Research On Intelligent Identification Method And Disposal Process Optimization Of Sticking Accidents In Drilling Process

During the drilling process,the effective and intelligent identification of sticking accidents based on relevant drilling engineering parameters and the use of workflow modeling to optimize the sticking accidents handling process represent the forefront of drilling safety technology.It has important scientific value and significance to carry out research in this area.The following research is conducted on the stuck drilling accidents during the drilling process.(1)The causes,preventive measures and control measures of each sticking accident in drilling process are analyzed in detail.FTA and ETA of drilling pipe sticking accident,mechanical sticking accident and loop sticking accident are established.Based on the analysis of five safety factors of preventive measures,including human,machine,material,rules and environment,the Bow-tie model of sticking accidents is established.These all provide guidance for the effective identification and disposal process of sticking accidents.(2)A classification model of sticking accidents based on generalized fuzzy neural network is established.This method combines FCM clustering sample optimization and GRNN data prediction and it can intelligently identify sticking accidents in the drilling process.It can also avoid the subjectivity of training sample selection when using neural networks for sticking accidents classification,and the impact of random clustering center on the stability and accuracy of FCM clustering results.(3)Based on the foregoing,through in-depth analysis of the sticking accidents symptoms and the logging data changes,the logging data of three major types of sticking accidents in Ya Cheng area were extracted and pre-processed.Compared to using FCM only,the application results of generalized fuzzy neural network of sticking accidents classification model shows that it can improve the classification accuracy rate of the pipe sticking accident from 86.7% to 100%,and improve the classification accuracy rate of mechanical sticking accident from 83.3% to 93%.(4)Considering the overall disposal process of sticking accidents and the theoretical application of stochastic Petri net workflow modeling,a SPN-based sticking accidents handling process model was established.Analysis of validity,safety,and activity of the model were analyzed by T-invariant and isomorphic Markov chain simultaneously.(5)According to the complexity of the treatment of sticking accidents,the fuzzy state theory and Markov fast transfer matrix are used to solve the steady state probability of the system.Quantitative analysis,including the busy and idle probability of the library,the utilization rate in the transition,and the average system delay are studied.Through calculation results,giving the suggestions for the parts of the disposal process that need to be optimized and improved to raise the handling efficiency of sticking accidents and reduce the loss of sticking accidents.