Mining area multi-dimensional data statistics and intelligent analysis management system

Abstract

The invention relates to the technical field of mining area management, solves the technical problems of low robustness, large data quantity and difficulty in efficient processing in data acquisition and data classification in the prior art, and particularly relates to a mining area multi-dimensional data statistics and intelligent analysis management system. The system is composed of a preprocessing module, a conversion module, a classification module, an optimization module, a risk fusion module and an early warning module, and the preprocessing module is used for collecting original data of a mining area and preprocessing the original data to obtain preprocessed data; the data are analyzed and classified through dynamic weights, and compared with a mode of using fixed weights in the prior art, fixed weight intervention can be avoided, neglecting of data features of the data can be avoided, the accuracy of data processing can be improved, and the data processing efficiency can be improved. In addition, help can be provided for subsequent steps, and the overall processing efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of mining area management, and in particular to a mining area multi-dimensional data statistics and intelligent analysis management system. Background Art

[0002] Multidimensional data from mining areas generally covers five core dimensions: geology, production, equipment, environment, and safety. Geological data comes from 3D geological modeling and remote sensing imagery, production data from production planning systems and ore metering devices, equipment data from IoT sensors and equipment monitoring systems, environmental data from environmental monitoring sensor networks and water quality testing instruments, and safety data from safety monitoring systems and personnel location tags. By processing these diverse data sets, multiple aspects of the mining area can be analyzed and managed. Existing technologies for acquiring multidimensional data generally rely on sensors and other devices, without any processing of the large amounts of data. Due to the sheer volume of data, processing is extremely time-consuming. Conventional data classification methods suffer from low robustness and are easily affected by abnormal data, impacting the accuracy of data classification. Summary of the Invention

[0003] In response to the shortcomings of the existing technology, the present invention provides a multi-dimensional data statistics and intelligent analysis management system for mining areas, which solves the technical problems of the existing technology in low robustness in data acquisition and data classification, and the difficulty in efficiently processing large amounts of data. It achieves the purpose of improving the robustness of data classification and being able to process large amounts of data quickly and efficiently.

[0004] To solve the above technical problems, the present invention provides the following technical solutions: a mining area multi-dimensional data statistics and intelligent analysis management system, the system comprising:

[0005] Pre-processing module, which is used to collect the raw data S of the mining area a , for the original data S a Performing preprocessing to obtain preprocessed data;

[0006] The conversion module is used to convert the preprocessed data into a preprocessing matrix Y through a matrix conversion method. ij , and according to the preprocessing matrix Y ij Calculate the forward solution and negative solution

[0007] Classification module, the classification module is used to and negative solution Calculate the approximation value T J , for the close value T J Arrange in descending order to obtain descending sequence J xl, and for the descending sequence J xl Classify and obtain a data classification set;

[0008] Optimization module, the optimization module is used to classify the data based on the set S1 j Calculate the weighted eigenvalue F used to build the preliminary tree model q , and optimize the preliminary tree model to obtain the optimized tree model sh;

[0009] Risk fusion module, the risk fusion module is used to obtain multiple probabilities by optimizing the tree model sh According to probability Calculate risk fusion value

[0010] The early warning module is used to integrate multiple risk values Calculate the warning threshold β used to generate warning information e and send early warning information to the management center of the mining area.

[0011] Furthermore, the preprocessing module includes:

[0012] Mean module, the mean module is used to convert the original data S of the mining area a Arrange into a data sequence, create a detection window with a window size of L on the data sequence, the step size of the detection window is μ, and calculate the mean of the data sequence in the detection window The calculation formula is:

[0013]

[0014] in, represents the mean of the a-th data sequence;

[0015] The standard deviation module is used to calculate the standard deviation according to the mean value. Calculate the standard deviation value B c , the calculation formula is:

[0016]

[0017] Among them, B c represents the cth standard deviation value;

[0018] The elimination module is used to eliminate the c Remove abnormal data from the data sequence and collect normal data as the data to be processed D l ;

[0019] like This is abnormal data, which is removed and the vacancy is marked as data to be filled Dt s ;

[0020] like It is normal data and is retained;

[0021] Filling module, the filling module is used to obtain the data to be filled Dt s Adjacent side data Dp s+1 and Dp s-1 , calculate the filling data T c , the calculation formula is:

[0022]

[0023] Among them, T c Indicates the cth filling data;

[0024] Adaptive module, the adaptive module is used to fill the data T c Fill in the data to be processed D l In the data to be processed D l Select any data d l , calculate the data d l The adaptive robust value L b , the calculation formula is:

[0025]

[0026] Among them, IQR(D l ) represents the data to be processed D l The quartile of , s represents a random number, and s∈[2,8];

[0027] The encryption weight module is used to adjust the weight according to the adaptive robust value L b Calculate the weighted smoothing value M p , the calculation formula is:

[0028]

[0029] Among them, k and Density (L b ) represent bandwidth and kernel density estimation respectively, Represents the adaptive robust value L b The average value, v s represents the encryption weight coefficient, N represents the encryption weight coefficient v s the number of

[0030] The set module is used to smooth the weighted value M p The collection is preprocessed data.

[0031] Furthermore, the conversion module includes:

[0032] The proportion module is used to integrate the preprocessed data into the preprocessing matrix Y ij , calculate the preprocessing matrix Y ij The proportion of Z b , the calculation formula is:

[0033]

[0034] Among them, y ij Represents the preprocessing matrix Y ij The elements in M represent the element y ij the number of

[0035] Dynamic module, the dynamic module is used to calculate the proportion value Z b Calculate the dynamic value d T , the calculation formula is:

[0036]

[0037] Among them, d T Indicates the Tth dynamic value;

[0038] Dynamic weight module, the dynamic weight module is used to calculate the weight of the dynamic value d T Calculate the dynamic weight value D q , the calculation formula is:

[0039]

[0040] Among them, D q Indicates the qth dynamic weight value;

[0041] The weight decision module is used to determine the weight of the dynamic weight value D q Calculate the weighted decision value Q for classifying multi-dimensional data of mining areas IJ , the calculation formula is:

[0042] Q IJ =D q ·Y ij

[0043] Among them, Q IJ represents the weighted decision value of row I and column J;

[0044] The positive and negative modules are used to determine the value Q according to the weight IJ Constructing a forward solution and negative solution The expression is:

[0045]

[0046] Among them, max(Q IJ) and min(Q IJ ) represent the maximum and minimum values of the weight decision value Q for each row respectively. IJ The maximum and minimum values of

[0047] Furthermore, the classification module includes:

[0048] Positive and negative Euclidean module, which is used to calculate the positive Euclidean distance and the negative Euclidean distance respectively according to the positive solution and the negative solution

[0049] Proximity module, which is used to calculate the proximity value T and the negative Euclidean distance The calculation formula is: J , and the calculation formula is:

[0050]

[0051] where T J represents the proximity value of the Jth;

[0052] Sequence module, which is used to sort the proximity value T J in descending order to obtain the descending sequence J xl , where l represents the ordinal number of the descending sequence, and the number of sequences L of the descending sequence J xl is obtained by the counting method;

[0053] Classification module, which is used to classify the descending sequence J xl sequentially according to the number of sequences L;

[0054] If l < 0.3L, then J xl is classified into the first category;

[0055] If 0.3L < l < 0.7L, then J xl is classified into the second category;

[0056] If 0.7L < l, then J xl is classified into the third category;

[0057] Classification set module, which is used to set the classified descending sequence J xl as the data classification set Sl j .

[0058] Furthermore, the calculation formulas of the positive Euclidean distance and the negative Euclidean distance are:

[0059]

[0060]

[0061] Among them, I represents and Q IJ The number of groups, and and Q IJ The number of groups.

[0062] Furthermore, the optimization module includes:

[0063] Classification feature module, which is used to define data classification sets Among them, sl xg Represents the first feature of the data, jf yg Represents the second feature of the data, U represents sl xg and jf yg the number of

[0064] The difference entropy module is used to classify the data into j Calculate the difference entropy C s , the calculation formula is:

[0065]

[0066] Among them, Sl jy Indicates that the second feature is a subset of jy, and jy∈yg, Ent(jf yg ) represents jf yg Information entropy of

[0067] Entropy ratio module, the entropy ratio module is used to calculate the entropy ratio of the difference value C s Calculate the entropy ratio S z , the calculation formula is:

[0068]

[0069] Among them, S z represents the zth entropy ratio;

[0070] The weighted feature module is used to z Calculate the weighted eigenvalue F after weighting q , the calculation formula is:

[0071] F q =S z ·sl xg

[0072] Among them, F q represents the qth weighted eigenvalue;

[0073] The preliminary model module is used to obtain the weighted eigenvalue F by the segmentation point generation method. q The split point is to select the maximum weighted eigenvalue F q The corresponding features and segmentation points are used to build a preliminary tree model;

[0074] Important coefficient module, which is used to calculate the important coefficient Zy of each node of the preliminary tree model x , the calculation formula is:

[0075]

[0076] Among them, depth(Jd) represents the depth value of the node, and Jd represents the node of the preliminary tree model;

[0077] A dynamic threshold module is used to calculate the weighted feature value F in the preliminary tree model. q Calculate the dynamic threshold Dz u , the calculation formula is:

[0078]

[0079] in, Represents the weighted eigenvalue F in the preliminary tree model q The mean of Represents the weighted eigenvalue F in the preliminary tree model q The standard deviation of

[0080] The pruning optimization module is used to optimize the pruning process according to the dynamic threshold Dz u Prune the preliminary tree model;

[0081] If Zy x ≥Dz u , no pruning is required;

[0082] If Zy x <Dz u , then prune;

[0083] The optimization model module is used to repeat step S48 to complete the pruning of the preliminary tree model and obtain the optimized tree model sh.

[0084] Furthermore, the risk fusion module includes:

[0085] Distribution module, the distribution module is used to obtain multiple distribution matrices by optimizing the tree model sh and in, Represents the probability that the optimized tree model sh believes that sample b belongs to category l;

[0086] Macro mean module, the macro mean module is used to calculate the macro mean H according to the optimized tree model sh z , the calculation formula is:

[0087]

[0088] in, represents the exact value of the optimized tree model sh, Represents the recall value of the optimized tree model sh;

[0089] Similarity value module, the similarity value module is used to Calculate the similarity value Xs r ;

[0090] Weight vector module, which is used to calculate the model weight vector M q , the calculation formula is:

[0091]

[0092] Where R represents H z the number of

[0093] Wind financing module, the wind financing module is used to calculate the wind financing module according to the weight vector M q Calculate the risk fusion value used to divide the risk probability The calculation formula is:

[0094]

[0095] in, Representing probability The risk fusion value after fusion.

[0096] Furthermore, the similarity value Xs r The calculation formula is:

[0097]

[0098] Among them, Xs r represents the rth similarity value, and Represents the distribution matrix The ath and dth distribution matrices, W and Q, represent and The number of

[0099] Furthermore, the early warning module includes:

[0100] Total risk module, the total risk module is used to fuse the values according to multiple risks Calculate the overall risk value Z f , the calculation formula is:

[0101]

[0102] Among them, Z f represents the fth overall risk value;

[0103] The early warning threshold module is used to calculate the overall risk value Z f Calculate the warning threshold β e , the calculation formula is:

[0104]

[0105] Among them, β e Indicates the e-th warning threshold;

[0106] The early warning judgment module is used to determine the early warning threshold β e Generate risk fusion value early warning information;

[0107] like This indicates a high risk probability and generates warning information;

[0108] like This indicates that the risk probability is low and no warning information is generated;

[0109] The sending module is used to send the early warning information to the management center of the mining area.

[0110] By means of the above technical solution, the present invention provides a multi-dimensional data statistics and intelligent analysis management system for mining areas, which has at least the following beneficial effects:

[0111] 1. The present invention can preprocess the acquired massive data through the data preprocessing step, thus avoiding the problem of low robustness in the preprocessed data. Moreover, the data can be more adaptively changed with the number of data types when processing the data. Finally, the data preprocessing is completed through weighted smoothing processing, thereby improving the efficiency of the preprocessing and increasing the robustness of the preprocessing.

[0112] 2. The present invention analyzes and classifies data through dynamic weights. Compared with the method of using fixed weights in the prior art, it can avoid fixed weight intervention, avoid ignoring the data characteristics of the data itself, reduce subjective bias, improve the accuracy of data processing, and provide assistance for subsequent steps to improve overall processing efficiency.

[0113] 3. The present invention uses a dynamic pruning method to avoid the problems of excessive data redundancy and low generalization ability in existing pruning methods. Through a dynamically adjustable pruning threshold, the generalization ability of the optimization tree model can be improved, redundant branches can be reduced, and the processing efficiency of the optimization tree model can be improved.

[0114] 4. The present invention drives weight distribution by model similarity and can quantify the redundancy between tree models. Compared with the existing methods, it can reduce the error rate with the same number of models, and at the same time avoid the problem of pseudo integration, thereby improving the accuracy of algorithm results and work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0115] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0116] Figure 1 This is a structural diagram of a mining area multi-dimensional data statistics and intelligent analysis management system of the present invention;

[0117] Figure 2 This is a structural block diagram of the preprocessing module of the present invention;

[0118] Figure 3 is a structural block diagram of the conversion module of the present invention;

[0119] Figure 4 This is a structural diagram of the classification module of the present invention;

[0120] Figure 5 This is a structural block diagram of the optimization module of the present invention;

[0121] Figure 6 This is a structural diagram of the risk fusion module of the present invention;

[0122] Figure 7 This is a structural block diagram of the early warning module of the present invention.

[0123] In the figure: 1. Preprocessing module; 11. Mean module; 12. Standard deviation module; 13. Elimination module; 14. Filling module; 15. Adaptive module; 16. Weight module; 17. Collection module; 2. Conversion module; 21. Proportion module; 22. Dynamic module; 23. Dynamic weight module; 24. Weight decision module; 25. Positive and negative module; 3. Classification module; 31. Positive and negative module; 32. Approximation module; 33. Sequence module; 34. Classification module; 35. Classification collection module; 4. Optimization module; 41. Classification feature module; 42. Difference entropy module; 43. Entropy ratio module; 44. Weighted feature module; 45. Preliminary model module; 46. Important coefficient module; 47. Dynamic threshold module; 48. Pruning optimization module; 49. Optimization model module; 5. Risk fusion module; 51. Distribution module; 52. Macro mean module; 53. Similarity value module; 54. Weight vector module; 55. Wind financing module; 6. Early warning module; 61. Total risk module; 62. Early warning threshold module; 63. Early warning judgment module; 64. Sending module. DETAILED DESCRIPTION

[0124] To make the above-mentioned objectives, features, and advantages of the present invention more clearly understood, the present invention is further described below in detail with reference to the accompanying drawings and specific embodiments. This will enable a full understanding of how this application uses technical means to solve technical problems and achieve technical effects, and to implement the invention accordingly.

[0125] Due to the low robustness of existing technologies in data acquisition and data classification, and the difficulty in efficiently processing large amounts of data, this embodiment proposes a multi-dimensional data statistics and intelligent analysis management system for mining areas, which can improve the robustness of data classification and can quickly and efficiently process large amounts of data, such as Figure 1 As shown in the figure, the system consists of a pre-processing module 1, a conversion module 2, a classification module 3, an optimization module 4, a risk fusion module 5, and an early warning module 6. The pre-processing module 1 is used to collect the original data S of the mining area. a , for the original data S a Preprocessing is performed to obtain preprocessed data; the conversion module 2 is used to convert the preprocessed data into a preprocessing matrix Y through a matrix conversion method ij , and according to the preprocessing matrix Y ij Calculate the forward solution and negative solution Classification module 3 is used to and negative solution Calculate the approximation value T J , for the close value T J Arrange in descending order to obtain descending sequence J xl , and for the descending sequence J xlClassification is performed to obtain a data classification set; the optimization module 4 is used to classify the data based on the data classification set S1. j Calculate the weighted eigenvalue F used to build the preliminary tree model q , and optimize the preliminary tree model to obtain the optimized tree model sh; the risk fusion module 5 is used to obtain multiple probabilities by optimizing the tree model sh According to probability Calculate risk fusion value Early warning module 6 is used to integrate multiple risk values Calculate the warning threshold β used to generate warning information e and send early warning information to the management center of the mining area.

[0126] In the present invention, the preprocessing module 1 is used to complete the processing of the original data S a The preprocessing module 2 converts the preprocessed data into a forward solution. and negative solution After calculation, the data is preliminarily classified through the classification module 3. Since the amount of initial data is huge and there are many types, it is necessary to divide the huge data into multiple categories through preliminary classification processing. In the present invention, only the processing of one type is used as an example for explanation. The processing methods of other types are the same. After being divided into multiple categories, the optimized optimization tree model sh is constructed through the optimization module 4, and then the risk fusion module 5 generates risk probability fusion values of different types of data based on the fusion of risk probabilities, and generates warning information for the data according to the risk probability fusion value through the warning module 6 and sends it to the management center. The data is analyzed and classified through dynamic weights. Compared with the method of using fixed weights in the prior art, it can avoid fixed weight intervention, avoid ignoring the data characteristics of the data itself, reduce subjective bias, improve the accuracy of data processing, and provide assistance for subsequent steps to improve overall processing efficiency.

[0127] Used to collect raw data of mining area a , for the original data S a The preprocessing module 1 for preprocessing the data is composed of a mean module 11, a standard deviation module 12, a removal module 13, a filling module 14, an adaptive module 15, a weight module 16 and a collection module 17. The mean module 11 is used to convert the original data S of the mining area into a Arrange into a data sequence, create a detection window with a window size of L on the data sequence, the step size of the detection window is μ, and calculate the mean of the data sequence in the detection window The calculation formula is:

[0128]

[0129] in, represents the mean of the a-th data sequence;

[0130] The standard deviation module 12 is used to calculate the standard deviation according to the mean value. Calculate the standard deviation value B c , the calculation formula is:

[0131]

[0132] Among them, B c represents the cth standard deviation value;

[0133] The elimination module 13 is used to eliminate the c Remove abnormal data from the data sequence and collect normal data as the data to be processed D l ;

[0134] like This is abnormal data, which is removed and the vacancy is marked as data to be filled Dt s ;

[0135] like The data is normal and retained. Since the original data of the mining area is greatly affected by the environment, the data fluctuates greatly, so a standard deviation value of 3 times B is set. c It is more appropriate to remove data.

[0136] The filling module 14 is used to obtain the data to be filled Dt s Adjacent side data Dp s+1 and Dp s-1 , calculate the filling data T c , the calculation formula is:

[0137]

[0138] Among them, T c Indicates the cth filling data;

[0139] The adaptive module 15 is used to fill the data T c Fill in the data to be processed D l In the data to be processed D l Select any data d l , calculate the data d l The adaptive robust value L b , the calculation formula is:

[0140]

[0141] Among them, IQR(D l ) represents the data to be processed D lThe quartile, s represents a random number, and s∈[2,8]; the quartile is a statistical method that arranges all values from small to large and divides them into four equal parts, and the values at the three dividing points are arranged. The quartile here represents the quartile at the 25% position.

[0142] The encryption module 16 is used to adjust the weight according to the adaptive robust value L b Calculate the weighted smoothing value M p , the calculation formula is:

[0143]

[0144] Among them, k and Density (L b ) represent bandwidth and kernel density estimation respectively, Represents the adaptive robust value L b The average value, v s represents the encryption weight coefficient, N represents the encryption weight coefficient v s the number of

[0145] The aggregation module 17 is used to convert the weighted smoothing value M p The collection is preprocessed data. Through the data preprocessing steps, the huge amount of data obtained can be preprocessed to avoid the problem of low robustness in the preprocessed data. When processing data, it can adapt to changes in the number of data types. Finally, the data preprocessing is completed through dense weighted smoothing processing, which improves the efficiency of preprocessing and increases the robustness of preprocessing.

[0146] Used to convert preprocessed data into preprocessing matrix Y through matrix conversion method ij , and according to the preprocessing matrix Y ij Calculate the forward solution and negative solution The conversion module 2 is composed of a proportion module 21, a dynamic module 22, a dynamic weight module 23, a weight decision module 24 and a positive and negative module 25. Among them, the proportion module 21 is used to integrate the preprocessed data into a preprocessing matrix Y ij , calculate the preprocessing matrix Y ij The proportion of Z b , the calculation formula is:

[0147]

[0148] Among them, y ij Represents the preprocessing matrix Y ij The elements in M represent the element y ij The number of preprocessed data is integrated into the preprocessing matrix Y ij The method converts the image into a matrix by flattening it into a vector.

[0149] The dynamic module 22 is used to calculate the proportion Z b Calculate the dynamic value d T , the calculation formula is:

[0150]

[0151] Among them, d T Indicates the Tth dynamic value;

[0152] The dynamic weight module 23 is used to calculate the dynamic weight of the T Calculate the dynamic weight value D q , the calculation formula is:

[0153]

[0154] Among them, D q Indicates the qth dynamic weight value;

[0155] The weight decision module 24 is used to determine the weight of the dynamic weight value D q Calculate the weighted decision value Q for classifying multi-dimensional data of mining areas IJ , the calculation formula is:

[0156] Q IJ =D q ·Y ij

[0157] Among them, Q IJ represents the weighted decision value of row I and column J;

[0158] The positive and negative module 25 is used to determine the value Q according to the weight IJ Constructing a forward solution and negative solution The expression is:

[0159]

[0160] Among them, max(Q IJ ) and min(Q IJ ) represent the weighted decision value Q of each row IJ The maximum and minimum values of the data can be preprocessed through the data preprocessing step, avoiding the problem of low robustness in the preprocessed data, and can be more adaptive to changes in the number of data types when processing data. Finally, the data preprocessing is completed through dense weight smoothing, which improves the efficiency of preprocessing and increases the robustness of preprocessing. Compared with the method of using fixed weights in the existing technology, the data is analyzed and classified through dynamic weights, which can avoid fixed weight intervention and avoid ignoring the data characteristics of the data itself.

[0161] For the forward solution and negative solutions Calculate the proximity value T J , for the proximity value T J Perform a descending order arrangement to obtain a descending sequence J xl , and for the descending sequence J xl Perform classification to obtain a classification module 3 of the data classification set, which consists of a positive and negative Euclidean module 31, a proximity module 32, a sequence module 33, a classification module 34, and a classification set module 35. Among them, the positive and negative Euclidean module 31 is used to calculate the positive Euclidean distance and negative solutions respectively calculate the positive Euclidean distance and negative Euclidean distance Positive Euclidean distance and negative Euclidean distance The calculation formulas for are:

[0162]

[0163] Among them, I respectively represent and Q IJ The number of groups, and and Q IJ The number of groups.

[0164] The proximity module 32 is used to calculate the proximity value T and negative Euclidean distance Calculate the proximity value T J , the calculation formula is:

[0165]

[0166] Among them, T J Represents the Jth proximity value;

[0167] The sequence module 33 is used to arrange the proximity value T J in descending order to obtain a descending sequence J xl , where l represents the ordinal number of the descending sequence, and the sequence quantity L of the descending sequence J xl is obtained by the counting method;

[0168] The classification module 34 is used to perform sequential classification on the descending sequence J xl according to the sequence quantity L;

[0169] If l < 0.3L, then classify J xl into the first category;

[0170] If 0.3L < l < 0.7L, then classify J xl into the second category;

[0171] If 0.7L < l, then classify J xlDivided into the third category; the classification here is a preliminary classification of the preprocessed data. The present invention only gives the situation of being divided into three categories. Under normal circumstances, it will be divided into multiple types. The classification after preprocessing can reduce the processing complexity of subsequent steps, reduce the amount of data processing, and only further process single data, thereby improving the efficiency of data processing.

[0172] The classification collection module 35 is used to sort the descending sequence J xl The set is the data classification set Sl j , data classification set Sl j The data in the data set is already classified. The data is analyzed and classified through dynamic weights. Compared with the method of using fixed weights in the existing technology, it can avoid fixed weight intervention, avoid ignoring the data characteristics of the data itself, reduce subjective bias, improve the accuracy of data processing, and provide assistance for subsequent steps to improve overall processing efficiency.

[0173] For classification based on data set S1 j Calculate the weighted eigenvalue F used to build the preliminary tree model q , and optimize the preliminary tree model to obtain the optimized tree model sh. The optimization module 4 consists of a classification feature module 41, a difference entropy module 42, an entropy ratio module 43, a weighted feature module 44, a preliminary model module 45, an important coefficient module 46, a dynamic threshold module 47, a pruning optimization module 48 and an optimization model module 49. Among them, the classification feature module 41 is used to define the data classification set Among them, sl xg Represents the first feature of the data, jf yg Represents the second feature of the data, U represents sl xg and jf yg The number of data classification sets Sl j Labeling of multiple categories and features is performed to facilitate calculations in subsequent steps.

[0174] The difference entropy module 42 is used to classify the data into j Calculate the difference entropy C s , the calculation formula is:

[0175]

[0176] Among them, Sl jy Indicates that the second feature is a subset of jy, and jy∈yg, Ent(jf yg ) represents jf yg Information entropy; Information entropy is a common method of calculating and branching tree models through probability, which will not be described here.

[0177] The entropy ratio module 43 is used to calculate the entropy ratio of the difference value C. sCalculate the entropy ratio S z , the calculation formula is:

[0178]

[0179] Among them, S z Represents the zth entropy ratio; the entropy ratio can be understood as the ratio of entropy values.

[0180] The weighted feature module 44 is used to determine the weighted feature according to the entropy ratio S z Calculate the weighted eigenvalue F after weighting q , the calculation formula is:

[0181] F q =S z ·sl xg

[0182] Among them, F q represents the qth weighted eigenvalue;

[0183] The preliminary model module 45 is used to obtain the weighted feature value F by the segmentation point generation method. q The split point is to select the maximum weighted eigenvalue F q The corresponding features and segmentation points are used to build a preliminary tree model;

[0184] The important coefficient module 46 is used to calculate the important coefficient Zy of each node of the preliminary tree model x , the calculation formula is:

[0185]

[0186] Among them, depth(Jd) represents the depth value of the node, and Jd represents the node of the preliminary tree model;

[0187] The dynamic threshold module 47 is used to determine the weighted feature value F in the preliminary tree model. q Calculate the dynamic threshold Dz u , the calculation formula is:

[0188]

[0189] in, Represents the weighted eigenvalue F in the preliminary tree model q The mean of Represents the weighted eigenvalue F in the preliminary tree model q The standard deviation is a commonly used calculation method in statistics and will not be described in detail here.

[0190] The pruning optimization module 48 is used to calculate the dynamic threshold Dz u Prune the preliminary tree model;

[0191] If Zyx ≥Dz u , no pruning is required;

[0192] If Zy x <Dz u , then prune;

[0193] The optimization model module 49 is used to repeat step S48 to complete the pruning of the preliminary tree model and obtain the optimized tree model sh. Through the dynamic pruning method, it can avoid the problem of excessive data redundancy and low generalization ability in the existing pruning method. Through the dynamically adjustable pruning threshold, it can improve the generalization ability of the optimized tree model, reduce redundant branches, and improve the processing efficiency of the optimized tree model.

[0194] Used to obtain multiple probabilities by optimizing the tree model sh According to probability Calculate risk fusion value The risk fusion module 5 is composed of a distribution module 51, a macro-mean module 52, a similarity module 53, a weight vector module 54 and a risk fusion module 55. The distribution module 51 is used to obtain multiple distribution matrices by optimizing the tree model sh. and in, Represents the probability that the optimized tree model sh believes that sample b belongs to category l;

[0195] The macro mean module 52 is used to calculate the macro mean H according to the optimized tree model sh z , the calculation formula is:

[0196]

[0197] in, represents the exact value of the optimized tree model sh, Represents the recall value of the optimized tree model sh;

[0198] The similarity value module 53 is used to calculate the similarity value based on multiple distribution matrices. Calculate the similarity value Xs r ; Similarity value Xs r The calculation formula is:

[0199]

[0200] Among them, Xs r represents the rth similarity value, and Represents the distribution matrix The ath and dth distribution matrices, W and Q, represent and The number of

[0201] The weight vector module 54 is used to calculate the model weight vector M q , the calculation formula is:

[0202]

[0203] Where R represents H z the number of

[0204] The wind financing module 55 is used to calculate the wind financing module 55 according to the weight vector M. q Calculate the risk fusion value used to divide the risk probability The calculation formula is:

[0205]

[0206] in, Representing probability The fused risk fusion value drives weight distribution through model similarity, which can quantify the redundancy between tree models. Compared with the existing methods, it can reduce the error rate with the same number of models, and at the same time avoid the problem of pseudo integration, thereby improving the accuracy of the algorithm results and work efficiency.

[0207] Used to fuse values based on multiple risks Calculate the warning threshold β used to generate warning information e The warning module 6 is composed of a total risk module 61, a warning threshold module 62, a warning judgment module 63 and a sending module 64, wherein the total risk module 61 is used to generate a warning signal based on multiple risk fusion values. Calculate the overall risk value Z f , the calculation formula is:

[0208]

[0209] Among them, Z f represents the fth overall risk value;

[0210] The early warning threshold module 62 is used to calculate the overall risk value Z f Calculate the warning threshold β e , the calculation formula is:

[0211]

[0212] Among them, β e Indicates the e-th warning threshold;

[0213] The early warning judgment module 63 is used to determine the early warning threshold β e Generate risk fusion value early warning information;

[0214] like This indicates a high risk probability and generates warning information;

[0215] like This indicates that the risk probability is low and no warning information is generated;

[0216] The sending module 64 is used to send the early warning information to the management center of the mining area. By driving the weight distribution by model similarity, the redundancy between tree models can be quantified. Compared with the existing methods, the error rate can be reduced with the same number of models. At the same time, the problem of pseudo integration can be avoided, thereby improving the accuracy of the algorithm results and improving work efficiency.

[0217] Each embodiment in this specification is described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the same or similar parts between the embodiments. For the above embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For relevant parts, refer to the partial description of the method embodiments.

[0218] The above embodiments provide a detailed introduction to the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A mining area multi-dimensional data statistics and intelligent analysis management system, characterized in that: The system includes: Pre-processing module (1), the pre-processing module (1) is used to collect the original data S of the mining area a , for the original data S a Performing preprocessing to obtain preprocessed data; The conversion module (2) is used to convert the preprocessed data into a preprocessed matrix Y through a matrix conversion method. ij , and according to the preprocessing matrix Y ij Calculate the forward solution and negative solution Classification module (3), the classification module (3) is used to classify and negative solution Calculate the approximation value T J , for the close value T J Arrange in descending order to obtain descending sequence J xl , and for the descending sequence J xl Classify and obtain a data classification set; Optimization module (4), the optimization module (4) is used to classify the data based on the set S1 j Calculate the weighted eigenvalue F used to build the preliminary tree model q , and optimize the preliminary tree model to obtain the optimized tree model sh; Risk fusion module (5), the risk fusion module (5) is used to obtain multiple probabilities by optimizing the tree model sh According to probability Calculate risk fusion value An early warning module (6) is used to generate a warning signal based on multiple risk fusion values. Calculate the warning threshold β used to generate warning information e and send early warning information to the management center of the mining area.

2. The intelligent analysis and management system according to claim 1, characterized in that: The pre-processing module (1) comprises: Mean module (11), the mean module (11) is used to convert the original data S of the mining area a Arrange into a data sequence, create a detection window with a window size of L on the data sequence, the step size of the detection window is μ, and calculate the mean of the data sequence in the detection window The calculation formula is: in, represents the mean of the a-th data sequence; The standard deviation module (12) is used to calculate the standard deviation according to the mean value. Calculate the standard deviation value B c , the calculation formula is: Among them, B c represents the cth standard deviation value; The elimination module (13) is used to eliminate the c Remove abnormal data from the data sequence and collect normal data as the data to be processed D l ; like This is abnormal data, which is removed and the vacancy is marked as data to be filled Dt s ; like It is normal data and is retained; Filling module (14), the filling module (14) is used to obtain the data to be filled Dt s Adjacent side data Dp s+1 and Dp s-1 , calculate the filling data T c , the calculation formula is: Among them, T c Indicates the cth filling data; Adaptive module (15), said adaptive module (15) is used to fill the data T c Fill in the data to be processed D l In the data to be processed D l Select any data d l , calculate the data d l The adaptive robust value L b , the calculation formula is: Among them, IQR(D l ) represents the data to be processed D l The quartile of , s represents a random number, and s∈[2,8]; The encryption weight module (16) is used to adjust the adaptive robustness value L b Calculate the weighted smoothing value M p , the calculation formula is: Among them, k and Density (L b ) represent bandwidth and kernel density estimation respectively, Represents the adaptive robust value L b The average value, v s represents the encryption weight coefficient, N represents the encryption weight coefficient v s the number of The collection module (17) is used to smooth the weighted value M p The collection is preprocessed data.

3. The intelligent analysis and management system according to claim 1, characterized in that: The conversion module (2) comprises: A proportion module (21) is used to integrate the preprocessing data into a preprocessing matrix Y ij , calculate the preprocessing matrix Y ij The proportion of Z b , the calculation formula is: Among them, y ij Represents the preprocessing matrix Y ij The elements in M represent the element y ij the number of Dynamic module (22), the dynamic module (22) is used to calculate the value of the ratio Z b Calculate the dynamic value d T , the calculation formula is: Among them, d T Indicates the Tth dynamic value; A dynamic weight module (23) is used to adjust the weight according to the dynamic value d T Calculate the dynamic weight value D q , the calculation formula is: Among them, D q Indicates the qth dynamic weight value; A weight decision module (24) is used to determine the weight of a q Calculate the weighted decision value Q for classifying multi-dimensional data of mining areas IJ , the calculation formula is: Q IJ =D q ·Y ij Among them, Q IJ represents the weighted decision value of row I and column J; The positive and negative module (25) is used to determine the weighted value Q IJ Calculate the forward solution and negative solution The expression is: Among them, max(Q IJ ) and min(Q IJ ) represent the weighted decision value Q of each row IJ The maximum and minimum values of .

4. The intelligent analysis and management system according to claim 1, characterized in that: The classification module (3) comprises: The positive and negative ohm module (31) is used to solve the problem according to the positive solution. and negative solution Calculate the positive Euclidean distance separately and negative Euclid The close module (32) is used to and negative Euclid Calculate the approximation value T J , the calculation formula is: Among them, T J represents the Jth closest value; Sequence module (33), the sequence module (33) is used to close the value T J Arrange in descending order to obtain descending sequence J xl , where l represents the ordinal number of the descending sequence, and the descending sequence J is obtained by counting method xl The number of sequences L; A classification module (34) is used to classify the descending sequence J according to the number of sequences L. xl Perform sequential classification; If l<0.3L, then J xl Divided into the first category; If 0.3L < l < 0.7L, then J xl is classified into the second category; If 0.7L < l, then classify J xl into the third category; Classification collection module (35), the classification collection module (35) is used to classify the descending sequence J xl The set is the data classification set Sl j .

5. The intelligent analysis and management system according to claim 4, characterized in that: The positive Euclid and negative Euclid The calculation formula is: Among them, I represents and Q IJ The number of groups, and and Q IJ The number of groups.

6. The intelligent analysis and management system according to claim 1, characterized in that: The optimization module (4) comprises: Classification feature module (41), the classification feature module (41) is used to define the data classification set Among them, sl xg Represents the first feature of the data, jf yg Represents the second feature of the data, U represents sl xg and jf yg the number of The difference entropy module (42) is used to classify the data into a set of j Calculate the difference entropy C s , the calculation formula is: Among them, Sl jy Indicates that the second feature is a subset of jy, and jy∈yg, Ent(jf yg ) represents jf yg Information entropy of Entropy ratio module (43), the entropy ratio module (43) is used to calculate the entropy ratio of the difference value C s Calculate the entropy ratio S z , the calculation formula is: Among them, S z represents the zth entropy ratio; The weighted feature module (44) is used to determine the weighted feature according to the entropy ratio S z Calculate the weighted eigenvalue F after weighting q , the calculation formula is: F q =S z ·sl xg Among them, F q represents the qth weighted eigenvalue; A preliminary model module (45) is used to obtain a weighted feature value F by a segmentation point generation method. q The split point is to select the maximum weighted eigenvalue F q The corresponding features and segmentation points are used to build a preliminary tree model; An important coefficient module (46) is used to calculate the important coefficient Zy of each node of the preliminary tree model. x , the calculation formula is: Among them, depth(Jd) represents the depth value of the node, and Jd represents the node of the preliminary tree model; A dynamic threshold module (47) is used to determine the weighted feature value F in the preliminary tree model. q Calculate the dynamic threshold Dz u , the calculation formula is: in, Represents the weighted eigenvalue F in the preliminary tree model q The mean of Represents the weighted eigenvalue F in the preliminary tree model q The standard deviation of The pruning optimization module (48) is used to optimize the pruning process according to the dynamic threshold Dz u Prune the preliminary tree model; If Zy x ≥Dz u , no pruning is required; If Zy x <Dz u , then prune; The optimization model module (49) is used to repeat step S48 to complete the pruning of the preliminary tree model and obtain the optimized tree model sh.

7. The intelligent analysis and management system according to claim 1, characterized in that: The risk fusion module (5) includes: A distribution module (51) is used to obtain multiple distribution matrices by optimizing the tree model sh and in, Represents the probability that the optimized tree model sh believes that sample b belongs to category l; A macro mean module (52) for calculating a macro mean H according to an optimized tree model sh z , the calculation formula is: in, represents the exact value of the optimized tree model sh, Represents the recall value of the optimized tree model sh; A similarity value module (53) is used to calculate the similarity value based on a plurality of distribution matrices. Calculate the similarity value Xs r ; A weight vector module (54) is used to calculate the model weight vector M q , the calculation formula is: Where R represents H z the number of Wind financing module (55), the wind financing module (55) is used to calculate the wind financing module according to the weight vector M q Calculate the risk fusion value used to divide the risk probability The calculation formula is: in, Representing probability The risk fusion value after fusion.

8. The intelligent analysis and management system according to claim 7, characterized in that: The similarity value Xs r The calculation formula is: Among them, Xs r represents the rth similarity value, and Represents the distribution matrix The ath and dth distribution matrices, W and Q, represent and The number of 9. The intelligent analysis and management system according to claim 1, characterized in that: The early warning module (6) comprises: A total risk module (61) is used to combine multiple risk values Calculate the overall risk value Z f , the calculation formula is: Among them, Z f represents the fth overall risk value; The early warning threshold module (62) is used to determine the overall risk value Z f Calculate the warning threshold β e , the calculation formula is: Among them, β e Indicates the e-th warning threshold; The early warning judgment module (63) is used to judge the early warning threshold value β e Generate risk fusion value early warning information; like This indicates a high risk probability and generates warning information; like This indicates that the risk probability is low and no warning information is generated; A sending module (64) is used to send the early warning information to the management center of the mining area.

Images