Control method of mechanical smoke emission in high-rise building corridor

The traditional method has a large control error in the corridor mechanical smoke method. Therefore, multi-task convolutional neural network-based high-rise building is proposed. Through exhaust principle of corridors, threshold set to predict volume corridors. movement corridors simulated according fire dynamics simulator determine turbulence state Input data into network complete Experimental results show that maximum accuracy this about 98%, and time always less than 1 second.