Transformer substation inspection robot for fire hazard grading early warning

The invention, which relates to the technical field of the automatic inspection robot, discloses a substation inspection robot for fire hazard grading early warning, comprising a detector module, a Kalman filtering module, a feature recognition module, and a decision reasoning system. The detector module is connected with the Kalman filtering module, the Kalman filtering module is connected with the feature recognition module, and the feature recognition module is connected with the decision reasoning system. According to the invention, the Kalman filtering algorithm is adopted to correct the collected data, the BP neural network can accurately identify different fire scenes through training of a small number of samples, the corresponding probability of occurrence of open fire and smoldering fire is given, and the reliability of the fire early warning system is improved. Secondly, a fuzzy reasoning algorithm is adopted in a fire early warning decision-making layer to consider external information of a monitor