CN109492928B - Method and device for constructing forest fire risk prediction model - Google Patents
Method and device for constructing forest fire risk prediction model Download PDFInfo
- Publication number
- CN109492928B CN109492928B CN201811406095.1A CN201811406095A CN109492928B CN 109492928 B CN109492928 B CN 109492928B CN 201811406095 A CN201811406095 A CN 201811406095A CN 109492928 B CN109492928 B CN 109492928B
- Authority
- CN
- China
- Prior art keywords
- information
- historical
- forest fire
- normalization
- meteorological data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Abstract
The application provides a method and a device for constructing a forest fire risk prediction model, wherein the method comprises the following steps: acquiring historical natural forest fire danger information, historical meteorological data information and vegetation type information of a target area; filtering the data records with missing items in the acquired information to obtain filtered information; smoothing the filtering information by using a logistic function to obtain smooth information; normalizing the smooth information to obtain normalization information, and corresponding the normalization information of the historical natural forest fire danger when the fire danger occurs to the normalization information of the historical meteorological data when the fire danger occurs; and training the long-time memory method model by taking the historical meteorological data normalization information and the vegetation type information as input and taking the historical natural forest fire risk normalization information mapped by the historical meteorological data normalization information as output until the trained model meets the preset precision requirement, thereby obtaining the forest fire risk prediction model. The prediction precision of the forest fire danger can be effectively improved.
Description
Technical Field
The application relates to the technical field of fire hazard prediction, in particular to a method and a device for constructing a forest fire hazard prediction model.
Background
Forest resources are one of the most important natural resources on the earth, on one hand, the forest is a main component of an earth ecosystem and has the functions of regulating climate, generating oxygen, preventing wind and fixing sand, eliminating noise and purifying environment, and on the other hand, the forest also provides a large amount of resources for production and life of people, for example, various forestry products such as raw wood, paper pulp making materials, artificial boards and the like and plant processing byproducts are provided for people. Therefore, the forest protection has irreplaceable economic significance, ecological significance and social significance. The prevention of natural fire danger of forest is a key research field of forest protection.
At present, with the national emphasis on forest fire prevention work and scientific progress, national standards of national forest fire danger division level and national forest fire danger weather level are formulated in 1992 and 1994 by combining scientific research units with the practice of national forest fire prevention work respectively, and the standards are issued by the ministry of forestry to be implemented as the standard of forest fire danger prediction in China. The method comprises the following steps of expressing the relation between a forest fire risk grade (p) and a fire risk factor (x) by using a logistic model of linear regression, wherein the linear expression is as follows:
in the formula (I), the compound is shown in the specification,
p is the forest fire danger grade;
X1、X2、...、Xnis a fire risk factor;
β1、β2、…、βnis the coefficient of the fire risk factor, beta0Is a constant.
However, according to the forest fire risk prediction method, because each fire risk factor generally has multi-dimensional complexity and is correlated with each other, the forest fire risk prediction adopting the logistic model of linear regression has certain limitation, and the precision of the prediction result is still to be improved; further, for the big data of the internet era, the logistic model of linear regression cannot meet the analysis and processing requirements of the big data.
Disclosure of Invention
In view of this, an object of the present application is to provide a method and an apparatus for constructing a forest fire prediction model, so as to improve the prediction accuracy of forest fire.
In a first aspect, an embodiment of the present application provides a method for constructing a forest fire risk prediction model, where the method includes:
acquiring historical natural forest fire danger information of a target area, wherein the historical natural forest fire danger information comprises fire area information and fire duration information;
acquiring historical meteorological data information and vegetation type information when the target area is in fire danger;
filtering data records with missing items in the historical natural forest fire danger information, the historical meteorological data information and the vegetation type information to obtain to-be-processed historical natural forest fire danger information, to-be-processed historical meteorological data information and vegetation type information;
smoothing the historical natural forest fire danger information to be processed and the historical meteorological data information to be processed by using a logistic function to obtain historical natural forest fire danger smoothing information and historical meteorological data smoothing information;
normalizing the historical natural forest fire danger smoothing information and the historical meteorological data smoothing information to obtain historical natural forest fire danger normalization information and historical meteorological data normalization information, corresponding the historical natural forest fire danger normalization information and the historical meteorological data normalization information when fire danger occurs, and constructing a mapping relation between the historical natural forest fire danger normalization information and the historical meteorological data normalization information;
and taking the historical meteorological data normalization information and the vegetation type information as the input of a long-short time memory method model, taking the historical natural forest fire risk normalization information mapped by the historical meteorological data normalization information as the output of the long-short time memory method model, and training the long-short time memory method model until the trained long-short time memory method model meets the preset precision requirement to obtain a forest fire risk prediction model.
Optionally, the to-be-processed historical natural forest fire danger information includes fire area information, and performing smoothing processing on the to-be-processed historical natural forest fire danger information by using a logistic function includes:
extracting fire area information in the historical natural forest fire danger information to be processed;
fitting the extracted fire passing area information according to the logistic function, and determining a fitting weight of the fire passing area information in the logistic function according to a fitting result;
and calculating the extracted fire area information in sequence according to the determined fitting weight of the logistic function to obtain fire area smooth information corresponding to the fire area information.
Optionally, the training the long-short term memory method model until the trained long-short term memory method model meets a preset precision requirement includes:
dividing a data set consisting of historical meteorological data normalization information and historical natural forest fire risk normalization information into a training set and a testing set by adopting a K-S algorithm;
and after the long-short time memory method model is trained by the training set, testing the trained long-short time memory method model by the testing set, stopping training the long-short time memory method model if the testing result meets the preset precision requirement, and continuing training the long-short time memory method model according to the training set until the testing result meets the preset precision requirement if the testing result does not meet the preset precision requirement.
Optionally, the method further comprises:
taking historical testing meteorological data normalization information and vegetation type information in a preset testing set as the input of the forest fire risk prediction model to obtain a prediction result, wherein the prediction result comprises: normalization information of the predicted fire passing area information and normalization information of the fire passing duration information;
drawing a loss function curve and a prediction result scatter diagram according to the prediction result and historical overfire area information and historical overfire duration information corresponding to the historical test meteorological data normalization information, and calculating an accuracy metric value;
and evaluating the precision of the forest fire risk prediction model according to the loss function curve, the prediction result scatter diagram and the accuracy metric value.
Optionally, the method further comprises:
acquiring current meteorological data information and vegetation type information of the target area;
and taking the current meteorological data information and vegetation type information as the input of the forest fire danger prediction model to obtain forest fire danger prediction information of the target area, wherein the forest fire danger prediction information comprises: normalization information of the fire area information and normalization information of the fire duration information.
Optionally, the historical meteorological data information and vegetation type information include: the day maximum air temperature, the day minimum air temperature, the day average air temperature, the day with the average air temperature of less than 18 ℃ in the previous month and day, the day with the average air temperature of more than 18 ℃ in the previous month and day, the total rainfall, the total snowfall, the thickness of the snow on the ground, the day maximum wind direction and the day maximum wind speed.
In a second aspect, an embodiment of the present application provides an apparatus for constructing a forest fire risk prediction model, where the apparatus includes:
the fire danger information acquisition module is used for acquiring historical natural forest fire danger information of a target area, wherein the historical natural forest fire danger information comprises fire area information and fire duration information;
the meteorological data information acquisition module is used for acquiring historical meteorological data information and vegetation type information when the fire danger occurs in the target area;
the information filtering module is used for filtering data records with missing items in the historical natural forest fire danger information, the historical meteorological data information and the vegetation type information to obtain to-be-processed historical natural forest fire danger information, to-be-processed historical meteorological data information and vegetation type information;
the smoothing processing module is used for smoothing the historical natural forest fire danger information to be processed, the historical meteorological data information to be processed and the vegetation type information by using a logistic function to obtain historical natural forest fire danger smoothing information and historical meteorological data smoothing information;
the normalization processing module is used for performing normalization processing on the historical natural forest fire danger smooth information and the historical meteorological data smooth information to obtain historical natural forest fire danger normalization information and historical meteorological data normalization information, corresponding the historical natural forest fire danger normalization information and the historical meteorological data normalization information when fire danger occurs, and constructing a mapping relation between the historical natural forest fire danger normalization information and the historical meteorological data normalization information;
and the model generation module is used for taking the historical meteorological data normalization information and the vegetation type information as the input of the long-short time memory method model, taking the historical natural forest fire risk normalization information mapped by the historical meteorological data normalization information as the output of the long-short time memory method model, and training the long-short time memory method model until the trained long-short time memory method model meets the preset precision requirement to obtain the forest fire risk prediction model.
Optionally, the smoothing module is specifically configured to:
extracting fire area information in the historical natural forest fire danger information to be processed;
fitting the extracted fire passing area information according to the logistic function, and determining a fitting weight of the fire passing area information in the logistic function according to a fitting result;
and calculating the extracted fire area information in sequence according to the determined fitting weight of the logistic function to obtain fire area smooth information corresponding to the fire area information.
Optionally, the apparatus further comprises:
the model evaluation module is used for taking historical test meteorological data normalization information and vegetation type information in a preset test set as the input of the forest fire risk prediction model to obtain a prediction result, and the prediction result comprises: normalization information of the predicted fire passing area information and normalization information of the fire passing duration information;
drawing a loss function curve and a prediction result scatter diagram according to the prediction result and historical overfire area information and historical overfire duration information corresponding to the historical test meteorological data normalization information, and calculating an accuracy metric value;
and evaluating the precision of the forest fire risk prediction model according to the loss function curve, the prediction result scatter diagram and the accuracy metric value.
Optionally, the apparatus further comprises:
the fire hazard prediction module is used for acquiring the current meteorological data information and vegetation type information of the target area;
and taking the current meteorological data information and vegetation type information as the input of the forest fire danger prediction model to obtain forest fire danger prediction information of the target area, wherein the forest fire danger prediction information comprises: normalization information of the fire area information and normalization information of the fire duration information.
According to the method and the device for constructing the forest fire danger prediction model, the historical natural forest fire danger information is represented by the fire area information and the fire duration information, interference of forest fire danger caused by human factors can be reduced, the obtained historical meteorological data information and the historical natural forest fire danger information are subjected to smoothing processing by using the logistic function, the accuracy of the forest fire danger prediction model can be effectively improved, normalization processing is performed on the smoothed information, and finally the long-time and short-time memory method model is input for training, so that the limitation of forest fire danger prediction of the logistic model adopting linear regression is avoided, and the precision of a prediction result can be effectively improved.
In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.
Fig. 1 is a schematic flow chart of a method for constructing a forest fire risk prediction model according to an embodiment of the present application;
FIG. 2 is a schematic structural diagram of an apparatus for constructing a forest fire risk prediction model according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of a computer device 300 according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.
Fig. 1 is a schematic flow chart of a method for constructing a forest fire risk prediction model according to an embodiment of the present application. As shown in fig. 1, the method includes:
in the embodiment of the application, a large amount of historical forest fire danger information is analyzed, wherein the fire passing duration information and the fire passing area information can directly reflect the occurrence scale and the occurrence probability of forest fire danger, the longer the fire passing duration is, the larger the fire passing area is, the higher the probability of occurrence of natural forest fire danger is considered, the forest fire danger with short fire passing duration and small fire passing area is greatly influenced by human factors, and the actual occurrence probability of forest fire danger is very low or even can be avoided. Therefore, in the embodiment of the application, the fire area information and the fire duration information are used for representing the historical natural forest fire danger information, so that on one hand, the quantity of the forest fire danger information needing to be processed can be effectively reduced, on the other hand, the interference of forest fire danger caused by human factors can be reduced, and the accuracy of prediction of the natural forest fire danger is improved.
In the embodiment of the application, the information of forest fire hazards occurring under natural conditions is acquired, and the information of forest fire hazards occurring under artificial conditions is not included, for example, the information of forest fire hazards occurring due to smoking or tomb sweeping.
102, acquiring historical meteorological data information and vegetation type information when the target area is in fire danger;
in the embodiment of the present application, as an optional embodiment, historical meteorological data information, vegetation type information, and historical natural forest fire risk information (historical forest fire data) of a target area in a predetermined time period may be collected from a network and related departments. For example, the historical meteorological data information and vegetation type information and historical natural forest fire information of the target area can be crawled from related websites by using a web crawler technology, and the historical meteorological data information and vegetation type information and historical natural forest fire information of the target area can be collected from related departments such as a forest farm management organization or a meteorological station.
In the embodiment of the present application, as an optional embodiment, the historical weather data information and the vegetation type information are historical weather data information, and the historical weather data information includes: the day maximum air temperature, the day minimum air temperature, the day average air temperature, the day with the average air temperature of lower than 18 ℃ in the previous month and day, the day with the average air temperature of higher than 18 ℃ in the previous month and day, the total rainfall, the total snowfall, the thickness of the snow accumulated on the ground, the day maximum wind direction, the day maximum wind speed and the like. As another optional embodiment, the historical meteorological data information and the vegetation type information may further include geographic data and vegetation data of the target area, wherein the geographic data includes: longitude, latitude, and altitude, vegetation data including forest coverage and major vegetation types.
103, filtering data records with missing items in the historical natural forest fire danger information, the historical meteorological data information and the vegetation type information to obtain to-be-processed historical natural forest fire danger information, to-be-processed historical meteorological data information and vegetation type information;
in the embodiment of the application, as massive multi-dimensional data is acquired or obtained, and noise, interference information, irrelevant information and incomplete information exist in the massive data, useless information (the noise, the interference information, the irrelevant information and the incomplete information) needs to be removed from the massive data so as to improve the reliability of the data and further improve the precision of a subsequently constructed forest fire prediction model.
In this embodiment, as an optional embodiment, taking historical weather data information and vegetation type information as an example, the filtering includes: the method comprises the steps of firstly, preliminarily cleaning historical meteorological data information and vegetation type information with data residual and incomplete items (incomplete information), carrying out extremum detection and time consistency detection on statistical values contained in the historical meteorological data information and the vegetation type information, and changing the detected error records.
in the embodiment of the present application, as an optional embodiment, taking fire area information as an example, performing smoothing processing on the to-be-processed historical natural forest fire risk information by using a logistic function includes:
a11, extracting fire passing area information in the historical natural forest fire danger information to be processed;
in the embodiment of the application, for the fire passing duration information contained in the historical natural forest fire danger information to be processed, the smoothing processing mode is the same as the smoothing processing mode of the fire passing area information. Similarly, the smoothing processing method is also the same for the parameters such as the daily maximum temperature and the daily minimum temperature in the historical weather data information and the vegetation type information.
A12, fitting the extracted fire area information according to the logistic function, and determining the fitting weight of the fire area information in the logistic function according to the fitting result;
in the embodiment of the present application, the logistic function formula is as follows:
in the formula (I), the compound is shown in the specification,
Yismoothing processing values of ith parameters in the historical natural forest fire danger information or the historical meteorological data information and the vegetation type information to be processed;
Kia value set for an industry standard;
αia first fitting weight value of the ith parameter;
Xiis the ith parameter.
In the embodiment of the application, the first fitting weight and the second fitting weight corresponding to the fire area information in the logistic function can be determined by fitting the extracted fire area information in the target area. According to a similar method, a first fitting weight and a second fitting weight corresponding to the fire duration information in the logistic function can be determined.
In the embodiment of the present application, as an optional embodiment, for the daily maximum air temperature, the corresponding logistic function is:
for daily minimum relative humidity, the corresponding logistic function is:
for the daily maximum wind speed, the corresponding logistic function is:
and A13, sequentially calculating the extracted fire area information according to the determined fitting weight of the logistic function to obtain fire area smoothing information corresponding to the fire area information.
In the embodiment of the application, after the fitting weight of the logistic function is determined, the information of the fire area can be smoothed according to the logistic function after the fitting weight is determined.
In the embodiment of the application, the relevant verification result shows that the smoothing processing is performed by using the logistic function, and the accuracy of the smoothing processing is highest.
in the embodiment of the application, when the forest fire danger prediction model is constructed, the input quantity is historical meteorological data smooth information and vegetation type information, and the output quantity is historical natural forest fire danger smooth information, but because various parameters in the historical meteorological data smooth information have huge numerical differences, if training is directly carried out according to various parameter values, before the optimal solution of the forest fire danger prediction model is found, the process of gradient reduction of the forest fire danger prediction model is very tortuous and time-consuming, and further the accuracy of prediction of the forest fire danger prediction model is influenced. Therefore, in the embodiment of the application, the normalization processing method is adopted to respectively perform normalization processing on the historical natural forest fire risk smoothing information and the historical meteorological data smoothing information. Thus, through normalization processing, dimensional influence among the parameters can be eliminated, the parameters are comparable, and comprehensive comparison and evaluation are suitable.
In the embodiment of the application, the normalization processing is performed by using the following dispersion normalization (Min-Max Scaling) normalization formula:
in the formula (I), the compound is shown in the specification,
z is normalization information of the ith parameter;
xifor the ith parameterSmoothing the information;
min(xi) The minimum smoothing information in the predetermined number of smoothing information before the smoothing information of the ith parameter;
max(xi) Is the maximum smoothing information among the predetermined number of smoothing information preceding the smoothing information of the i-th parameter.
In the embodiment of the present application, each parameter may be normalized to be within the [0, 1] interval by the normalization processing.
According to the method and the device, the mapping relation between the historical natural forest fire risk normalization information and the historical meteorological data normalization information is established according to dates. For example, according to the fire passing duration information in the historical natural forest fire danger information, historical daily meteorological data information in the fire passing duration information is obtained, and historical natural forest fire danger normalization information corresponding to the historical natural forest fire danger information and historical meteorological data normalization information corresponding to the obtained historical daily meteorological data information are constructed. If the historical solar meteorological data information does not have corresponding historical natural forest fire danger information, the fire area information and the fire duration information in the historical natural forest fire danger information mapped by the historical solar meteorological data information can be set to be zero.
And 106, taking the historical meteorological data normalization information and the vegetation type information as input of a long-short time memory method model, taking historical natural forest fire risk normalization information mapped by the historical meteorological data normalization information as output of the long-short time memory method model, training the long-short time memory method model until the trained long-short time memory method model meets preset precision requirements, and obtaining a forest fire risk prediction model.
In the embodiment of the application, a Long Short Term Memory (LSTM) model can solve the Long-Term dependence of a Recurrent Neural Network (RNN) model, and the LSTM model improves the Long-Term dependence by adding an input gate, an output gate and a forgetting gate structure in a cell.
In the embodiment of the application, when the long-time memory method model is trained, a data set consisting of historical meteorological data normalization information, historical natural forest fire risk normalization information and vegetation type information can be divided into a training set and a testing set. The trained vegetation type information is vegetation type information subjected to filtering processing. For example, using the Kennard-Stone (K-S) algorithm, 70% of the data set is divided into training sets and 30% of the data set is divided into test sets. The long-short time memory method model is trained through a training set, the trained long-short time memory method model is tested through a testing set, if a testing result meets the preset precision requirement, the long-short time memory method model can be stopped from being trained, and if the testing result does not meet the preset precision requirement, the long-short time memory method model continues to be trained until the testing result meets the preset precision requirement.
In the embodiment of the application, the obtained historical meteorological data information and the historical natural forest fire danger information are subjected to smoothing processing by using the logistic function, the accuracy of a forest fire danger prediction model can be effectively improved, then the smoothed information is subjected to normalization processing, and finally the LSTM model is input for training, so that the limitation of forest fire danger prediction of a logistic model adopting linear regression is avoided, and the precision of a prediction result can be effectively improved; furthermore, the LSTM model can adapt to the time sequence of historical meteorological data information, vegetation type information and historical natural forest fire information, can obtain a better fitting effect when modeling mass data, and meets the analysis and processing requirements of big data.
In the embodiment of the application, as an optional embodiment, the accuracy of the forest fire risk prediction model can be evaluated. Thus, the method further comprises:
taking historical testing meteorological data normalization information and vegetation type information in a preset testing set as the input of the forest fire risk prediction model to obtain a prediction result, wherein the prediction result comprises: normalization information of the predicted fire passing area information and normalization information of the fire passing duration information;
drawing a loss function curve and a prediction result scatter diagram according to the prediction result and historical overfire area information and historical overfire duration information corresponding to the historical test meteorological data normalization information, and calculating an accuracy metric value;
and evaluating the precision of the forest fire risk prediction model according to the loss function curve, the prediction result scatter diagram and the accuracy metric value.
In this embodiment, as an optional embodiment, the accuracy metric includes: mean square error value, correlation coefficient, standard deviation, cross validation standard deviation, prediction standard deviation, relative analysis error, and the like. And determining the precision of the forest fire risk prediction model through the loss function curve, the prediction result scatter diagram and the calculated accuracy metric values.
In the embodiment of the application, as another optional embodiment, a forest fire danger prediction model can be used for forest fire danger prediction. Thus, the method further comprises:
acquiring current meteorological data information and vegetation type information of the target area;
and taking the current meteorological data information and the vegetation type information as the input of the forest fire danger prediction model to obtain the forest fire danger prediction information of the target area.
In the embodiment of the application, the current meteorological data information can be smoothed and normalized. The forest fire danger prediction information comprises: normalization information of the fire passing area prediction information and normalization information of the fire passing duration prediction information. Therefore, the probability of forest fire danger in the target area is represented by utilizing the normalization information of the fire area prediction information and the normalization information of the fire duration prediction information, and the prediction result is more visual.
Fig. 2 is a schematic structural diagram of an apparatus for constructing a forest fire risk prediction model according to an embodiment of the present application. As shown in fig. 2, the apparatus includes:
the fire danger information acquisition module 201 is used for acquiring historical natural forest fire danger information of a target area, wherein the historical natural forest fire danger information comprises fire area information and fire duration information;
in the embodiment of the application, the information of the fire passing area and the information of the fire passing duration are used for representing the historical natural forest fire danger information, so that the interference of forest fire danger caused by human factors can be reduced.
A meteorological data information obtaining module 202, configured to obtain historical meteorological data information and vegetation type information when a fire risk occurs in the target area;
in the embodiment of the application, the network crawler technology is utilized to crawl historical meteorological data information, vegetation type information and historical natural forest fire danger information of a target area from a related website, and historical meteorological data information, vegetation type information and historical natural forest fire danger information of the target area are collected from a forest farm management mechanism, a meteorological station and related department.
In the embodiment of the present application, as an optional embodiment, the historical weather data information and the vegetation type information are historical weather data information, and the historical weather data information includes: the day maximum air temperature, the day minimum air temperature, the day average air temperature, the day with the average air temperature of lower than 18 ℃ in the previous month and day, the day with the average air temperature of higher than 18 ℃ in the previous month and day, the total rainfall, the total snowfall, the thickness of the snow accumulated on the ground, the day maximum wind direction, the day maximum wind speed and the like.
The information filtering module 203 is used for filtering data records with missing items in the historical natural forest fire danger information, the historical meteorological data information and the vegetation type information to obtain to-be-processed historical natural forest fire danger information, to-be-processed historical meteorological data information and vegetation type information;
the smoothing module 204 is used for smoothing the historical natural forest fire danger information to be processed and the historical meteorological data information to be processed by using a logistic function to obtain historical natural forest fire danger smoothing information and historical meteorological data smoothing information;
the normalization processing module 205 is configured to perform normalization processing on the historical natural forest fire risk smoothing information and the historical meteorological data smoothing information to obtain historical natural forest fire risk normalization information and historical meteorological data normalization information, and to correspond the historical natural forest fire risk normalization information when a fire occurs to the historical meteorological data normalization information when the fire occurs, so as to construct a mapping relationship between the historical natural forest fire risk normalization information and the historical meteorological data normalization information;
in the embodiment of the application, a normalization processing method is adopted to respectively perform normalization processing on the historical natural forest fire risk smoothing information and the historical meteorological data smoothing information. Thus, through normalization processing, dimensional influence among the parameters can be eliminated, the parameters are comparable, and comprehensive comparison and evaluation are suitable.
In the embodiment of the application, the dispersion standardization normalization formula is used for normalization.
And the model generation module 206 is configured to use the historical meteorological data normalization information and the vegetation type information as input of a long-short time memory method model, use historical natural forest fire risk normalization information mapped by the historical meteorological data normalization information as output of the long-short time memory method model, and train the long-short time memory method model until the trained long-short time memory method model meets preset precision requirements, so as to obtain a forest fire risk prediction model.
In this embodiment, as an optional embodiment, the training of the long-short time memory method model until the trained long-short time memory method model meets the preset accuracy requirement includes:
dividing a data set consisting of historical meteorological data normalization information and historical natural forest fire risk normalization information into a training set and a testing set by adopting a K-S algorithm;
and after the long-short time memory method model is trained by the training set, testing the trained long-short time memory method model by the testing set, stopping training the long-short time memory method model if the testing result meets the preset precision requirement, and continuing training the long-short time memory method model according to the training set until the testing result meets the preset precision requirement if the testing result does not meet the preset precision requirement.
In this embodiment, as an optional embodiment, the smoothing module 204 is specifically configured to:
extracting fire area information in the historical natural forest fire danger information to be processed;
fitting the extracted fire passing area information according to the logistic function, and determining a fitting weight of the fire passing area information in the logistic function according to a fitting result;
and calculating the extracted fire area information in sequence according to the determined fitting weight of the logistic function to obtain fire area smooth information corresponding to the fire area information.
In this embodiment, as an optional embodiment, the apparatus further includes:
a model evaluation module (not shown in the figure) configured to use historical test meteorological data normalization information and vegetation type information in a preset test set as inputs of the forest fire risk prediction model to obtain a prediction result, where the prediction result includes: normalization information of the predicted fire passing area information and normalization information of the fire passing duration information;
drawing a loss function curve and a prediction result scatter diagram according to the prediction result and historical overfire area information and historical overfire duration information corresponding to the historical test meteorological data normalization information, and calculating an accuracy metric value;
and evaluating the precision of the forest fire risk prediction model according to the loss function curve, the prediction result scatter diagram and the accuracy metric value.
In this embodiment, as another optional embodiment, the apparatus further includes:
a fire risk prediction module (not shown in the figure) for acquiring the current meteorological data information and vegetation type information of the target area;
and taking the current meteorological data information and vegetation type information as the input of the forest fire danger prediction model to obtain forest fire danger prediction information of the target area, wherein the forest fire danger prediction information comprises: normalization information of the fire area information and normalization information of the fire duration information.
As shown in fig. 3, an embodiment of the present application provides a computer device 300 for executing the method for constructing a forest fire prediction model in fig. 1, the device includes a memory 301, a processor 302, and a computer program stored on the memory 301 and executable on the processor 302, wherein the processor 302 implements the steps of the method for constructing a forest fire prediction model when executing the computer program.
Specifically, the memory 301 and the processor 302 can be general-purpose memories and processors, and are not limited to specific examples, and the processor 302 can execute the above method for constructing the forest fire risk prediction model when executing the computer program stored in the memory 301.
Corresponding to the method for constructing a forest fire prediction model in fig. 1, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, performs the steps of the method for constructing a forest fire prediction model.
In particular, the storage medium can be a general-purpose storage medium, such as a removable disk, a hard disk, or the like, and when executed, the computer program on the storage medium can execute the above method for constructing the forest fire risk prediction model.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and there may be other divisions when actually implemented, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some communication interfaces, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus once an item is defined in one figure, it need not be further defined and explained in subsequent figures, and moreover, the terms "first", "second", "third", etc. are used merely to distinguish one description from another and are not to be construed as indicating or implying relative importance.
Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present application, and are used for illustrating the technical solutions of the present application, but not limiting the same, and the scope of the present application is not limited thereto, and although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope disclosed in the present application; such modifications, changes or substitutions do not depart from the spirit and scope of the present disclosure, which should be construed in light of the above teachings. Are intended to be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (7)
1. A method for constructing a forest fire risk prediction model is characterized by comprising the following steps:
acquiring historical natural forest fire danger information of a target area, wherein the historical natural forest fire danger information comprises fire area information and fire duration information;
acquiring historical meteorological data information and vegetation type information when the target area is in fire danger;
filtering data records with missing items in the historical natural forest fire danger information, the historical meteorological data information and the vegetation type information to obtain to-be-processed historical natural forest fire danger information, to-be-processed historical meteorological data information and vegetation type information;
smoothing the historical natural forest fire danger information to be processed and the historical meteorological data information to be processed by using a logistic function to obtain historical natural forest fire danger smoothing information and historical meteorological data smoothing information;
normalizing the historical natural forest fire danger smoothing information and the historical meteorological data smoothing information to obtain historical natural forest fire danger normalization information and historical meteorological data normalization information, corresponding the historical natural forest fire danger normalization information and the historical meteorological data normalization information when fire danger occurs, and constructing a mapping relation between the historical natural forest fire danger normalization information and the historical meteorological data normalization information;
taking the historical meteorological data normalization information and the vegetation type information as input of a long-short time memory method model, taking historical natural forest fire risk normalization information mapped by the historical meteorological data normalization information as output of the long-short time memory method model, and training the long-short time memory method model until the trained long-short time memory method model meets the preset precision requirement to obtain a forest fire risk prediction model;
the to-be-processed historical natural forest fire danger information comprises fire area information, and the smoothing processing of the to-be-processed historical natural forest fire danger information by utilizing a logistic function comprises the following steps:
extracting fire area information in the historical natural forest fire danger information to be processed;
fitting the extracted fire passing area information according to the logistic function, and determining a fitting weight of the fire passing area information in the logistic function according to a fitting result;
calculating the extracted fire area information in sequence according to the determined fitting weight of the logistic function to obtain fire area smooth information corresponding to the fire area information;
training the long-short time memory method model until the trained long-short time memory method model meets the preset precision requirement, and the method comprises the following steps:
dividing a data set consisting of historical meteorological data normalization information and historical natural forest fire risk normalization information into a training set and a testing set by adopting a K-S algorithm;
after the long-short time memory method model is trained by the training set, the trained long-short time memory method model is tested by the testing set, if the testing result meets the preset precision requirement, the training of the long-short time memory method model is stopped, and if the testing result does not meet the preset precision requirement, the long-short time memory method model continues to be trained according to the training set until the testing result meets the preset precision requirement;
normalization processing is carried out by using the following dispersion normalization formula:
in the formula (I), the compound is shown in the specification,
z is normalization information of the ith parameter;
xismoothing information for the ith parameter;
min(xi) The minimum smoothing information in the predetermined number of smoothing information before the smoothing information of the ith parameter;
max(xi) Is the maximum smoothing information among the predetermined number of smoothing information preceding the smoothing information of the i-th parameter.
2. The method of claim 1, wherein the method further comprises:
taking historical testing meteorological data normalization information and vegetation type information in a preset testing set as the input of the forest fire risk prediction model to obtain a prediction result, wherein the prediction result comprises: normalization information of the predicted fire passing area information and normalization information of the fire passing duration information;
drawing a loss function curve and a prediction result scatter diagram according to the prediction result and historical overfire area information and historical overfire duration information corresponding to the historical test meteorological data normalization information, and calculating an accuracy metric value;
and evaluating the precision of the forest fire risk prediction model according to the loss function curve, the prediction result scatter diagram and the accuracy metric value.
3. The method of claim 1, wherein the method further comprises:
acquiring current meteorological data information and vegetation type information of the target area;
and taking the current meteorological data information and vegetation type information as the input of the forest fire danger prediction model to obtain forest fire danger prediction information of the target area, wherein the forest fire danger prediction information comprises: normalization information of the fire area information and normalization information of the fire duration information.
4. The method of claim 1, wherein the historical meteorological data information and vegetation type information comprises: the day maximum air temperature, the day minimum air temperature, the day average air temperature, the day with the average air temperature of less than 18 ℃ in the previous month and day, the day with the average air temperature of more than 18 ℃ in the previous month and day, the total rainfall, the total snowfall, the thickness of the snow on the ground, the day maximum wind direction and the day maximum wind speed.
5. An apparatus for constructing a forest fire prediction model, the apparatus comprising:
the fire danger information acquisition module is used for acquiring historical natural forest fire danger information of a target area, wherein the historical natural forest fire danger information comprises fire area information and fire duration information;
the meteorological data information acquisition module is used for acquiring historical meteorological data information and vegetation type information when the fire danger occurs in the target area;
the information filtering module is used for filtering data records with missing items in the historical natural forest fire danger information, the historical meteorological data information and the vegetation type information to obtain to-be-processed historical natural forest fire danger information, to-be-processed historical meteorological data information and vegetation type information;
the smoothing processing module is used for smoothing the historical natural forest fire danger information to be processed and the historical meteorological data information to be processed by utilizing a logistic function to obtain the historical natural forest fire danger smoothing information and the historical meteorological data smoothing information;
the normalization processing module is used for performing normalization processing on the historical natural forest fire danger smooth information and the historical meteorological data smooth information to obtain historical natural forest fire danger normalization information and historical meteorological data normalization information, corresponding the historical natural forest fire danger normalization information and the historical meteorological data normalization information when fire danger occurs, and constructing a mapping relation between the historical natural forest fire danger normalization information and the historical meteorological data normalization information;
the model generation module is used for taking the historical meteorological data normalization information and the vegetation type information as input of a long-time memory method model, taking historical natural forest fire risk normalization information mapped by the historical meteorological data normalization information as output of the long-time memory method model, training the long-time memory method model until the trained long-time memory method model meets preset precision requirements, and obtaining a forest fire risk prediction model;
the smoothing module is specifically configured to:
extracting fire area information in the historical natural forest fire danger information to be processed;
fitting the extracted fire passing area information according to the logistic function, and determining a fitting weight of the fire passing area information in the logistic function according to a fitting result;
calculating the extracted fire area information in sequence according to the determined fitting weight of the logistic function to obtain fire area smooth information corresponding to the fire area information;
training the long-short time memory method model until the trained long-short time memory method model meets the preset precision requirement, and the method comprises the following steps:
dividing a data set consisting of historical meteorological data normalization information and historical natural forest fire risk normalization information into a training set and a testing set by adopting a K-S algorithm;
after the long-short time memory method model is trained by the training set, the trained long-short time memory method model is tested by the testing set, if the testing result meets the preset precision requirement, the training of the long-short time memory method model is stopped, and if the testing result does not meet the preset precision requirement, the long-short time memory method model continues to be trained according to the training set until the testing result meets the preset precision requirement;
normalization processing is carried out by using the following dispersion normalization formula:
in the formula (I), the compound is shown in the specification,
z is normalization information of the ith parameter;
xismoothing information for the ith parameter;
min(xi) The minimum smoothing information in the predetermined number of smoothing information before the smoothing information of the ith parameter;
max(xi) Is the maximum smoothing information among the predetermined number of smoothing information preceding the smoothing information of the i-th parameter.
6. The apparatus of claim 5, wherein the apparatus further comprises:
the model evaluation module is used for taking historical test meteorological data normalization information and vegetation type information in a preset test set as the input of the forest fire risk prediction model to obtain a prediction result, and the prediction result comprises: normalization information of the predicted fire passing area information and normalization information of the fire passing duration information;
drawing a loss function curve and a prediction result scatter diagram according to the prediction result and historical overfire area information and historical overfire duration information corresponding to the historical test meteorological data normalization information, and calculating an accuracy metric value;
and evaluating the precision of the forest fire risk prediction model according to the loss function curve, the prediction result scatter diagram and the accuracy metric value.
7. The apparatus of claim 5, wherein the apparatus further comprises:
the fire hazard prediction module is used for acquiring the current meteorological data information and vegetation type information of the target area;
and taking the current meteorological data information and vegetation type information as the input of the forest fire danger prediction model to obtain forest fire danger prediction information of the target area, wherein the forest fire danger prediction information comprises: normalization information of the fire area information and normalization information of the fire duration information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811406095.1A CN109492928B (en) | 2018-11-23 | 2018-11-23 | Method and device for constructing forest fire risk prediction model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811406095.1A CN109492928B (en) | 2018-11-23 | 2018-11-23 | Method and device for constructing forest fire risk prediction model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109492928A CN109492928A (en) | 2019-03-19 |
| CN109492928B true CN109492928B (en) | 2021-08-06 |
Family
ID=65697695
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811406095.1A Active CN109492928B (en) | 2018-11-23 | 2018-11-23 | Method and device for constructing forest fire risk prediction model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109492928B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542128A (en) * | 2010-11-19 | 2012-07-04 | 浙江农林大学 | Forest fire danger accurate zoning model for southern collective forest areas |
| CN102663082A (en) * | 2012-04-06 | 2012-09-12 | 昆明理工大学 | Forest fire forecasting method based on data mining |
| CN107085904A (en) * | 2017-03-31 | 2017-08-22 | 上海事凡物联网科技有限公司 | Forest fire danger class decision method and system based on single classification SVM |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9424521B2 (en) * | 2013-09-27 | 2016-08-23 | Transvoyant, Inc. | Computer-implemented systems and methods of analyzing spatial, temporal and contextual elements of data for predictive decision-making |
-
2018
- 2018-11-23 CN CN201811406095.1A patent/CN109492928B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542128A (en) * | 2010-11-19 | 2012-07-04 | 浙江农林大学 | Forest fire danger accurate zoning model for southern collective forest areas |
| CN102663082A (en) * | 2012-04-06 | 2012-09-12 | 昆明理工大学 | Forest fire forecasting method based on data mining |
| CN107085904A (en) * | 2017-03-31 | 2017-08-22 | 上海事凡物联网科技有限公司 | Forest fire danger class decision method and system based on single classification SVM |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109492928A (en) | 2019-03-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bakkensen et al. | Climate shocks, cyclones, and economic growth: bridging the micro-macro gap | |
| Russo et al. | Neural network forecast of daily pollution concentration using optimal meteorological data at synoptic and local scales | |
| Hennemuth et al. | Statistical methods for the analysis of simulated and observed climate data: applied in projects and institutions dealing with climate change impact and adaptation | |
| Brocca et al. | Application of a model-based rainfall-runoff database as efficient tool for flood risk management | |
| Caggiano et al. | Building loss in WUI disasters: evaluating the core components of the wildland–urban interface definition | |
| Mutahara et al. | Development of a sustainable livelihood security model for storm-surge hazard in the coastal areas of Bangladesh | |
| González-Álvarez et al. | Isohyetal maps of daily maximum rainfall for different return periods for the Colombian Caribbean Region | |
| Coffield et al. | Using remote sensing to quantify the additional climate benefits of California forest carbon offset projects | |
| CN113887972A (en) | Comprehensive drought monitoring and evaluating method based on hydrological process | |
| Guo et al. | The effect of wind averaging time on wind erosivity estimation | |
| Welker et al. | Comparing an insurer's perspective on building damages with modelled damages from pan-European winter windstorm event sets: a case study from Zurich, Switzerland | |
| Vakhare et al. | Time Series Analysis and Forecasting of Temperatures Records in Aurangabad District of Maharashtra | |
| Das et al. | Two decades of ensemble flood forecasting: a state-of-the-art on past developments, present applications and future opportunities | |
| Zonno et al. | Environmental and ambient vibration monitoring of historical adobe buildings: applications in emblematic andean churches | |
| Morikawa | Remote sensing tools for evaluating poverty alleviation projects: A case study in Tanzania | |
| Dunn et al. | Comparing extremes indices in recent observational and reanalysis products | |
| Kussul et al. | Fire danger assessment using moderate-spatial resolution satellite data | |
| CN112016744B (en) | Forest fire prediction method and device based on soil moisture and storage medium | |
| CN109492928B (en) | Method and device for constructing forest fire risk prediction model | |
| Ávila et al. | Seasonal streamflow forecast in the Tocantins River Basin, Brazil: an evaluation of ECMWF-SEAS5 with multiple conceptual hydrological models | |
| Zhao et al. | Spatiotemporal characteristics of droughts and their propagation during the past 67 years in Northern Thailand | |
| CN116091103A (en) | Method, device, electronic equipment and medium for measuring and calculating periodic environment remediation | |
| Bagalwa et al. | Spatial and seasonal patterns of rainfall erosivity in the Lake Kivu region: Insights from a meteorological observatory network | |
| CN106952026A (en) | A kind of social stability methods of risk assessment to large complicated engineering project | |
| CN111105148B (en) | Off-job probability evaluation method, apparatus and computer readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |