通过试验建立了锆合金管材超声检测参数的优化模型, 模型选取三层RBP网络, 其中输入层为转速、送进速度和重复频率, 输出层为人工缺陷重复检测的标准偏差。构建的网络为2-8-1, 网络测试误差(均方差)为0.004。通过检测参数优化模型, 确定出9 mm×0.6 mm(直径×壁厚)的核用Zr-4管材超声检测的最佳参数为管材送进速度7 m·min-1, 探头旋转速度5 100 rpm, 仪器重复频率15 kHz。将模型确定的检测工艺参数应用于生产检测中, 结果表明, 利用参数优化模型为优化超声检测工艺参数提供了一种新方法。

Optimization model of zirconium alloy tube ultrasonic testing parameter is established in this paper and the model employs a three layer RBF network. The input layer consists of rotation speed, feed speed and repetition frequency, whereas the output layer is of the standard deviation of the repeat test for artificial defect. The constructed network is of 2-8-1 type and the results of RBF module show that the mean square error is 0.004. The model is optimized by ultrasonic testing parameter, and it is determined that the optimal ultrasonic testing parameters for 9 mm×0.6 mm nuclear Zr-4 tube are 7 m·min-1 as the feed speed, 5 100 rpm as the rotation speed, 15 kHz as the repetition frequency. The result shows that the parameter optimization model is the new way to optimize the ultrasonic testing parameter.