基于信号时频特征对金属跌落零件的报警方法

THE ALARM METHOD ABOUT FALLING PARTS IN NUCLEAR POWER STATION BASED ON SIGNAL TIME FREQUENCY CHARACTERISTICS

针对核电站一回路系统中检测金属跌落零件误报警率高的问题，提出了基于信号时频域二维特征的神经网络报警方法。用改进的ＢＰ算法对模拟跌落零件试验的结果进行了处理，表明该报警方法是可行的。

On the problem of alarm when parts are falling down in the nuclear power station,the Artificial Neural Network(ANN) method has been firts applied in realizing the alarm of falling parts based on the time frequency characteristics of impact signal.The data from experiments justify the method.