitable for neural network processing,and the data computation was simplified.The rough set parameters and block neural network parameters were encoded,and the improved harmony search algorithm and the measured data were used for processing the monitoring model.The simulation results show that,this monitoring model can effectively identify the leak signal with high precision,which meets the actual needs.