Fault diagnosis in modern industrial and information systems is critical for ensuring equipment reliability and operational safety, but traditional methods have difficulty in effectively capturing spatiotemporal dependencies and fault-sensitive features in multi-sensor data, especially rarely considering dynamic features between multi-sensor data. To address these challenges, this study proposes DyGAT-FTNet, a novel graph neural network model tailored to multi-sensor fault detection. The model dynamically constructs association graphs through a learnable dynamic graph construction mechanism, enabling automatic adjacency matrix generation based on time–frequency features derived from the short-time Fourier transform (STFT). Additionally, the dynamic graph attention network (DyGAT) enhances the extraction of spatiotemporal dependencies by dynamically assigning node weights. The time–frequency graph pooling layer further aggregates time–frequency information and optimizes feature representation.Experimental evaluations on two benchmark multi-sensor fault detection datasets, the XJTUSuprgear dataset and SEU dataset, show that DyGAT-FTNet significantly outperformed existing methods in classification accuracy, with accuracies of 1.0000 and 0.9995, respectively, highlighting its potential for practical applications.

It is important to estimate the source term for a gas emission event in atmosphere. Optimization method is one of the useful tools to identify the source parameters by solving the inverse problem. Swarm intelligent optimization (SIO) algorithm has been used to estimate the source term successfully. However, there are still some issues for the SIO method. Therefore, an active firefly algorithm (AFA) method was proposed to improve the estimation performance of common passive firefly algorithm (PFA) for source estimation. Then, the release experiment cases were used to test the AFA method. The comparison results prove that AFA has much higher computation efficiency than PFA and PSO as well as higher estimation accuracy. Further, different effect factors on the performance of AFA were discussed. Compared with common PFA, the estimation results of AFA are more robust with different population number, and the estimation accuracy of AFA with less population scale is better than that of PFA. The estimation of AFA with less generation is better than that of PFA, and the computation efficiency of AFA is improved significantly. Finally, the effect of sensor numbers on AFA method was discussed. The estimation accuracy increases with the sensor numbers. AFA can still obtain a better estimation result with less sensor amount than PFA. Hence, the proposed AFA method has better performance than the commonly used SIO method to estimate the atmospheric source term. It is a potentially useful method for gas emission inverse problem.