Search

CN-122020276-A - Multi-model fusion-based multi-step prediction method for drilling pressure in geological drilling process

CN122020276ACN 122020276 ACN122020276 ACN 122020276ACN-122020276-A

Abstract

The invention relates to the field of intelligent monitoring and control of geological drilling processes, and discloses a multi-step prediction method for the drilling pressure of a geological drilling process based on multi-model fusion, which comprises the steps of collecting data of the geological drilling process and preprocessing the data of the geological drilling process; A smooth regression model, a jump classification model and a weight on bit variation regression model are respectively built by using a plurality of lightweight gradient lifting tree models, a plurality of prediction models are fused by adopting a weighted least square method, and weight on bit prediction values of a plurality of depth points in the drilling process are generated through multi-step prediction based on the fused models. The method effectively improves the overall accuracy and robustness of the weight-on-bit prediction, can improve the exploration efficiency and reduce the operation cost, and has important engineering application value and economic significance.

Inventors

  • DU SHENG
  • HUANG CHENG

Assignees

  • 中国地质大学（武汉）

Dates

Publication Date
20260512
Application Date
20251219

Claims (8)

  1. 1. The multi-step prediction method for the drilling pressure in the geological drilling process based on multi-model fusion is characterized by comprising the following steps of: S1, collecting data of a geological drilling process and preprocessing the data of the geological drilling process; S2, respectively establishing a smooth regression model, a jump classification model and a weight on bit variation regression model by using a plurality of lightweight gradient lifting tree models; s3, fusing a plurality of prediction models by adopting a weighted least square method; And S4, generating predicted values of the bit pressure of a plurality of depth points in the drilling process through multi-step prediction based on the fused model.
  2. 2. The multi-model fusion-based geological drilling process weight-on-bit multi-step prediction method according to claim 1, wherein the step S1 specifically comprises: S11, collecting time sequence data in the sensor drilling process, including drilling depth, weight on bit, drilling speed, riser pressure, surface torque, rotating speed, mud pit flow and mud pit density parameters S12, performing preliminary cleaning on the acquired time sequence data, and eliminating sensor faults and abnormal data; S13, resampling the time sequence data according to the drilling depth, wherein each drilling depth of 1 meter is used as a data point, and each data point represents a specific depth and the corresponding time sequence data.
  3. 3. The multi-model fusion-based geological drilling process weight-on-bit multi-step prediction method according to claim 1, wherein the step S2 specifically comprises the following steps: S21, constructing a smooth regression model by using a lightweight gradient lifting tree, inputting historical geological drilling process data taking drilling depth as a main shaft, performing model training, and outputting a preliminary predicted value of the weight on bit, wherein the preliminary predicted value of the weight on bit is expressed at a specific depth point; S22, constructing a jump classification model by using a lightweight gradient lifting tree, wherein input data is historical geological drilling process data taking drilling depth as a main shaft, model training is carried out, and output of the model is a classification label for judging whether abnormal fluctuation or abrupt change exists in the drilling pressure; S23, constructing a weight-on-bit variation regression model by using the lightweight gradient lifting tree, wherein input data is historical geological drilling process data taking drilling depth as a main shaft, model training is carried out, and output of the model is variation of the weight-on-bit, namely the amplitude of the weight-on-bit variation.
  4. 4. The multi-model fusion-based geological drilling process drilling pressure multi-step prediction method according to claim 3, wherein in the steps S21-S23, for the corresponding model training, multi-step training comprising m steps is adopted, each step is independently trained to train 1 model, and the corresponding prediction conditions of m depth points are obtained.
  5. 5. The multi-model fusion-based geological drilling process weight-on-bit multi-step prediction method according to claim 4, wherein the step S3 is specifically as follows: And S31, carrying out weighted fusion on the corresponding models to obtain a final predicted value d j of each depth point, wherein j=1, 2, 3, & m is as follows: Wherein the method comprises the steps of As a preliminary weight-on-bit predictor for the smooth regression model, Predicting probability values for weight-on-bit transitions of the transition classification model, The weight change prediction value is a weight change prediction value of the weight change regression model, and w 1 ,w 2 ,w 3 is the weights of the sliding regression model, the jump classification model and the weight change regression model respectively; S32, determining the weight w 1 ,w 2 ,w 3 by optimizing the prediction error of each model according to the least square method, wherein the specific calculation is as follows: Where q i is the mean square error of the ith model over the training set, i=1, 2, 3.
  6. 6. The multi-model fusion-based geological drilling process weight-on-bit multi-step prediction method as claimed in claim 5, wherein the step S4 is specifically as follows: And (3) inputting n-step historical drilling data of different drilling wells to the fused model in the step (S3), and carrying out corresponding m-step drilling pressure prediction on depth points of different drilling stages according to the input historical data of the well by the model.
  7. 7. A computer device, comprising at least one or more processors and a memory storing one or more computer programs, wherein the processors invoke the computer programs to implement the steps of the multi-step prediction method for weight on bit for a geological drilling process based on multi-model fusion as set forth in any one of claims 1-6.
  8. 8. A computer storage device storing a computer program which is invoked by a processor to perform the steps of the multi-step prediction method of geological drilling process based on multi-model fusion as set forth in any one of claims 1-6.

Description

Multi-model fusion-based multi-step prediction method for drilling pressure in geological drilling process Technical Field The invention belongs to the field of intelligent monitoring and control of geological drilling processes, and particularly relates to a multi-step prediction method for drilling pressure in a geological drilling process based on multi-model fusion. Background The geological drilling process is a key link in mineral resource exploration, the weight on bit is used as an important index for measuring the drilling state and stability, and accurate prediction of the weight on bit has important significance for improving exploration efficiency, shortening exploration period and reducing cost. In order to accurately predict the change of the weight-on-bit in the drilling process, a multi-step prediction method is used for continuously predicting the weight-on-bit at a plurality of depth points by combining drilling depth and related parameters. The method can predict the weight-on-bit change of a plurality of depth points in the future from the current depth point, help operators know the change trend of the weight-on-bit in real time, and provide guidance for subsequent drilling operation. Through multi-step prediction, operators can not only master the current drilling state, but also pre-judge the change trend of the drilling pressure in advance, so that the drilling operation plan is optimized, and the operation efficiency and the safety are improved. Most current weight-on-bit prediction methods rely primarily on time-series data processing techniques to predict weight-on-bit for each time step. Although these methods can reflect the real-time state of the drilling process to some extent, they often have an overfitting to the local data, which makes it difficult to capture complex dynamic changes in the drilling process comprehensively and accurately. Therefore, the geological drilling process weight-on-bit prediction method combined with drilling depth data is developed, noise and interference in time sequence data can be reduced, spatial features are better extracted, and the geological drilling process weight-on-bit prediction method has important practical significance. Currently, single-step prediction methods are the most common means in weight-on-bit prediction, and generally rely on local predictions of current drilling depth or time points. However, this approach is often limited to analysis of the current drilling conditions, and it is difficult to adequately reflect long-term trends and dynamic changes in depth during drilling. Therefore, the multi-step prediction method is developed, the drilling pressure can be predicted in multiple steps at multiple depth points, operators can be helped to recognize risks or changes possibly occurring in the future drilling process in advance, and the operation plan is optimized. Disclosure of Invention The invention aims to solve the technical problem that the drilling operation efficiency is low because the trend and the depth of the drilling process cannot be fully reflected by the existing single-step weight on bit prediction. In order to solve the technical problems, the invention provides a multi-step prediction method for the drilling pressure in the geological drilling process based on multi-model fusion. The invention provides a multi-step prediction method for drilling pressure in a geological drilling process based on multi-model fusion, which specifically comprises the following steps: S1, collecting data of a geological drilling process and preprocessing the data of the geological drilling process; S2, respectively establishing a smooth regression model, a jump classification model and a weight on bit variation regression model by using a plurality of lightweight gradient lifting tree models; s3, fusing a plurality of prediction models by adopting a weighted least square method; And S4, generating predicted values of the bit pressure of a plurality of depth points in the drilling process through multi-step prediction based on the fused model. A computer device at least comprises one or more processors and a memory storing one or more computer programs, wherein the processors call the computer programs to realize the multi-step prediction method of the geological drilling process drilling pressure based on multi-model fusion. A computer storage device stores a computer program that is invoked by a processor to implement the steps of the multi-step prediction method for weight-on-bit for a multi-model fusion-based geological drilling process. The technical scheme provided by the invention has the following beneficial effects: 1. The prediction precision is remarkably improved, namely three lightweight gradient lifting tree models are integrated by smooth regression, jump classification and weight on bit change regression, the capturing capability of different models on different characteristics in the drilling process is comprehe