Study On Gas Emission Prediction Method Based On Improved Gravitational Search Algorithm-KELM

China's coal production and consumption are among the highest in the world.With the development of science and technology,the demand for coal is also high,and the production of coal is indeed at the same risk.In the process of coal mining,gas es caping will occur,causing accidents such as suffocation and gas explosion in underground people.At the same time,it will cause a series of secondary accidents such as roof accidents,water permeable accidents and fire accidents,causing heavy casualties and property losses.Therefore,accurate prediction of the amount of gas emission has important practical and social significance.The number of influencing factors of gas emission is large and the relationship is complex.There is information overlap between various influencing factors,which has nonlinearity and correlation.Therefore,the established model should take this into account to ensure the accuracy of prediction.This paper proposes a method for predicting gas emission in combination with kerne l extreme learning machine network and principal component analysis.The improved universal gravitation algorithm is used to optimize the kernel parameters and output weights of kernel extreme learning machines,and the kernel extreme based on improved uni versal gravitation algorithm is established.A prediction model of gas emission from learning machine and principal component analysis.Firstly,the theoretical basis and limitations of the kernel extreme learning machine are analyzed,and the universal gr avitational search algorithm is proposed to optimize its parameters.Then the basic theory of the universal gravitational search algorithm is analyzed,and its advantages and disadvantages are analyzed.Then the improved method of the universal gravitational search algorithm is proposed,that is,the reverse learning mechanism is used to initialize the population,and the uniform distribution of the initial population is expanded.The Tent chaotic map is used to increase the diversity of the population.And introduce the memory and social exchange ideas of the particle swarm algorithm,improve the position and speed update of the gravitational search algorithm,enhance the traversal search ability of the gravitational search algorithm,and avoid falling into local extremum.The simulation is verified with the actual case,and the comparison experiment is carried out with the universal gravitational kernel extreme learning machine and the particle swarm kernel extreme learning machine network model.The results show that the gas emission of the kernel extreme learning machine and principal component analysis based on the improved universal gravitational search algorithm is presented.The prediction model prediction error is within 0.10m~3/min.Compared with the other two models,the prediction accuracy is improved by nearly 7.5 times and 15.4 times respectively,and the convergence is earlier than the other two models.It shows that the method has better prediction accuracy and convergence speed for coal mine gas emission.The paper has 14 pictures,11 tables and 56 references.