A Method for Predicting Oil Well Fluid Level Based on Dynamic Integrated Modeling

Abstract

A dynamic integrated modeling based prediction method for fluid level in oil wells, belonging to the field of information technology; including: obtaining samples and sorting them by weight; performing model training on the samples, retraining sub-models whose errors exceed a threshold, and calculating sub-models whose errors do not exceed the threshold Sub-model weight; sub-model weighted output integrated model; use survey data to calculate whether the error of the integrated model exceeds the threshold, if yes, form a new training set, and retrain the original sub-model with the error exceeding the threshold as a new sub-model, otherwise , keep the atomic model as a new sub-model, and weight the new sub-model as a new integrated model; collect auxiliary variables in real time, input the integrated model to predict the integrated oil well fluid level; the present invention adopts iterative integrated modeling, which is more effective than a single model Higher prediction accuracy, the weighted output of the sub-model reduces the influence of individual errors, and the generalization of the model is strong; the change of the sample weight in the sub-model of the weak learning machine can improve the adaptability of the new model to the sample.

Description

technical field [0001] The invention belongs to the field of information technology, and in particular relates to a method for predicting the fluid level of oil wells based on dynamic integrated modeling. Background technique [0002] In the actual production process of the oil field, in order to achieve the maximum liquid production, the pumping unit needs to adjust its pumping frequency according to the changing parameters of the oil well to make it reach a reasonable working state. The dynamic liquid level of an oil well is the liquid level depth of the annular space of the oil well casing during the production process. It is one of the important production guidance data of the oil field and an important reference index reflecting the working state of the pumping unit. Only by grasping the data information of the dynamic fluid level in time can the pumping unit operate in a reasonable operation mode. At present, the measurement of dynamic liquid level in most oil wells s...

AI Technical Summary

Problems solved by technology

At present, the measurement of dynamic liquid level in most oil wells still adopts traditional manual measurement methods, such as echo measurement, pressure measurement method and buoy method, etc. Traditional manual measurement has the disadvantages of large errors, low efficiency, and poor real-time performance. Meet current production requirements

In recent years, with the wide application of soft-sensing methods in various industries, people have proposed a single-model intelligent algorithm soft-sensing method. However, this method has poor generalization ability, prone to over-fitting and precision in practical applications. Not high enough, and in the actual production process, the prediction accuracy of the model will decrease with the dynamic operation of production, so the model needs to be updated to improve the dynamic performance of the model

The traditional update method is to update the model by adding training sample data, which will make the performance of the model gradually deteriorate as the training progresses and cannot meet the production needs

Images