Multi-feature Fusion Flame Detection Algorithm Based on BP Neural Network

In recent years, in order to ensure the safety of industrial boilers in production and improve the utilization rate of coal resources, a series of technical regulations on the detection of industrial boilers and related industrial emission regulations have been issued. In this paper, the traditional flame detection method has the problems of low accuracy, high failure rate and high maintenance cost caused by complicated detection equipment. A multi-feature fusion flame detection algorithm based on BP Neural Network is designed. For flame images with flickering characteristics, during the preprocessing of the data set, the principle of retaining more flame features is to use the sample matrix of four types of flame features, are used for training, and the proposed flame detection algorithm is applied to the actual flame sample test matrix to verify the timeliness of the algorithm proposed.