Intelligent IoT system Full Chain Fault Diagnosis Model Based on Deep reinforcement learning

A Deep Reinforcement Learning (DR-L) based Intelligent IoT system (IIOTS) full chain fault diagnosis model is proposed to address the issue of traditional fault diagnosis methods not being suitable for power IIOTS full chain fault diagnosis with large and complex data volume. Firstly, an analysis was conducted on the IIOTS architecture and the operational principles of the power IIOTS, and based on this, a full chain fault diagnosis approach for the power IIOTS was proposed. Then, by analyzing D-L and R-L, a Deep Q-Network (DQN) based IIOTS full chain fault diagnosis model was proposed. Finally, a comparative analysis was conducted between the proposed model and the other three comparative models under the same conditions through simulation experiments. The results indicate that the proposed model has the highest accuracy and Fl score when using SCADA system datasets and TS2000 simulation datasets. The accuracy on the two datasets reached $\mathbf{9 7. 5 4 \%}$ and $\mathbf{9 5. 3 3 \%}$, respectively, while the FI score achieved $\mathbf{9 6. 3 2 \%}$ and $\mathbf{9 4. 1 2 \%}$, respectively. Its performance is superior to the other three comparative models.