A sliding orientation intelligent control method based on gru

Abstract

The invention discloses a GRU-based sliding directional intelligent control method, comprising: 1: acquiring initial sample data on site in sliding drilling, classifying, integrating and storing in a database; 2: performing data cleaning on the initial sample data in the database , after cleaning, standardize the initial sample data, and obtain standardized sample data after processing; 3: Build a GRU network model; Step 4: Collect real-time data of sliding drilling and input it into the GRU network model, generate automatic control parameters, and use automatic control parameters The input torsion system realizes intelligent control. The invention acquires, fuses and stores field data, cleans and standardizes the data, obtains effective and high-quality data, designs a data generator, uses field data for GRU model training, forms a GRU network model, and is based on the GRU network model Generate automatic control parameters for sliding orientation operation to realize intelligent decision-making and control of torsion system.

Description

technical field [0001] The invention belongs to the technical field of oil and gas drilling (drilling) engineering, and in particular relates to a GRU-based sliding directional intelligent control method, which is used for sliding directional intelligent control, further improves the intelligent level of sliding drilling directional operations, and realizes intelligent decision-making for sliding directional operations . Background technique [0002] With the rise of technologies such as big data and artificial intelligence, the concept of artificial intelligence in directional well drilling has been gradually proposed in the past two years. In order to improve single well production, oil and gas recovery and development efficiency, the scale of directional drilling operations such as horizontal wells, multilateral wells and extended-reach wells is increasing. And model optimization, simulating the decision-making behavior of directional experts, making directional well dri...

AI Technical Summary

Problems solved by technology

Although this method realizes the function of determining the control parameters of the torsional drill string through historical operation data, in the actual application process, this technology still has the following defects: the amount of sliding drilling data is huge, and the model training process is prone to gradient explosion and gradient disappearance. problems, which did not propose targeted solutions

The model learning data features is slow and inefficient, and may even fail to learn any data features. It is difficult to quickly and efficiently obtain a neural network model that can be used to guide drilling operations

Moreover, the accuracy and performance of the model are poor, and there is a safety risk in direct application to control the torsion system

Images