An early intelligent diagnosis model for drilling overflow based on GA–BP algorithm

Abstract

Aiming at the problems existing in the current drilling overflow accident, this paper uses standpipe pressure (SPP) and casing pressure (CP) monitoring data based on the pressure wave transmission to eliminate the impact of pressure wave jitter on the overall trend of SPP and CP monitoring data by eigenvalue extraction of the monitoring data of SPP and CP. The GA–BP (Genetic algorithm and BP neural network) is used to establish the functional relationship between the expected outputs of well overflow diagnostic model and several characteristic parameters. The model reduces the error caused by the selection of single feature parameters, and also the convergence speed of BP neural network is accelerated by genetic algorithm, and avoids falling into local extreme value. The early diagnosis of drilling overflow is achieved, and probability of wrong judgment and miss judgment of drilling kick is reduced. Finally, the drilling data of YY well in a certain oilfield is taken as the research object. The experimental results show that the error of the GA–BP model is 73.73% less and the accuracy is higher than the BP neural network model. The early-stage intelligent diagnosis model of drilling overflow based on GA–BP algorithm can timely and effectively diagnoses the accident of drilling overflow.