A Surrogate Model-Based Method for Multipurpose Overload Robot's Healthy Status Prediction

Due to the extreme environment and large loads, the key components of the multipurpose overload robot sustained greater damage while performing maintenance tasks of the vacuum in the China Fusion Engineering Test Reactor (CFETR). It is necessary to monitor and predict the health status of these key compositions, avoiding safety incidents and economic losses. However, the lack of historical data poses a challenge in building performance evaluation models, while this simulation also requires a lot of computing power and time due to the system's complexity. A real-time high-fidelity monitoring and prediction methods are urgently required to meet the requirements of safe maintenance. Therefore, an effective surrogate model-based structure health status prediction method is proposed. At first, the Kriging model is built to represent the relationship between loads and stress of the CFETR multipurpose overload robot (CMOR)'s joint. Then, the rainflow counting method and S-N curve are leveraged to evaluate the cumulative damage according to the above stress spectrum. Finally, the joint 4# of CMOR is taken as an example to demonstrate the method, the model performs well and a key point's cumulative damage is calculated.