Abstract：The paper presents a new method that based on the insensitivity of L1 norm to separate off-group points data in inverse theory to remove off-group points data when logging model is set up.The method first builds up the initial value for equation parameters by L1 norm,then removes off-group points data iteratively by using maximum distance method.The final log mode builds up a log interpretation model by using Uussian hypothesis(least square method)in data without off-group points data.The application results show that the method can finish automatic deletion of off-group points data,having objective deletingstandard.The application of the method to set up log appreciation model satisfies demand for precision and its operation is suitable to standardize.
