A Multi-model Fusion Method Based on 0-1 Programming and Its Application in Early Warning of Coal Mill Failure

In practical operation of the thermal power plant, fault on coal mill is one of the principal factors that hinder the power plant from completing the power generating tasks on schedule. Therefore, early warning on coal mill failures is of great significance to guarantee the stability of thermal power plant and reliability of power supply to the grid system. In this paper, a multi-model fusion method based on 0-1 programming for early warning of the coal mill failure is proposed and evaluated. For this method, the background information and main features of the mill failure problems are taken into consideration, and an optimization algorithm is implemented to combine the results of multiple models. In the optimization algorithm, the sum of the false alarm cost and false negative cost is the objective, and whether accept the pre-warning or not is the variable. The method is evaluated in practical situations, and numerical test shows that the total cost of false alarm and false negative of multi-model fusion method is more promising than that of each single model. The validity is proved through the result.