The traditional power automation equipment maintenance and inspection, usually according to the fixed cycle, basically do not consider the current and future state of the equipment, leading to the maintenance should not necessary, the needed can not be maintenance, the dislocation phenomenon is more serious, the economy is poor, and bring safety risks. To improve this situation, an automated equipment status evaluation system based on the primary and secondary fusion model is proposed. This system gets through the long-standing information island of power grid enterprises, integrates the dispatching management system, power monitoring system, state estimation, defect management and other business systems, and constructs a primary and secondary fusion model. Based on this model, the training sample is generated online, which eliminates manual annotation and greatly improves the efficiency. The automatically generated training sample is used to train the random forest model and update the evaluation and prediction of the current state and future state of the automation equipment, formulate the maintenance and inspection plan, effectively solve the dislocation problem, and improve the economy and safety.