Prediction Of Water Inrush From Coal Mine Floor Based On LLE-FOA-SVC

The demand for coal and mining output have been rising daily in recent years as a result of society’s quickening development,which puts people’s lives and the security of their property at significant risk from deepening and intensifying coal mining operations as well as frequent water inrush issues.Mine water hazard prediction is still a significant issue that needs to be addressed in the process of coal mine production and construction in China,despite the fact that the advancement of modern science and technology has led to improvements in coal mine production and construction.This is because coal mines have harsh hydrogeological conditions and complex mining conditions.Small sample sizes,nonlinearity,and high dimensionality of the coal mine water inrush data cause problems with incomplete feature extraction and difficulty choosing the parameters of the prediction algorithm,which leads to insufficient prediction accuracy.Therefore,this thesis builds a novel water inrush prediction model by introducing a machine learning algorithm based on the sample characteristics of coal mine floor water inrush prediction.The following are the primary study findings:(1)An improved approach involves using the Locally Linear Embedding algorithm to reduce the dimensionality of the dataThis research examines the most popular dimensionality reduction techniques and performs simulation experiments in light of the drawbacks of high dimensionality and numerous noise components in coal mine water inrush data.The findings demonstrate that the LLE approach is suited for handling nonlinear,high-dimensional water inrush data and overcomes the drawbacks of the principal component analysis method in nonlinear data.It also better maintains the local linearity of the data.As a result,the LLE technique is used in this thesis to preprocess the data,and a prediction model based on LLE-SVC in conjunction with a support vector classification machine is constructed.The data’s dimension and noise characteristics are effectively reduced by the model,and the model’s operational efficiency is increased,but the prediction accuracy still has to be increased.(2)An improved approach involves utilizing the fruit fly optimization algorithm to optimize the parameters of SVCThe complex parameter selection for the SVC method is the subject of this thesis’ s analysis of frequently employed optimization algorithms and simulation tests evaluating their performance.The outcomes demonstrate that the FOA method effectively addresses the blindness and unpredictability of artificially selected parameters in the past and has the advantages of a simple structure,powerful global optimization capability,and a difficult time falling into local minima,among others.Thus,a water inrush prediction model based on FOA-SVC is created in this article by optimizing the key SVC parameters using the FOA method.Although the model successfully increases forecast accuracy,its operational efficiency is low.(3)A prediction model for water inrush from coal mine floor based on LLE-FOASVC is builtIn order to exploit the complementing benefits of the three techniques,dimensionality reduction,optimization,and classification are integrated in this study to create a novel combination model.The prediction model developed in this thesis is contrasted with the conventional method and the modern advanced method through a simulation experiment.The advancedness and applicability of the model are assessed by contrasting and analyzing the two indicators of running time and prediction accuracy.The findings demonstrate that the LLE-FOA-SVC model developed in this study,which combines the benefits of the first two models,has the best prediction effect and is more appropriate for the field of coal mine floor water inrush prediction due to its high prediction accuracy and quick operation speed.This article introduces machine learning algorithms to construct a prediction model for coal mine floor water inrush based on LLE-FOA-SVC,in order to solve the problem of feature extraction and parameter selection in water inrush prediction.Through experimental comparison,the model achieves a good balance between prediction accuracy and operating speed,which is of great significance for solving the problem of water inrush in coal mine production.