IoT event data monitoring and prediction considering within- and between-event correlations
提出一种双时间框架的非参数霍克斯过程模型，用于处理物联网事件数据中的事件内和事件间相关性，并基于改进算法构建预测与监测框架，通过仿真和风电机组数据验证了有效性。
In the era of Industry 4.0, with significant advancements in data acquisition, Internet of Things (IoT) systems are now capable of capturing a vast array of complex and diverse event data. These data often highlight sudden system anomalies that occur amidst stable signals. To address the correlations within and between events, we introduce a novel nonparametric Hawkes process model with a dual temporal framework (DTNHawkes). This model permits the baseline intensity and triggering functions to fluctuate over time, thereby transcending the limitations inherent in traditional methodologies. Furthermore, this work develops a prediction and monitoring framework based on intensity functions estimated via a modified algorithm (DTPEM-EL), which integrates penalized expectation–maximization with the Euler–Lagrange equation. Extensive simulation experiments, conducted across a variety of scenarios, along with case studies utilizing real-world wind turbine data, have validated the efficacy of the proposed model and estimation algorithm.