GA-BP算法应用于分布式温度状态实时识别

Real-Time Recognition of Temperature State Based on GA-BP Algorithm

研究地铁车辆火灾温度识别问题,分析地铁车辆车载大功率设备表面温度场及相关特殊位置温度场,温度是表明设备正常运行的一个重要条件,适时获取设备表面的实时温度状态信息有利于安全,但大功率设备的分散性,被测设备温度测点位置选择比较多及测点温度的相关性,同时设备的温度状态信息又受多种因素影响。由于引入了神经网络非线性拟合能力,建立GA-BP算法对任务分解神经网络组采用分布式单片机系统,通过搭建实验平台与仿真软件实现各个大功率设备测点温度状态模式的实时识别,对表面测点与特殊点相关识别及温度加强报警识别。试验结果表明,可以满足识别要求,为应用提供了可靠的安全保证依据。

Deeply study surface and special area temperature field of high power device of subway vehicle.Temperature is the important index related with normal state of device.In order to observe and recognize temperature state,based on no-line fitting ability of neural network,optimal GA-BP algorithm and task decomposition,this paper built a neural network and applied it in distributed embedded MCU system.By constructing a platform of experiment,making use of experts experience and simulation software,real-time recognition of temperature state pattern was realized,and the recognition between surface point and special point and alarming recognition of temperature feature value were nhanced.The experiments show a satisfied result.