CN-121997155-A - Coal mine fault control degree evaluation method and system

CN121997155ACN 121997155 ACN121997155 ACN 121997155ACN-121997155-A

Abstract

The invention discloses a coal mine fault control degree evaluation method and system, and relates to the technical field of fault identification reliability evaluation. The method comprises the steps of fusing a mining area characteristic set and a label set, establishing a training set and a set to be predicted, establishing a support vector machine two-class model and training, inputting a test set into the trained model, outputting posterior probability values P 1 of sample points, determining a dynamic threshold according to the posterior probability values, mapping the posterior probability values into fault point categories by using the dynamically determined threshold, evaluating the quality of break points, and evaluating the fault risk level of a region with dense distribution of fault points in the mining area by using the ratio of the break points with different quality. The method can not only maintain the objective and high-efficiency of AI, but also restore and even surpass the reliability quantitative evaluation capability of manual experience.

Inventors

REN KE
ZHU YUZHEN
SUN CHAO
LI NA
ZHANG WENYAN

Assignees

空天信息大学(筹)
山东省煤田地质规划勘察研究院

Dates

Publication Date: 20260508
Application Date: 20260120

Claims (6)

1. The coal mine fault control degree evaluation method is characterized by comprising the following specific steps of: Extracting seismic attribute values and constructing a full-area feature set; Making a tag set by counting the exposure information of drilling holes and roadways in a mining area, fusing the tag set with the whole area feature set, and establishing a training set A and a set B to be predicted, wherein the number of fault sample points of the training set A is as follows The number of non-fault sample points is ; Building a support vector machine two-class model, dividing the training set into a first training set and a first testing set, training the support vector machine two-class model by using the first training set, wherein the first testing set is used for testing the support vector machine two-class model until the training precision reaches a preset threshold value, and outputting the trained support vector machine two-class model; Inputting the first test set into a trained support vector machine two-class model, outputting by using a support vector machine posterior probability model, and recording posterior probability values P 1 of each sample point of the first test set output by the model; Ordering the posterior probability values P 1 and recording the th Size of the individual value And (d) Size of +1 values ; Predicting by using the set B to be predicted through a trained support vector machine two-class model, outputting by using a support vector machine posterior probability model, and recording posterior probability values P 2 of each sample point of the set to be predicted, which is output by the model; Evaluating the quality of the break points according to the probability value P 2 , and dividing the reliability of the break points; And evaluating fault risk levels in the mining area according to the breakpoint duty ratios of different qualities.
2. The method for evaluating the control degree of the coal mine fault according to claim 1, wherein the specific steps of extracting the seismic attribute values and constructing the full-area feature set are as follows: Acquiring an original seismic data volume of a target stratum of a mining area; And determining the position of a phase axis of the seismic reflection wave of the target horizon according to logging data, tracking the target horizon of the original seismic data body to obtain horizon information of the target stratum, extracting seismic attribute values along the target horizon of the original seismic data body, and constructing the total-area feature set.
3. A method of evaluating the extent of fault control in a coal mine as claimed in claim 1, in which the seismic attribute values include variance, curvature, instantaneous amplitude.
4. The method for evaluating the control degree of the coal mine fault according to claim 1, wherein the mathematical expression for evaluating and judging the quality of the break point is: ; Wherein, the The breakpoint evaluation result is represented by a, a represents a reliable fault point, B represents a more reliable fault point, C represents an unreliable fault point, and D represents a non-fault point.
5. The method for evaluating the control degree of a coal mine fault as claimed in claim 4, wherein the mathematical expression for evaluating the fault risk level in the mine area is: ; Wherein, the The fault evaluation result is represented by 2, 1, 0, a number of class A breakpoints, B number of class B breakpoints and C number of class C breakpoints.
6. The coal mine fault control degree evaluation system is characterized by comprising a feature extraction module, a data set construction module, a model training module, a posterior probability quantification output module, a dynamic threshold determination module, a prediction module, a fault point reliability classification module and a fault risk grade rating module; the feature extraction module is used for extracting seismic attribute values and constructing a full-area feature set; the data set construction module is used for counting the exposure information of the drill holes and the roadways in the mining area to manufacture a tag set, fusing the tag set with the full-area feature set, and establishing a training set A and a set B to be predicted, wherein the number of fault sample points of the training set A is as follows The number of non-fault sample points is ; The model training module is used for constructing a support vector machine two-class model, dividing the training set into a first training set and a first testing set, training the support vector machine two-class model by using the first training set, and outputting the trained support vector machine two-class model by using the first testing set until the training precision reaches a preset threshold; The posterior probability quantization output module is used for inputting the first test set into a trained support vector machine two-class model, outputting by using the support vector machine posterior probability model, and recording posterior probability values P 1 of each sample point of the first test set output by the model; The dynamic threshold determining module is used for sorting the posterior probability values P 1 and recording the th Size of the individual value And (d) Size of +1 values ; The prediction module is used for predicting the set B to be predicted through a trained support vector machine two-class model, outputting the set B to be predicted through a support vector machine posterior probability model, and recording posterior probability values P 2 of each sample point of the set B to be predicted, which is output by the model; The fault point reliability classification module is used for evaluating the quality of the break points according to the probability value P 2 and dividing the fault point reliability; And the fault risk level rating module is used for evaluating fault risk levels in the mining area through breakpoint duty ratios of different qualities.

Description

Coal mine fault control degree evaluation method and system Technical Field The invention relates to the technical field of fault identification reliability evaluation, in particular to a coal mine fault control degree evaluation method and system. Background In the field of mineral exploitation, accurate detection of geological structures such as faults is a key for guaranteeing safety and optimizing design. At present, the mainstream fault identification and reliability evaluation method is mainly based on the following two types of technologies, namely a traditional seismic interpretation technology, wherein the method relies on an interpreter to synthesize a seismic profile and various attribute bodies for manual identification and tracking. For reliability evaluation, the traditional method is completed through comprehensive judgment of an interpreter, and the evaluation mode is highly dependent on personal experience and subjective judgment of the interpreter, is difficult to quantify and standardize, and causes poor reproducibility, difficult inheritance and low efficiency of an evaluation result. The second category is a conventional artificial intelligence fault identification technology, and in order to overcome the defect of manual interpretation, automatic fault identification is started by using a machine learning algorithm. Such methods achieve a leap in efficiency by learning patterns from seismic attributes. However, existing AI schemes generally go to another extreme, a "binarization" simplification, in order to pursue high recognition accuracy. They typically output only one "fault" or "non-fault" tag. The method has the following defects that uncertainty information is lost, reliability grading cannot be achieved, and management requirements of quantitative grading of risks in mine safety production cannot be met. Therefore, the current state of the art presents a prominent contradiction that the traditional method has qualitative evaluation but subjectively low efficiency, and the emerging AI method has high efficiency and objectivity but no quantitative evaluation. Thus, there is an urgent need for new methods for mining area safety production that maintain the objective and efficient AI, and that restore and even exceed the reliability quantitative evaluation capabilities of human experience, for those skilled in the art. Disclosure of Invention The invention aims to provide a coal mine fault control degree evaluation method and system, which can not only keep the objectivity and high efficiency of an AI algorithm, but also restore and even surpass the reliability quantitative evaluation capability of manual experience so as to solve the problems in the background technology. In order to achieve the purpose, the invention provides the scheme that on the one hand, the method for evaluating the fault control degree of the coal mine comprises the following specific steps: Extracting seismic attribute values and constructing a full-area feature set; Making a tag set by counting the exposure information of drilling holes and roadways in a mining area, fusing the tag set with the whole area feature set, and establishing a training set A and a set B to be predicted, wherein the number of fault sample points of the training set A is as follows The number of non-fault sample points is; Building a support vector machine two-class model, dividing the training set into a first training set and a first testing set, training the support vector machine two-class model by using the first training set, wherein the first testing set is used for testing the support vector machine two-class model until the training precision reaches a preset threshold value, and outputting the trained support vector machine two-class model; Inputting the first test set into a trained support vector machine two-class model, outputting by using a support vector machine posterior probability model, and recording posterior probability values P 1 of each sample point of the first test set output by the model; Ordering the posterior probability values P 1 and recording the th Size of the individual valueAnd (d)Size of +1 values; Predicting by using the set B to be predicted through a trained support vector machine two-class model, outputting by using a support vector machine posterior probability model, and recording posterior probability values P 2 of each sample point of the set to be predicted, which is output by the model; Evaluating the quality of the break points according to the probability value P 2, and dividing the reliability of the break points; And evaluating fault risk levels in the mining area according to the breakpoint duty ratios of different qualities. Preferably, the specific steps of extracting the seismic attribute value and constructing the full-area feature set are as follows: Acquiring an original seismic data volume of a target stratum of a mining area; And determining the position of a phase axis of t