Goaf mixed water source identification method based on random forest model

Through the goaf mixed water source identification method based on the random forest model, the prediction is made using the integrated results of multiple decision trees, which solves the problems of low accuracy and low efficiency of water source identification in goaf, and achieves efficient, accurate and automated Water source identification adapts to changes in complex hydrogeological conditions in goaf.

CN119028472APending Publication Date: 2024-11-26ORDOS HUAXING ENERGY CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411096610.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-26

Smart Images

  • Figure CN119028472A_ABST
    Figure CN119028472A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of mine goaf mixed water source identification, and particularly relates to a goaf mixed water source identification method based on a random forest model, and the method comprises the following specific steps: S1, preprocessing chemical data of goaf mixed water, and determining characteristic parameters; s2, constructing a random forest model, selecting characteristic variables, training the model to identify the source of the mixed water in the goaf, then combining the hydrochemical data and the random forest model, providing an efficient, accurate and automatic mixed water source identification method in the goaf, having significant technical progress and practical application value, and having a wide application prospect. The problems that in the prior art, goaf mixed water source recognition precision is low, efficiency is low and the like are solved, theoretical guidance is provided for technologies such as timely and effective drainage and closure of goaf water, and reliable technical support is provided for safe production of a coal mine.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of mixed water source identification in a mine goaf, and in particular to a mixed water source identification method in a goaf based on a random forest model. Background Art

[0002] At present, my country's coal resources still dominate China's energy structure system. With the expansion of coal mining, the formation of goaf is inevitable. Due to the particularity of its structure, it is often easy to accumulate water, causing water damage. Accurately identifying the source of water accumulation in goaf is crucial for the safe production and governance of coal mines.

[0003] There are many technical means for the identification of mine water sources. The evolution mechanism of water accumulation in goaf is relatively complex, and the water chemical characteristics are relatively complex compared with other water-filled aquifers. The water accumulation in goaf is quite different from that of other water-filled aquifers. Therefore, the single analysis of its water chemical characteristics to determine the water source is easily restricted by many factors, such as the interference of complex hydrogeological conditions, the interference of individual abnormal data, and the interference of human factors, which leads to deviations in the identification results. In addition, some methods have a cumbersome identification process and low efficiency, making it difficult to quickly and accurately identify the water source.

[0004] To this end, a mixed water source identification method for goaf areas based on random forest model is proposed to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a method for identifying mixed water sources in goaf areas based on a random forest model to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for identifying mixed water sources in goaf areas based on a random forest model. The specific steps of the method for identifying mixed water sources in goaf areas based on a random forest model are as follows:

[0007] S1. Preprocess the chemical data of mixed water in the goaf to determine characteristic parameters;

[0008] S2, build a random forest model, select feature variables, and train the model to identify the source of mixed water in the goaf;

[0009] S3. Use the trained random forest model to predict the water chemistry of the goaf that needs to be identified and identify the source of mixed water;

[0010] S4. Verify the model prediction results and adjust the model parameters to improve the recognition accuracy.

[0011] Preferably, the preprocessing of the water chemical data of the mixed water in the goaf in S1 needs to ensure that the water quality indicators are within a certain range. The purpose of the pretreatment system includes adjusting the water quality indicators, such as pH value not greater than 5.0, and strive to be lower than 3.0; turbidity lower than 5; removing residual chlorine or similar oxides, such as ozone, and ensuring that there are no other chemicals that may cause membrane pollution or degradation; appropriate sampling methods need to be adopted. For the sampling of mixed water in the goaf, a suitable sampling method should be selected according to the specific situation. For example, for the collection of sewage and wastewater, direct collection or the use of specific tools to sample in-depth in the monitoring well; for surface water or groundwater, different sampling methods are selected according to the different depths and properties of the water body, such as directly drawing surface water or using tools to reach a predetermined depth and then draw; necessary chemical treatment is required, and the pretreatment also includes chemical treatment such as digestion, separation and enrichment of water samples. The purpose of digestion is to destroy organic matter, eliminate interference, dissolve suspended matter, etc., so that the water sample is clear and transparent; separation and enrichment are to remove or reduce interfering substances and improve the detection limit and signal-to-noise ratio of the substance to be tested.

[0012] Preferably, in S1, the characteristic parameters of the mixed water in the goaf include flow rate, temperature, pH value, dissolved oxygen, chemical oxygen demand (COD), biochemical oxygen demand (BOD), total solids, suspended matter, heavy metal content, and radioactive substance content; the flow rate reflects the flow velocity and volume of the water body, which has an important influence on the evaluation of the self-purification ability of the water body and the ability to transport pollutants; the temperature affects the growth and chemical reaction rate of aquatic organisms, and has a direct impact on water quality; the pH value indicates the acidity and alkalinity of water, which affects the survival of aquatic organisms and the direction of chemical reactions; the dissolved oxygen reflects the self-purification ability and ecological health status of the water body, which is crucial to the survival of aquatic organisms; the chemical oxygen demand (COD) and biochemical oxygen demand (BOD) respectively indicate the content of organic matter in the water, which are important indicators for evaluating the degree of organic pollution in the water body; the total solids and suspended matter reflect the turbidity of the water body and possible sediment problems; the heavy metal content includes lead, mercury, and cadmium, which have long-term effects on the environment and human health; the radioactive substance content is an important indicator for evaluating whether the water body is affected by nuclear pollution.

[0013] Preferably, the characteristic variables of the random forest model in S2 include but are not limited to Na+, Mg2+, Ca2+, Cl-, SO42- and HCO3-.

[0014] Preferably, the random forest model algorithm in the model training in S2 specifically includes:

[0015] The random forest algorithm includes multiple decision trees, each of which independently predicts the mixed water source in the goaf. The final classification result is obtained by using the trees h1(X), h2(X),…, h2(X) as classifiers in the random forest (RF). N(X) is voted and the training subset is obtained randomly according to the distribution of vectors Y and X. The definition of the marginal function is shown in formula (1):

[0016]

[0017] Among them, I(·) represents the index function, av N (·) denotes the mean, and the margin measures the extent to which the average vote of X,Y for the correct class exceeds the average vote of any other class. The larger the margin, the more prepared it is for classification. The generalization error of the random forest classifier is described by the equation;

[0018] PE * =P X,Y (mg(X,Y)<0) (2)

[0019] Formula (2) explains {h(X,θ k )|k=1,2,…} follows the rule of large numbers. The value of N is large enough for the model, and the classifier has enough trees. At the same time, it is proved that the upper limit of the generalization error is convergent, because the random vector θ,…,PE * , which is determined by formula (3);

[0020]

[0021] in, represents the upper limit of the generalization error, ρ represents the average correlation coefficient between trees, and s represents the average classification performance of the decision tree. Formula (4) shows that the larger the average correlation coefficient, the larger the upper limit of the generalization error. Similarly, the larger the average classification amount, the larger the upper limit of the generalization error. That is to say, the classification performance is affected by two factors, one is the overall performance of the trees, and the other is the diversity between the trees.

[0022] Preferably, the adjustment step of the training model in S2 includes adjusting the hyperparameters of the random forest model, such as the number of decision trees, the maximum depth and the minimum number of sample splits, to optimize the model performance.

[0023] Preferably, the specific step S3 is to use the trained random forest model to predict the newly collected water chemistry data, identify the mixed water source in the goaf, verify the model prediction results, and use the cross-validation method to evaluate the accuracy and robustness of the model.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] (1) Improve identification accuracy: By combining water chemistry data and the random forest model, the source of mixed water in the goaf can be more accurately identified, reducing errors in human judgment and improving identification accuracy and reliability;

[0026] (2) Efficient processing of complex data: The hydrochemical characteristics of water in the goaf are more complex than those of other aquifers, and the hydrochemical characteristics of the aquifers are quite different, which leads to data volatility. The random forest model can effectively process high-dimensional and complex data and use the integrated results of multiple decision trees for prediction, avoiding the overfitting problem that is prone to occur in a single model, making the model more robust and generalizable.

[0027] (3) Automated analysis process: The present invention uses a random forest model to perform automated data analysis and water source identification, which greatly reduces manual intervention, improves work efficiency, and reduces labor costs;

[0028] (4) Real-time prediction and updating: The method of the present invention can predict and update in real time when new data is input, so that the model can dynamically adapt to the changes in the mixed water source in the goaf and provide timely and accurate identification results;

[0029] In summary, the present invention provides an efficient, accurate and automated method for identifying mixed water sources in goaf areas by combining water chemistry data and random forest models. It has significant technological progress and practical application value, and solves the problems of low accuracy and low efficiency in identifying mixed water sources in goaf areas in the prior art. It provides theoretical guidance for technologies such as timely and effective drainage and interception of goaf water, and provides reliable technical support for safe production in coal mines. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flow chart of the identification method of the present invention;

[0031] Figure 2 Schematic diagram of random forest classification algorithm;

[0032] Figure 3 To construct the feature weight map of each feature parameter of the random forest model;

[0033] Figure 4 This is the confusion matrix diagram for the test set results. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments 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.

[0035] In the description of the present invention, it is necessary to understand that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0036] See also Figure 1-4 , the present invention provides a technical solution:

[0037] A mixed water source identification method for goaf area based on random forest model, the specific steps of the mixed water source identification method for goaf area based on random forest model are as follows:

[0038] S1. Preprocess the chemical data of mixed water in the goaf to determine characteristic parameters;

[0039] S2, build a random forest model, select feature variables, and train the model to identify the source of mixed water in the goaf;

[0040] S3. Use the trained random forest model to predict the water chemistry of the goaf that needs to be identified and identify the source of mixed water;

[0041] S4. Verify the model prediction results and adjust the model parameters to improve the recognition accuracy.

[0042] The pretreatment of the water chemical data of the mixed water in the goaf in S1 needs to ensure that the water quality indicators are within a certain range. The purpose of the pretreatment system includes adjusting the water quality indicators, such as pH value not greater than 5.0, and strive to be lower than 3.0; turbidity lower than 5; removing residual chlorine or similar oxides, such as ozone, and ensuring that there are no other chemicals that may cause membrane pollution or degradation; appropriate sampling methods need to be adopted. For the sampling of mixed water in the goaf, the appropriate sampling method should be selected according to the specific situation. For example, for the collection of sewage and wastewater, direct collection or the use of specific tools to conduct in-depth sampling in the monitoring well; for surface water or groundwater, different sampling methods are selected according to the different depths and properties of the water body, such as directly drawing surface water or using tools to reach a predetermined depth before drawing; necessary chemical treatment is required, and the pretreatment also includes chemical treatment such as digestion, separation and enrichment of water samples. The purpose of digestion is to destroy organic matter, eliminate interference, dissolve suspended matter, etc., so that the water sample is clear and transparent; separation and enrichment are to remove or reduce interfering substances and improve the detection limit and signal-to-noise ratio of the substance to be tested.

[0043] In S1, the characteristic parameters of the mixed water in the goaf include flow rate, temperature, pH value, dissolved oxygen, chemical oxygen demand (COD), biochemical oxygen demand (BOD), total solids, suspended matter, heavy metal content, and radioactive substance content; the flow rate reflects the flow velocity and volume of the water body, which has an important impact on the evaluation of the self-purification ability of the water body and the ability to transport pollutants; the temperature affects the growth of aquatic organisms and the rate of chemical reactions, and has a direct impact on water quality; the pH value indicates the acidity and alkalinity of water, which affects the survival of aquatic organisms and the direction of chemical reactions; the dissolved oxygen reflects the self-purification ability and ecological health status of the water body, which is crucial to the survival of aquatic organisms; the chemical oxygen demand (COD) and biochemical oxygen demand (BOD) respectively represent the content of organic matter in the water, which are important indicators for evaluating the degree of organic pollution in the water body; the total solids and suspended matter reflect the turbidity of the water body and possible sediment problems; the heavy metal content includes lead, mercury, and cadmium, which have long-term effects on the environment and human health; the radioactive substance content is an important indicator for evaluating whether the water body is affected by nuclear pollution.

[0044] The characteristic variables of the random forest model in S2 include but are not limited to Na+, Mg2+, Ca2+, Cl-, SO42- and HCO3-.

[0045] The random forest model algorithm in the model training in S2 specifically includes:

[0046] The random forest algorithm includes multiple decision trees, each of which independently predicts the mixed water source in the goaf. The final classification result is obtained by using the trees h1(X), h2(X),…, h2(X) as classifiers in the random forest (RF). N (X) is voted and the training subset is obtained randomly according to the distribution of vectors Y and X. The definition of the marginal function is shown in formula (1):

[0047]

[0048] Among them, I(·) represents the index function, av N (·) denotes the mean, and the margin measures the extent to which the average vote of X,Y for the correct class exceeds the average vote of any other class. The larger the margin, the more prepared it is for classification. The generalization error of the random forest classifier is described by the equation;

[0049] PE * =P X,Y (mg(X,Y)<0) (2)

[0050] Formula (2) explains {h(X,θ k)|k=1,2,…} follows the rule of large numbers. The value of N is large enough for the model, and the classifier has enough trees. At the same time, it is proved that the upper limit of the generalization error is convergent, because the random vector θ,…,PE * , which is determined by formula (3);

[0051]

[0052] in, represents the upper limit of the generalization error, ρ represents the average correlation coefficient between trees, and s represents the average classification performance of the decision tree. Formula (4) shows that the larger the average correlation coefficient, the larger the upper limit of the generalization error. Similarly, the larger the average classification amount, the larger the upper limit of the generalization error. That is to say, the classification performance is affected by two factors, one is the overall performance of the trees, and the other is the diversity between the trees.

[0053] As an ensemble learning algorithm, this classifier has better overall performance due to the good performance of each tree and the appropriate number of trees. In addition, the coverage ability of the random forest space depends on the differences between trees. In order to achieve the differences between trees, the following two methods are used: (a) Generate multiple subsets of training data; (b) Structurally, at each node, grow trees with random features by random selection, and then split a small group of input variables according to the maximum size without pruning. In summary, the two parameters of ntree and mtry should be determined when applying to identify abnormal patterns of bivariate autocorrelation processes. The steps to establish a random forest model are as follows.

[0054] (1) Based on the Bootstrap method, multiple training subsets are sampled from the training set;

[0055] (2) Generate each corresponding tree based on each training data subset. First, randomly select features from the original feature set. Second, grow the tree with the selected random features to the maximum value at each node without pruning. Note that in this process, the capacity of the feature subset is usually constant.

[0056] (3) Test the performance of the RF classifier using the test data set and obtain the classification results from each tree;

[0057] (4) The final classification will be the result with the most votes among the classification results.

[0058] The adjustment step of the training model in S2 includes adjusting the hyperparameters of the random forest model, such as the number of decision trees, the maximum depth, and the minimum number of sample splits, to optimize the model performance.

[0059] The specific step of S3 is to use the trained random forest model to predict the newly collected water chemistry data, identify the mixed water source in the goaf, verify the model prediction results, and use the cross-validation method to evaluate the accuracy and robustness of the model.

[0060] Example:

[0061] See attached Figure 1 and Figure 2 The present invention utilizes Na + Mg 2+ , Ca 2+ , Cl - 、SO4 2- and HCO 3- The ion content is used as the characteristic variable, and the classification model is constructed using the random forest algorithm. The random forest model constructs multiple decision trees and uses the integrated results of multiple decision trees for classification. The model is trained through the training sample set, and the model parameters are adjusted to obtain the best performance. After the random forest model is established, the water chemistry data of the goaf water is input into the model as the prediction set to obtain the classification results of the goaf water;

[0062] This embodiment is shown in Table 1. Four types of water are mainly selected, including Tangjiahui Coal Mine Ordovician ash water, 6 coal seam top and bottom water, and 9 coal seam bottom water, totaling 110 samples, as training sets and test sets to establish the model. The sample quantity information is described in Table 1. + Mg 2+ , Ca 2+ , Cl - 、SO4 2- and HCO 3- The item is used as the independent variable, and the category is used as the dependent variable for random forest modeling. As can be seen from Table 1, a total of 110 samples are involved in the analysis as training sets and prediction sets.

[0063] Table 1 Summary of basic information of random forest classification

[0064]

[0065] The feature weight shows the importance of each title's contribution to the model, and its sum is 1. As can be seen from the figure: K + Na + The proportion is 21.41%, which has the highest weight and plays a key role in model construction; Ca2+ accounts for 19.27%; Mg 2+ The proportion is 16.38%; Cl - The proportion of the above four characteristics is 15.67%; the proportion of the above four characteristics is 72.73%; the remaining two HCO 3- 、SO4 2-The proportions are 14.26% and 13.01% respectively.

[0066] The results of the training set and the prediction set obtained by running the model with the software are shown in Table 2 and Table 3. The concepts in the table are explained as follows: First: accuracy, the proportion of samples with correct prediction results to the total samples, the higher the accuracy, the better; second: precision, the proportion of positive samples in the predicted results, the higher the index, the better; third: recall, the proportion of positive samples predicted as positive samples, the higher the recall, the better; fourth: F1-score, which is a comprehensive evaluation index that combines precision and recall. It is the harmonic mean of precision and recall; fifth: both precision and recall are better, but the two are often contradictory, so F1-score is often used to comprehensively evaluate the effect of the classifier. Its value range is 0 to 1. The closer to 1, the better the effect, but it should be evaluated in combination with the actual situation. It is often used to evaluate the improvement effect of the model.

[0067] Table 2. Evaluation results of training set model

[0068] category Accuracy Recall f1-score Number of samples 1.0 1.00 1.00 1.00 21 2.0 1.00 1.00 1.00 28 3.0 1.00 1.00 1.00 15 4.0 1.00 1.00 1.00 24 Accuracy 1.00 88 average value 1.00 1.00 1.00 88 Average value (comprehensive) 1.00 1.00 1.00 88

[0069] Table 3. Test set model evaluation results

[0070] category Accuracy Recall f1-score Number of samples 1.0 0.33 0.50 0.40 2 2.0 0.83 1.00 0.91 5 3.0 1.00 1.00 1.00 5 4.0 1.00 0.80 0.89 10 Accuracy 0.86 22 average value 0.79 0.82 0.80 22

[0071] Next, you can view the confusion matrix of the test data, which is the intersection of the model predictions and the actual situation, see Figure 3 : When the confusion matrix is ​​generated, the larger the value of the lower right triangle diagonal, the better, which means that the predicted value and the true value are completely consistent. In the above figure, except for 2 samples in 4 categories being judged as category 1 and 1 sample in category 1 being judged as category 2, the rest are all correct, which means that the random forest model performs well on the test data.

[0072] Table 4 shows the main parameters of the random forest model, where the training set ratio is 0.8, the number of decision trees is 100, the minimum number of samples for node splitting is 2, the minimum number of samples for leaf nodes is 1, and the maximum depth is unlimited. The final model has an accuracy of 86.36% on the test set, a precision (comprehensive) of 90.15%, a recall (comprehensive) of 86.36%, and an F1-score (comprehensive) of 0.87. The model effect is acceptable.

[0073] Table 4 Summary of model parameters

[0074]

[0075] After the random forest model was established, the goaf water described in Table 2 was input into the model as the prediction set, and the classification results of goaf water were shown in Table 5. It can be seen that 51 samples originated from the roof water of the 6th coal seam, 25 from the Ordovician ash water, 24 from the 9th coal floor water, and 13 from the 6th coal floor water. It can be judged that the mixed water in the goaf mainly originated from the roof water of the 6th coal seam, followed by the Ordovician ash water, the 9th coal floor water, and the 6th coal floor water.

[0076] Table 5 Prediction set evaluation results

[0077] category Number of samples 1(Aohui Water) 25 2(6 coal roof water) 51 3 (6 coal bottom plate water) 13 4 (9 coal bottom plate water) 24

[0078] In summary, the present invention combines the water ion index of the goaf to construct a random forest model, and provides a mixed water source identification method for the goaf based on the random forest model. First, the water chemical data of the goaf of the coal mine is collected, and the variables related to water source identification are selected, such as Na + Mg 2+ , Ca 2+ , Cl - 、SO4 2- and HCO 3- Etc., construct a feature vector for water source identification, then use the above example concentration as a feature variable to construct a random forest model, train the model to identify the mixed water source in the goaf, then use the trained random forest model to predict the water chemistry of the goaf that needs to be identified, identify the mixed water source, and finally verify the model prediction results, and adjust the model parameters to improve the recognition accuracy. Finally, through example simulation, the results show that the mixed water in the goaf of Tangjiahui Coal Mine mainly comes from the 6th coal roof water, followed by the Ordovician ash water, the 9th coal floor water and the 6th coal floor water. The discrimination effect of the present invention is good and the accuracy is high.

[0079] The above is a detailed introduction to the mixed water source identification method for goaf areas based on the random forest model provided by the present invention. Specific examples are used in this article 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 idea; at the same time, for general technical personnel in this field, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0080] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention; therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is limited by the attached claims rather than the above description. Therefore, it is intended to include all changes within the meaning and scope of the equivalent elements of the claims in the present invention, and any figure marks in the claims should not be regarded as limiting the claims involved.

[0081] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for identifying mixed water sources in goaf areas based on a random forest model, characterized in that: The specific steps of the mixed water source identification method in goaf area based on random forest model are as follows: S1. Preprocess the chemical data of mixed water in the goaf to determine characteristic parameters; S2, build a random forest model, select feature variables, and train the model to identify the source of mixed water in the goaf; S3. Use the trained random forest model to predict the water chemistry of the goaf that needs to be identified and identify the source of mixed water; S4. Verify the model prediction results and adjust the model parameters to improve the recognition accuracy.

2. The method for identifying mixed water sources in goaf areas based on a random forest model according to claim 1, characterized in that: The pretreatment of the water chemical data of the mixed water in the goaf in S1 needs to ensure that the water quality indicators are within a certain range. The purpose of the pretreatment system includes adjusting the water quality indicators, such as pH value not greater than 5.0, and strive to be lower than 3.0; turbidity lower than 5; removing residual chlorine or similar oxides, such as ozone, and ensuring that there are no other chemicals that may cause membrane pollution or degradation; appropriate sampling methods need to be adopted. For the sampling of mixed water in the goaf, the appropriate sampling method should be selected according to the specific situation. For example, for the collection of sewage and wastewater, direct collection or the use of specific tools to conduct in-depth sampling in the monitoring well; for surface water or groundwater, different sampling methods are selected according to the different depths and properties of the water body, such as directly drawing surface water or using tools to reach a predetermined depth before drawing; necessary chemical treatment is required, and the pretreatment also includes chemical treatment such as digestion, separation and enrichment of water samples. The purpose of digestion is to destroy organic matter, eliminate interference, dissolve suspended matter, etc., so that the water sample is clear and transparent; separation and enrichment are to remove or reduce interfering substances and improve the detection limit and signal-to-noise ratio of the substance to be tested.

3. The method for identifying mixed water sources in goaf areas based on a random forest model according to claim 1, characterized in that: In S1, the characteristic parameters of the mixed water in the goaf include flow rate, temperature, pH value, dissolved oxygen, chemical oxygen demand (COD), biochemical oxygen demand (BOD), total solids, suspended matter, heavy metal content, and radioactive substance content; the flow rate reflects the flow velocity and volume of the water body, which has an important impact on the evaluation of the self-purification ability of the water body and the ability to transport pollutants; the temperature affects the growth of aquatic organisms and the rate of chemical reactions, and has a direct impact on water quality; the pH value indicates the acidity and alkalinity of water, which affects the survival of aquatic organisms and the direction of chemical reactions; the dissolved oxygen reflects the self-purification ability and ecological health status of the water body, which is crucial to the survival of aquatic organisms; the chemical oxygen demand (COD) and biochemical oxygen demand (BOD) respectively represent the content of organic matter in the water, which are important indicators for evaluating the degree of organic pollution in the water body; the total solids and suspended matter reflect the turbidity of the water body and possible sediment problems; the heavy metal content includes lead, mercury, and cadmium, which have long-term effects on the environment and human health; the radioactive substance content is an important indicator for evaluating whether the water body is affected by nuclear pollution.

4. The method for identifying mixed water sources in goaf areas based on a random forest model according to claim 1, characterized in that: The characteristic variables of the random forest model in S2 include but are not limited to Na+, Mg2+, Ca2+, Cl-, SO42- and HCO3-.

5. The method for identifying mixed water sources in goaf areas based on a random forest model according to claim 1, characterized in that: The random forest model algorithm in the model training in S2 specifically includes: The random forest algorithm includes multiple decision trees, each of which independently predicts the mixed water source in the goaf. The final classification result is obtained by using the trees h1(X), h2(X),…, h2(X) as classifiers in the random forest (RF). N (X) is voted and the training subset is obtained randomly according to the distribution of vectors Y and X. The definition of the marginal function is shown in formula (1): Among them, I(·) represents the index function, av N (·) denotes the mean, and the margin measures the extent to which the average vote of X,Y for the correct class exceeds the average vote of any other class. The larger the margin, the more prepared it is for classification. The generalization error of the random forest classifier is described by the equation; ON * =P X,Y (mg(X,Y)<0) (2) Formula (2) explains {h(X,θ k )|k=1,2,…} follows the rule of large numbers. The value of N is large enough for the model, and the classifier has enough trees. At the same time, it is proved that the upper limit of the generalization error is convergent, because the random vector θ,…,PE * , which is determined by formula (3); Among them, ε represents the upper limit of generalization error, ρ represents the average correlation coefficient between trees, and s represents the average classification performance of decision trees. Formula (4) shows that the larger the average correlation coefficient, the larger the upper limit of generalization error. Similarly, the larger the average classification amount, the larger the upper limit of generalization error. That is to say, the classification performance is affected by two factors, one is the overall performance of trees, and the other is the diversity between trees.

6. The method for identifying mixed water sources in goaf areas based on a random forest model according to claim 1, characterized in that: The adjustment step of the training model in S2 includes adjusting the hyperparameters of the random forest model, such as the number of decision trees, the maximum depth, and the minimum number of sample splits, to optimize the model performance.

7. The method for identifying mixed water sources in goaf areas based on a random forest model according to claim 1, characterized in that: The specific step of S3 is to use the trained random forest model to predict the newly collected water chemistry data, identify the mixed water source in the goaf, verify the model prediction results, and use the cross-validation method to evaluate the accuracy and robustness of the model.