在智能制造的浪潮下，工业机器人作为实现智能制造的重要设备，其应用市场需求将持续增加。随着对工业机器人的依赖性增加，工业机器人的状态与生产质量和效益息息相关，因此非常有必要对工业机器人进行监控维护以保持高性能。然而，工业机器人结构精密，人工维护成本高，效率低，因此催生了对机器人进行远程监控的迫切需求。目前机器人监控研究现状中存在物理世界和信息世界相互割裂、单向连接的问题。而数字孪生的出现，能够为解决上述问题提供思路。因此，本文结合数字孪生技术，建立了一个集成模拟、状态分析、控制的工业机器人远程监控系统。主要研究工作如下：1. 理论层面——设计了工业机器人远程监控的数字孪生系统模型架构。本文在数字孪生五维模型的基础上，提出了一种适用于工业机器人远程监控的数字孪生模型，为建立基于数字孪生的工业机器人远程监控平台实现提供架构支撑。2. 技术层面——研究与实现了数据采集与传输系统，为实现数字孪生提供数据基础。本文以现有的工业机器人上料工位为案例，分别依次从数据采集、数据传输两个方面阐述了实现数据采集与传输系统的实现过程。3. 技术层面——提出了关于工业机器人的数字孪生虚拟模型建模方法，利用真实数据驱动仿真实现机器人生产状态可视化监控。结合数字孪生虚拟模型的三大要求：监控的完整性、模型的超写实性以及仿真的流畅性，本文提出了虚拟模型建模方法，最后用真实数据驱动实时仿真。4. 技术层面——提出了一种基于无监督的工业机器人异常检测方法，作为数字孪生服务核心，从而实现机器人的设备实时状态监测。本文结合机器人数据的特点，基于变分自编码器进行改进，提出了适用于工业机器人异常检测的模型（SWCVAE），最后通过现有机器人工作站对模型进行测试，结果表明模型是可行的。5. 综合以上核心技术功能，设计与实现了基于数字孪生的工业机器人远程监控平台。本文将以上核心功能整合封装，并建立友好界面与用户进行交互。经过项目的实际测试，最后该系统能够同时实现对机器人的作业状态监视和机器人自身状态监测，又可以实现在必要情况下远程控制工业机器人，为学术界和工业界研究和实施工业机器人远程监控系统提供参考。
Under the wave of intelligent manufacturing, industrial robots, as important devices for intelligent manufacturing, will continue to increase their application market demand. As the dependence on industrial robots increases, the status of industrial robots is closely related to production quality and efficiency. Therefore, it is very necessary to monitor and maintain industrial robots to maintain high performance. However, industrial robots have sophisticated structures, high manual maintenance costs, and low efficiency, which has created an urgent need for remote monitoring of robots. At present, the current situation of robot monitoring research has the problem that the physical world and the information world are separated from each other and connected in one direction. The emergence of digital twins can provide ideas for solving the above problems. Therefore, this article combines digital twin technology to establish an industrial robot remote monitoring system that integrates simulation, status analysis, and control. The main research work is as follows:1. Theoretical level-Design the digital twin system model architecture for remote monitoring of industrial robots. Based on the five-dimensional digital twin model, this paper proposes a digital twin model suitable for remote monitoring of industrial robots, which provides architectural support for the establishment of a digital twin-based remote monitoring platform for industrial robots.2. Technical level-research and implement a data acquisition and transmission system to provide a data foundation for the realization of digital twins. This article takes the existing industrial robot loading station as a case, and elaborates the realization process of the data collection and transmission system from two aspects: data collection and data transmission.3. Technical level-a digital twin virtual model modeling method for industrial robots is proposed, and real data-driven simulation is used to achieve visual monitoring of robot production status. Combining the three requirements of the digital twin virtual model: the integrity of monitoring, the super-realism of the model and the fluency of simulation, this paper proposes a virtual model modeling method, and finally uses real data to drive real-time simulation.4. Technical level-An unsupervised industrial robot anomaly detection method is proposed as the core of the digital twin service, so as to realize the real-time status monitoring of the robot's equipment. Based on the characteristics of robot data, this paper proposes a model suitable for industrial robot anomaly detection (SWCVAE) based on the improvement of the variational autoencoder. Finally, the model is tested through the existing robot workstation, and the results show that the model is feasible.5. Based on the above core technical functions, a remote monitoring platform for industrial robots based on digital twins is designed and implemented. This article integrates and encapsulates the above core functions, and establishes a friendly interface to interact with users. After the actual test of the project, finally the system can realize the monitoring of the robot's working status and the robot's own status at the same time, and can realize the remote control of industrial robots if necessary, and research and implement the industrial robot remote monitoring system for academia and industry. for reference.
