Multi Sensor Data Fusion Algorithm Based on Multi Neural Network Decision Model and D-S Evidence Theory

A data fusion model TFIT-DS based on multi neural network decision model and D-S evidence is proposed to address the uncertainty and redundancy of sensor data. This model utilizes the D-S evidence theory to fuse the decision information output from the attention based Transformer model and the InceptionTime model with convolutional neural network structure at the decision level. Through the depth combination structure, the model can fully perceive the characteristics of sensor data in different situations, thus achieving a new breakthrough on the basis of high-precision based on a single mode. According to the experimental outcomes, the model is able to effectively fuse different sensor information and has high accuracy in bearing fault identification.