Electronic olfactory power tower fire monitoring method based on ai

Abstract

The application discloses an AI-based electronic olfaction power tower fire monitoring method, which comprises the following steps: arranging layered electronic olfaction nodes on a power tower, collecting odor data, and forming an odor time sequence; controlling the multi-directional air inlets of each node to be sealed and opened in a preset order, and constructing an irreplaceable incident chain; performing directional flow guiding according to the primary and secondary incident directions, and shunting the odor samples to a double-cavity structure to obtain a double-cavity directional response sequence; inputting the double-cavity directional response sequence into an improved LSTMFCN model to generate a pyrolysis memory debt chain; sealing the primary air inlets along the irreplaceable incident chain in sequence to construct a source-evidence locking spectrum and determine the propagation order; and outputting a fire risk result and a candidate fire area according to the incident direction, the incident chain debt chain and the locking spectrum. Through directional sealing incident inversion, double-cavity odor memory modeling and source-evidence locking verification, the application realizes accurate identification and source determination of early weak pyrolysis odor of a power tower fire.

Images