Research On Forest Fire Prediction And Detection Based On Internet Of Things

Forest fire is by no means just a general natural disaster since it is a calamity that seriously damages the ecological environment and threatens the safety of human life and property.Forest fire is so sudden and destructive that it becomes one of the greatest threats to global ecological security.Forest fire is serious in China.For natural and historical reasons,forest fire occurs frequently and causes serious damage to our forest resources and ecological environment,not only burning down forests and reducing forest density but also destroying forest structure.The frequent occurrence of forest fire,which is one of the four main sources of air pollution,adds to the current haze and also reduces the function of wind and sand control due to the destruction of vegetation.At present,global warming and reduced rainfall have led to unusually frequent weather extremes,and desertification,dust storms,and forest fires are becoming increasingly serious,making it urgent to protect forest resources.The high incidence and destructive nature of forest fire determine the importance of forest fire prevention and control,and countries around the world spare no effort in this regard.The many factors that cause forest fire are complex and difficult to quantify accurately,which is the main reason why forest fire research has increased rather than decreased,despite nearly a century of forest fire research.In recent years,the rapid development of the Internet of Things(Io T),big data,and artificial intelligence technology represented by machine learning have provided an opportunity to address forest fire level prediction.The Internet of Things(Io T)enables large-scale real-time access to all-dimensional,all-time phase,and all-scale in situ data,providing the possibility of monitoring variable factors including the water content of forest combustibles for forest fire occurrence prediction.At the same time,with the help of machine learning technology,the relationship between forest fire drivers and forest fire occurrence can be analyzed from an inverse perspective,avoiding the problems of low accuracy and poor explanatory power faced by traditional generalized linear models parameterized by constant values.This paper investigates technologies for forestry Io T and forest fire monitoring with the help of Io T and machine learning,mainly including forestry Io T data transmission,forest fire monitoring based on Io T and sound spectrum analysis,and a forest fire meteorological rating prediction model based on Io T and support vector machines,specifically:(1)Research on Io T data transmission model in forest environmentIn forestry Io T technology applications,solar energy conversion systems are often used to obtain energy and provide energy for Io T systems.However,due to the development of conversion technology,the external energy absorbed by the solar panel does not guarantee the continuous operation of the node,so the node usually works at a low duty cycle most of the time.In addition,the node is heavily dependent on the external environment for energy,and when the external environment changes,the node’s energy supply changes accordingly.Therefore,the sensors must constantly adjust their duty cycle to adapt to the energy consumption,and the network routing is in dynamic adjustment mode.This paper proposes a communication scheme with a hybrid asynchronous and synchronous MAC protocol to reduce end-to-end(E2E)delay,where the sensor will switch between asynchronous and synchronous systems to adapt its duty cycle according to its energy supply and the actual situation.(2)Research on sound spectrum analysis technology for forest fire based on the Io TForest fire is accompanied by sound signal generation,causing changes in the original sound pressure or noise of the forest,and there is a clear difference in the impact on forest noise between the occurrence of surface fire and crown fire.The noise amplitude of crown fire is significantly higher than that of surface fire,and in the spectral analysis,crown fire has significant variation characteristics in the low-frequency region,while surface fire has significant variation characteristics in the high-frequency region.This paper proposes a way to identify surface fire and crown fire based on sound spectrum and establishes an automatic analysis model based on sound spectrum analysis technology,which can carry out automatic spectrum analysis of sound data in big data systems to monitor the occurrence of the forest fire.And by studying the sound spectrum characteristics of surface fire and crown fire,surface fire and crown fire are identified promptly by analyzing the sound spectrum of the monitored forest areas.(3)Research on meteorological prediction technology of forest fire risk based on Internet of things and Support Vector MachineMeteorological parameters in forest areas are one of the main factors for the occurrence of forest fire,which are closely related to meteorological factors such as temperature,wind speed,humidity,and rainfall.This paper proposes a system for predicting the meteorological rating of forest fires using wireless sensor networks(WSNs)to collect meteorological parameters and a Support Vector Machine(SVM)to assess the relationship between forest fire danger rating and meteorological data.