Research on the operating status prediction of workshop equipment based on digital twin and k-nearest neighbor algorithm
-
摘要: 由于传统车间设备运行状态预测不能有效利用历史数据进行学习,实时响应能力有限,难以在复杂调度环境中取得良好效果,因此文章提出一种数字孪生与k-近邻算法相结合的车间设备运行状态预测模型。构建车间设备实体在信息空间的数字孪生模型,并建立设备实体与模型之间的映射关系,从而获取实时特征数据,即设备的运行状态特征数据。运用k-近邻算法计算实时特征数据与历史数据之间的欧几里得距离,即计算设备当前运行状态与历史已知状态的相似度,最终通过前k个距离所对应的设备历史运行状态数据,预测设备的当前运行状态。该模型的本质是通过数字孪生的实时数据采集,获取指定设备运行状态特征数据,运用k-近邻算法预测设备的实时运行状态。相较以往研究,本研究贡献在于提高设备实时运行状态预测的准确率。如果将数字孪生、k-近邻算法与具备自我学习能力的相关算法相结合,模型的预测效果会更好。
-
关键词:
- k-近邻算法 /
- 机器学习 /
- 数字孪生 /
- 车间设备运行状态预测
Abstract: Due to the inability of traditional workshop equipment operating status prediction to effectively utilize historical data for learning, limited real-time response ability, and difficulty in achieving good results in complex scheduling environments, a workshop equipment operating status prediction model combining digital twin and k-nearest neighbor algorithm is proposed. A digital twin model of workshop equipment entities in the information space is constructed, and then, a mapping relationship between equipment entities and models is also established, in order to obtain real-time feature data, namely equipment operating status feature data. The k-nearest neighbor algorithm is used to calculate the euclidean distance between the real-time feature data and the historical data, that is, to calculate the similarity between the current operating status of the equipment and the known historical status, and finally, based on the historical operating status data of the equipment corresponding to the first k distances, the current operating status of the equipment is predicted. The essence of this model is to collect real-time data from digital twin, obtain characteristic data of designated equipment operating status, and use k-nearest neighbor algorithm to predict the real-time operating status of equipment. Compared to previous studies, the contribution of this study is to improve the accuracy of equipment real-time operating status prediction.If digital twin and k-nearest neighbor algorithm are combined with relevant algorithms with self-learning ability, the predictive performance of the model will be better. -
表 1 机械分析数据库中设备故障模式
故障模式代码 故障模式分类 1 连接件故障 2 轴承失效 3 机械松动 4 基础变形 5 不平衡 6 无故障/正常运行 表 2 部分初始数据
振动速度/
(mm/h)加速度/
(g/s)位移/
(mm/h)震动频率/
(Hz/s)温度/
℃状态
标签40920 8.326976 0.953952 3.225536 33.6 6 14488 7.153469 1.673904 4.054865 37.8 6 26052 1.441871 0.805124 0 56.2 2 75136 13.147394 0.428964 3.336213 60.8 1 38344 1.669788 0.134296 5.632872 70.0 1 72993 10.141740 1.032955 0 5.6 3 35948 6.830792 1.213192 2.758132 36.5 6 42666 13.276369 0.543880 2.774165 33.4 6 67497 8.631577 0.749278 0 10.9 1 35483 12.273169 1.508053 3.656185 40.7 6 表 3 归一化处理后的部分数据
振动速度/
(mm/h)加速度/
(g/s)位移/
(mm/h)震动频率/
(Hz/s)温度/
℃0.513766 0.170320 0.262181 0.126801 0.087159 0.089599 0.154426 0.785277 0.159402 0.098054 0.611167 0.172689 0.915245 0 0.145785 0.012578 0.000000 0.195477 0.131151 0.157717 0.110241 0.187926 0.287082 0.221437 0.181582 0.812113 0.705201 0.681085 0 0.014526 0.729712 0.490545 0.960202 0.108426 0.094682 0.130301 0.133239 0.926158 0.109056 0.086641 0.557755 0.722409 0.780811 0 0.028275 0.437051 0.247835 0.131156 0.143729 0.105577 表 4 模型预测部分结果及准确率
振动速度/
(mm/h)加速度/
(g/s)位移/
(mm/h)震动频率/
(Hz/s)温度/
℃状态
标签预测
结果0.513766 0.170320 0.262181 0.126801 0.087159 6 6 0.089599 0.154426 0.785277 0.159402 0.098054 6 6 0.611167 0.172689 0.915245 0 0.145785 2 2 0.012578 0.000000 0.195477 0.131151 0.157717 1 1 0.110241 0.187926 0.287082 0.221437 0.181582 1 1 0.812113 0.705201 0.681085 0 0.014526 3 3 0.729712 0.490545 0.960202 0.108426 0.094682 6 6 0.130301 0.133239 0.926158 0.109056 0.086641 6 6 0.557755 0.722409 0.780811 0 0.028275 1 1 0.437051 0.247835 0.131156 0.143729 0.105577 6 6 ······(100 rows × 5 columns)
模型预测准确率为0.96 -
[1] 陶飞,张辰源,张贺,等. 未来装备探索:数字孪生装备[J]. 计算机集成制造系统,2022,28(1):1-16. [2] 魏一雄,郭磊,陈亮希,等. 基于实时数据驱动的数字孪生车间研究及实现[J]. 计算机集成制造系统,2021,27(2):352-363. doi: 10.13196/j.cims.2021.02.004 [3] 丁华,杨亮亮,杨兆建,等. 数字孪生与深度学习融合驱动的采煤机健康状态预测[J]. 中国机械工程,2020,31(7):815-823. doi: 10.3969/j.issn.1004-132X.2020.07.007 [4] 刘义,刘晓冬,焦曼,等. 基于数字孪生的智能车间管控[J]. 制造业自动化,2020,42(7):148-152. doi: 10.3969/j.issn.1009-0134.2020.07.035 [5] 李宝帅,叶春明. 深度强化学习算法求解作业车间调度问题[J]. 计算机工程与应用,2021,57(23):248-254. doi: 10.3778/j.issn.1002-8331.2105-0299 [6] 肖鹏飞,张超勇,孟磊磊,等. 基于深度强化学习的非置换流水车间调度问题[J]. 计算机集成制造系统,2021,27(1):192-205. doi: 10.13196/j.cims.2021.01.018 [7] Latif H ,Starly B . A simulation algorithm of a digital twin for manual assembly process[J]. Procedia Manufacturing,2020,48:932-939. [8] Li J Z,Pang D ,Zheng Y ,et al. A flexible manufacturing assembly system with deep reinforcement learning[J]. Control Engineering Practice,2022,118:1-8. [9] 姜万录,赵岩,李振宝,等. 多模型Stacking集成学习的旋转机械故障诊断方法[J]. 液压与气动,2023,47(4):46-58. [10] Xiong J B ,Zhang Q H ,Peng Z P ,et al. A diagnosis method for rotation machinery faults based on dimensionless indexes combined with k-nearest neighbor algorithm[J]. Mathematical Problems in Engineering,2015,2015(10):1-9. [11] 王正家,刘鸣,何嘉奇,等. 基于2DSVD特征提取的工业设备故障诊断[J]. 组合机床与自动化加工技术,2021(12):49-52,57. doi: 10.13462/j.cnki.mmtamt.2021.12.012 [12] 李长伟,雷文平,董辛旻,等. GA-CFS结合案例推理的轴承故障诊断[J]. 机械设计与制造,2023(1):26-29. [13] Elshenawy L M ,Mahmoud T A ,Chakour C . Fault detection and diagnosis strategy based on k-nearest neighbors and fuzzy C-means clustering algorithm for industrial processes[J]. Journal of the Franklin Institute,2022,359(13):7115-7139. [14] Liu Z F,Zhang Y Z,Zhang C X ,et al. Real-time workshop digital twin scheduling platform for discrete manufacturing[J]. Journal of Physics:Conference Series,2021,1884(1):1-6. [15] Maschler B ,Braun D ,Jazdi N ,et al. Transfer learning as an enabler of the intelligent digital twin[J]. Procedia CIRP,2021,100:127-132. [16] Choi S H ,Park K B ,Dong H R ,et al. An integrated mixed reality system for safety-aware human-robot collaboration using deep learning and digital twin generation[J]. Robotics and Computer-Integrated Manufacturing,2022,73:1-19. [17] Xia M ,Shao H ,Williams D ,et al. Intelligent fault diagnosis of machinery using digital twin-assisted deep transfer learning[J]. Reliability Engineering & System Safety,2021,215:102258. [18] Lee S M ,Park S Y ,Choi B H . Application of domain-adaptive convolutional variational autoencoder for stress-state prediction[J]. Knowledge-based Systems, 2022,248:1-12. [19] 冯继伟,孙开畅. 基于因子分解机的水利工程事件风险状态预测[J]. 水利水电技术(中英文),2021,52(12):178-184. [20] Jiang L,Zhao T S,Feng C X,et al. Improvement of random forest by multiple imputation applied to tower crane accident prediction with missing data[J]. Engineering Construction & Architectural Management,2023,30(3):1222-1242. [21] 王旭,钟炜. 一种污水处理厂设备预测及辅助维护智能平台[J]. 中国给水排水,2023,39(10):121-125. [22] Amzile K ,Habachi M . Assessment of support vector machine performance for default prediction and credit rating[J]. Banks and Bank Systems,2022,17(1):161-175. [23] 郭亮,李长根,高宏力,等. 大数据背景下基于特征学习的机械设备剩余寿命预测[J]. 西南交通大学学报,2021,56(4):730-735,768. [24] 何淑波,项薇,石钟淼. 基于机器学习的电动汽车电池系统的风险预警[J]. 中国安全科学学报,2023,33(2):159-165.