Analysis of Influencing Factors of Fracturing Effect Based on Artificial Neural Network

The GGY Oilfield in northern Shaanxi, a major oil and gas producing area in China, faces challenges in accurately predicting production due to complex interactions among various factors. Traditional mathematical methods often lead to significant deviations from actual outcomes and require substantial time and energy. Therefore, there is a pressing need for a new, efficient approach to analyze the factors influencing fracturing effects. To address this issue, a study was conducted on 300 oil wells in different blocks, including FFC, FZ, GGY, GT, LQ, MKY, QHF, and YD. The study focused on reservoir geological parameters and field construction parameters, using the average daily oil production of wells post-fracturing as a measure of fracturing effectiveness. Various correlation coefficient methods were employed to assess the relationship between influencing factors and daily yield, pinpointing the key control factors. Abnormal data was identified and removed using the box plot method. Subsequently, a BP artificial neural model was developed using Python, with the main control factors as inputs and average daily production as output. Through parameter optimization and extensive training, the model achieved a prediction accuracy of 90%, closely matching actual field data. 1 INTRODUCTION 1.1 Purpose And Significance Of The Paper Hydraulic fracturing technology is an important method for enhancing production and injection in oil and gas well production. This technology utilizes high-pressure equipment on the ground to inject a mixture of water, sand, and chemical reagents into the wellbore at a high flow rate, thereby creating pressure fractures in the oil reservoir. Subsequently, it continuously injects fracturing fluid into the wellbore to expand the gaps and then injects fracturing fluid with proppant again. Finally, inject replacement fluid and push the proppant into the gap to support the crack. The fracturing fluid injected will naturally degrade and flow out of the wellhead, leaving several fracture channels. These channels have a high flow capacity, which can reduce permeability resistance near the bottom of the well and increase oil and gas production. [1]. In recent years, with the advancement of hydraulic fracturing technology as a crucial method for oil and gas extraction, significant oil fields in China have implemented extensive fracturing production, resulting in the accumulation of a large amount of production data. According to research on predicting fracturing effects, only 6-7 out of 10 operations are often successful [2]. In our country, the above success rate is acceptable because the cost of treating the bottomhole area can always be compensated for by increasing the production of crude oil; However, the cost that can be compensated for by increasing oil production is only brought about by the wells with high production before processing the bottom zone of the oil layer. The increased oil production from low production oil wells that have been exploited for a long time is not enough to compensate for the cost of hydraulic fracturing construction in the oil layer. Therefore, it is urgent to update traditional methods for predicting fracturing effects and establish a scientific method for analyzing the impact of fracturing effects using existing production, geological, and construction data. However, the fracturing effect is influenced by various factors, and the relationship between these factors and the fracturing effect is very complex. Traditional mathematical methods are challenging to use for such problems and often have low accuracy. Based on this, an artificial intelligence system has emerged for evaluating and analyzing fracturing effects using artificial neural network methods. [5].