Top-blown furnace system sensor fault detection and data reconstruction method based on space-time cooperation

According to the sensor fault detection and data reconstruction method based on space-time cooperation, the sensor fault is detected by means of the space correlation, and the fault sensor data reconstruction is carried out by means of the time correlation, so that the system ensures sufficient fault-tolerant capability within a certain time range, and the system reliability is improved. A spatial correlation model based on MIC-GCN is provided, a graph theory is introduced into a top-blown furnace sensor fault diagnosis scene, a sensor network knowledge graph is constructed according to maximum mutual information, multi-scale spatial features are captured by means of a graph convolution mechanism, and then the fault state of the sensor is captured; and secondly, designing a time-space collaborative STG-Transform graph-level prediction model by introducing a graph-level time sequence correlation relationship, and on the basis of positioning the fault sensor, performing data reconstruction on the fault sensor b