Prediction The Cementing Quality Based On ANN And GT And A Study On Construction Parameter Optimization Design

In the petroleum well drilling, the cementing operation is critical to the entire drilling whether success or not .Many factors affect cementing perfection rate. At present, the cementing construction plan is mainly designed by cementing experts according to their many years'accumulation experience and massive computations, and the main influence factors that affect the cementing quality are also designed by experts'experience. So it does not provide a theoretical support for cementing operation. And the designed cementing construction plan cannot obtain the design result beforehand, and the questions in the plan are also unable to be corrected timely. Therefore, it is necessary to carry out the multi- factor cementing quality research and inquire about the drilling cementing process as well as the geological condition of the block and such factors that affect the cementing quality.This article handles the historical data of cementing before analysis to filter the abnormal value. The key factors are found according to the experts'experiences that affect the cementing quality. Using fuzzy comprehensive evaluation and gray relational analysis to carry out an analysis of the 9 key factors, these factors are analyzed separately to decide the influence of the cementing quality.Before establishing the cementing quality prediction model, the principal component analysis method is carried on to analyze the 9 key factors in order to reduce the input dimension of the model. This article has established four cementing quality prediction models:1. LMBP artificial neutral network cementing quality prediction model;2. Gray System GM(1,N) cementing quality prediction model;3. Gray neural network prediction model.Based on the cementing quality prediction models, this article has also conducted the research on cementing construction parameter optimum design, before cementing construction, the prediction value of cementing quality can be obtained by the cementing quality prediction model, and the controllable parameter can be adjusted according to the prediction value, making the cementing quality achieve superiorly.