Identification of Multiple Fire Sources in the Utility Tunnel Based on a Constrained Particle Swarm Optimization Algorithm

Fire disaster is one of the most dangerous disasters in the utility tunnel with plenty of high-voltage and communication cables. Fire source identification is an important part of fire protection in utility tunnel fires. The particle swarm optimization (PSO) algorithm based on limited temperature observations was applied in the multiple fire sources identification problem, and a constrained PSO algorithm is developed for performance improvement. The fire characteristics could be estimated simultaneously, including the fire source location, the maximum temperature value, and the attenuation coefficient. Based on these parameters, the whole temperature distribution of the tunnel could be predicted correspondingly. The feasibility, superiority, and robustness of the proposed algorithm were demonstrated in numerical and experimental scenarios. Results showed that the proposed constrained algorithm could identify the double fire sources with high accuracy, and the identified locations were gathered around the actual ones in comparison with the basic algorithm. The fire source locations and fire states could be estimated under noisy and disturbance situations within an acceptance error. When the measurement noises varied from 0.02 to 0.10, the temperature prediction error of each measurement point changed from [0.1°C, 5.4°C] to [7.3°C, 36.8°C]. Additionally, the closer the distance between fire source and sensors is, and the more sensors allocated, the higher the prediction accuracy is.