Research On Drilling Parameter Optimization Method Based On FSVR-FWA

Oil and natural gas are important strategic resources to maintain the development of human society and play an important role in national energy security.Drilling is an important step to exploit oil and natural gas.With the development of oil resources,the drilling depth is deepening,the drilling difficulty is increasing,and the drilling cost is increasing day by day.Relevant experts and scholars are working on how to improve the rate of drilling and mining efficiency.With the development of artificial intelligence technology,drilling technology is gradually integrated with new technology,so that drilling engineering develops towards the direction of intelligence.Based on this development trend,this paper takes mechanical drilling rate and drilling parameters as the research objective,and uses machine learning related intelligent algorithms to predict and optimize drilling parameters affecting mechanical drilling rate,so as to obtain the optimal value of drilling parameters and mechanical drilling rate,so as to improve the drilling rate.Specific research contents are as follows:Firstly,the random forest algorithm is used to rank the importance of drilling parameters related to ROP,and the features with high importance were selected for feature extraction.Then,aiming at the problem that the isolated data in drilling data interferes with the drilling rate prediction model,the isolation degree of all sample data points is detected through the local anomaly factor isolation degree detection method,and the local anomaly factor value of sample data is obtained to define the isolation degree of each group of data.Secondly,fuzzy factor values are assigned to each group of sample data according to local anomaly factors,and a new data set with fuzzy factor is constructed.A small fuzzy factor value is attached to the isolated data points to reduce the interference to the drilling rate prediction model.The normal data points are assigned a larger fuzzy factor value to make them play a larger role in training the model.A fuzzy support vector regression prediction model for mechanical drilling rate was built.The degree of sample deviation from regression interval was described by fuzzy factor,and the new data set with fuzzy factor was used to train the fuzzy support vector regression model.Finally,based on the prediction of mechanical drilling rate by fuzzy support vector regression,the related controllable drilling parameters are optimized.Fireworks algorithm was selected as the optimization model to increase the population diversity and local coverage ability in the optimization process.In addition,the optimization ability of fireworks algorithm is improved to improve the adaptive ability of the algorithm.The controllable parameters of drilling were optimized by the improved fireworks algorithm,so as to improve the drilling rate and drilling efficiency.In this paper,Python language is used for programming,and the model is analyzed and verified.The final experimental results show that compared with other models such as traditional support vector regression,the fuzzy support vector regression prediction model in this paper has stronger anti-jamming ability,better fitting accuracy and generalization ability.Compared with standard fireworks algorithm and particle swarm optimization algorithm,the improved fireworks algorithm has better searching ability and optimization results for drilling parameters optimization.This study has a certain significance for promoting the development of drilling technology and building intelligent oil field.