We apply tipping point analysis to measurements of electronic components commonly used in applications in the automotive or aviation industries and demonstrate early warning signals based on scaling properties of resistance time series. The analysis utilises the statistical physics framework with stochastic modelling by representing the measured time series as a composition of deterministic and stochastic components estimated from measurements. The early warning signals are observed much earlier than those estimated from conventional techniques, such as threshold-based failure detection, or bulk estimates used in Weibull failure analysis. The introduced techniques may be useful for predictive maintenance of power electronics, with industrial applications. We suggest that this approach can be applied to various electromagnetic measurements in power systems and energy applications.

中文翻译：

---

具有预测性维护故障预警信号的电阻数据的临界点分析

我们将临界点分析应用于汽车或航空工业应用中常用的电子元件的测量，并展示基于电阻时间序列缩放特性的预警信号。该分析利用具有随机建模的统计物理框架，将测量的时间序列表示为根据测量估计的确定性和随机分量的组合。观察到的预警信号比通过传统技术估计的要早得多，例如基于阈值的故障检测或威布尔故障分析中使用的批量估计。引入的技术可用于具有工业应用的电力电子设备的预测性维护。