The Research On Grading Pre-warning System Of Early Fire Detection

The fire parameters gathered by the detectors are unable to know in advance, and they are non-constitutive signal. The traditional survey methods merely carry on the judgment and the recognition through gathering sole fire characteristic parameter information. So they are disturbed by the environment inevitably. The question about the system's high mistakenly reporting rate is prominent. In recent years, the accuracy of fire alarm becomes better, while the sensibility and reliability of detectors are improved. But it can't match the request of the automation of the fire detection system. It can decrease mistakenly reporting only by describing fire inherent characteristic completely and exactly. Thus through researching the fire development process and the present fire surveying methods, and combining with fire detection system's own characteristic, A new theory of grading pre-warning in the early fire stage is proposed. The multi-sensor data fusion technology and Dynamic Fuzzy Neural Network algorithm are used in the grading pre-warning system of early fire detection.Because the early fire characteristic signals are not single, dividing three pre-warning grades based on the characteristic signals. Combining with smell sensors, photoelectric smoke sensors and carbon monoxide sensors and using multi-sensor in signal early extraction, signal identification and signal judgment. Using the Dynamic Fuzzy Neural Network theory to construct early fire signal classification pre-warning system model, make pattern recognition and detect the gas of early fire. Making grading pre-warning based on the concentrations and concentration changing rates of gas and smoke.Dynamic Fuzzy Neural Network used in detecting early signs of fire is based on its fast constringent speed and the dynamic of network structural, besides that the Dynamic Fuzzy Neural Network has the characteristic that can effectively make system model. Using the Dynamic Fuzzy Neural Network to construct the model of early fire detecting signals grading pre-warning system then pick up and identify the early fire signals. The simulation of MATABL can show the approximation curve of network model output and system expected output. Besides that according to train the network can get the three stages fire probabilities of the early fire. Comparing the errors with the results of RBF algorithm can prove that using the Dynamic Fuzzy Neural Network to construct the model of early fire detecting signals grading pre-warning system is feasibility, additionally network testing results prove that using the Dynamic Fuzzy Neural Network to get the early fire probabilities are close to the expected probabilities, furthermore it can realize grading pre-warning of early fire.