The equipment system's status seriously affects the calculation's reliability and the availability of industrial equipment. System status is usually determined by the operating characteristics of the equipment. Criteria of abnormal status are determined by expert review, which is not objective. Due to the lack of objective, comprehensive health assessments of equipment systems, staff cannot perform predictive maintenance of the system. In this paper, an anomaly detection algorithm is proposed to evaluate the operating state data of the detection equipment system, which is used to solve the predictive maintenance problem of the system. We adopt the DBSCAN algorithm for detecting the operating status of the equipment system and marking out the historical data operating status to replace the assessment criteria created by experts. The abnormal status is applied to the GRU model as a train data label to predict the abnormal status of the equipment system. Compared with other mainstream algorithms, the universality and accuracy of the abnormal status prediction model we proposed are verified in multi-dimensional time series.