Research On Inversion Methods Of Parameters In Probability Integral Model

As one of the main energy sources for human life and economic development,coal plays a very important role in the world energy consumption structure.Although coal mining can obtain effective resources,it is also accompanied by disasters such as surface subsidence,road and railway deformation,building damage,and so on,which are threatening human's life and property.Studying the law and characteristics of surface deformation caused by coal mining has important guiding significance and practical value for guiding disaster control and prevention,early planning and disaster warning.The probability integral model is an important mathematical model to describe the surface deformation law caused by coal mining.There are a certain number of parameters to be sought in the model.The accuracy of the parameter inversion directly affects the fitting effect of the probability integral model,which in turn affects the analysis of surface deformation characteristics.The corresponding mathematical methods are usually used for parameter estimation.Common methods include optimization algorithms and intelligent optimization algorithms.In this paper,the inversion method of parameters in the probability integral model is taken as the research object,and the related research is carried out in combination with the development trend of the nonlinear function parameter estimation theory.The main work is summarized as follows:(1)A new intelligent optimization algorithm,the intrusive weed optimization algorithm,is introduced,and the inversion method of parameters in the probability integral model based on the intrusive weed optimization algorithm is established.This method simulates the process of weed seeds from germination to competitive extinction.The process of continuous optimization and competitive exclusion ensures that the parameters obtained by the inversion have strong robustness,reliability and adaptability.By simulating the performance of the simulation data in different medium errors,different gross errors and different proportions of observation points,the results show that the invasive weed optimization algorithm has certain ability to resist gross error and anti-observation value,and the new method The absolute error,relative error and error in unit weight of inversion parameters are better than genetic algorithm and particle swarm optimization algorithm,and its multiple runs have strong stability.(2)Aiming at the problem that the current optimization algorithm and intelligent optimization algorithm have higher initial value requirements or non-unique parameters in the parameter inversion of probability integral model,a parameter inversion algorithm based on BFGS algorithm is proposed.The iterative process of the algorithm avoids the Hesse matrix and its inverse matrix of the objective function by establishing the approximate matrix of the Hesse matrix.The theoretical solution parameters have strong stability and reduce the computational complexity and operational efficiency.At the same time,for the parameters to be sought,the partial derivatives of the eight parameters to be sought with respect to the objective function are derived.The simulation results show that the BFGS algorithm solves the defect that the initial value of the parameter is too high and the result of the genetic algorithm is unstable,and it has certain resistance to the gross error and the missing observation data of the observatory.At the same time,compared with the model vector method and the genetic algorithm,the accuracy of the parameters calculated by the BFGS algorithm is relatively high,and the stability and reliability of the parameters are significantly improved.(3)The experimental data of a mining area in Inner Mongolia was used to perform parameter inversion based on the template vector method,different intelligent optimization algorithms and optimization algorithm(BFGS algorithm).The experimental results show that the error of the unit weight of the parameter inversion fitting curve of BFGS algorithm is smaller than other algorithms,showing good statistical characteristics.In addition,the error in the unit weight of the inversion parameters of different methods in the engineering case is obviously larger than that of the simulation experiment.The main reason for the geological mining conditions in the mining area is that the surface of the mining area is undulating and is affected by the gully terrain of the mining area.Therefore,the probabilistic integral model is used for fitting,and there is systematic system function error,which makes the influence of different parameter inversion methods on the overall fitting accuracy weak.(4)According to the probability integral model function and the theoretical content of this paper,the mine subsidence analysis and parameter prediction software system is established.The software system integrates the method module of curve drawing and parameter inversion to provide corresponding products and results for mining subsidence analysis.Figures [24] Tables [14] References [79]...