Neural Network Parameter Optimization

The optimization of drilling parameters is a key technology in drilling engineering.Due to the uncertainties in the drilling process,the precise model in the drilling engineering is difficult to establish,resulting in a large difference between the simulation results and the actual situation;secondly,the control system is becoming more and more complicated,and the drilling parameters optimization is studied with the pure mathematical model,and the various links are coupled.Said that it has been very complicated,to optimize the control of more and more sophisticated equipment will inevitably limit the application of the model.The development of new drilling technologies,as well as the continuous improvement of the automation level and the performance of monitoring instruments,have also brought new issues to the optimization of drilling parameters.Drilling parameters optimization must not only consider the drilling speed and drilling cost,but also dynamically optimize and comprehensively analyze the drilling parameters in the drilling process.Through the research on drill parameter optimization theory and on-site related research,it is found that the feedback on the drilling site is composed of a large amount of data.Therefore,this paper attempts to transfer the control algorithm to the learning algorithm and proposes the sensing layer,transmission layer,application layer,and control.The neural network-based drilling parameter optimization architecture composed of layers.The Cerebellar Model Articulation controller(CMAC)model was used for drilling parameter optimization through comparative analysis.A neural network-based drilling parameter optimization algorithm was proposed.The language of the algorithm uses Python,and Anaconda is used as an integrated development environment(IDE)platform for debugging and compilation.Use Pandas to implement data transfer,querying,updating,and managing related databases.Sk-Learn completes and normalizes drilling parameters.The neural network model is designed and trained to improve the prediction accuracy and shorten the training time by improving the network structure and optimizing the activation function for the situation where the accuracy of the test set is insufficient.Based on the actual drilling data collected in the field investigation,this paper optimized the drilling parameters and verified the effectiveness of the optimization method through simulation.In addition,the optimization experiment was conducted in the well area under construction,which resulted in a slight increase in ROP and verified the feasibility of the optimization method.