| 基于TCN-LSTM神经网络的线缆性能衰退预测方法 |
| |
| 引用本文： | 王发麟,俞威.基于TCN-LSTM神经网络的线缆性能衰退预测方法[J].科学技术与工程,2023,23(35):15109-15116. |
| |
| 作者姓名： | 王发麟 俞威 |
| |
| 作者单位： | 南昌航空大学 |
| |
| 基金项目： | 江西省自然科学基金重点资助项目(20212ACB202005) |
| |
| 摘 要： | 针对复杂机电产品线缆在长时间使用后会出现性能衰退现象，而导致运作出现的安全问题后可以对线缆衰退进行预警预知以预防突发事故，提出了基于时序卷积网络结合长短期记忆网络的复杂机电产品线缆性能衰退预测方法。首先通过对线缆的衰退原因进行分析，依据分析结果选取参数指标来确定数据集，将其中反映线缆性能状态的数据在时序卷积网络下进行时序特征的提取。最后将这些特征数据与经过皮尔逊相关系数分析确定的相关性、影响系数足够高的非时序数据一同通过长短期记忆网络进行训练预测，得到预测结果。通过案例分析以及不同神经网络之间预测的结果对比，验证了构建的方法的预测效果同实际情况贴合程度高。
|
| 关 键 词： | 复杂机电产品 线缆性能衰退 时序卷积网络 长短期记忆网络 皮尔逊相关系数 |
| 收稿时间： | 2023/1/5 0:00:00 |
| 修稿时间： | 2023/9/28 0:00:00 |
Prediction Method Research of Cable Performance Degradation State for Complex Mechatronic Products Based on TCN-LSTM |
| |
| Wang Falin,Yu Wei.Prediction Method Research of Cable Performance Degradation State for Complex Mechatronic Products Based on TCN-LSTM[J].Science Technology and Engineering,2023,23(35):15109-15116. |
| |
| Authors: | Wang Falin Yu Wei |
| |
| Affiliation: | Nanchang Hangkong University |
| |
| Abstract: | For complex mechatronic products cable in a long time after the use of performance degradation will lead to operational problems and product quality and safety, in order to make early warning prediction of cable decline state and prevent unexpected accidents of complex mechatronic products, a cable performance degradation state prediction method based on time-series convolutional fusion long and short-term memory network for complex mechatronic products is proposed. First, by analyzing the causes of cable decline, the data set is determined by selecting parameter indicators based on the analysis results, and the data reflecting the performance state of the cable is extracted under the temporal convolutional network for the temporal features. Finally, these feature data are trained together with non-time-series data with sufficiently high correlation and influence coefficients determined by Pearson correlation coefficient analysis to predict through a long and short-term memory network to obtain prediction results. The case results show and compare that the prediction results fit well with the actual situation, and for the data set containing time series data, the prediction results of the model with the addition of the time-series network part are higher than those of the other individual networks. |
| |
| Keywords: | complex mechatronic products cable performance degradation temporal convolutional network long and short-term memory network Pearson correlation coefficient |
|
| 点击此处可从《科学技术与工程》浏览原始摘要信息 |
|
点击此处可从《科学技术与工程》下载全文 |
