Intelligent optimized earthquake multi-attribute fusion method based on crack model

The invention relates to an intelligent optimized multi-attribute fusion method based on a crack model. The method comprises the following steps that S1) earthquake and logging data is loaded and calibrated; S2) a logging curve is corrected manually to ensure that the logging data matches the earthquake data; S3) a crack density curve is calculated, and different logging response and different earth physical response caused by different sizes of cracks are analyzed; S4) according to the earth physical response of the cracks of different sizes, logging is combined with earthquake to establish a crack network forward model, and optimized crack sensitive attributes are analyzed; S5) a crack sensitive parameter is introduced, and earthquake data is trained near a well point; and S6) the trained nonlinear relation is applied to the whole earthquake data body, and the aim of predicting the crack variable parameter is achieved. The fusion method has the advantages that the multiple crack attributes are combined and optimized via nonlinear expansion of the algorithm, and the multi-solution performance of prediction of cracks of different sizes is effectively reduced via multiple times of training learning and probability estimation.