Residual current circuit breaker with overcurrent (RCBO) is an electrical appliance which involves the safety of users. In order to predict its lifetime, a combined prediction method of Genetic Algorithm-Back Propagation (GA-BP) and support vector machine of grid search (GS-SVM) is proposed. First, the constant stress accelerated degradation test (ADT) is carried out for the RCBO, and the ADT data are statistically analyzed to establish the performance degradation trajectory. Then, the GA-BP model and GS-SVM model are established to predict the degradation trend of the residual operating current value. Then, based on the GA-BP prediction model and GS-SVM prediction model, the optimal weighted combined prediction based on the least square method is established. The combined prediction model is used to predict the pseudo-failure life under each accelerated stress, and the Arrhenius accelerated model is used to estimate the pseudo-failure life by the least square method. The service lifetime of RCBO under normal temperature (25 °C) is predicted. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

中文翻译：

---

基于GA-BP和GS-SVM组合模型的带过流保护的剩余电流断路器寿命预测

带过电流的漏电断路器（RCBO）是一种涉及用户安全的电器。为了预测其寿命，提出了一种遗传算法-反向传播（GA-BP）和网格搜索支持向量机（GS-SVM）相结合的预测方法。首先，对RCBO进行恒应力加速退化试验（ADT），对ADT数据进行统计分析，建立性能退化轨迹。然后，建立GA-BP模型和GS-SVM模型来预测剩余工作电流值的退化趋势。然后，基于GA-BP预测模型和GS-SVM预测模型，建立基于最小二乘法的最优加权组合预测。组合预测模型用于预测各加速应力下的伪失效寿命，Arrhenius加速模型通过最小二乘法估计伪失效寿命。预测了常温（25°C）下RCBO的使用寿命。© 2022 日本电气工程师学会。由 Wiley Periodicals LLC 出版。