<p>Tight oil and gas resources occupy a considerable proportion in China’s petroleum resource reserves, and the dense lithology and poor physical properties of tight reservoir reservoir reservoirs determine that the exploitation of tight reservoirs becomes very difficult. Large volume fracturing technology is the core technology for the successful development of tight reservoirs, which aims to improve the inflow capacity of the reservoir and expand the drainage area, and has been widely used in the field. There are many factors affecting the production capacity of wells after fracturing, which makes it difficult to predict the production capacity of volumetric fracturing wells. Therefore In this paper, based on the horizontal well capacity prediction model of machine learning, BP neural network and support vector machine are used to predict the capacity of three decreasing stages, and by comparing with the data generated by the seepage mathematical model of volumetrically fractured horizontal wells, the root-mean-square error (RMSE) and the coefficient of determination (R2) are computed, and the improved method combining the two algorithmic models, i.e., the combined prediction model, is proposed. The results show that it is better than the single prediction model of BP neural network and support vector machine in terms of prediction accuracy and stability. Using the established combined prediction model method to give two examples of the application of volume fracturing horizontal well production prediction in tight oil reservoirs. The method of optimal design of fracturing parameters for volumetric fracturing horizontal wells is established, and combined with the actual data of a horizontal well in a tight oil reservoir, the optimal design of fracturing parameters is carried out for the completed well to be fractured M well.</p>