Method and system for identifying areas prone to thunder damage

Abstract

The invention discloses a method and a system for identifying a region prone to thunder damage, and relates to the technical field of thunder damage region identification, historical thunder damage feature data of a region to be identified is acquired, data preprocessing is performed on the historical thunder damage feature data to obtain feature factor data, the feature factor data is adopted to construct a plurality of thunder damage feature sets, and thunder damage regions are identified according to the thunder damage feature sets. The lightning damage feature set comprises a classification training set and a classification test set, constructing a decision tree model corresponding to each classification training set based on a decision tree construction algorithm, inputting the classification test set into each decision tree model for area identification, and determining an easily-occurring lightning damage area corresponding to the to-be-identified area. The problems that in the prior art, various correlation analysis methods are used for mining the influence of altitude topography or meteorological conditions of different regions on thunderstorm activity characteristic parameters, but a correlation mining method easily ignores the sequence relation between items, analysis on the difference of characteristic parameter laws of different time sequences of thunderstorm activities is lacked, and the accuracy of thunderstorm activity characteristic parameter laws is poor are solved. And the accuracy of lightning damage risk assessment is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of mine damage area identification, and in particular to a method and system for identifying a mine damage prone area. Background Art

[0002] Lightning faults are an important factor affecting the reliability of power supply of transmission lines, especially in areas with frequent lightning activities. The installation of overhead ground wires, grounding, line arresters, lightning rods and other means are still the main means of lightning protection for transmission lines. However, the cost of installing these lightning protection equipment is high. Therefore, assessing the lightning risk in different regions and carrying out differentiated lightning protection design for lines in areas prone to lightning damage has become an important method to reduce the cost of lightning protection for transmission lines.

[0003] At present, the existing technology mainly uses various correlation analysis methods to explore the impact of altitude terrain or meteorological conditions in different regions on the characteristic parameters of thunderstorm activity. However, the correlation mining method tends to ignore the sequential relationship between items and lacks analysis of the differences in the regularity of characteristic parameters of thunderstorm activities at different time series, which reduces the accuracy of lightning risk assessment. Summary of the invention

[0004] The present invention provides a method and system for identifying areas prone to lightning damage, which solves the technical problem that the prior art mainly uses various correlation analysis methods to explore the influence of altitude terrain or meteorological conditions in different regions on characteristic parameters of thunderstorm activities, but the correlation mining method easily ignores the sequential relationship between items and lacks analysis of the differences in the regularity of characteristic parameters of thunderstorm activities at different time series, thereby reducing the accuracy of lightning damage risk assessment.

[0005] The present invention provides a method for identifying a lightning-prone area, comprising:

[0006] Acquire historical lightning damage characteristic data of the area to be identified, and perform data preprocessing on the historical lightning damage characteristic data to obtain characteristic factor data;

[0007] Using the characteristic factor data, constructing a plurality of lightning damage feature sets, wherein the lightning damage feature sets include a classification training set and a classification test set;

[0008] Based on the decision tree construction algorithm, construct a decision tree model corresponding to each of the classification training sets;

[0009] The classification test sets corresponding to the classification training sets are respectively input into the decision tree models for regional identification, so as to determine the lightning-prone areas corresponding to the areas to be identified.

[0010] Optionally, the step of using the characteristic factor data to construct a plurality of lightning damage feature sets includes:

[0011] Based on the entropy weight algorithm, construct an initial decision tree model corresponding to the characteristic factor data;

[0012] Inputting each characteristic factor in the characteristic factor data into the initial decision tree model to obtain a global weight corresponding to each characteristic factor;

[0013] Screening out characteristic factors with a preset first screening number value from all the characteristic factors as first characteristic factors, and taking the remaining characteristic factors as second characteristic factors;

[0014] Sort all the first characteristic factors in sequence according to the global weights corresponding to the first characteristic factors to obtain a characteristic sequence;

[0015] The first feature factors of the second screening quantity value preset before screening out from the feature sequence are used as the first feature set, and the remaining first feature factors of the feature sequence are used as the third feature factors;

[0016] A plurality of classification training sets and a plurality of classification test sets are constructed according to the first feature set, all the third feature factors and all the second feature factors.

[0017] Optionally, the step of constructing multiple classification training sets and multiple classification test sets according to the first feature set, all the third feature factors and all the second feature factors includes:

[0018] Selecting the minimum value of the global weights corresponding to the first feature factors in the first feature set as a reference weight;

[0019] Taking each of the second characteristic factors and each of the third characteristic factors as intermediate characteristic factors;

[0020] Determining whether the global weight value of each of the intermediate feature factors is greater than or equal to the reference weight;

[0021] When the global weight value of the intermediate characteristic factor is greater than or equal to the reference weight value, the intermediate characteristic factor is determined as an important characteristic factor;

[0022] Determine a corresponding decision quantity value according to the first feature set, all the third feature factors, all the important feature factors and a preset classification accuracy function;

[0023] The first feature set and all important feature factors are selected using a sampling method with replacement to construct a classification training set and a classification test set for the decision quantity value.

[0024] Optionally, the step of inputting the classification test sets corresponding to the classification training sets into the decision tree models for region identification to determine the lightning-prone area corresponding to the area to be identified includes:

[0025] Inputting the classification test sets corresponding to the classification training sets into the decision tree models for interval recognition, and outputting the initial decision results corresponding to the decision tree models;

[0026] The average method is used to average all the initial decision results to obtain the lightning-prone area corresponding to the area to be identified.

[0027] Optionally, the step of determining a corresponding decision quantity value according to the first feature set, all the third feature factors, all the important feature factors and a preset classification accuracy function comprises:

[0028] Performing mean processing on the global weights corresponding to the respective third characteristic factors to obtain a target mean;

[0029] Calculate the standard deviation of the global weights corresponding to each of the third characteristic factors;

[0030] Multiplying the standard deviation by a preset standard constant value to obtain a target multiplication value;

[0031] Performing difference processing on the target mean and the target multiplication value to obtain a target difference;

[0032] Determine whether the global weight corresponding to each of the third characteristic factors is less than the target difference;

[0033] When the global weight corresponding to the third characteristic factor is less than the target difference, the third characteristic factor is determined as an irrelevant factor;

[0034] The first feature set, all the irrelevant factors, all the third feature factors and all the important feature factors are input into a preset classification accuracy function to obtain corresponding decision quantity values.

[0035] Optionally, the step of acquiring historical lightning damage characteristic data of the area to be identified and performing data preprocessing on the historical lightning damage characteristic data to obtain characteristic factor data includes:

[0036] Acquire historical lightning damage characteristic data of the area to be identified, and grid the historical lightning damage characteristic data to obtain a plurality of grid data;

[0037] Normalizing all the grid data to obtain a plurality of intermediate grid data;

[0038] Sorting all the intermediate grid data from large to small according to the density of ground lightning in the intermediate grid data to obtain a grid sequence;

[0039] Filter out the intermediate grid data of a third preset screening quantity value from the grid sequence as the grid data set;

[0040] The grid data set is discretized to obtain characteristic factor data.

[0041] The second invention of the present invention provides a system for identifying areas susceptible to lightning damage, comprising:

[0042] A preprocessing module is used to obtain historical lightning damage characteristic data of the area to be identified, and perform data preprocessing on the historical lightning damage characteristic data to obtain characteristic factor data;

[0043] A first construction module is used to construct a plurality of lightning damage feature sets using the feature factor data, wherein the lightning damage feature sets include a classification training set and a classification test set;

[0044] A second construction module is used to construct a decision tree model corresponding to each of the classification training sets based on a decision tree construction algorithm;

[0045] The identification module is used to input the classification test sets corresponding to each of the classification training sets into each of the decision tree models for regional identification, and determine the lightning-prone area corresponding to the area to be identified.

[0046] A third aspect of the present invention provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for identifying a lightning-prone area as described in any one of the above items.

[0047] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the method for identifying a lightning-prone area as described in any one of the above items.

[0048] It can be seen from the above technical solutions that the present invention has the following advantages:

[0049] The present invention aims at identifying areas prone to lightning damage, takes historical lightning damage characteristic data of the area to be identified as the analysis object, constructs multiple decision tree models based on the historical lightning damage characteristic data, and determines the lightning damage prone area corresponding to the area to be identified according to the identification results of each decision tree model, and can accurately identify the lightning damage prone area of ​​the area to be identified. Compared with the correlation analysis method, the present invention performs regional identification based on multiple decision tree models, can quickly process large-scale data sets, and can obtain accurate identification results, which is more conducive to the assessment of lightning damage risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0051] Figure 1 A flowchart of a method for identifying a lightning-prone area provided in Embodiment 1 of the present invention;

[0052] Figure 2 A flowchart of a method for identifying a lightning-prone area provided in Embodiment 2 of the present invention;

[0053] Figure 3 A flowchart of the steps of discretization processing provided in the second embodiment of the present invention;

[0054] Figure 4 A structural block diagram of a system for identifying areas susceptible to lightning damage provided by Embodiment 3 of the present invention;

[0055] Figure 5 This is a structural block diagram of a computer device provided in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0056] The embodiment of the present invention provides a method and system for identifying areas prone to lightning damage, which is used to solve the technical problem that the existing technology mainly uses various correlation analysis methods to explore the influence of altitude terrain or meteorological conditions in different regions on the characteristic parameters of thunderstorm activity, but the correlation mining method easily ignores the sequential relationship between items and lacks analysis of the differences in the regularity of characteristic parameters of thunderstorm activities at different time series, thereby reducing the accuracy of lightning damage risk assessment.

[0057] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] See also Figure 1 , Figure 1 This is a flowchart of a method for identifying areas prone to lightning damage provided in Example 1 of the present invention.

[0059] The present invention provides a method for identifying a lightning-prone area, comprising:

[0060] Step 101, obtaining historical lightning damage characteristic data of the area to be identified, and performing data preprocessing on the historical lightning damage characteristic data to obtain characteristic factor data;

[0061] Historical lightning damage characteristic data refers to the ground lightning data, meteorological data and topographic data of the lightning location system in the area to be identified within the preset time. The preset time can be one day, 6 hours, 12 hours or 30 minutes.

[0062] In an embodiment of the present invention, historical lightning damage characteristic data of a region to be identified is obtained, and the historical lightning damage characteristic data is gridded, normalized and discretized in sequence to obtain characteristic factor data.

[0063] Step 102: using the characteristic factor data to construct a plurality of lightning damage feature sets, wherein the lightning damage feature sets include a classification training set and a classification test set;

[0064] In an embodiment of the present invention, an initial decision tree model corresponding to the feature factor data is generated based on the entropy weight method, and each feature factor in the feature factor data is input into the initial decision tree model to obtain the global weight corresponding to each feature factor. According to the global weight of each feature factor, multiple classification training sets and classification test sets are constructed.

[0065] In another embodiment, an initial decision tree model corresponding to the feature factor data is generated based on the ID3 algorithm, each feature factor in the feature factor data is input into the initial decision tree model to obtain the global weight corresponding to each feature factor, and multiple classification training sets and classification test sets are constructed according to the global weight of each feature factor.

[0066] Step 103: construct a decision tree model corresponding to each classification training set based on a decision tree construction algorithm;

[0067] In an embodiment of the present invention, a CART algorithm (decision tree construction algorithm) is used to perform decision tree modeling on each classification training set; for example, there are m training sample sets, and decision tree models are constructed for the m training sample sets respectively to obtain m classification results; specifically, each decision tree selects v features from u features of the input variable, generally u=v, and then uses information entropy (entropy) and Gini index (Gini) as node splitting criteria.

[0068] It is worth mentioning that the CART algorithm is a decision tree-based machine learning algorithm, mainly used for classification and regression tasks. It builds a decision tree by recursively dividing the data set into smaller subsets, each of which is based on maximizing information gain or reducing the impurity of the target variable.

[0069] Step 104: Input the classification test sets corresponding to the classification training sets into the decision tree models to perform regional identification, and determine the lightning-prone areas corresponding to the areas to be identified.

[0070] In an embodiment of the present invention, the classification test sets corresponding to the classification training sets are respectively input into the decision tree models for regional identification, and multiple identification results are output. The average method is used to average the multiple identification results to obtain the lightning-prone area corresponding to the area to be identified.

[0071] It should be noted that the averaging method refers to performing an average operation on multiple recognition results.

[0072] In an embodiment of the present invention, historical lightning damage feature data of a region to be identified is obtained, and data preprocessing is performed on the historical lightning damage feature data to obtain feature factor data, and the feature factor data is used to construct multiple lightning damage feature sets, wherein the lightning damage feature set includes a classification training set and a classification test set, and a decision tree model corresponding to each classification training set is constructed based on a decision tree construction algorithm, and the classification test set corresponding to each classification training set is respectively input into each decision tree model for regional identification, and the lightning damage prone area corresponding to the region to be identified is determined. The prior art uses various correlation analysis methods to mine the influence of altitude terrain or meteorological conditions in different regions on the characteristic parameters of thunderstorm activities, but the correlation mining method easily ignores the order relationship between items, lacks analysis of the regular differences of characteristic parameters of thunderstorm activities at different time series, and reduces the accuracy of lightning damage risk assessment. The present invention is aimed at the identification of areas prone to lightning damage, takes the historical lightning damage feature data of the region to be identified as the analysis object, constructs multiple decision tree models based on the historical lightning damage feature data, and determines the lightning damage prone area corresponding to the region to be identified according to the identification results of each decision tree model, and can accurately identify the lightning damage prone area of ​​the region to be identified. Compared with the correlation analysis method, the method of the present invention performs region identification based on multiple decision tree models, can quickly process large-scale data sets, and can obtain accurate identification results, which is more conducive to the assessment of mine damage risks.

[0073] See also Figure 2 , Figure 2 This is a flowchart of a method for identifying areas prone to lightning damage provided in Embodiment 2 of the present invention.

[0074] The present invention provides a method for identifying a lightning-prone area, comprising:

[0075] Step 201, obtaining historical lightning damage characteristic data of the area to be identified, and performing data preprocessing on the historical lightning damage characteristic data to obtain characteristic factor data;

[0076] Further, step 201 includes the following sub-steps:

[0077] S11, obtaining historical lightning damage characteristic data of the area to be identified, and gridding the historical lightning damage characteristic data to obtain a plurality of grid data;

[0078] In an embodiment of the present invention, historical lightning damage characteristic data of the area to be identified is obtained (the historical lightning damage characteristic data includes but is not limited to temperature, humidity, wind speed, topographic data, lightning strike location, lightning current amplitude, polarity, number of return strokes, etc.), and the historical lightning damage characteristic data is gridded to obtain multiple grid data.

[0079] It should be noted that the purpose of gridding is to use various types of data in each grid as samples for subsequent model building and training.

[0080] It is worth mentioning that when gridding, taking into account the size of the high-altitude area under study and the characteristics of various types of original data, a 1km square grid is used to count the density of ground lightning (unit: times / square kilometer / year), and at the same time, a 1km square grid is used to count the proportion of positive ground lightning, and a 1km square grid is used to count the mode value of the number of ground lightning return strokes (which can represent that the number of return strokes of the ground lightning points in this grid is mostly a few times) (a smaller grid is used to refine the characteristics of the above parameters, and if a larger grid is used, the characteristics of the above parameters may not be obvious), and a 5km square grid is used to count the probability density distribution of the lightning current amplitude, and the lightning current amplitude with the highest probability density is taken as the representative value of the grid (the larger the grid, the more Obtaining more flash points, the statistics of the probability density distribution of their amplitudes will be more meaningful and operational, and due to the random dispersion of lightning current amplitudes, there is no need to take small grids for feature refinement), taking a 3km square grid to count the average value of topographic data obtained by the geographic information system (the grid size is a multiple of the grid size of the geographic information system data, which is convenient for operation, and the grid size can be adjusted according to the actual source data situation), taking a 3km square grid to count the average value of various meteorological data (the grid size is consistent with the aforementioned terrain data, which is convenient for subsequent models to use it as input together with topographic data), if it is a vector (such as wind speed, etc.), it will be decomposed in the spatial coordinate system to form east-west and north-south components.

[0081] S12, normalizing all grid data to obtain a plurality of intermediate grid data;

[0082] In the embodiment of the present invention, each grid data is normalized to obtain a plurality of intermediate grid data.

[0083] It is worth mentioning that the normalization process is as follows: Normalized value = (actual value - minimum value) / maximum value

[0084] It should be noted that the purpose of normalization is to eliminate different dimensions.

[0085] S13, sorting all the intermediate grid data from large to small according to the density of ground-to-ground lightning in the intermediate grid data to obtain a grid sequence;

[0086] In the embodiment of the present invention, the intermediate grid data are sorted from large to small according to the density of ground lightning in each intermediate grid data to obtain a grid sequence.

[0087] S14, selecting intermediate grid data of a third preset screening quantity value from the grid sequence as a grid data set;

[0088] In the embodiment of the present invention, the first 30% of the intermediate grid data are screened out from the grid sequence as the grid data set.

[0089] S15. Discretize the grid data set to obtain characteristic factor data.

[0090] In the embodiment of the present invention, refer to Figure 3 As shown in the figure, the grid data set is used as a minefield area, and the topographic features and meteorological factor parameters (collectively referred to as environmental factors) of the minefield area are counted. The numerical distribution range of each environmental factor in the entire study area is counted. If the numerical distribution range of each environmental factor in the minefield area accounts for more than 50% of the entire numerical distribution range, it is determined that the environmental factor has a low degree of influence on the occurrence of lightning ground flashes. The environmental factor is discretized according to the distribution proportion of 0-25%, 25%-50%, 50%-75%, and 75%-100%, that is, it is discretized into four categories of X1-X4. The factor value is at the maximum value of 0-25% and is divided into X1, and so on. If the numerical distribution range of each environmental factor in a mine-prone area accounts for less than 50% of the entire numerical distribution range, the variance of the numerical distribution of the environmental factor is calculated. If the variance is less than or equal to 0.05, it is discretized into six categories from X1 to X6; if the variance is greater than 0.05 and less than or equal to 0.5, it is discretized into eight categories from X1 to X8; if the variance is greater than 0.5, it is discretized into ten categories from X1 to X10.

[0091] Step 202: using the characteristic factor data to construct a plurality of lightning damage feature sets, wherein the lightning damage feature sets include a classification training set and a classification test set;

[0092] Further, step 202 includes the following sub-steps:

[0093] S21. Based on the entropy weight algorithm, construct an initial decision tree model corresponding to the feature factor data;

[0094] In the embodiment of the present invention, an entropy weight algorithm is used to construct an initial decision tree model corresponding to the characteristic factor data (the characteristic factor data includes multiple characteristic factors, and the characteristic factors may be topography, meteorological factor parameters, etc.).

[0095] It should be noted that the entropy weight algorithm is specifically:

[0096] Let N be the total number of nodes in the tree τ. First, assign a local weight to the feature factor j:

[0097]

[0098] Q(i,j)=exp{-(E l +E r )}

[0099] Among them, Q(i,j) is the splitting quality of characteristic factor j, i is the splitting node of characteristic factor j, and E l is the left child of node i, E r is the right child node of node i, N is the total number of nodes in the initial decision tree model, and w τ (j) is the local weight of feature factor j.

[0100] Let δ τ is the sampling error of the initial decision tree model τ, then the normalized weight of the initial decision tree model τ is:

[0101]

[0102] Among them, δ τ is the sampling error of the initial decision tree model τ, γ τ is the normalized weight of the initial decision tree model τ.

[0103] The global weight of feature factor j is:

[0104]

[0105] Among them, w(j) is the global weight of feature factor j.

[0106] S22, inputting each characteristic factor in the characteristic factor data into the initial decision tree model to obtain a global weight corresponding to each characteristic factor;

[0107] In the embodiment of the present invention, each characteristic factor in the characteristic factor data is used as an input of the initial decision tree model to obtain a global weight corresponding to each characteristic factor.

[0108] S23, selecting characteristic factors with a preset first screening number value from all characteristic factors as first characteristic factors, and selecting the remaining characteristic factors as second characteristic factors;

[0109] In the embodiment of the present invention, characteristic factors of a preset first screening number value are randomly screened out from all characteristic factors as first characteristic factors, and the remaining characteristic factors are used as second characteristic factors.

[0110] S24, sorting all first characteristic factors in sequence according to the global weights corresponding to the first characteristic factors to obtain a characteristic sequence;

[0111] In the embodiment of the present invention, the first characteristic factors are sorted from large to small according to the global weights corresponding to the first characteristic factors to obtain a characteristic sequence.

[0112] S25, selecting the first characteristic factors of the second screening quantity value preset before from the characteristic sequence as the first characteristic set, and taking the remaining first characteristic factors of the characteristic sequence as the third characteristic factors;

[0113] In an embodiment of the present invention, the first feature factors of the second screening quantity value preset before screening out from the feature sequence are used as the first feature set (the first feature set is an important feature set), and the remaining first feature factors of the feature sequence are used as the third feature factors (the third feature factors are unimportant feature factors).

[0114] S26. Construct multiple classification training sets and multiple classification test sets based on the first feature set, all third feature factors and all second feature factors.

[0115] Further, S26 includes the following sub-steps:

[0116] S261, selecting the minimum value of the global weights corresponding to the first feature factors in the first feature set as a reference weight;

[0117] In the embodiment of the present invention, the smallest global weight in the first feature set is selected as the reference weight.

[0118] S262, taking each second characteristic factor and each third characteristic factor as an intermediate characteristic factor;

[0119] S263, judging whether the global weight value of each intermediate characteristic factor is greater than or equal to the reference weight value;

[0120] In the embodiment of the present invention, each second characteristic factor and each third characteristic factor are used as an intermediate characteristic factor, and it is determined whether the global weight value corresponding to each intermediate characteristic factor is greater than or equal to the reference weight value.

[0121] S264. When the global weight value of the intermediate characteristic factor is greater than or equal to the reference weight value, the intermediate characteristic factor is determined as an important characteristic factor;

[0122] In the embodiment of the present invention, when the global weight value of the intermediate feature factor is greater than or equal to the reference weight value, the intermediate feature factor is marked as an important feature factor.

[0123] S265, determining a corresponding decision quantity value according to the first feature set, all third feature factors, all important feature factors and a preset classification accuracy function;

[0124] Furthermore, S265 includes the following sub-steps:

[0125] A1. Perform mean processing on the global weights corresponding to each third characteristic factor to obtain the target mean;

[0126] In the embodiment of the present invention, the mean values ​​of the third characteristic factors are calculated as the target mean values.

[0127] A2. Calculate the standard deviation of the global weights corresponding to each third characteristic factor;

[0128] In the embodiment of the present invention, the standard deviation of the global weight of each third characteristic factor is calculated.

[0129] A3. Multiply the standard deviation by the preset standard constant value to obtain the target multiplication value.

[0130] A4. Perform difference processing on the target mean and the target multiplication value to obtain the target difference;

[0131] In the embodiment of the present invention, the standard deviation is multiplied by a preset standard constant value to obtain a target multiplication value, and the target mean is subtracted from the target multiplication value to obtain a target difference value (μ n -2σ n ). Among them, μ n is the target mean, σ n is the standard deviation.

[0132] A5. Determine whether the global weight corresponding to each third characteristic factor is less than the target difference;

[0133] In the embodiment of the present invention, it is determined whether the global weight corresponding to each third characteristic factor is less than the target difference (μ n -2σ n ).

[0134] A6. When the global weight corresponding to the third characteristic factor is less than the target difference, the third characteristic factor is determined as an irrelevant factor;

[0135] In the embodiment of the present invention, when the global weight corresponding to the third eigenfactor is smaller than the target difference, the third eigenfactor is determined as an irrelevant factor (ie, the third eigenfactor smaller than the target difference is eliminated).

[0136] A7. Input the first feature set, all irrelevant factors, all third feature factors and all important feature factors into a preset classification accuracy function to obtain corresponding decision quantity values.

[0137] In the embodiment of the present invention, the first feature set, all irrelevant factors, all third feature factors and all important feature factors are used as inputs of a preset classification accuracy function to obtain corresponding decision quantity values.

[0138] It should be noted that the classification accuracy function is specifically:

[0139]

[0140] dη≈λ(θΔB+lq u Δu+lq v Δv)

[0141]

[0142] Among them, u is the sum of the number of the first feature set and all important feature factors, v is the sum of the number of all third factors, q is the probability that node i selects the third feature factor or the first feature factor in the first feature set, r is the probability that node i selects the third feature factor, f is the selection constant, Δu is the number of important feature factors, Δv is the sum of the number of irrelevant factors and important feature factors, B is the initial decision number value, N av is the number of decision nodes, η s is the decision strength value, ρ′ is the common probability value of the tree, ρ is the common probability value of the node, η c is the common probability value of the node pair, η is the classification accuracy value, θ is the first intermediate value, l is the second intermediate value, dη is the total differential of the classification accuracy value, q u Find the partial derivative of q with respect to u, q v Find the partial derivative of q with respect to v, Δr is, ΔB is the decision change,

[0143] S266. Use the sampling method with replacement to select data from the first feature set and all important feature factors, and construct a classification training set and a classification test set for decision quantity values.

[0144] In an embodiment of the present invention, the Bootstrap method (sampling with replacement) is used to randomly select 80% of the sample data from the first feature set and all important feature factors to construct a classification training set for decision quantity values, and the remaining 20% ​​to construct a classification test set.

[0145] Step 203: construct a decision tree model corresponding to each classification training set based on a decision tree construction algorithm;

[0146] In the embodiment of the present invention, the CART algorithm (decision tree construction algorithm) is used to perform decision tree modeling on each classification training set to obtain a decision tree model corresponding to each classification training set.

[0147] Step 204: input the classification test sets corresponding to the classification training sets into the decision tree models for interval recognition, and output the initial decision results corresponding to the decision tree models;

[0148] In the embodiment of the present invention, the classification test set is input into each decision tree model for interval recognition, and the initial decision result corresponding to each decision tree model is obtained.

[0149] Step 205: use the averaging method to average all the initial decision results to obtain the lightning-prone area corresponding to the area to be identified.

[0150] In the embodiment of the present invention, an averaging method is used to average all initial decision results to obtain the lightning-prone area corresponding to the area to be identified.

[0151] In one embodiment, when new historical lightning damage characteristic data is received, the random forest model is updated using the method of steps 201 to 205 above, thereby obtaining lightning damage-prone areas in the area to be identified.

[0152] In an embodiment of the present invention, historical lightning damage feature data of a region to be identified is obtained, and data preprocessing is performed on the historical lightning damage feature data to obtain feature factor data, and the feature factor data is used to construct multiple lightning damage feature sets, wherein the lightning damage feature set includes a classification training set and a classification test set, and a decision tree model corresponding to each classification training set is constructed based on a decision tree construction algorithm, and the classification test set corresponding to each classification training set is respectively input into each decision tree model for regional identification, and the lightning damage prone area corresponding to the region to be identified is determined. The prior art uses various correlation analysis methods to mine the influence of altitude terrain or meteorological conditions in different regions on the characteristic parameters of thunderstorm activities, but the correlation mining method easily ignores the order relationship between items, lacks analysis of the regular differences of characteristic parameters of thunderstorm activities at different time series, and reduces the accuracy of lightning damage risk assessment. The present invention is aimed at the identification of areas prone to lightning damage, takes the historical lightning damage feature data of the region to be identified as the analysis object, constructs multiple decision tree models based on the historical lightning damage feature data, and determines the lightning damage prone area corresponding to the region to be identified according to the identification results of each decision tree model, and can accurately identify the lightning damage prone area of ​​the region to be identified. Compared with the correlation analysis method, the method of the present invention performs region identification based on multiple decision tree models, can quickly process large-scale data sets, and can obtain accurate identification results, which is more conducive to the assessment of mine damage risks.

[0153] In the embodiment of the present invention, the specific implementation process of step 201 is similar to step 101 and will not be repeated here.

[0154] See also Figure 4 , Figure 4This is a structural block diagram of a system for identifying areas prone to lightning damage provided by Embodiment 3 of the present invention.

[0155] The present invention provides a system for identifying areas susceptible to lightning damage, comprising:

[0156] The preprocessing module 301 is used to obtain historical lightning damage characteristic data of the area to be identified, and perform data preprocessing on the historical lightning damage characteristic data to obtain characteristic factor data;

[0157] The first construction module 302 is used to construct a plurality of lightning damage feature sets using feature factor data, wherein the lightning damage feature sets include a classification training set and a classification test set;

[0158] The second construction module 303 is used to construct a decision tree model corresponding to each classification training set based on a decision tree construction algorithm;

[0159] The identification module 304 is used to input the classification test sets corresponding to each classification training set into each decision tree model for regional identification, and determine the lightning-prone area corresponding to the area to be identified.

[0160] Furthermore, the preprocessing module 301 includes:

[0161] The gridding submodule is used to obtain the historical lightning damage characteristic data of the area to be identified, and grid the historical lightning damage characteristic data to obtain multiple grid data;

[0162] The normalization submodule is used to normalize all grid data to obtain multiple intermediate grid data;

[0163] The first sorting submodule is used to sort all the intermediate grid data from large to small according to the density of ground lightning in the intermediate grid data to obtain a grid sequence;

[0164] A first screening submodule is used to screen out intermediate grid data of a third pre-set screening quantity value from the grid sequence as a grid data set;

[0165] The discretization submodule is used to discretize the grid data set to obtain characteristic factor data.

[0166] Furthermore, the first building block 302 includes:

[0167] The first construction submodule is used to construct an initial decision tree model corresponding to the feature factor data based on an entropy weight algorithm;

[0168] The global weight submodule is used to input each characteristic factor in the characteristic factor data into the initial decision tree model to obtain the global weight corresponding to each characteristic factor;

[0169] A second screening submodule is used to screen out the characteristic factors with a preset first screening quantity value from all the characteristic factors as the first characteristic factors, and to use the remaining characteristic factors as the second characteristic factors;

[0170] The second arrangement submodule is used to sort all the first characteristic factors from large to small according to the global weights corresponding to the first characteristic factors to obtain a characteristic sequence;

[0171] A third screening submodule is used to screen out first feature factors of a preset second screening quantity value from the feature sequence as a first feature set, and to use the remaining first feature factors of the feature sequence as third feature factors;

[0172] The analysis submodule is used to construct multiple classification training sets and multiple classification test sets according to the first feature set, all third feature factors and all second feature factors.

[0173] Furthermore, the analysis submodule includes:

[0174] A first selection unit, used to select the minimum value of the global weights corresponding to the first feature factors in the first feature set as a reference weight;

[0175] A second selection unit is used to use each second characteristic factor and each third characteristic factor as an intermediate characteristic factor;

[0176] A first analysis unit, used to determine whether the global weight value of each intermediate characteristic factor is greater than or equal to the reference weight value;

[0177] When the global weight value of the intermediate feature factor is greater than or equal to the reference weight value, the intermediate feature factor is determined as an important feature factor;

[0178] A second analysis unit is used to determine a corresponding decision quantity value according to the first feature set, all third feature factors, all important feature factors and a preset classification accuracy function;

[0179] The construction unit is used to select data from the first feature set and all important feature factors by adopting a sampling method with replacement, and to construct a classification training set and a classification test set of decision quantity values.

[0180] Furthermore, the second analysis unit comprises:

[0181] The first analysis subunit is used to average the global weights corresponding to the third characteristic factors to obtain a target average;

[0182] The second analysis subunit is used to calculate the standard deviation of the global weight corresponding to each third characteristic factor;

[0183] The standard deviation is multiplied by a preset standard constant value to obtain a target multiplication value;

[0184] The third analysis subunit is used to perform difference processing on the target mean and the target multiplication value to obtain a target difference;

[0185] The fourth analysis subunit is used to determine whether the global weight corresponding to each third characteristic factor is less than the target difference;

[0186] A fifth analysis subunit, configured to determine the third characteristic factor as an irrelevant factor when the global weight corresponding to the third characteristic factor is less than the target difference;

[0187] The first feature set, all irrelevant factors, all third feature factors and all important feature factors are input into a preset classification accuracy function to obtain corresponding decision quantity values.

[0188] Furthermore, the identification module 304 includes:

[0189] The recognition submodule is used to input the classification test sets corresponding to each classification training set into each decision tree model for interval recognition, and output the initial decision results corresponding to each decision tree model;

[0190] The mean submodule is used to process all the initial decision results by averaging method to obtain the lightning-prone area corresponding to the area to be identified.

[0191] See also Figure 5 , Figure 5 This is a structural block diagram of a computer device provided in Embodiment 4 of the present invention.

[0192] An electronic device according to an embodiment of the present invention includes: a memory 401 and a processor 402, wherein the memory 402 stores a computer program; when the computer program is executed by the processor 402, the processor 402 executes a method for identifying a lightning-prone area according to any of the above embodiments.

[0193] The memory 401 can be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk or a ROM. The memory 401 has a storage space 403 for a program code 413 for executing any method step in the above method. For example, the storage space 403 for the program code can include individual program codes 413 for implementing the various steps in the above method respectively. These program codes can be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks. The program code can be compressed, for example, in an appropriate form. These codes, when run by a computing and processing device, cause the computing and processing device to perform the various steps in the method described above. These program codes can be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks. The program code can be compressed, for example, in an appropriate form. When these codes are executed by a computing and processing device, they cause the computing and processing device to execute the various steps of the method for identifying a lightning-prone area described above.

[0194] Embodiment 5 of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method for identifying a lightning-prone area as described in any of the above embodiments is implemented.

[0195] Embodiment 6 of the present invention further provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer executes a method for identifying a lightning-prone area as described in any of the above embodiments.

[0196] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0197] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0198] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0199] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0200] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0201] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying areas prone to lightning damage, characterized in that: include: Acquire historical lightning damage characteristic data of the area to be identified, and perform data preprocessing on the historical lightning damage characteristic data to obtain characteristic factor data; Using the characteristic factor data, constructing a plurality of lightning damage feature sets, wherein the lightning damage feature sets include a classification training set and a classification test set; Based on the decision tree construction algorithm, construct a decision tree model corresponding to each of the classification training sets; The classification test sets corresponding to the classification training sets are respectively input into the decision tree models for regional identification, so as to determine the lightning-prone areas corresponding to the areas to be identified.

2. The method for identifying areas susceptible to lightning damage according to claim 1, characterized in that: The step of using the characteristic factor data to construct a plurality of lightning damage feature sets comprises: Based on the entropy weight algorithm, construct an initial decision tree model corresponding to the characteristic factor data; Inputting each characteristic factor in the characteristic factor data into the initial decision tree model to obtain a global weight corresponding to each characteristic factor; Screening out characteristic factors with a preset first screening number value from all the characteristic factors as first characteristic factors, and taking the remaining characteristic factors as second characteristic factors; Sort all the first characteristic factors in sequence according to the global weights corresponding to the first characteristic factors to obtain a characteristic sequence; The first feature factors of the second screening quantity value preset before screening out from the feature sequence are used as the first feature set, and the remaining first feature factors of the feature sequence are used as the third feature factors; A plurality of classification training sets and a plurality of classification test sets are constructed according to the first feature set, all the third feature factors and all the second feature factors.

3. The method for identifying areas susceptible to lightning damage according to claim 2, characterized in that: The step of constructing a plurality of classification training sets and a plurality of classification test sets according to the first feature set, all the third feature factors and all the second feature factors comprises: Selecting the minimum value of the global weights corresponding to the first feature factors in the first feature set as a reference weight; Taking each of the second characteristic factors and each of the third characteristic factors as intermediate characteristic factors; Determining whether the global weight value of each of the intermediate feature factors is greater than or equal to the reference weight; When the global weight value of the intermediate characteristic factor is greater than or equal to the reference weight value, the intermediate characteristic factor is determined as an important characteristic factor; Determine a corresponding decision quantity value according to the first feature set, all the third feature factors, all the important feature factors and a preset classification accuracy function; The first feature set and all important feature factors are selected using a sampling method with replacement to construct a classification training set and a classification test set for the decision quantity value.

4. The method for identifying areas susceptible to lightning damage according to claim 1, characterized in that: The step of inputting the classification test sets corresponding to the classification training sets into the decision tree models to perform regional identification and determine the lightning-prone areas corresponding to the areas to be identified includes: Inputting the classification test sets corresponding to the classification training sets into the decision tree models for interval recognition, and outputting the initial decision results corresponding to the decision tree models; The average method is used to average all the initial decision results to obtain the lightning-prone area corresponding to the area to be identified.

5. The method for identifying areas susceptible to lightning damage according to claim 3, characterized in that: The step of determining the corresponding decision quantity value according to the first feature set, all the third feature factors, all the important feature factors and the preset classification accuracy function comprises: Performing mean processing on the global weights corresponding to the respective third characteristic factors to obtain a target mean; Calculate the standard deviation of the global weights corresponding to each of the third characteristic factors; Multiplying the standard deviation by a preset standard constant value to obtain a target multiplication value; Performing difference processing on the target mean and the target multiplication value to obtain a target difference; Determine whether the global weight corresponding to each of the third characteristic factors is less than the target difference; When the global weight corresponding to the third characteristic factor is less than the target difference, the third characteristic factor is determined as an irrelevant factor; The first feature set, all the irrelevant factors, all the third feature factors and all the important feature factors are input into a preset classification accuracy function to obtain corresponding decision quantity values.

6. The method for identifying areas susceptible to lightning damage according to claim 1, characterized in that: The step of obtaining historical lightning damage characteristic data of the area to be identified and performing data preprocessing on the historical lightning damage characteristic data to obtain characteristic factor data includes: Acquire historical lightning damage characteristic data of the area to be identified, and grid the historical lightning damage characteristic data to obtain a plurality of grid data; Normalizing all the grid data to obtain a plurality of intermediate grid data; Sorting all the intermediate grid data from large to small according to the density of ground lightning in the intermediate grid data to obtain a grid sequence; Filter out the intermediate grid data of a third preset screening quantity value from the grid sequence as the grid data set; The grid data set is discretized to obtain characteristic factor data.

7. A system for identifying areas prone to lightning damage, characterized in that: include: A preprocessing module is used to obtain historical lightning damage characteristic data of the area to be identified, and perform data preprocessing on the historical lightning damage characteristic data to obtain characteristic factor data; A first construction module is used to construct a plurality of lightning damage feature sets using the feature factor data, wherein the lightning damage feature sets include a classification training set and a classification test set; A second construction module is used to construct a decision tree model corresponding to each of the classification training sets based on a decision tree construction algorithm; The identification module is used to input the classification test sets corresponding to each of the classification training sets into each of the decision tree models for regional identification, and determine the lightning-prone area corresponding to the area to be identified.

8. An electronic device, characterized in that: The method comprises a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the method for identifying a lightning-prone area according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the method for identifying a lightning-prone area as described in any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program comprises program instructions, wherein when the program instructions are executed by a computer, the computer is caused to execute the method for identifying a lightning-prone area according to any one of claims 1 to 6.

Images