Study Of Tunnel Ventilation Control Based On Time Series Analysis

In recent years，the establishment of the road system is coming to the peak period，more and more highway tunnels were put into operation. How to ensure the safeoperation of the tunnel has become a major issue. New methods of road tunnelventilation control technologies based on time series analysis are proposed in this paper.Firstly, this paper introduced the highway tunnel ventilation control objectives andtunnel ventilation control system structure，and introduced briefly the tunnel ventilationcontrol methods current has been applied. According to the characteristics of tunnelenvironmental parameters time series, a combination prediction algorithm based onEmpirical Mode Decomposition was proposed. We used this algorithm to predict COconcentration and VI value in the tunnel, and compared with these traditional time seriesforecasting algorithms. The results showed that the proposed new method in predictingaccuracy and stability have achieved good results.Meanwhile, the time series data segment similarity search technology was studied,found the similar data segment from the existing database to predict the future trends ofthe environmental parameters in the tunnel. A similarity search technology based onmorphological features was proposed, and used the time series match method to detectVI poisoning case. Finally, the principle and new algorithm of reinforcement learningwas researched. By means of new reinforcement learning algorithm improved thetraditional PID controller. For the controlled object, created reinforcement learning PIDcontroller suit the tunnel environment, and the traditional PID controller was compared,the results show that reinforcement learning PID controller performance had improvedsignificantly.