煤层注水渗透特征的参数反演

为克服实验室测定方法的条件限制和方法本身存在“以点代面”的不足，提出用可变容差最优化方法根据注水试验观测资料反演煤体导水参数和贮水参数的方案，建立了煤体湿润的观测值(水分增值)与计算值的相对偏差最小为搜索目标函数。提出利用对己知参数的正向结果的反演计算来验证反演算法的精度。通过用不同初值对自身的反演，结果表明，反演收敛快，精度高，稳定性好，对初值的选择要求不高。在进行抗噪性实例检验计算中，迭代21次，调用有限元数值模型42次后参数既趋向稳定；湿润分布(水分增值)的最大绝对误差不超过0．03％，计算精度满足工程要求。这项研究，能够实现利用现场注水的结果不断修正模型参数，使模型不断适应煤层条件的变化。

Parameter reverse-solving of infiltrating character during water injection into coal seams

To overcome the condition restricts of measuring methods in laboratories and the shortcoming of using a point to represent a surface that exists in the method itself; the author puts forward an optimal search method in which the tolerance is variable. According to the observed materials in water injection experiment, the schemes of coal water flowing and water storage parameters can be reverse-solved, and the objective function which optimum searching is that the deviation between the observed values of wetness in coal body (the water content increment) and the computing values is minimum is established, at the same time the iterative convergence criterion is also determined. Simulations of the uniqueness and convergence of reverse-solving at different initial values are done, the computing result show that reverse-solving is a quick and stable method, and the demand on initial value is not very strict. In the calculation of the examination on antinoise capability example, after 21 times iterations and calculation 42 times by using the numerical model of finite element, the parameters turn to be stable. The maximum absolute error of humid distribution (the water content increment) is less than 0.03%, which can meet the demand of engineering calculating precision. The study can correct model parameters gradually according to the result of local water injection, and then the model can adapt to the variations of coal seams continuously. The study has provided an efficient means that can instruct water injection into coal seams.