LRAD测量管道内α污染活度的BP网络模拟

Application of BP neural network for LRAD-based alpha contamination monitoring inside pipes

针对风速、流量、探源距、测试管管长和管径等因素对核设施退役中管道内α污染测量造成的非线性影响,采用控制变量法开展长距离α测量(Long range alpha detector,LRAD)模拟装置下的多参数影响实验,初步分析了各种因素对系统测量值的影响特征,建立了以影响因素和测量值为输入、放射源活度为输出的BP神经网络模型,分别对948和100组数据进行了模型建立和实例检验,结果说明,预测平均相对误差为3.4218×10–4,实例平均相对误差为2.217×10–2。应用BP网络模型模拟LRAD装置下的α活度是可行且有效的。

Factors of airspeed,flux,activity,source position,pipe length and pipe diameter affect nonlinearly source activity readout of the Long Range Alpha Detection（LRAD）.In this paper,multiparameter influence experiment is carried out using variable-control method,aiming at studying relationships between the readout and each of the factors.The back propagation（BP） neural network model is established to overcome the nonlinear effects of the factors on the readout,with the readout and the multiparameters being the input,and the source activity being the output.Experiment data of 948 groups are used for BP neural network forecasting,with an average relative error of 3.4218×10–4.And in a 100-group test,an average relative error of 2.217×10–2 is obtained.It shows that with this method source radioactivity in pipes can be simulated.