基于BP神经网络的钢丝绳断丝损伤定量检测系统的设计
Design of Quantitative Detection System of Broken Wire for Steel Rope Based on BP Neural Network
-
摘要: 针对传统的钢丝绳断丝损伤定量检测系统检测精度不高的问题,提出了一种基于BP神经网络的钢丝绳断丝损伤定量检测系统的设计方案。该系统由漏磁检测与处理电路获取钢丝绳损伤信号,由光码盘控制单片机对损伤信号进行等空间采样,经单片机处理后的损伤信号再上传至工控机,由工控机调用Matlab软件进行BP神经网络的训练,得到权重矩阵和阈值矩阵,然后由单片机程序进行BP神经网络的前向计算,从而实现钢丝绳断丝损伤的判定。检测结果表明,该系统对钢丝绳断丝损伤的识别率达到了86.9%,具有一定的实用性。Abstract: In view of the problem of low detection precision existed in traditional quantitative detection system of broken wire for steel rope, the paper proposed a design scheme of quantitative detection system of broken wire for steel rope based on BP neural network. The system uses flux leakage detection and processing circuit to get damage signal of steel rope, and the signal is sampled with equal-space by single-chip microcomputer controlled by encoder. Then the signal is uploaded to industrial computer, and the industrial computer calls Matlab software to train it with BP neural network to get weight matrix and thre shold matrix, then a forward calculation of BP neural network is made by single-chip microcomputer program, so as to realize judgment of broken wire of steel rope. The detection result showed that the recognition rate of broken wire of steel rope detected by the system is 86.9%, which has certain practicality.