In porous media, part of the overburden pressure is borne by the fluid filling the voids and in proportion to that, the fluid enters the wall of the voids, which is called the pore pressure. The pore pressure is one of the important geomechanical parameters. Therefore, full knowledge of this parameter in the reservoir can significantly help in geomechanical studies of the field, well wall stability, design of slit layer operations, etc. Machine learning methods based on support vectors have good capabilities in estimating the spatial model Unknown parameters in oil fields have been shown. However, selecting the appropriate kernel function is a challenging issue in the implementation of support vector machines that significantly affect the performance of these algorithms. In order to select the optimal value for these two parameters in the present problem, the gray wolf optimization algorithm is used. Finally, by optimizing the value of the fine parameters and the kernel parameter for the proposed problem, the value of 13.9 for the penalty parameter (C) and the value of 2.63 as the best The value of the kernel parameter selected g Using these two values, the backup vector machine regression algorithm was trained again with these values using training data, and finally the trained model was performed on the test data and the amount of pore pressure was predicted.