Prediction method of rock burst hazard level based on local weighted c4.5 algorithm

Abstract

The invention provides a method for predicting the hazard level of rock burst based on a local weighted C4.5 algorithm, and relates to the technical field of rock burst prediction. This method first uses the MDLP method to discretize the continuous attribute data in the sample data, then uses the local weighting method to select the training set and calculate the sample weight, uses the sample weight to calculate the information gain rate of each attribute, and selects the sample attribute according to the information gain rate as C4.5 The root node of the decision tree and the split attributes of other branch nodes. Finally, the decision tree is pessimistically pruned by using the sample weight instead of the number of samples to realize the prediction of the rock burst hazard level in the predicted area. The method for predicting the hazard level of rock burst based on the local weighted C4.5 algorithm provided by the present invention overcomes the disadvantage of using information gain to select node split attributes in the ID3 algorithm, and avoids the problem of over-fitting. The prediction accuracy of the model is high.

Description

technical field [0001] The invention relates to the technical field of rock burst prediction, in particular to a method for predicting a rock burst hazard level based on a local weighted C4.5 algorithm. Background technique [0002] Rock burst is a dynamic phenomenon characterized by sudden, sharp and violent damage caused by the release of deformation energy of the coal and rock mass around mine shafts and stopes. It is one of the major disasters that affect the safety of coal mine production. All countries are threatened by rock burst to varying degrees. In recent years, developed countries have shut down rock burst mines due to energy structure adjustment and safety considerations. my country has become the main victim of rock burst and the main country for rock burst prevention nation. [0003] Prediction and evaluation of rock burst is a key step in the prevention and control of rock burst based on the study of the mechanism of rock burst, but because the mechanism of ro...

AI Technical Summary

Problems solved by technology

[0003] Prediction and evaluation of rock burst is a key step in the prevention and control of rock burst based on the study of the mechanism of rock burst, but because the mechanism of rock burst is not fully understood, especially for deep rock burst The research on the mechanism of rock burst is still in its infancy, which increases the difficulty of rock burst prediction

At present, the methods for predicting rock burst mainly include rock mechanics methods and geophysical methods, among which rock mechanics methods include drilling cuttings method, mining stress detection method, etc., and geophysical methods include geoacoustic monitoring, microseismic monitoring, electromagnetic radiation monitoring, etc. method; in addition, with the development of artificial intelligence, there have been some methods for predicting rock burst using intelligent algorithms, such as: neural network method, Bayes discriminant analysis method, support vector machine, etc. A large number of research results have been obtained in this method, but there are still some problems, such as the neural network generally requires more samples, but the sample size used for rock burst prediction is less, the Bayes method requires high independence between data, and In reality, it is difficult for the sampling data of rock burst to meet the independence requirements, and the above method does not consider the overfitting problem of the model, etc.

Images