Algorithm for selecting data analysis model for predicting the performance of industrial robots

Due to the intensive pace of development of systems for data collection, accumulation and analysis, more and more methods, approaches and systems are being created for decision-making in the field of predictive maintenance in modern robotic industries in order to increase productivity and efficiency of resource use (time, finances and material resources). Maintaining fixed assets of production is crucial to ensuring safe, efficient and continuous production. Modern equipment is fitted with a variety of monitoring systems, self-diagnosis and intelligent sensors that allow collecting a significant amount of primary data that may contain useful knowledge. The article presents an approach to developing an algorithm for selecting machine learning models when analyzing data on the performance of industrial manipulators as part of the predictive maintenance process. The developed algorithm makes it possible to reduce the time spent on training data analysis models (including machine learning and artificial neural networks) by selecting arrays of data collected from a fleet of equipment (for example, industrial robots) that have the greatest degree of similarity relative to the data collected from single equipment; this helps to avoid training additional data analysis models with satisfactory test results. Data was collected from four different industrial robots. The following methods were used for the analysis: linear model, convolutional neural network, multilayer perceptron. The algorithm of dynamic transformation of the timeline was used to assess the degree of similarity.

1. Mohammadi N., Taylor J. Knowledge discovery in smart city digital twins. Proceedings of the 53rd Hawaii International Conference on System Sciences. 2020. DOI: 10.24251/HICSS.2020.204.

2. Bodo R., Bertocco M., Bianchi A. Fault classification driven by maintenance management for smart maintenance applications. 2020 IEEE International Workshop on Metrology for Industry 4.0 & IoT. 2020;27–32. DOI: 10.1109/MetroInd4.0IoT48571.2020.9138294.

3. Ortiz G., Caravaca J.A., Garcia-de-Prado A. et al. Real-time context-aware microservice architecture for predictive analytics and smart decision-making. IEEE Access. 2019;7:183177–183194. DOI: 10.1109/ACCESS.2019.2960516.

4. Silvestrin L.P., Hoogendoorn M., Koole G. A comparative study of state-of-the-art machine learning algorithms for predictive maintenance. 2019 IEEE Symposium Series on Computational Intelligence (SSCI). 2019;760–767. DOI: 10.1109/SSCI44817.2019.9003044.

5. Raja H.A., Asad B., Vaimann T. et al. Custom simplified machine learning algorithms for fault diagnosis in electrical machines. 2022 International Conference on Diagnostics in Electrical Engineering (Diagnostika). 2022;1–4. DOI: 10.1109/Diagnostika55131.2022.9905174.

6. Schelter S., Biesmann F., Januschowski T. et al. On challenges in machine learning model management. IEEE Data Engineering Bulletin. 2015.

7. Ashmore R., Calinescu R., Paterson C. Assuring the machine learning lifecycle: Desiderata, methods, and challenges. ACM Computing Surveys (CSUR). 2021;54(5):1–39. DOI: 10.1145/3453444.

8. Naing Y.T., Raheem M., Batcha N.K. Feature selection for customer churn prediction: a review on the methods & techniques applied in the Telecom industry. 2202 IEEE International Conference on Distributed Computing and Electrical Circuits and Electronics (ICDCECE). 2022;1–5. DOI: 10.1109/ICDCECE53908.2022.9793315.

9. Dsouza J., Velan S. Preventive maintenance for fault detection in transfer nodes using machine learning. 2019 International conference on computational intelligence and knowledge economy (ICCIKE). 2019;401–404. DOI: 10.1109/ICCIKE47802.2019.9004230.

10. Singh P., Agrawal S., Chakraborty A. Multi-Classifier Predictive Maintenance Strategy for a Manufacturing Plant. 2021 International Conference on Maintenance and Intelligent Asset Management (ICMIAM). 2021;1–4. DOI: 10.1109/ICMIAM54662.2021.9715224.

11. Chianese R., Cicala L., Angelino C.V. et al. A Risk and Priority Model for Cost-Benefit Analysis and Work Scheduling within Predictive Maintenance Scenarios. 2021 26th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA). 2021;1–4.

12. Alimova I.S., Solovev V.D., Batyrshin I.Z. Comparative analysis of similarity measures based on the transformation of sliding approximations in time series classification problems. Proceedings of the Institute of System Programming of the Russian Academy of Sciences. 2016;28(6):207–222. DOI: 10.15514/ISPRAS-2016-28(6)-15. (In Russ.).