Intelligent evaluation method and system for ecological problem of mine tailings

Abstract

According to the mine tailing ecological problem intelligent evaluation method and system provided by the invention, the tailing ecological restoration estimation result is debugged based on the independent tailing ecological restoration data debugging coefficient, and the independent tailing ecological restoration data result corresponding to the target mine matter tag is obtained; the tailing ecological restoration estimation result is obtained by evaluating the characteristics of the tailing ecological information and reflects the quality of the tailing ecological information on the objective level, and the tailing ecological restoration estimation result is debugged according to the deviation corresponding to the significant characteristics of the mine matters; independent tailing ecological restoration data results obtained through debugging are close to evaluation results obtained based on mine items, that is, independent tailing ecological restoration data are carried out on tailing ecological information by combining subjectivity of the mine items and objective quality of the tailing ecological information; the independent tailing ecological restoration data result meets the uniqueness of individual mine items, so that the accuracy of the independent tailing ecological restoration data result is improved.

Description

Technical Field

[0001] This application relates to the technical field of intelligent evaluation of tailings ecological problems. Specifically, it relates to an intelligent evaluation method and system for tailings ecological problems in mines. Background Art

[0002] After mining, there are many ecological damage problems in mines, such as pollution of soil and water sources by metal slag or dust pollution and other pollution situations. Currently, relevant technical personnel are used for sampling and analysis to evaluate the sample pollution information in the area, but a comprehensive evaluation of pollution information cannot be carried out. Therefore, it is necessary to intelligently identify and evaluate the ecological environment. However, how to evaluate the effect after real-time evaluation and repair is a problem that is difficult to solve at present. Summary of the Invention

[0003] To improve the technical problems existing in the related art, this application provides an intelligent evaluation method and system for tailings ecological problems in mines.

[0004] In a first aspect, an intelligent evaluation method for tailings ecological problems in mines is provided. The method includes: Obtain the tailings ecological information to be evaluated and the target mine matter label; Extract the tailings ecological restoration element features corresponding to the tailings ecological information to be evaluated; Combine the tailings ecological restoration element features to determine the tailings ecological restoration prediction result of the tailings ecological information to be evaluated; Through an independent tailings ecological restoration data optimization model, combine the tailings ecological restoration element features and the splicing result of the mine matter significance features corresponding to the target mine matter label to determine the independent tailings ecological restoration data debugging coefficient corresponding to the target mine matter label; the mine matter significance features are data reflecting the significance characteristics of mine matters; Based on the tailings ecological restoration prediction result and the independent tailings ecological restoration data debugging coefficient, determine the independent tailings ecological restoration data result corresponding to the target mine matter label.

[0005] In this application, the method further includes: Obtain a number of reference tailings ecological information corresponding to the target mine matter label, and each reference tailings ecological information is annotated with an independent tailings ecological restoration data evaluation value corresponding to the target mine matter label; Load the several reference tailings ecological information into the unique mine matter salience feature extraction model respectively, and obtain the unique mine matter salience features corresponding to the several reference tailings ecological information through the unique mine matter salience feature extraction model; the unique mine matter salience feature is the salience feature of the reference tailings ecological information corresponding to the unique mine matter salience feature. Obtain the maximum value and the minimum value of the independent tailings ecological restoration data corresponding to each reference tailings ecological information and the target mine matter label. For each reference tailings ecological information, based on the evaluation value of the independent tailings ecological restoration data corresponding to the target mine matter label, the maximum value of the independent tailings ecological restoration data, and the minimum value of the independent tailings ecological restoration data, obtain the uniqueness degree of the target mine matter label to the reference tailings ecological information. Based on the uniqueness degree of the target mine matter label to each reference tailings ecological information, and the unique mine matter salience features corresponding to each reference tailings ecological information, obtain the mine matter salience feature corresponding to the target mine matter label.

[0006] It can be understood that In this application, extracting the tailings ecological restoration element features corresponding to the tailings ecological information to be evaluated includes: Obtain the tailings ecological restoration element feature extraction model. Load the tailings ecological information to be evaluated into the tailings ecological restoration element feature extraction model, and output the tailings ecological restoration element features through the tailings ecological restoration element feature extraction model.

[0007] In this application, the tailings ecological restoration element feature extraction model is jointly trained with the unique mine matter salience feature extraction model; the tailings ecological restoration element feature extraction model and the unique mine matter salience feature extraction model have a common standard local model, and respectively include corresponding output local models; the steps of jointly training the tailings ecological restoration element feature extraction model and the unique mine matter salience feature extraction model include: Obtain the first tailings ecological information example set, the second tailings ecological information example set, the tailings ecological restoration element feature extraction model, and the unique mine matter salience feature extraction model; each first tailings ecological information example in the first tailings ecological information example set has a tailings ecological restoration element feature configuration directory, and each second tailings ecological information example in the second tailings ecological information example set has a unique mine matter salience feature configuration directory. Load the first tailings ecological information example into the tailings ecological restoration element feature extraction model, extract features from the first tailings ecological information example through the standard local model of the tailings ecological restoration element feature extraction model, and output the regression analysis result of the tailings ecological restoration element features through the output local model of the tailings ecological restoration element feature extraction model; Load the second tailings ecological information example into the unique mine matter significance feature extraction model, extract features from the second tailings ecological information example through the standard local model of the unique mine matter significance feature extraction model, and output the regression analysis result of the unique mine matter significance features through the output local model of the unique mine matter significance feature extraction model; Based on the regression analysis result of the tailings ecological restoration element features and the tailings ecological restoration element feature configuration directory, as well as the regression analysis result of the unique mine matter significance features and the unique mine matter significance feature configuration directory, jointly train the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model.

[0008] In this application, the step of jointly training the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model based on the regression analysis result of the tailings ecological restoration element features and the tailings ecological restoration element feature configuration directory, as well as the regression analysis result of the unique mine matter significance features and the unique mine matter significance feature configuration directory, includes: Build a first evaluation index algorithm based on the distinction between the regression analysis result of the tailings ecological restoration element features and the tailings ecological restoration element feature configuration directory, and build a second evaluation index algorithm based on the distinction between the regression analysis result of the unique mine matter significance features and the unique mine matter significance feature configuration directory; Obtain the first model pyramid coefficient by minimizing the first evaluation index algorithm, and obtain the second model pyramid coefficient by minimizing the second evaluation index algorithm; Combine the first model pyramid coefficient and the second model pyramid coefficient to jointly train the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model.

[0009] In this application, the training steps of the independent tailings ecological restoration data optimization model include: Obtain a third tailings ecological information example corresponding to the example mine matter label and the independent tailings ecological restoration data optimization model; there is an independent tailings ecological restoration data debugging coefficient configuration directory corresponding to the example mine matter label in the third tailings ecological information example; Extract the example of the tailings ecological restoration element characteristics corresponding to the third tailings ecological information example; Obtain the example of the significance characteristics of the mine matters corresponding to the example mine matter label; Through the independent tailings ecological restoration data optimization model, combine the splicing results of the example of the tailings ecological restoration element characteristics and the example of the significance characteristics of the mine matters to obtain the regression analysis result of the debugging coefficient of the independent tailings ecological restoration data corresponding to the example mine matter label; Based on the regression analysis result of the debugging coefficient of the independent tailings ecological restoration data and the configuration directory of the debugging coefficient of the independent tailings ecological restoration data, train the independent tailings ecological restoration data optimization model.

[0010] In this application, the third tailings ecological information example corresponds to more than one example mine matter label; The steps for obtaining the configuration directory of the debugging coefficient of the independent tailings ecological restoration data include: Obtain the example of the result of the independent tailings ecological restoration data for each of the example mine matter labels corresponding to the third tailings ecological information example; Based on each example of the result of the independent tailings ecological restoration data, obtain the example of the mean value of the independent tailings ecological restoration data; Based on each example of the result of the independent tailings ecological restoration data and the example of the mean value of the independent tailings ecological restoration data, obtain the example of the debugging coefficient of the independent tailings ecological restoration data corresponding to each example mine matter label, and determine each example of the debugging coefficient of the independent tailings ecological restoration data as the configuration directory of the debugging coefficient of the independent tailings ecological restoration data corresponding to the third tailings ecological information example and each example mine matter label.

[0011] In this application, the obtaining of the example of the significance characteristics of the mine matters corresponding to the example mine matter label includes: Obtain the example of the unique significance characteristics of the mine matters corresponding to the third tailings ecological information example; Obtain the example of the result of the independent tailings ecological restoration data corresponding to the example mine matter label; Based on the example of the result of the independent tailings ecological restoration data and the example of the unique significance characteristics of the mine matters, obtain the example of the significance characteristics of the mine matters corresponding to the example mine matter label.

[0012] In this application, the steps for obtaining the example of the tailings ecological restoration element characteristics and the example of the unique significance characteristics of the mine matters corresponding to the third tailings ecological information example include: Obtain the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model; Load the third tailings ecological information example into the tailings ecological restoration element feature extraction model, extract features from the third tailings ecological information example through the standard local model of the tailings ecological restoration element feature extraction model, and output the tailings ecological restoration element feature example through the output local model of the tailings ecological restoration element feature extraction model; Load the third tailings ecological information example into the unique mine matter significance feature extraction model, extract features from the third tailings ecological information example through the standard local model of the unique mine matter significance feature extraction model, and output the unique mine matter significance feature example through the output local model of the unique mine matter significance feature extraction model.

[0013] In this application, the example mine matter label corresponds to no less than two third tailings ecological information examples; the independent tailings ecological restoration data result example is an independent tailings ecological restoration data evaluation value example; obtaining the mine matter significance feature example corresponding to the example mine matter label based on the independent tailings ecological restoration data result example and the unique mine matter significance feature example includes: Obtain the independent tailings ecological restoration data maximum value example and the independent tailings ecological restoration data minimum value example of no less than two of the third tailings ecological information examples; For one of the third tailings ecological information examples, obtain the uniqueness degree of the example mine matter label to the third tailings ecological information example based on the independent tailings ecological restoration data evaluation value example, the independent tailings ecological restoration data maximum value example, and the independent tailings ecological restoration data minimum value example corresponding to the example mine matter label; Based on the uniqueness degree of the example mine matter label to each of the third tailings ecological information examples, and the unique mine matter significance feature examples corresponding to each of the third tailings ecological information examples, obtain the mine matter significance feature example corresponding to the example mine matter label.

[0014] In this application, the feature layer of the unique mine matter significance feature example is several; obtaining the mine matter significance feature example corresponding to the example mine matter label based on the uniqueness degree and the unique mine matter significance feature example includes: Combine the uniqueness degree to adjust the feature values of the unique mine matter significance feature example in each layer to obtain the mine matter significance feature example corresponding to the example mine matter label.

[0015] In this application, the tailings ecological information to be evaluated is several; the independent tailings ecological restoration data result is an independent tailings ecological restoration data evaluation value; the method further includes: Obtain the independent tailings ecological restoration data evaluation value of each of the tailings ecological information to be evaluated by the target mine matter label; Select a preset number of target tailings ecological information from each of the tailings ecological information to be evaluated according to the independent tailings ecological restoration data evaluation value; Output the target tailings ecological information to the terminal where the target mine matter label is located.

[0016] In a second aspect, a smart evaluation system for mine tailings ecological problems is provided, including a processor and a memory that communicate with each other. The processor is configured to read and execute a computer program from the memory to implement the above method.

[0017] For a smart evaluation method and system for mine tailings ecological problems provided by an embodiment of the present application, first determine the estimated result of tailings ecological restoration according to the characteristics of tailings ecological restoration elements of the tailings ecological information to be evaluated, and determine the independent tailings ecological restoration data debugging coefficient according to the characteristics of tailings ecological restoration elements and the significance characteristics of mine matters corresponding to the target mine matter label. Then, based on the independent tailings ecological restoration data debugging coefficient, debug the estimated result of tailings ecological restoration to obtain the independent tailings ecological restoration data result corresponding to the target mine matter label. In this way, since the estimated result of tailings ecological restoration is a result obtained by evaluating the characteristics of the tailings ecological information itself, it reflects the quality of the tailings ecological information at the objective level. And by debugging the estimated result of tailings ecological restoration according to the deviation corresponding to the significance characteristics of the mine matter, the obtained independent tailings ecological restoration data result approaches the evaluation result obtained by the mine matter based on itself, that is, combining the subjectivity of the mine matter and the objective quality of the tailings ecological information to perform independent tailings ecological restoration data on the tailings ecological information, so that the independent tailings ecological restoration data result meets the uniqueness of individual mine matters, thereby improving the accuracy of the independent tailings ecological restoration data result. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application, and thus should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1 It is a flowchart of a smart evaluation method for mine tailings ecological problems provided by an embodiment of the present application. DETAILED DESCRIPTION

[0020] To better understand the above technical solution, the technical solution of the present application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. Without conflict, the technical features in the embodiments of the present application and the embodiments can be combined with each other.

[0021] Please refer to Figure 1 , which shows an intelligent evaluation method for the ecological problems of mine tailings. This method may include the technical solutions described in the following steps 202-step 210.

[0022] Step 202, obtain the tailings ecological information to be evaluated and the target mine matter label.

[0023] Among them, the tailings ecological information to be evaluated is the ecological information of the restoration location, such as: the metal mineral information and soil information of this area, etc. In the actual operation process, it can be obtained manually or through an artificial intelligence network.

[0024] Among them, the target mine matter label is the mine matter label of the target mine matter. The mine matter label is used to describe the mine location information of the mine matter. By simulating the perception of the target mine matter on the tailings ecological information, the tailings ecological information to be evaluated is evaluated, so that the evaluation result obtained is consistent with the target mine matter.

[0025] Specifically, an intelligent evaluation application for mine tailings ecological problems runs on the terminal, and the target mine matter is registered with the target mine matter label based on the intelligent evaluation application for mine tailings ecological problems.

[0026] Exemplarily, in this project, the tailings ecological information to be evaluated and the target mine matter label are obtained simultaneously by artificial and artificial intelligence, and the accuracy of the data obtained by artificial intelligence can be evaluated.

[0027] Among them, for the comprehensive treatment project of the mine geological environment in the coal mining subsidence area, various types of data such as county meteorology, hydrology, basic geology, hydro-engineering environmental geology, geological disasters, social economy, construction planning, town and village-level area planning, and satellite images were comprehensively collected. Through means such as ground surveys, drilling, mountain engineering, remote sensing interpretation, testing and experiments, and new technologies and methods such as InSAR and UAV aerial photography, a large amount of first-hand data was obtained, laying a solid foundation for the preparation of the results. The project made full use of the existing achievement data of geological disasters, engineering geology, hydrogeology, environmental geology, geotechnical engineering investigation, etc. within and around Aba County, combined with digital elevation models, remote sensing images and interpretation results, and systematically sorted out the main disaster-causing conditions related to the formation of geological disasters in Aba County, such as topography, geological structure, engineering geological rock groups, meteorology and hydrology, slope structure, easily collapsible and slippery strata, and human engineering activities. Based on the investigation of disaster-causing geological conditions, the exploration of typical geological disasters, the comprehensive review of existing geological disasters and potential hazard points, the investigation of newly added geological disasters and potential hazards, and the review of the effectiveness of treatment projects, a relatively comprehensive analysis and summary were carried out on the disaster-causing geological conditions, disaster-forming models, formation mechanisms, and development and distribution laws of different disaster types in Aba County, identifying 28 newly added hazard sources and 47 slopes with high risks and above. After the restoration, the restoration status was detected by this method, and the accuracy reached more than 90%.

[0028] Step 204: Extract the characteristics of the tailings ecological restoration elements corresponding to the tailings ecological information to be evaluated.

[0029] Exemplarily, the characteristics of the tailings ecological restoration elements are understood as the characteristics of factors such as topography, climate characteristics, hydrological conditions, physical, chemical and biological characteristics of the soil, topsoil conditions, and potential pollution.

[0030] In a possible embodiment, step 204 includes: obtaining a model for extracting the characteristics of the tailings ecological restoration elements; loading the tailings ecological information to be evaluated into the model for extracting the characteristics of the tailings ecological restoration elements, and outputting the characteristics of the tailings ecological restoration elements through the model for extracting the characteristics of the tailings ecological restoration elements.

[0031] Exemplarily, the feature extraction model can be understood as a convolutional neural network in the prior art.

[0032] Among them, the model for extracting the characteristics of the tailings ecological restoration elements is a machine learning model trained with the ability to extract the characteristics of the tailings ecological restoration elements. The model for extracting the characteristics of the tailings ecological restoration elements can be trained through examples of tailings ecological information and a configuration directory of the characteristics of the tailings ecological restoration elements.

[0033] It can be understood that for a general machine learning model with the ability to extract tailings ecological restoration element features, the tailings ecological restoration element features extracted by it meet the requirements of the embodiments of the present application for tailings ecological restoration element features. A general machine learning model with the ability to extract tailings ecological restoration element features can be used as the tailings ecological restoration element feature extraction model of the embodiments of the present application.

[0034] Specifically, the computer device obtains the tailings ecological information to be evaluated and the tailings ecological restoration element feature extraction model. The computer device loads the tailings ecological information to be evaluated into the tailings ecological restoration element feature extraction model to obtain the tailings ecological restoration element features output by the tailings ecological restoration element feature extraction model.

[0035] Exemplarily, in the present application, the tailings ecological restoration element features are for comprehensively managing and ecologically restoring the occupation and damage of mine slag to land and vegetation resources, mine geological safety disasters and other ecological problems.

[0036] Step 206, determine the tailings ecological restoration prediction result of the tailings ecological information to be evaluated according to the tailings ecological restoration element features.

[0037] Exemplarily, the pre-evaluation in the tailings ecological restoration prediction result can be understood as evaluating all the features in the tailings ecological restoration element features. Here, the evaluation is not based on the influence weight, and only the theoretical influencing factors are evaluated in this step.

[0038] Generally, the following technical methods can be used to obtain the tailings ecological restoration prediction result: Analytic Hierarchy Process (AHP): This method is used to determine the weights of each signal. The elements related to the decision-making problem are decomposed into levels such as objectives, criteria, and solutions through a hierarchical structure model, and qualitative and quantitative analyses are carried out on this basis.

[0039] Long Short-Term Memory (LSTM) network: This is a technology for training a prediction model based on historical data. The model parameters are optimized through cross-validation to predict the future state of the device.

[0040] Cross-validation: This is a model evaluation technology. The dataset is divided into multiple subsets for multiple trainings and tests to ensure the stability of the model performance. Common cross-validation methods include k-fold cross-validation and leave-one-out cross-validation.

[0041] Data mining: Through data cleaning, integration, and preprocessing, the accuracy and consistency of the data are ensured to provide a reliable data basis for strategic evaluation. Data mining techniques also include confusion matrices for evaluating the performance of classification models.

[0042] Determine an independent tailings ecological restoration data debugging coefficient corresponding to the target mine matter label according to the characteristics of tailings ecological restoration elements and the significance characteristics of the mine matters corresponding to the target mine matter label.

[0043] Among them, the significance characteristics of the mine matter are data reflecting the significance characteristics of the mine matter, and the significance characteristics of the mine matter are correlated with the effect of the mine matter on tailings ecological information.

[0044] In this application, considering that there are differences in the evaluation criteria of different mine matters for tailings ecological information, and the significance of the mine matter is an important factor affecting the uniqueness of the mine matter, the coefficient of the significance characteristics of the mine matter is introduced. The significance characteristics of the mine matter are coefficients associated with the effect of the mine matter on tailings ecological information.

[0045] In a possible embodiment, the significance characteristics of the mine matter may include several feature layers, and each feature layer corresponds to a significance characteristic. The feature value of each feature layer can be used to represent the score of the mine matter in the corresponding significance characteristic. This score is not only related to the significance characteristic of the mine matter itself but also related to the effect of the mine matter on tailings ecological information.

[0046] Among them, the independent tailings ecological restoration data debugging coefficient is used to debug the predicted result of tailings ecological restoration, so that the obtained independent tailings ecological restoration data result matches the restoration effect of the target mine matter.

[0047] In a possible embodiment, the significance characteristics of the mine matter corresponding to the target mine matter label can be determined in advance. The computer device can obtain the significance characteristics of the mine matter corresponding to the loaded target mine matter label, or extract the significance characteristics of the mine matter corresponding to the target mine matter label from the pre-stored significance characteristics of the mine matter.

[0048] In a possible embodiment, step 208 includes: obtaining an independent tailings ecological restoration data optimization model; through the independent tailings ecological restoration data optimization model, obtaining an independent tailings ecological restoration data debugging coefficient according to the characteristics of tailings ecological restoration elements and the significance characteristics of the mine matter.

[0049] Among them, the independent tailings ecological restoration data optimization model is a machine learning model trained with the ability to calculate the independent tailings ecological restoration data debugging coefficient. The independent tailings ecological restoration data optimization model can be trained through examples of tailings ecological restoration element characteristics, examples of significance characteristics of mine matters, and an independent tailings ecological restoration data debugging coefficient configuration directory.

[0050] In a possible embodiment, the computer device obtains the independent tailings ecological restoration data debugging coefficient according to the splicing result of the tailings ecological restoration element characteristics and the mine matter significance characteristics through the independent tailings ecological restoration data optimization model.

[0051] Specifically, the computer device obtains the tailings ecological restoration element characteristics, the mine matter significance characteristics, and the independent tailings ecological restoration data optimization model. The computer device first determines the splicing result of the tailings ecological restoration element characteristics and the mine matter significance characteristics, and then loads the splicing result into the independent tailings ecological restoration data optimization model to obtain the independent tailings ecological restoration data debugging coefficient output by the independent tailings ecological restoration data optimization model. Alternatively, the computer device loads the tailings ecological restoration element characteristics and the mine matter significance characteristics into the independent tailings ecological restoration data optimization model. The independent tailings ecological restoration data optimization model first determines the splicing result of the tailings ecological restoration element characteristics and the mine matter significance characteristics, and then outputs the independent tailings ecological restoration data debugging coefficient according to the splicing result.

[0052] Step 210, based on the tailings ecological restoration prediction result and the independent tailings ecological restoration data debugging coefficient, determine the independent tailings ecological restoration data result corresponding to the target mine matter label.

[0053] Among them, during the specific implementation process, the quality of the restoration can also be monitored to ensure the construction quality of the project and the effect after the later restoration.

[0054] Among them, the independent tailings ecological restoration data result is the implementation evaluation result of the simulated target mine matter on the tailings ecological information to be evaluated. This step is for the current tailings restoration situation (such as: xxx town, xxx city, xxx province).

[0055] Specifically, the computer device adjusts the tailings ecological restoration prediction result according to the independent tailings ecological restoration data debugging coefficient to obtain the independent tailings ecological restoration data result corresponding to the target mine matter label.

[0056] In a possible embodiment, there is an independent tailings ecological restoration data model in one embodiment. It can be seen that the independent tailings ecological restoration data model includes a tailings ecological restoration element feature extraction model, a restoration effect evaluation model, and an independent tailings ecological restoration data optimization model. First, the tailings ecological information to be evaluated is loaded into the tailings ecological restoration element feature extraction model, and the tailings ecological restoration element features are output through the tailings ecological restoration element feature extraction model; then the tailings ecological restoration element features are loaded into the restoration effect evaluation model, and the tailings ecological restoration prediction result is output through the restoration effect evaluation model; then through the independent tailings ecological restoration data optimization model, an independent tailings ecological restoration data debugging coefficient is obtained according to the tailings ecological restoration element features and the mine matter significance features of the target mine matter label; finally, based on the tailings ecological restoration prediction result and the independent tailings ecological restoration data debugging coefficient, the independent tailings ecological restoration data result corresponding to the target mine matter label is determined.

[0057] In the above intelligent evaluation method for mine tailings ecological problems, first, the tailings ecological restoration prediction result is determined according to the tailings ecological restoration element features of the tailings ecological information to be evaluated, and the independent tailings ecological restoration data debugging coefficient is determined according to the tailings ecological restoration element features and the mine matter significance features corresponding to the target mine matter label. Then, the tailings ecological restoration prediction result is debugged based on the independent tailings ecological restoration data debugging coefficient to obtain the independent tailings ecological restoration data result corresponding to the target mine matter label. In this way, since the tailings ecological restoration prediction result is the result obtained by evaluating the characteristics of the tailings ecological information itself, it reflects the restoration effect of the tailings ecological information at the objective level. And by debugging the tailings ecological restoration prediction result according to the mine matter significance features, the obtained independent tailings ecological restoration data result approaches the evaluation result obtained by the mine matter based on itself, that is, the independent tailings ecological restoration data of the tailings ecological information is combined with the subjective evaluation of the mine matter and the objective evaluation quality of the tailings ecological information, so that the independent tailings ecological restoration data result meets the single characteristics of individual mine matters, thereby improving the accuracy of the independent tailings ecological restoration data result.

[0058] In a possible embodiment, there are several pieces of tailings ecological information to be evaluated; the independent tailings ecological restoration data result is the independent tailings ecological restoration data evaluation value; the method further includes: obtaining the independent tailings ecological restoration data evaluation value of the target mine matter label for each piece of tailings ecological information to be evaluated; selecting a preset number of target tailings ecological information from each piece of tailings ecological information to be evaluated according to the independent tailings ecological restoration data evaluation value; and outputting the target tailings ecological information to the terminal where the target mine matter label is located.

[0059] Specifically, the computer device obtains a number of tailings ecological information to be evaluated corresponding to the tailings ecological information evaluation task, obtains the independent tailings ecological restoration data evaluation value of the target mine matter label for each tailings ecological information to be evaluated, selects a preset number of target tailings ecological information from high to low according to the independent tailings ecological restoration data evaluation value, and outputs the target tailings ecological information to the terminal where the target mine matter label is located. In this way, in each tailings ecological information evaluation task, the computer device preferentially displays the tailings ecological information that meets the target mine matter to the target mine matter.

[0060] This application relates to the significant features of mine matters. The significant features of mine matters can be determined in advance or determined according to the reference tailings ecological information in the application scenario. The following provides a method for obtaining the significant features of mine matters corresponding to the target mine matter label.

[0061] In a possible embodiment, the process of obtaining the significant features of mine matters in one embodiment. It can be seen that obtaining the significant features of mine matters corresponding to the target mine matter label includes: Step 502, obtain more than one reference tailings ecological information corresponding to the target mine matter label, and each reference tailings ecological information is annotated with the independent tailings ecological restoration data result corresponding to the target mine matter label; the independent tailings ecological restoration data result is the independent tailings ecological restoration data evaluation value.

[0062] In a possible embodiment, the computer device outputs more than one reference tailings ecological information to the target mine matter label and receives more than one reference tailings ecological information returned by the target mine matter label and annotated with the independent tailings ecological restoration data result. Alternatively, the computer device directly receives more than one reference tailings ecological information corresponding to the target mine matter label, and each reference tailings ecological information is annotated with the independent tailings ecological restoration data result corresponding to the target mine matter label.

[0063] In a possible embodiment, the reference tailings ecological information may be only annotated with the independent tailings ecological restoration data evaluation value corresponding to the target mine matter label, or may be simultaneously annotated with the independent tailings ecological restoration data evaluation value corresponding to the target mine matter label and the independent tailings ecological restoration data evaluation value corresponding to other mine matter labels.

[0064] Step 504, obtain a unique mine matter significant feature extraction model, load more than one reference tailings ecological information into the unique mine matter significant feature extraction model respectively, and obtain the unique mine matter significant features corresponding to each of the more than one reference tailings ecological information through the unique mine matter significant feature extraction model.

[0065] Exemplarily, a unique mine matter is understood in this application as a mine mountain body that needs to be monitored.

[0066] Among them, for a piece of reference tailings ecological information, the significant features of a unique mine matter are the significant features unique to this reference tailings ecological information. The significant features of a unique mine matter may include several feature layers, each feature layer corresponding to a significant characteristic, and the feature value of each feature layer is used to represent the score of the mine matter in the corresponding significant characteristic, and this score is only related to the significant characteristic of the mine matter itself.

[0067] It can be understood that the feature value of the significant features of a unique mine matter is only related to the significant characteristic of the mine matter itself, and the feature value of the aforementioned significant features of the mine matter is not only related to the significant characteristic of the mine matter itself, but also related to the effect of the mine matter on the tailings ecological information.

[0068] Among them, the significant feature extraction model of a unique mine matter is a machine learning model trained with the ability to extract the significant features of a unique mine matter. The significant feature extraction model of a unique mine matter can be trained through examples of tailings ecological information and a configuration directory of the significant features of a unique mine matter.

[0069] Specifically, the computer device obtains the reference tailings ecological information and the significant feature extraction model of a unique mine matter. The computer device loads more than one piece of reference tailings ecological information into the significant feature extraction model of a unique mine matter respectively, and obtains the significant features of a unique mine matter corresponding to each of the more than one piece of reference tailings ecological information output by the significant feature extraction model of a unique mine matter.

[0070] Step 506, obtain the maximum value and the minimum value of the independent tailings ecological restoration data corresponding to each reference tailings ecological information and the target mine matter label.

[0071] Specifically, the computer device selects the maximum value and the minimum value of the independent tailings ecological restoration data from the evaluation values of the independent tailings ecological restoration data corresponding to each reference tailings ecological information and the target mine matter label.

[0072] For example, obtain 5 reference tailings ecological information corresponding to the target mine matter label: Reference tailings ecological information A, Reference tailings ecological information B, Reference tailings ecological information C, Reference tailings ecological information D, and Reference tailings ecological information E. The evaluation value of the independent tailings ecological restoration data corresponding to the reference tailings ecological information A and the target mine matter label is 1, the evaluation value of the independent tailings ecological restoration data corresponding to the reference tailings ecological information B and the target mine matter label is 2, the evaluation value of the independent tailings ecological restoration data corresponding to the reference tailings ecological information C and the target mine matter label is 3, the evaluation value of the independent tailings ecological restoration data corresponding to the reference tailings ecological information D and the target mine matter label is 4, and the evaluation value of the independent tailings ecological restoration data corresponding to the reference tailings ecological information E and the target mine matter label is 5. Then the maximum value of the independent tailings ecological restoration data is 5, and the minimum value of the independent tailings ecological restoration data is 1.

[0073] Step 508, for one of the reference tailings ecological information, based on the evaluation value of the independent tailings ecological restoration data corresponding to the target mine matter label, the maximum value of the independent tailings ecological restoration data, and the minimum value of the independent tailings ecological restoration data, obtain the uniqueness degree of the target mine matter label to this reference tailings ecological information.

[0074] In a possible embodiment, the computer device obtains a first difference between the maximum value of the independent tailings ecological restoration data and the minimum value of the independent tailings ecological restoration data, obtains a second difference between the evaluation value of the independent tailings ecological restoration data corresponding to the target mine matter label on the reference tailings ecological information and the minimum value of the independent tailings ecological restoration data, and determines the ratio between the second difference and the first difference as the uniqueness degree of the target mine matter label to this reference tailings ecological information.

[0075] Step 510, based on the uniqueness degree of the target mine matter label to each reference tailings ecological information, and the unique mine matter significance features corresponding to each reference tailings ecological information, obtain the mine matter significance features corresponding to the target mine matter label.

[0076] In this application, the significance features of the mine matter are adjusted by the uniqueness degree of the mine matter to the tailings ecological information, so that the feature value of the significance features of the mine matter is related to the effect of the mine matter on the tailings ecological information.

[0077] In a possible embodiment, for one of the reference tailings ecological information, the computer device adjusts the feature values of each layer of the unique mine matter saliency feature corresponding to the reference tailings ecological information according to the uniqueness of the reference tailings ecological information with respect to the target mine matter label, so as to obtain the mine matter saliency feature of the target mine matter label with respect to the reference tailings ecological information. The computer device obtains the average value or weighted average value of each layer of the mine matter saliency feature of the target mine matter label with respect to each reference tailings ecological information, so as to obtain the mine matter saliency feature corresponding to the target mine matter label.

[0078] In this embodiment, the mine matter saliency feature of the mine matter is obtained based on the uniqueness of the mine matter with respect to the reference tailings ecological information and the unique mine matter saliency feature of the reference tailings ecological information. The obtained mine matter saliency feature is related to the effect of the mine matter on the tailings ecological information. Subsequently, the evaluation result is determined based on the mine matter saliency feature and the tailings ecological restoration element feature, which can improve the accuracy of the evaluation result.

[0079] In a possible embodiment, it can be seen that the independent tailings ecological restoration data model includes a tailings ecological restoration element feature extraction model, a unique mine matter saliency feature extraction model, a restoration effect evaluation model, and an independent tailings ecological restoration data optimization model.

[0080] Specifically, first, the computer device obtains more than one reference tailings ecological information corresponding to the target mine matter label, and each reference tailings ecological information is annotated with an independent tailings ecological restoration data evaluation value corresponding to the target mine matter label. The computer device loads more than one reference tailings ecological information into the unique mine matter saliency feature extraction model respectively, and obtains the unique mine matter saliency feature corresponding to each of the more than one reference tailings ecological information through the unique mine matter saliency feature extraction model. The computer device obtains the mine matter saliency feature corresponding to the target mine matter label according to the independent tailings ecological restoration data evaluation value annotated on each reference tailings ecological information and corresponding to the target mine matter label, and the unique mine matter saliency feature corresponding to each reference tailings ecological information.

[0081] Secondly, after the computer device obtains the mine matter salience features corresponding to the target mine matter label, it loads the tailings ecological information to be evaluated into the tailings ecological restoration element feature extraction model, and outputs the tailings ecological restoration element features through the tailings ecological restoration element feature extraction model. Then, it loads the tailings ecological restoration element features into the restoration effect evaluation model, and outputs the estimated tailings ecological restoration result through the restoration effect evaluation model. Furthermore, through the independent tailings ecological restoration data optimization model, it obtains the independent tailings ecological restoration data debugging coefficient according to the tailings ecological restoration element features and the mine matter salience features of the target mine matter label. Finally, based on the estimated tailings ecological restoration result and the independent tailings ecological restoration data debugging coefficient, it determines the independent tailings ecological restoration data result corresponding to the target mine matter label.

[0082] In this embodiment, an independent tailings ecological restoration data model is built by using both the objective factors of the tailings ecological information and the subjective factors of the mine matters, that is, the personalization and popularization of the tailings ecological information are modeled by using different mine matter salience features, so that the independent tailings ecological restoration data result obtained through the independent tailings ecological restoration data model can better meet the uniqueness of individual mine matters.

[0083] This application relates to a training method for an independent tailings ecological restoration data model, which will be specifically described through the following embodiments.

[0084] First, for the tailings ecological restoration element feature extraction model and the unique mine matter salience feature extraction model, the tailings ecological restoration element feature extraction model and the unique mine matter salience feature extraction model can be trained separately, that is, the tailings ecological restoration element feature extraction model is trained through the tailings ecological information examples and the tailings ecological restoration element feature configuration directory, and the unique mine matter salience feature extraction model is trained through the tailings ecological information examples and the unique mine matter salience feature configuration directory. The tailings ecological restoration element feature extraction model and the unique mine matter salience feature extraction model can also be trained jointly. The following provides a method for joint training.

[0085] In a possible embodiment, the tailings ecological restoration element feature extraction model is jointly trained with the unique mine matter significance feature extraction model; the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model have a common standard local model, and each includes a corresponding output local model; the steps of jointly training the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model include: obtaining a first set of tailings ecological information examples, a second set of tailings ecological information examples, the tailings ecological restoration element feature extraction model, and the unique mine matter significance feature extraction model; each first tailings ecological information example in the first set of tailings ecological information examples has a tailings ecological restoration element feature configuration directory, and each second tailings ecological information example in the second set of tailings ecological information examples has a unique mine matter significance feature configuration directory; loading the first tailings ecological information examples into the tailings ecological restoration element feature extraction model, extracting features from the first tailings ecological information examples through the standard local model of the tailings ecological restoration element feature extraction model, and outputting the tailings ecological restoration element feature regression analysis result through the output local model of the tailings ecological restoration element feature extraction model; loading the second tailings ecological information examples into the unique mine matter significance feature extraction model, extracting features from the second tailings ecological information examples through the standard local model of the unique mine matter significance feature extraction model, and outputting the unique mine matter significance feature regression analysis result through the output local model of the unique mine matter significance feature extraction model; jointly training the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model based on the tailings ecological restoration element feature regression analysis result and the tailings ecological restoration element feature configuration directory, as well as the unique mine matter significance feature regression analysis result and the unique mine matter significance feature configuration directory.

[0086] Among them, the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model have a common standard local model, and each includes a corresponding output local model. The number of the standard local models can be one or two. When the number of the standard local models is one, the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model share one standard local model, and this standard local model is connected to two output local models; when the number of the standard local models is two, the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model each correspond to one standard local model, the two standard local models are each connected to one output local model, and the model structures of the two standard local models are the same.

[0087] Specifically, taking two standard local models as an example, the computer device loads the first tailings ecological information example into the tailings ecological restoration element feature extraction model, extracts features from the first tailings ecological information example through the standard local model of the tailings ecological restoration element feature extraction model, and outputs the tailings ecological restoration element feature regression analysis result through the output local model of the tailings ecological restoration element feature extraction model. Then, the computer device loads the second tailings ecological information example into the unique mine matter significance feature extraction model, extracts features from the second tailings ecological information example through the standard local model of the unique mine matter significance feature extraction model, and outputs the unique mine matter significance feature regression analysis result through the output local model of the unique mine matter significance feature extraction model. Then, the computer device jointly trains the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model based on the distinction between the tailings ecological restoration element feature regression analysis result and the tailings ecological restoration element feature configuration directory, and the distinction between the unique mine matter significance feature regression analysis result and the unique mine matter significance feature configuration directory.

[0088] In a possible embodiment, jointly training the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model based on the tailings ecological restoration element feature regression analysis result and the tailings ecological restoration element feature configuration directory, and the unique mine matter significance feature regression analysis result and the unique mine matter significance feature configuration directory includes: building a first evaluation index algorithm based on the distinction between the tailings ecological restoration element feature regression analysis result and the tailings ecological restoration element feature configuration directory, and building a second evaluation index algorithm based on the distinction between the unique mine matter significance feature regression analysis result and the unique mine matter significance feature configuration directory; obtaining a first model pyramid coefficient by minimizing the first evaluation index algorithm, and obtaining a second model pyramid coefficient by minimizing the second evaluation index algorithm; combining the first model pyramid coefficient and the second model pyramid coefficient to jointly train the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model.

[0089] Among them, the evaluation index algorithm can be understood as a loss function in this application. The first evaluation index algorithm is set in advance, and the second evaluation index algorithm is obtained through debugging according to the actual situation.

[0090] Among them, the specific calculation method of the first evaluation index algorithm is as follows: For a set of predicted values \(\hat{y}_i\) and true values \(y_i\), where \(i = 1,2,\cdots,n\), the calculation formula of the mean square error loss function is: \(MSE=\frac{1}{n}\sum_{i = 1}^{n}(\hat{y}_i - y_i)^2\) Calculation steps 1. Calculate the difference between the predicted value and the true value: For each data point, calculate \(\hat{y}_i - y_i\).

[0091] 2. Square the difference: Square the difference obtained in the previous step, i.e., \((\hat{y}_i - y_i)^2\). This can avoid the cancellation of positive and negative errors and emphasize the impact of larger errors at the same time.

[0092] 3. Sum: Sum up the squared differences of all data points to get \(\sum_{i = 1}^{n}(\hat{y}_i -y_i)^2\).

[0093] 4. Calculate the average: Divide the sum result by the number of data points \(n\) to get the mean squared error MSE.

[0094] The mean squared error loss function is often used in regression problems to measure the average difference between the predicted value and the true value of the model. The smaller its value, the better the prediction effect of the model. Specifically, the computer device constructs the first evaluation index algorithm based on the distinction between the regression analysis results of the tailings ecological restoration element characteristics and the tailings ecological restoration element characteristics configuration directory. For example, the first evaluation index algorithm is constructed through the Euclidean distance between the regression analysis results of the tailings ecological restoration element characteristics and the tailings ecological restoration element characteristics configuration directory; and the second evaluation index algorithm is constructed based on the distinction between the regression analysis results of the unique mine matter significance characteristics and the unique mine matter significance characteristics configuration directory. For example, the second evaluation index algorithm is constructed through the Euclidean distance between the regression analysis results of the unique mine matter significance characteristics and the unique mine matter significance characteristics configuration directory. Then, the computer device obtains the first model pyramid coefficient by minimizing the first evaluation index algorithm and obtains the second model pyramid coefficient by minimizing the second evaluation index algorithm. Then, the computer device combines the first model pyramid coefficient and the second model pyramid coefficient to jointly train the tailings ecological restoration element characteristics extraction model and the unique mine matter significance characteristics extraction model. For example, the sum of the first model pyramid coefficient and the second model pyramid coefficient is obtained, and the model coefficients of the tailings ecological restoration element characteristics extraction model and the unique mine matter significance characteristics extraction model are updated by backpropagation using the sum of the first model pyramid coefficient and the second model pyramid coefficient.

[0095] In this embodiment, since the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model share a common standard local model, the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model can adopt a co-training method, thereby saving computer resources and improving the model training speed. Moreover, by obtaining the objective factors of tailings ecological information and the subjective factors of mine matters in the same model, it has practical value for the research of independent tailings ecological restoration data.

[0096] In a possible embodiment, the steps of extracting the tailings ecological restoration element feature example corresponding to the fourth tailings ecological information example include: obtaining the tailings ecological restoration element feature extraction model; loading the fourth tailings ecological information example into the tailings ecological restoration element feature extraction model, and extracting the features of the fourth tailings ecological information example through the tailings ecological restoration element feature extraction model to obtain the tailings ecological restoration element feature example.

[0097] Specifically, the computer device loads the fourth tailings ecological information example into the tailings ecological restoration element feature extraction model, and extracts the tailings ecological restoration element feature example corresponding to the fourth tailings ecological information example through the tailings ecological restoration element feature extraction model.

[0098] It can be understood that by weighted summing the evaluation value probability distribution regression analysis result and the corresponding evaluation value, the tailings ecological restoration prediction result can be obtained.

[0099] In a possible embodiment, the training steps of the independent tailings ecological restoration data optimization model include: obtaining the third tailings ecological information example corresponding to the example mine matter label and the independent tailings ecological restoration data optimization model; there is an independent tailings ecological restoration data debugging coefficient configuration directory corresponding to the example mine matter label for the third tailings ecological information example; extracting the tailings ecological restoration element feature example corresponding to the third tailings ecological information example; obtaining the mine matter significance feature example corresponding to the example mine matter label; through the independent tailings ecological restoration data optimization model, according to the splicing result of the tailings ecological restoration element feature example and the mine matter significance feature example, obtaining the independent tailings ecological restoration data debugging coefficient regression analysis result corresponding to the example mine matter label; based on the independent tailings ecological restoration data debugging coefficient regression analysis result and the independent tailings ecological restoration data debugging coefficient configuration directory, training the independent tailings ecological restoration data optimization model.

[0100] In a possible embodiment, the computer device first constructs an independent tailings ecological restoration data optimization model composed of a fully connected model, and then trains the independent tailings ecological restoration data optimization model using the third tailings ecological information example set and the independent tailings ecological restoration data debugging coefficient configuration directory.

[0101] In a possible embodiment, the third tailings ecological information example corresponds to more than one example mine matter label; the steps for obtaining the independent tailings ecological restoration data debugging coefficient configuration directory include: obtaining the independent tailings ecological restoration data result examples of each example mine matter label corresponding to the third tailings ecological information example; obtaining the independent tailings ecological restoration data mean example based on each independent tailings ecological restoration data result example; obtaining the independent tailings ecological restoration data debugging coefficient examples corresponding to each example mine matter label based on each independent tailings ecological restoration data result example and the independent tailings ecological restoration data mean example, and determining each independent tailings ecological restoration data debugging coefficient example as the independent tailings ecological restoration data debugging coefficient configuration directory corresponding to the third tailings ecological information example and each example mine matter label.

[0102] Specifically, since there is an evaluation value corresponding to one example mine matter label in the third tailings ecological information example, for one of the third tailings ecological information examples, the average value of the evaluation values corresponding to this third tailings ecological information example is obtained. For one of the example mine matter labels, the difference between its corresponding evaluation value and the average value of the evaluation values is the independent tailings ecological restoration data debugging coefficient of this example mine matter label.

[0103] Specifically, the computer device loads the third tailings ecological information example into the tailings ecological restoration element feature extraction model, and extracts the tailings ecological restoration element feature example from the third tailings ecological information example through the tailings ecological restoration element feature extraction model. Then, the computer device obtains the mine matter significance feature example corresponding to the example mine matter label. Then, the computer device obtains the regression analysis result of the independent tailings ecological restoration data debugging coefficient corresponding to the example mine matter label through the independent tailings ecological restoration data optimization model according to the concatenation result of the tailings ecological restoration element feature example and the mine matter significance feature example. Then, the computer device constructs an evaluation index algorithm based on the difference between the regression analysis result of the independent tailings ecological restoration data debugging coefficient and the independent tailings ecological restoration data debugging coefficient configuration directory, and trains the independent tailings ecological restoration data optimization model in the direction of minimizing the evaluation index algorithm.

[0104] This application involves the mine matter significance feature example. The following provides a method for obtaining the mine matter significance feature example corresponding to an example mine matter label.

[0105] In a possible embodiment, obtaining the example of the significance feature of the mine matter corresponding to the example mine matter label includes: obtaining the example of the unique significance feature of the mine matter corresponding to the example of the third tailings ecological information; obtaining the example of the result of the independent tailings ecological restoration data corresponding to the example mine matter label; and obtaining the example of the significance feature of the mine matter corresponding to the example mine matter label based on the example of the result of the independent tailings ecological restoration data and the example of the unique significance feature of the mine matter.

[0106] In a possible embodiment, the example mine matter label corresponds to no less than two examples of the third tailings ecological information; the example of the result of the independent tailings ecological restoration data is the example of the evaluation value of the independent tailings ecological restoration data; and obtaining the example of the significance feature of the mine matter corresponding to the example mine matter label based on the example of the result of the independent tailings ecological restoration data and the example of the unique significance feature of the mine matter includes: obtaining the example of the maximum value of the independent tailings ecological restoration data and the example of the minimum value of the independent tailings ecological restoration data of no less than two examples of the third tailings ecological information; for one example of the third tailings ecological information, obtaining the uniqueness degree of the example mine matter label to the example of the third tailings ecological information based on the example of the evaluation value of the independent tailings ecological restoration data, the example of the maximum value of the independent tailings ecological restoration data, and the example of the minimum value of the independent tailings ecological restoration data corresponding to the example mine matter label; and obtaining the example of the significance feature of the mine matter corresponding to the example mine matter label based on the uniqueness degree of the example mine matter label to each example of the third tailings ecological information and the example of the unique significance feature of the mine matter corresponding to each example of the third tailings ecological information.

[0107] In a possible embodiment, the number of feature layers of the example of the unique significance feature of the mine matter is several; and obtaining the example of the significance feature of the mine matter corresponding to the example mine matter label based on the uniqueness degree and the example of the unique significance feature of the mine matter includes: adjusting the feature values of the example of the unique significance feature of the mine matter in each layer according to the uniqueness degree to obtain the example of the significance feature of the mine matter corresponding to the example mine matter label.

[0108] For the process of obtaining the example of the significance feature of the mine matter corresponding to the example mine matter label above, reference can be specifically made to the method of obtaining the significance feature of the target mine matter label described above, which will not be elaborated here.

[0109] This application also provides an application scenario, which applies the intelligent evaluation method for the ecological problems of mine tailings described above. Specifically, the application of the intelligent evaluation method for the ecological problems of mine tailings in this application scenario is as follows: Step 802, obtain the tailings ecological information to be evaluated and the target mine matter label.

[0110] Step 804: Obtain the tailings ecological restoration element feature extraction model, load the tailings ecological information to be evaluated into the tailings ecological restoration element feature extraction model, and output the tailings ecological restoration element features through the tailings ecological restoration element feature extraction model.

[0111] Step 806: Obtain the restoration effect evaluation model, load the tailings ecological restoration element features into the restoration effect evaluation model, and output the predicted result of tailings ecological restoration through the restoration effect evaluation model.

[0112] Step 808: Obtain the independent tailings ecological restoration data optimization model. Through the independent tailings ecological restoration data optimization model, according to the splicing result of the tailings ecological restoration element features and the significance features of mine matters, obtain the debugging coefficient of the independent tailings ecological restoration data.

[0113] Among them, the method for obtaining the significance features of mine matters is as follows: The computer device obtains more than one reference tailings ecological information corresponding to the target mine matter label, and each reference tailings ecological information is annotated with the evaluation value of the independent tailings ecological restoration data corresponding to the target mine matter label; Then, the computer device obtains the unique mine matter significance feature extraction model, loads more than one reference tailings ecological information into the unique mine matter significance feature extraction model respectively, and obtains the unique mine matter significance features corresponding to each of the more than one reference tailings ecological information through the unique mine matter significance feature extraction model; Then, the computer device obtains the maximum value and the minimum value of the independent tailings ecological restoration data corresponding to each reference tailings ecological information and the target mine matter label; Then, for one of the reference tailings ecological information, the computer device obtains the uniqueness degree of the target mine matter label to this reference tailings ecological information based on the evaluation value of the independent tailings ecological restoration data, the maximum value of the independent tailings ecological restoration data, and the minimum value of the independent tailings ecological restoration data corresponding to the target mine matter label; Then, the computer device obtains the significance features of the target mine matter label based on the uniqueness degree of the target mine matter label to each reference tailings ecological information and the unique mine matter significance features corresponding to each reference tailings ecological information.

[0114] Step 810: Determine the independent tailings ecological restoration data result corresponding to the target mine matter label based on the predicted result of tailings ecological restoration and the debugging coefficient of the independent tailings ecological restoration data.

[0115] In a possible embodiment, the training of the feature extraction model for tailings ecological restoration elements. For the feature extraction model of tailings ecological restoration elements and the significance feature extraction model of unique mine matters, load the first tailings ecological information example into the feature extraction model of tailings ecological restoration elements, and output the regression analysis result of the features of tailings ecological restoration elements through the feature extraction model of tailings ecological restoration elements. Load the second tailings ecological information example into the significance feature extraction model of unique mine matters, and output the regression analysis result of the significance features of unique mine matters through the significance feature extraction model of unique mine matters. Build the first evaluation index algorithm based on the distinction between the regression analysis result of the features of tailings ecological restoration elements and the configuration directory of the features of tailings ecological restoration elements, and build the second evaluation index algorithm based on the distinction between the regression analysis result of the significance features of unique mine matters and the configuration directory of the significance features of unique mine matters. Obtain the first model pyramid coefficient by minimizing the first evaluation index algorithm, and obtain the second model pyramid coefficient by minimizing the second evaluation index algorithm. Combine the first model pyramid coefficient and the second model pyramid coefficient to jointly train the feature extraction model of tailings ecological restoration elements and the significance feature extraction model of unique mine matters.

[0116] Among them, the training of the feature extraction model is understood as the training of a convolutional neural network. A convolutional neural network (Convolutional Neural Network, CNN) mainly includes a convolutional layer, a pooling layer, a fully connected layer, etc. The following are the calculation methods of each layer: Convolutional layer Convolution operation: Perform a convolution operation by sliding the convolution kernel on the input data. Assume that the input data is a three-dimensional tensor X with a shape of (H_{in}, W_{in}, C_{in}), representing height, width, and number of channels respectively. The shape of the convolution kernel K is (h, k, C_{in}, C_{out}), where h and k are the height and width of the convolution kernel, and C_{out} is the number of output channels. For each position (i, j) in the input data, the value of the output Y of the convolutional layer at the corresponding position (i', j') is calculated as: Y(i', j', c)=\sum_{c' = 0}^{C_{in}-1}\sum_{m = 0}^{h - 1}\sum_{n = 0}^{k - 1}X(i'+m, j'+n, c')\times K(m, n, c', c) Among them, the value ranges of i' and j' are determined according to the size, stride, and padding method of the convolution kernel.

[0117] Bias addition: Usually, a bias term b with a shape of (1, 1, C_{out}) is added to the result of the convolution operation. The final output of the convolutional layer is Z = Y + b.

[0118] Pooling layer Max pooling: Divide the input data into several non - overlapping sub - regions, and take the maximum value in each sub - region as the output of that region. For example, for a 2×2 max - pooling kernel with a stride of 2, if the shape of the input data X is (H_{in}, W_{in}, C_{in}), then the shape of the output data Y is (\frac{H_{in}}{2},\frac{W_{in}}{2},C_{in}), where Y(i,j,c)=\max_{m = 0}^{1}\max_{n = 0}^{1}X(2i + m,2j + n,c).

[0119] Average pooling: Similar to max pooling, but take the average value in each sub - region as the output.

[0120] Fully - connected layer The fully - connected layer unfolds the output of the previous layer into a one - dimensional vector, and then performs calculations through matrix multiplication and bias addition. Suppose the length of the vector output by the previous layer is n, the shape of the weight matrix W of the fully - connected layer is (n,m), and the shape of the bias vector b is (1,m), then the output Y of the fully - connected layer is Y = XW + b, where X is the one - dimensional vector after unfolding the previous layer, and Y is a one - dimensional vector of length m.

[0121] In practical applications, a convolutional neural network usually contains multiple convolutional layers, pooling layers and fully - connected layers, and combines these layers to achieve feature extraction and classification or regression of the input data.

[0122] In a possible embodiment, the training of the repair effect evaluation model in an embodiment. Load the fourth tailings ecological information example into the tailings ecological restoration element feature extraction model, and output the tailings ecological restoration element feature example through the tailings ecological restoration element feature extraction model. Then, load the tailings ecological restoration element feature example into the repair effect evaluation model, and obtain the evaluation value probability distribution regression analysis result through the repair effect evaluation model. Then, build an evaluation index algorithm based on the difference between the evaluation value probability distribution regression analysis result and the evaluation value probability distribution configuration directory, and train the repair effect evaluation model in the direction of minimizing the evaluation index algorithm.

[0123] In a possible embodiment, the training of an independent tailings ecological restoration data optimization model in one embodiment. Load the third tailings ecological information example into the tailings ecological restoration element feature extraction model, and output the tailings ecological restoration element feature example through the tailings ecological restoration element feature extraction model. Then, load the third tailings ecological information example into the unique mine matter significance feature extraction model, and output the unique mine matter significance feature example through the unique mine matter significance feature extraction model. Based on the unique mine matter significance feature example, obtain the mine matter significance feature example corresponding to the example mine matter label. Then, through the independent tailings ecological restoration data optimization model, according to the splicing result of the tailings ecological restoration element feature example and the mine matter significance feature example, obtain the regression analysis result of the independent tailings ecological restoration data debugging coefficient corresponding to the example mine matter label. Then, based on the distinction between the regression analysis result of the independent tailings ecological restoration data debugging coefficient and the independent tailings ecological restoration data debugging coefficient configuration directory, build an evaluation index algorithm, and train the independent tailings ecological restoration data optimization model in the direction of minimizing the evaluation index algorithm.

[0124] On this basis, an intelligent evaluation device for mine tailings ecological problems is provided. The device includes: A knowledge fragment extraction module, configured to perform knowledge fragment extraction processing on the example collapse hazard data to obtain the secondary knowledge fragment and the important knowledge fragment of the example collapse hazard data; A label construction module, configured to construct secondary description labels according to the secondary knowledge fragments respectively corresponding to a plurality of the example collapse hazard data, and construct important description labels according to the important knowledge fragments respectively corresponding to a plurality of the example collapse hazard data; A knowledge fragment obtaining module, configured to perform knowledge fragment extraction processing on the to-be-processed collapse hazard data to obtain the secondary knowledge fragment and the important knowledge fragment of the to-be-processed collapse hazard data; A label analysis module, configured to perform similarity analysis processing on the secondary description labels according to the secondary knowledge fragment of the to-be-processed collapse hazard data to obtain similar secondary description labels, and perform similarity analysis processing on the important description labels according to the important knowledge fragment of the to-be-processed collapse hazard data to obtain similar important description labels; An information determination module, configured to determine at least some of the example collapse hazard data corresponding to the similar secondary description labels and the example collapse hazard data corresponding to the similar important description labels as the collapse hazard warning information of the to-be-processed collapse hazard data.

[0125] On the basis described above, an intelligent evaluation system for ecological problems of mine tailings is shown, including a processor and a memory that communicate with each other. The processor is configured to read and execute a computer program from the memory to implement the above method.

[0126] On the basis described above, a computer-readable storage medium is further provided, and the computer program stored thereon implements the above method when running.

[0127] In summary, based on the above solution, first determine the estimated result of tailings ecological restoration according to the characteristics of tailings ecological restoration elements of the tailings ecological information to be evaluated as needed, and determine the independent tailings ecological restoration data debugging coefficient according to the characteristics of tailings ecological restoration elements and the significance characteristics of mine matters corresponding to the target mine matter label. Then, based on the independent tailings ecological restoration data debugging coefficient, debug the estimated result of tailings ecological restoration to obtain the independent tailings ecological restoration data result corresponding to the target mine matter label. In this way, since the estimated result of tailings ecological restoration is the result obtained by evaluating the characteristics of the tailings ecological information itself, it reflects the quality of the tailings ecological information at the objective level. And by debugging the estimated result of tailings ecological restoration according to the deviation corresponding to the significance characteristics of mine matters, the obtained independent tailings ecological restoration data result approaches the evaluation result obtained by the mine matter based on itself, that is, combining the subjectivity of the mine matter and the objective quality of the tailings ecological information to perform independent tailings ecological restoration data on the tailings ecological information, so that the independent tailings ecological restoration data result meets the uniqueness of individual mine matters, thereby improving the accuracy of the independent tailings ecological restoration data result.

[0128] It should be understood that the above-described system and its modules can be implemented in various ways. For example, in some embodiments, the system and its modules can be implemented by hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in the memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art can understand that the above methods and systems can be implemented using computer-executable instructions and / or included in processor control code. For example, such code is provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The system and its modules of the present application can be implemented not only by a hardware circuit of a programmable hardware device such as a very large scale integrated circuit or a gate array, a semiconductor such as a logic chip or a transistor, or a programmable logic device such as a field programmable gate array, but also by software implemented by various types of processors, or by a combination of the above hardware circuits and software (for example, firmware).

[0129] It should be noted that the beneficial effects that may be produced by different embodiments are different. In different embodiments, the beneficial effects that may be produced can be any one or several combinations of the above, or any other beneficial effects that may be obtained.

Claims

1. An intelligent evaluation method for ecological problems of mine tailings, characterized in that, The method includes: Obtaining the tailings ecological information to be evaluated and the target mine matter label; Extracting the tailings ecological restoration element features corresponding to the tailings ecological information to be evaluated; Combining the tailings ecological restoration element features to determine the tailings ecological restoration prediction result of the tailings ecological information to be evaluated; Through an independent tailings ecological restoration data optimization model, combining the splicing result of the tailings ecological restoration element features and the mine matter significance features corresponding to the target mine matter label, determining the independent tailings ecological restoration data debugging coefficient corresponding to the target mine matter label; the mine matter significance features are data reflecting the significance characteristics of mine matters; Based on the tailings ecological restoration prediction result and the independent tailings ecological restoration data debugging coefficient, determining the independent tailings ecological restoration data result corresponding to the target mine matter label.

2. The method according to claim 1, wherein The method further includes: Obtaining a plurality of reference tailings ecological information corresponding to the target mine matter label, and each reference tailings ecological information is annotated with an independent tailings ecological restoration data evaluation value corresponding to the target mine matter label; Loading the plurality of reference tailings ecological information into a unique mine matter significance feature extraction model respectively, and obtaining the unique mine matter significance features corresponding to the plurality of reference tailings ecological information through the unique mine matter significance feature extraction model; the unique mine matter significance features are the significance features of the reference tailings ecological information corresponding to the unique mine matter significance features; Obtaining the maximum value of the independent tailings ecological restoration data and the minimum value of the independent tailings ecological restoration data corresponding to each reference tailings ecological information and the target mine matter label; For each reference tailings ecological information, obtaining the uniqueness degree of the target mine matter label to the reference tailings ecological information based on the independent tailings ecological restoration data evaluation value, the independent tailings ecological restoration data maximum value, and the independent tailings ecological restoration data minimum value corresponding to the target mine matter label; Based on the uniqueness degree of the target mine matter label to each reference tailings ecological information and the unique mine matter significance features corresponding to each reference tailings ecological information, obtaining the mine matter significance features corresponding to the target mine matter label.

3. The method according to claim 2, characterized in that, The extracting the tailings ecological restoration element features corresponding to the tailings ecological information to be evaluated includes: Obtaining a tailings ecological restoration element feature extraction model; Loading the tailings ecological information to be evaluated into the tailings ecological restoration element feature extraction model, and outputting the tailings ecological restoration element features through the tailings ecological restoration element feature extraction model.

4. The method according to claim 3, characterized in that, The tailings ecological restoration element feature extraction model is jointly trained with the unique mine matter significance feature extraction model; the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model are common standard local models, and each includes a corresponding output local model; The steps of jointly training the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model include: Obtain a first set of example tailings ecological information, a second set of example tailings ecological information, the tailings ecological restoration element feature extraction model, and the unique mine matter significance feature extraction model; each first example tailings ecological information in the first set of example tailings ecological information has a tailings ecological restoration element feature configuration directory, and each second example tailings ecological information in the second set of example tailings ecological information has a unique mine matter significance feature configuration directory; Load the first set of example tailings ecological information into the tailings ecological restoration element feature extraction model, perform feature extraction on the first set of example tailings ecological information through the standard local model of the tailings ecological restoration element feature extraction model, and output the tailings ecological restoration element feature regression analysis result through the output local model of the tailings ecological restoration element feature extraction model; Load the second set of example tailings ecological information into the unique mine matter significance feature extraction model, perform feature extraction on the second set of example tailings ecological information through the standard local model of the unique mine matter significance feature extraction model, and output the unique mine matter significance feature regression analysis result through the output local model of the unique mine matter significance feature extraction model; Based on the tailings ecological restoration element feature regression analysis result and the tailings ecological restoration element feature configuration directory, as well as the unique mine matter significance feature regression analysis result and the unique mine matter significance feature configuration directory, jointly train the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model.

5. The method according to claim 4, wherein The step of jointly training the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model based on the tailings ecological restoration element feature regression analysis result and the tailings ecological restoration element feature configuration directory, as well as the unique mine matter significance feature regression analysis result and the unique mine matter significance feature configuration directory, includes: Build a first evaluation index algorithm based on the difference between the tailings ecological restoration element feature regression analysis result and the tailings ecological restoration element feature configuration directory, and build a second evaluation index algorithm based on the difference between the unique mine matter significance feature regression analysis result and the unique mine matter significance feature configuration directory; Obtain the first model pyramid coefficient by minimizing the first evaluation index algorithm, and obtain the second model pyramid coefficient by minimizing the second evaluation index algorithm; Combine the first model pyramid coefficient and the second model pyramid coefficient to jointly train the tailings ecological restoration element feature extraction model and the unique mine matter significance feature extraction model.

6. The method according to claim 2, wherein The training steps of the independent tailings ecological restoration data optimization model include: Obtain a third set of example tailings ecological information corresponding to the example mine matter label and the independent tailings ecological restoration data optimization model; the third set of example tailings ecological information has an independent tailings ecological restoration data debugging coefficient configuration directory corresponding to the example mine matter label; Extract the example of the tailings ecological restoration element characteristics corresponding to the third tailings ecological information example; Obtain the example of the significance characteristics of the mine matters corresponding to the example mine matter label; Through the independent tailings ecological restoration data optimization model, combine the splicing results of the example of the tailings ecological restoration element characteristics and the example of the significance characteristics of the mine matters to obtain the regression analysis result of the independent tailings ecological restoration data debugging coefficient corresponding to the example mine matter label; Train the independent tailings ecological restoration data optimization model based on the regression analysis result of the independent tailings ecological restoration data debugging coefficient and the configuration directory of the independent tailings ecological restoration data debugging coefficient; Among them, the third tailings ecological information example corresponds to more than one example mine matter label; The obtaining steps of the configuration directory of the independent tailings ecological restoration data debugging coefficient include: Obtain the example of the independent tailings ecological restoration data result corresponding to each example mine matter label of the third tailings ecological information example; Obtain the example of the independent tailings ecological restoration data mean based on each example of the independent tailings ecological restoration data result; Based on each example of the independent tailings ecological restoration data result and the example of the independent tailings ecological restoration data mean, obtain the example of the independent tailings ecological restoration data debugging coefficient corresponding to each example mine matter label, and determine each example of the independent tailings ecological restoration data debugging coefficient as the configuration directory of the independent tailings ecological restoration data debugging coefficient corresponding to the third tailings ecological information example and each example mine matter label.

7. The method according to claim 6, wherein The obtaining of the example of the significance characteristics of the mine matters corresponding to the example mine matter label includes: Obtain the example of the unique significance characteristics of the mine matters corresponding to the third tailings ecological information example; Obtain the example of the independent tailings ecological restoration data result corresponding to the example mine matter label; Based on the example of the independent tailings ecological restoration data result and the example of the unique significance characteristics of the mine matters, obtain the example of the significance characteristics of the mine matters corresponding to the example mine matter label; Among them, the obtaining steps of the example of the tailings ecological restoration element characteristics and the example of the unique significance characteristics of the mine matters corresponding to the third tailings ecological information example include: Obtain the tailings ecological restoration element characteristic extraction model and the unique mine matter significance characteristic extraction model; Load the third tailings ecological information example into the tailings ecological restoration element characteristic extraction model, perform feature extraction on the third tailings ecological information example through the standard local model of the tailings ecological restoration element characteristic extraction model, and output the example of the tailings ecological restoration element characteristics through the output local model of the tailings ecological restoration element characteristic extraction model; Load the third tailings ecological information example into the unique mine matter significance characteristic extraction model, perform feature extraction on the third tailings ecological information example through the standard local model of the unique mine matter significance characteristic extraction model, and output the example of the unique mine matter significance characteristics through the output local model of the unique mine matter significance characteristic extraction model.

8. The method according to claim 7, characterized in that The described example mine matter label corresponds to no less than two third tailings ecological information examples; the independent tailings ecological restoration data result example is an independent tailings ecological restoration data evaluation value example; based on the independent tailings ecological restoration data result example and the unique mine matter significance feature example, obtaining the mine matter significance feature example corresponding to the example mine matter label includes: Obtaining the maximum value example of the independent tailings ecological restoration data and the minimum value example of the independent tailings ecological restoration data of no less than two of the third tailings ecological information examples; For one of the third tailings ecological information examples, based on the independent tailings ecological restoration data evaluation value example corresponding to the example mine matter label, the independent tailings ecological restoration data maximum value example, and the independent tailings ecological restoration data minimum value example, obtaining the uniqueness degree of the example mine matter label to the third tailings ecological information example; Based on the uniqueness degree of the example mine matter label to each of the third tailings ecological information examples and the unique mine matter significance feature examples corresponding to each of the third tailings ecological information examples, obtaining the mine matter significance feature example corresponding to the example mine matter label; Among them, the feature layer of the unique mine matter significance feature example is several; obtaining the mine matter significance feature example corresponding to the example mine matter label based on the uniqueness degree and the unique mine matter significance feature example includes: Combining the uniqueness degree, adjusting the feature values of the unique mine matter significance feature example in each layer to obtain the mine matter significance feature example corresponding to the example mine matter label.

9. The method according to claim 1, wherein The tailings ecological information to be evaluated is several; the independent tailings ecological restoration data result is an independent tailings ecological restoration data evaluation value; the method further includes: Obtaining the independent tailings ecological restoration data evaluation value of the target mine matter label for each of the tailings ecological information to be evaluated; Selecting a preset number of target tailings ecological information from each of the tailings ecological information to be evaluated according to the independent tailings ecological restoration data evaluation value; Outputting the target tailings ecological information to the terminal where the target mine matter label is located.

10. An intelligent evaluation system for ecological problems of mine tailings, characterized in that, Including a processor and a memory that communicate with each other, the processor is configured to read and execute a computer program from the memory to implement the method according to any one of claims 1-9.

Images