| Fire promotes the civilization and progress of human society，while fire disaster givesmankind a great deal of harm. With the rapid development of economy and the growingprosperity of industry in modern society, various electrical products are emerged in anendless stream, which provides greater possibilities for the occurrence of the fire. In recentyears, there always occur a large number of electrical fires, and the disastrous&major firewith great loss always caused by electrical fire. Due to the limitations of detectiontechnology, signal processing method and theory research, the false positives and falsenegatives always happened in the process of electrical fire monitoring.Paper deeply analyzed the electrical fire formation mechanism, based on the analysisof arc (spark) and high temperature are the fire source’s fundamental form of electrical fire,thorough studied the different load form’s AC arc fault voltage and current waveformscharacteristics through a lot of experiments, found out that AC arc has potential"zero-off-phenomenon" in the process of combustion.The arc fault’s characteristics of"zero-off phenomenon" broaden the thinking of arc fault detection. Proposed the methodof the use of arc fault detection and monitoring to predict electrical fire.Analyzed arc fault current signal through wavelet function, constructed orthogonalquadratic spline wavelet as wavelet function, used porous algorithm’s dyadic wavelettransform to achieve fast wavelet transform algorithm. If use wavelet analysis, thecharacteristic of "cycle zero-off" of arc fault is expressed as periodic singular points,therefore proposed a new algorithm of periodic singular points to detect arc fault, andanalyzed the feasibility and effectiveness of the arc fault detection new algorithm.On the basis of arc faults’ detection, multiple-parameter real-time monitor theelectrical fire scene’s main feature signals, using multi-information fusion technology tocomplete the integration of fire information characteristics, realized electrical firerecognition. Designed information fusion’s three-layer model and the experimentalsimulation is carried out by using our country’s standard fire data and typical disturbancedata, the simulation results showed that the fusion model can realize the electrical fire’s quick and accurate alarm, effectively avoid the false positives and false negatives ofelectrical fire.Distributed control concept is used to complete electric fire early warning system’sdesign based on arc fault and multiple information fusion technology. The whole system isdivided into upper and lower machine, and hypogynous machine is also divided into hostmachine and slave machine. The hypogynous machine’s main task is signal acquisition,pretreatment and transmission, and the host machine can complete some signal processingand judgment; the upper machine’s main task is the realization of all kinds of signalprocessing algorithms, data storage and the realization of monitoring system picture. Afterrepeated testing and debugging, the electrical fire forecast system has been used in severalapplication units, realized the prevention of electrical fire.The paper finished some research work in electrical fire prediction and made someprogress, but there still has many hot spots worthwhile to research about electrical fire, forexample, there have further developments in arc fault further fusion with other electricalfire parameters, the optimization of electrical fire’s fusion model structure, the use of newdetection technology and detector to extend the capabilities of existing systems, the leadof new ways of detecting technology of electrical fire by new technology in otherfields(such as laser image size grouping, laser forward/backward scattering applications,etc), electrical fire monitoring technology in further combine with automation, moderncommunication technology, intelligent building technology to make electrical firedetection system more and more automatic, openness and modularization, etc. |
PDF Full Text Request