Abstract: Infrared gas sensors are easily affected by the ambient temperature during operation, resulting in unsatisfactory detection accuracy. To solve the temperature compensation problem of infrared gas sensors, the ISSA was introduced to optimize the BP neural network. First, a model of the sparrow search algorithm is established in MATLAB to optimize BP neural network and improve its optimization ability; the method of TENT chaotic mapping and dynamic adaptive weight factor is added to optimize the sparrow search algorithm. After a performance test of the benchmark function, the ISSA demonstrated excellent global search ability, stable performance, and a fast optimization speed. Finally, the ISSA-BP algorithm was used for the actual temperature compensation of infrared sensors, and the results showed that the ISSA-BP temperature compensation algorithm can effectively compensate for the temperature of infrared gas sensors.