Research On Control Strategy Of Ship Fire Intelligence Alarm System

In recent years,our country attaches great importance to the shipbuilding industry,a large number of advanced ships put into use,But there are more and more ship accidents.Among them,ship fire is always one of the most important factors that threaten the safety of ships,and has been paid much attention by all countries.Ships in the design process,due to certain needs,often the cabin and channel space design is very narrow,which leaves hidden dangers for crew.Once somewhere occur a fire,it will inevitably spread to other compartments,causing greater loss of property.Therefore,understanding and taking action as soon as possible is the key to reduce fire losses.Methods using fire alarm control system,can detect fire effectively and timely,once confirmed man-made fire,the fire alarm controller will start the linkage to put out the fire,achieve detective defensive effect.But the fire alarm control system exits the problem of high false alarm and non-respond rate.It is necessary that designing a rational intellectual control algorithm reduces error alarm and non-respond rate.Therefore,the study of ship fire alarm system is important to detect fire in time.Based on the analysis of the characteristics of ship fire and the development of fire alarm system at home and abroad,a two reasoning model of fire detection is designed.The model uses artificial neural network and D-S evidence theory fusion algorithm,through the relevant samples collected by the quantized using LM algorithm for neural network training,neural network training is completed,the test samples are input to the network,then the network's output input to the D-S theory and get the final result.The model was different from the fire alarm controller that detect and analysis twice,can effectively reduce the occurrence of error alarm and non-respond,and simulation results verify the feasibility of the model.Although the model can solve the defects of the controller effectively,the method in the detection of fire,do not take into the danger problems of current cabin,meanwhile the neural network has many disadvantages,such as,with the increase of samples,the generalization ability decline,so using the fuzzy control algorithm designs a fire danger assessment model,the different compartments of fire been taken into the judgment,according to the data collecting by the detector,then fuzzification,at last getting the final fire risk grade through fuzzy inference.Simulation verifies the feasibility of the model.In order to verify the feasibility of the fire twice reasoning model,the hardware platform of fire alarm control system is built.Using the existing laboratory equipment,with C++programming language,the control part of the fire alarm controller is been programmed theinterface,the detector data can be real-time display on the interface.At the same time,the fusion algorithm of BP neural network and D-S evidence fusion is added to the interface,and the fire is judged by collecting the data of different compartments.The fire,smoldering fire different respectively were tested,the test results show that the model can judge the fire quickly and effectively that reduce the occurrence of fire alarm controller the false alarm rate.