Multi-space environment fire source positioning method and system based on Attention-LSTM

The invention discloses a multi-space fire source positioning method based on Attention-LSTM (Long Short Term Memory). Belongs to the technical field of fire source positioning. The method comprises the following steps: acquiring a multi-space long-time measuring point temperature data set through a temperature acquisition module; abnormal values in the data set are detected through an abnormal value detection module; supplementing the data set by using a data complementation module and carrying out normalization processing; and inputting the data into a fire source position prediction module to obtain a fire position prediction value. According to the invention, the Attention-LSTM neural network is used to detect the temperature of the multi-space environment, the importance degree of different time data can be noticed, so that the position of a fire source can be intelligently identified in a high-precision manner, the fire-fighting intelligence of a city is realized, the efficiency of information transmiss