CN-121980730-A - Method and equipment for predicting residual life of oil and gas drilling equipment

CN121980730ACN 121980730 ACN121980730 ACN 121980730ACN-121980730-A

Abstract

The embodiment of the application provides a method and equipment for predicting the residual life of oil and gas drilling equipment. The method combines the machine vision technology, and extracts the damage area of the surface abrasion defect of the equipment part through an improved image segmentation model. Based on a defect evolution rule, a degradation characteristic is built by using a damage area, so that the built degradation model can practically reflect the health state of equipment parts. The optimal estimation of the degradation rate state estimation function is obtained by solving the linear degradation model based on the Kalman filter which is iteratively updated by the expected maximum algorithm, so that subjectivity of parameter setting is avoided, and the probability density function of the residual life of the part is obtained by simulating the wiener process and is used for predicting the residual life of the part. The application improves the accuracy of life prediction.

Inventors

ZHOU TAOTAO
WANG KEJIAN
MA JUNYUAN
ZHENG WENPEI
GAO FUMIN

Assignees

中国石油大学(北京)

Dates

Publication Date: 20260505
Application Date: 20251128

Claims (11)

1. A method of predicting the remaining life of an oil and gas drilling apparatus, comprising: Collecting image data in the historical time of the parts, and acquiring the abrasion area of the image data according to an improved image segmentation model; Acquiring degradation characteristics of the parts in historical time according to the abrasion area; Establishing a linear degradation model according to the degradation characteristics, wherein the linear degradation model comprises a system equation based on degradation rate and a state equation based on the degradation characteristics; solving the linear degradation model through a Kalman filter which is iteratively updated based on a desired maximum algorithm to obtain an optimal estimation of the degradation rate state estimation function; And obtaining a probability density function of the residual life of the part through a simulated wiener process according to the degradation characteristic and the optimal estimation of the degradation rate state estimation function, wherein the probability density function is used for predicting the residual life of the part.
2. The prediction method according to claim 1, wherein the obtaining the optimal estimate of the degradation rate state estimation function by solving the linear degradation model by a kalman filter iteratively updated based on a desired maximum algorithm comprises: Solving the linear degradation model through a Kalman filter to obtain an iterative solution of function parameters in the degradation rate state estimation function, wherein the function parameters comprise the expectation of posterior estimation, the variance of posterior estimation and the Kalman gain; updating initialization parameters of the Kalman filter through an expected maximum algorithm combined by an RTS smoothing algorithm according to the iterative solution of the function parameters; Returning to the step of solving the linear degradation model through the Kalman filter according to the updated initialization parameters until the expected maximum algorithm meets the convergence condition or reaches the iteration times, and taking the iteration solution of the function parameters as the optimal estimation of the function parameters; and obtaining the optimal estimation of the degradation rate state estimation function according to the optimal estimation of the function parameter.
3. The prediction method according to claim 2, wherein the initialization parameters of the kalman filter include an expectation of an initial degradation state of a state estimation function of the degradation rate, a variance of the initial degradation state, system process noise, and a diffusion coefficient; The initialization parameters of the Kalman filter are updated according to the expected maximum algorithm combined by the RTS smoothing algorithm by the iterative solution of the function parameters, and the method comprises the following steps: Obtaining a first expected value, a second expected value and a third expected value through an RTS smoothing algorithm according to an iterative solution of a state estimation function of the degradation rate, wherein the first expected value is an expected value of the degradation rate at the current moment in the ith iteration, the second expected value is an expected value of the square of the degradation rate at the current moment in the ith iteration, and the third expected value is an expected value of the product of the degradation rate at the current moment and the degradation rate at the last moment in the ith iteration; taking the first expected value, the second expected value and the third expected value as the expectation of E steps of the expected maximum algorithm, and acquiring the updated expectation of the initial degradation state, the variance of the initial degradation state, the system process noise and the diffusion coefficient through M steps of the expected maximum algorithm.
4. The prediction method according to claim 1, wherein the degradation rate-based system equation is: Wherein, the Is part at The degradation rate of time; is a drift coefficient; the state equation based on the degradation characteristic is as follows: wherein y k is the degradation characteristic at the kth time; is the diffusion coefficient; To conform to Is a normal distribution of (c).
5. The method of predicting as set forth in claim 1, wherein said obtaining degradation characteristics of the component over the historical time based on the wear area includes: acquiring the degradation characteristic according to the wear area and the historical time through a degradation characteristic formula, wherein the degradation characteristic formula is as follows: Wherein y k is the degradation characteristic at time t k , The worn area of the component at the moment in time, Is a constant value, and is used for the treatment of the skin, Is a characteristic coefficient.
6. The prediction method according to any one of claims 1 to 5, wherein the obtaining the probability density function of the remaining life of the component by the simulated wiener process based on the degradation characteristics and the optimal estimation of the degradation rate state estimation function includes: obtaining the probability density function of the remaining life according to the following formula: Wherein, the A probability density function of the remaining life of the component, Is damage area data of the parts, w is failure threshold value of the parts, In order for the diffusion coefficient to be the same, For an optimal estimation of the degradation rate state estimation function, For the length of time to time at time k.
7. The prediction method according to claim 1, wherein the acquiring the wear area of the image data according to the improved image segmentation model comprises: identifying an initial wear area of the image data based on the improved image segmentation model; Dividing the initial wear area into an initial inner wear area and an initial edge wear area; Processing the initial internal abrasion area according to a preset loss function to obtain an internal abrasion area; constructing an edge loss function according to a preset structure similarity function, and dividing the initial edge abrasion area into a plurality of edge abrasion image blocks; processing the plurality of edge abrasion image blocks according to the edge loss function to obtain an edge abrasion area; the inner wear area and the edge wear area are integrated as the wear area of the image data.
8. The prediction method according to claim 1, wherein the image segmentation model improvement process includes: acquiring an image sample data set of the part, and performing image enhancement processing on a plurality of image samples in the image sample data set to obtain a plurality of enhanced image samples; dividing the plurality of enhanced image samples to obtain a training set and a testing set; training a preset image segmentation model according to the training set to obtain an original image segmentation model; testing the original image segmentation model according to the test set to finish the test of the original image segmentation model; Extracting a backbone network and an original decoder of the original image segmentation model; extracting a feature extractor in the backbone network, and configuring the feature extractor to obtain a configured feature extractor; After each transposed convolutional layer in the original decoder, establishing a characteristic compensation path process according to a preset residual error connection mechanism; And improving the original image segmentation model according to the configured feature extractor and the feature compensation path process to obtain an improved image segmentation model.
9. The prediction method according to claim 7, wherein the processing the plurality of edge wear image blocks according to the edge loss function to obtain a calculation formula of an edge wear area includes: Wherein L is the loss value of the edge abrasion area, k is the number of image blocks; The value of (2) is 0, which indicates that the image does not contain the edge part of the part, and the calculation result of the image position does not account for loss and neutralization; And Respectively representing local average values of the ith image block in the prediction mask and the real mask; And Representing standard deviations of the ith image block in the prediction mask and the true mask, respectively; Representing the covariance of the ith image block between the prediction mask and the true mask; And Is a numerical stability constant and is used for preventing the situation that the denominator is zero.
10. A device for predicting the remaining life of an oil and gas drilling apparatus, comprising: the segmentation module is used for collecting image data in the historical time of the parts and acquiring the abrasion area of the image data according to the improved image segmentation model; the characteristic module is used for acquiring degradation characteristics of the parts in historical time according to the abrasion area; The model module is used for establishing a linear degradation model according to degradation characteristics, wherein the linear degradation model comprises a system equation based on degradation rate and a state equation based on the degradation characteristics; the iteration module is used for obtaining the optimal estimation of the optimal solution degradation rate state estimation function of the degradation rate state estimation function through solving the linear degradation model through a Kalman filter which is iteratively updated based on an expected maximum algorithm; The prediction module is used for obtaining a probability density function of the residual life of the part through a simulated wiener process according to the degradation characteristics and the optimal estimation of the degradation rate state estimation function of the optimal solution of the degradation rate state estimation function, and the probability density function is used for predicting the residual life of the part.
11. An electronic device is characterized by comprising a memory and a processor; the memory stores computer-executable instructions; A processor executing computer-executable instructions stored in a memory, causing the processor to perform the method of any one of claims 1-9.

Description

Method and equipment for predicting residual life of oil and gas drilling equipment Technical Field The application relates to the technical field of oil and gas drilling, in particular to a method and equipment for predicting the residual life of oil and gas drilling equipment. Background The oil gas drilling equipment is key equipment for oil gas resource exploration and development, is closely related to links such as development production, internal and external gathering and delivery, terminal distribution and the like, and once unintended failure of key parts occurs, unexpected shutdown of the equipment is caused, and immeasurable loss can be caused. Therefore, there is a need for applying effective health management techniques to critical equipment components, predicting the remaining life of oil and gas drilling equipment, enabling the interrelation between the maintenance activity schedule and the corresponding maintenance resource management to be fully considered, and ultimately completing maintenance at minimal cost. Many data-driven health management approaches have been developed, most of which use degradation data such as structural noise, pretension, motor current signals, etc., that rely on indirect wear effect detection. However, the detection means based on the indirect abrasion effect inevitably brings errors due to the introduction of other physical quantities, influences the accuracy of prediction, and has great application limitation on complex operation environments of oil and gas drilling. Aiming at the defects, the application provides a method and equipment for predicting the residual life of oil and gas drilling equipment. Disclosure of Invention The application provides a method and equipment for predicting the residual life of oil and gas drilling equipment, which are used for improving the accuracy of life prediction. In a first aspect, the present application provides a method for predicting the remaining life of an oil and gas drilling apparatus, comprising: collecting image data in the historical time of the parts, and acquiring the abrasion area of the image data according to the improved image segmentation model; acquiring degradation characteristics of the parts in historical time according to the abrasion area; establishing a linear degradation model according to degradation characteristics, wherein the linear degradation model comprises a system equation based on degradation rate and a state equation based on the degradation characteristics; Solving the linear degradation model through a Kalman filter based on the iterative updating of the expected maximum algorithm to obtain the optimal estimation of the degradation rate state estimation function; and obtaining a probability density function of the residual life of the part through a simulated wiener process according to the optimal estimation of the degradation characteristic and the degradation rate state estimation function, wherein the probability density function is used for predicting the residual life of the part. In one possible design, the obtaining the optimal estimate of the degradation rate state estimation function by solving the linear degradation model with a kalman filter iteratively updated based on a desired maximum algorithm includes: Solving the linear degradation model through a Kalman filter to obtain an iterative solution of function parameters in the degradation rate state estimation function, wherein the function parameters comprise the expectation of posterior estimation, the variance of posterior estimation and the Kalman gain; updating initialization parameters of the Kalman filter through an expected maximum algorithm combined by an RTS smoothing algorithm according to the iterative solution of the function parameters; Returning to the step of solving the linear degradation model through the Kalman filter according to the updated initialization parameters until the expected maximum algorithm meets the convergence condition or reaches the iteration times, and taking the iteration solution of the function parameters as the optimal estimation of the function parameters; and obtaining the optimal estimation of the degradation rate state estimation function according to the optimal estimation of the function parameter. In one possible design, the initialization parameters of the kalman filter include an expectation of an initial degradation state of a state estimation function of the degradation rate, a variance of the initial degradation state, a system process noise, and a diffusion coefficient; The initialization parameters of the Kalman filter are updated according to the expected maximum algorithm combined by the RTS smoothing algorithm by the iterative solution of the function parameters, and the method comprises the following steps: Obtaining a first expected value, a second expected value and a third expected value through an RTS smoothing algorithm according to an iterative solution of