Pipeline defect prediction and evaluation method based on AKF and SR-UPF

Abstract

The invention provides a pipeline defect prediction and evaluation method and system based on AKF and SR-UPF. The method comprises the following steps: obtaining observation data of a pipeline; based on the observation data, a growth rule of pipeline defects is obtained through a preset mechanism model; introducing random disturbance in the defect growth process through a preset random process model based on the growth rule of the pipeline defect, and generating a pipeline defect prediction result; wherein the mechanism model carries out dynamic estimation and correction on parameter uncertainty and noise distribution through adaptive Kalman filtering, and the random process model carries out particle weight and distribution optimization through square root unscented particle filtering. The problems that in the prior art, a pipeline corrosion and crack growth reliability evaluation method is insufficient in prediction precision and low in calculation efficiency are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline defect assessment, and in particular to a pipeline defect prediction and assessment method based on AKF and SR-UPF. Background Art

[0002] With the widespread adoption of pipeline transportation technology, infrastructure such as oil and gas pipelines and water supply pipelines has become a crucial support for energy and resource transportation. However, long-serving pipelines are prone to defects such as corrosion and crack propagation under complex operating conditions, resulting in reduced pipeline strength and reliability. In severe cases, this can lead to major safety incidents such as leaks and explosions. Therefore, accurately predicting pipeline defect growth trends and assessing their health and reliability are crucial for ensuring the safe operation of engineering structures. Currently, modeling methods for pipeline defect growth fall into two main categories: mechanism models and stochastic process modeling.

[0003] Mechanistic models, based on physical equations or empirical formulas, such as the Paris model of crack propagation and the exponential model of corrosion rate, make predictions by describing deterministic laws governing defect growth. However, in practical applications, the parameters of mechanistic models rely on theoretical assumptions and experimental results, failing to fully account for environmental changes, differences in material properties, and uncertainties in loading conditions. Furthermore, these models are prone to cumulative errors during long-term predictions, resulting in significant deviations between predicted results and actual defect growth patterns, making it difficult to meet the requirements for high-precision predictions.

[0004] Stochastic process modeling, such as the gamma process, inverse Gaussian process, and Wiener process, describes the uncertain behavior of defect growth by introducing random perturbations. These methods can effectively address nonlinear and random issues, addressing the shortcomings of mechanistic models. However, the construction of stochastic process models relies on internal detection data, which is often subject to noise. Furthermore, particle filtering methods often face problems such as particle degradation and high computational complexity, resulting in insufficient accuracy and poor real-time performance in practical applications.

[0005] Therefore, existing mechanistic models and stochastic process modeling methods both have limitations. Using them alone makes it difficult to accurately and dynamically predict and evaluate pipeline defect growth states. To address this issue, filtering technology has been increasingly applied to defect growth modeling. By dynamically calibrating the model's predicted state with actual observed data, filtering methods effectively improve the accuracy and robustness of state estimation.

[0006] However, traditional filtering methods (such as standard Kalman filtering and extended Kalman filtering) have limited performance when dealing with nonlinear systems and uncertain noise, and are difficult to adapt to complex defect growth processes. Therefore, an improved method is needed that can combine adaptive weighted Kalman filtering (AKF) and square-root unscented particle filtering (SUP) to dynamically correct and optimize the mechanism model and stochastic process modeling, respectively. Summary of the Invention

[0007] The present invention provides a pipeline defect prediction and assessment method based on AKF and SR-UPF, which is used to solve the problems of insufficient prediction accuracy and low calculation efficiency in the pipeline corrosion and crack growth reliability assessment methods in the prior art.

[0008] The present invention provides a pipeline defect prediction and assessment method based on AKF and SR-UPF, comprising: Obtain pipeline observation data; Obtaining the growth law of pipeline defects through a preset mechanism model based on the observation data; Based on the growth law of pipeline defects, random disturbances in the defect growth process are introduced through a preset random process model to generate pipeline defect prediction results; The mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, and the random process model optimizes particle weights and distribution through square root unscented particle filtering.

[0009] According to a pipeline defect prediction and assessment method based on AKF and SR-UPF provided by the present invention, the mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, specifically including: The mechanism model represents the defect growth state evolution and observation process through linear state equations and observation equations; Based on the linear state equation and observation equation, the state value is predicted and updated through adaptive Kalman filtering, and the parameter uncertainty and noise distribution of the mechanism model are dynamically estimated and corrected to obtain a corrected mechanism model.

[0010] According to a pipeline defect prediction and assessment method based on AKF and SR-UPF provided by the present invention, the state value is predicted and updated by adaptive Kalman filtering based on the linear state equation and observation equation, specifically including: Based on the linear state equation and observation equation, the current state is predicted by adaptive Kalman filtering to obtain the state value; The state value is compared with the latest observation data, and the state value is corrected and updated according to the comparison result.

[0011] According to a pipeline defect prediction and assessment method based on AKF and SR-UPF provided by the present invention, the random process model optimizes particle weights and distribution through square root unscented particle filtering, specifically including: The random process model uses the posterior probability density function obtained by unscented transformation as the prior probability density function of the particle filter through the square root unscented particle filter, resamples the particle filter, and optimizes the particle weight and distribution.

[0012] According to a pipeline defect prediction and assessment method based on AKF and SR-UPF provided by the present invention, the random process model uses the posterior probability density function obtained by unscented transformation as the prior probability density function of the particle filter through square root unscented particle filtering, and resamples the particle filter, including: Initialize filtering conditions and generate particle swarm; Obtain sigma points by performing unscented transformation based on particle swarm; Perform point set prediction based on sigma point weights, generate predicted values, calculate observed values ​​based on predicted values, and perform measurement updates; Based on the measurement update results, the sampling update particles are calculated and the weights are updated. After resampling, the state variable estimation value and the error covariance matrix after unscented sampling filtering calculation are output.

[0013] According to a pipeline defect prediction and assessment method based on AKF and SR-UPF provided by the present invention, the calculation process of the sigma point weight is: Where: is the weight coefficient for predicting the update of state variables, is the weight coefficient for the updated state variable covariance prediction, l, a, b are all unscented transformation parameters, and , n is the dimension of the state vector, α It is usually a positive number not greater than 1.

[0014] The present invention also provides a pipeline defect prediction and assessment system based on AKF and SR-UPF, the system comprising: Data acquisition module, used to obtain pipeline observation data; A defect prediction module, configured to obtain the growth law of pipeline defects through a preset mechanism model based on the observation data; The disturbance setting module is used to introduce random disturbances in the defect growth process based on the growth law of pipeline defects through a preset random process model to generate pipeline defect prediction results; The mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, and the random process model optimizes particle weights and distribution through square root unscented particle filtering.

[0015] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the pipeline defect prediction and assessment method based on AKF and SR-UPF as described above is implemented.

[0016] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the pipeline defect prediction and assessment method based on AKF and SR-UPF as described above is implemented.

[0017] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-described pipeline defect prediction and assessment methods based on AKF and SR-UPF.

[0018] The present invention provides a pipeline defect prediction and assessment method based on AKF and SR-UPF. The method uses adaptive Kalman filtering to process the uncertainty and noise interference of mechanism model parameters, dynamically adjusts the weight and noise covariance matrix, and enables the mechanism model to be adjusted in real time according to observation data, thereby improving the accuracy and adaptability of prediction. The method also uses square root unscented particle filtering to filter and correct the state estimation of the random process. Through weight optimization and particle compression, the particle degradation problem is effectively solved, the computational complexity is reduced, and the filtering efficiency and modeling accuracy are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 It is a flow chart of the pipeline defect prediction and evaluation method based on AKF and SR-UPF provided by the present invention.

[0021] Figure 2 This is a schematic diagram of the module connection of the pipeline defect prediction and evaluation system based on AKF and SR-UPF provided by the present invention.

[0022] Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention.

[0023] Reference numerals: 110 : data acquisition module; 120 : defect prediction module; 130 : disturbance setting module; 310 : processor; 320 : communication interface; 330 : memory; 340 : communication bus. DETAILED DESCRIPTION

[0024] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0025] The following combination Figure 1 The present invention describes a pipeline defect prediction and assessment method based on AKF and SR-UPF, including: step 100, obtaining pipeline observation data.

[0026] Step 200: Obtain the growth law of pipeline defects through a preset mechanism model based on the observation data.

[0027] Step 300: Based on the growth law of pipeline defects, random disturbances in the defect growth process are introduced through a preset random process model to generate pipeline defect prediction results.

[0028] The mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, and the random process model optimizes particle weights and distribution through square root unscented particle filtering.

[0029] In the present invention, the mechanism model is dynamically corrected, and the uncertainty and noise interference of the mechanism model parameters are processed through adaptive Kalman filtering (AKF). The weights and noise covariance matrix are dynamically adjusted so that the mechanism model can be adjusted in real time according to the observed data, thereby improving the accuracy and adaptability of the prediction. The random process modeling is optimized, and the state estimation of the random process is filtered and corrected using the square root unscented particle filter (SR-UPF). Through weight optimization and particle compression, the particle degradation problem is effectively solved, the computational complexity is reduced, and the filtering efficiency and modeling accuracy are improved. When predicting defect growth, the defect growth mechanism model and random process corrected based on filtering technology provide more accurate defect growth trend prediction and reliability assessment results, providing accurate and reliable support for engineering structure health monitoring, maintenance decision-making and life prediction.

[0030] In the defect growth modeling process, parameter uncertainty and noise interference in the mechanism model lead to a decrease in prediction accuracy. To address this problem, an adaptive Kalman filter (AKF) is introduced to dynamically correct and optimize the mechanism model.

[0031] Linear state equation and observation equation description,The state evolution and observation process of defect growth can be expressed as linear state equation and observation equation, respectively.

[0032] .

[0033] .

[0034] Where, is the system state (such as the depth of defect growth, crack length, etc.), is the external control variable (such as load, environmental factors), is the process noise; For observation data (such as defect data obtained by internal inspection equipment), is the observation noise.

[0035] The implementation of an adaptive Kalman filter (AKF) is as follows: A Kalman filter is a recursive filter used to estimate the system state, combining a dynamic model of the system with measurement data. The mathematical model of the Kalman filter consists of two main steps: prediction and update. The prediction step uses the system's dynamic model and an estimate of the previous state to predict the current state of the system, resulting in a state value. The update step compares the predicted state value with the latest observed data and modifies the predicted state value based on the difference between the two.

[0036] In the prediction step, the predicted update of the system state vector is given by: .

[0037] Where: is the state estimation vector at time k; is the state transfer matrix; is the control input matrix; is the control input vector; is the covariance matrix of the state estimation error; is the covariance matrix of the process noise.

[0038] In the update step, the state estimate is corrected using the latest observation data. The updated state estimate vector and covariance matrix are given by the following formula: .

[0039] Where, is the Kalman gain matrix; is the observation matrix; is the updated error covariance matrix; is the observation noise covariance; is the observation vector at time .

[0040] Dynamic Adjustment (covariance matrix of process noise) and (Observation noise covariance) In response to changes in actual measurements, an adaptive Kalman filter is constructed that adjusts parameter values ​​using the innovation covariance.

[0041] Building an innovation sequence for: .

[0042] Constructing innovation covariance for: .

[0043] If the actual covariance of the innovation series deviates significantly from ,show It needs to be readjusted. The adjustment strategy is: .

[0044] Where: is the smoothing factor, The adjustment is based on the long-term performance of the forecast error, similar.

[0045] To address the nonlinear and random nature of the defect growth process, the square root unscented particle filter (SR-UPF) method is used to modify the random process modeling. The random process model is based on the following nonlinear state equation and observation equation.

[0046] .

[0047] Where, is the system state (such as the depth of defect growth, crack length, etc.), is the external control variable (such as load, environmental factors), is the process noise; For observation data (such as defect data obtained by internal inspection equipment), is the observation noise.

[0048] The implementation of the square root unscented particle filter (SR-UPF) is as follows: Particle filters are well-suited for nonlinear problems, but as the number of importance samples increases, particle degeneration occurs, with the weights of most particles decreasing and their contribution to the posterior probability density approaching zero. To avoid this, the posterior probability density function obtained through the unscented transformation is used as the prior probability density function for the particle filter, resampling the particle filter. The UKF (Unscented Kalman Filter) selects sampling points based on the prior mean and prior mean square error. This process requires the covariance matrix to be positive definite, so the square root unscented transformation is chosen, and the QR transform is used instead of the Cholesky transform to ensure non-negativity.

[0049] The Unscented Kalman Filter (UKF) is an algorithm for state estimation in nonlinear systems and an extension of the Kalman filter. It addresses nonlinear problems through an unscented transform, avoiding the linearization errors found in the traditional Extended Kalman Filter (EKF), resulting in higher accuracy and stability.

[0050] The core of the UKF is the unscented transform, which approximates the probability distribution by selecting a set of deterministic sample points (called sigma points). After being transformed by a nonlinear function, these points can preserve the mean and covariance of the original distribution, thereby more accurately describing the state changes of the nonlinear system.

[0051] Sigma point selection, based on the mean and covariance of the current state, a set of Sigma points are symmetrically selected to cover the distribution range of the state.

[0052] Prediction (prediction phase) propagates the Sigma points through the nonlinear state equation to obtain the predicted Sigma points. The mean and covariance of the predicted state are calculated.

[0053] The update (correction phase) propagates the predicted Sigma points through the nonlinear observation equation to obtain the predicted observation value. The Kalman gain is calculated and combined with the actual observation value to update the state estimate and covariance.

[0054] QR transform, usually referring to QR decomposition and its application in algorithms, is the process of decomposing a matrix into an orthogonal matrix (Q) and an upper triangular matrix (R).

[0055] The Cholesky transform is a method for decomposing a symmetric positive definite matrix into the product of a lower triangular matrix and its transpose. It is widely used in numerical computing, optimization, statistics, and machine learning, and is particularly suitable for efficiently solving symmetric positive definite linear equations.

[0056] The calculation steps of the square root unscented particle filter include: Initialize the filtering conditions, hour, The prior probability density of , the variance is , generate the initial value of particles , and then generate a particle swarm , , its distribution conforms to the normal distribution, and its mean is , the variance is .

[0057] Perform an untraceable transformation on x. According to the sampling symmetry principle, take the sigma point through Cholesky transformation and get: .

[0058] Where, Sigma points are generated sample points for unscented transformation calculation. is the covariance matrix of the system state, which represents the error range (uncertainty) of the state estimation. k is the secondary scale factor, usually n The value of k= 3 -n , chol() is Cholesky decomposition, a decomposition method of the covariance matrix, used to generate sigma points.

[0059] The calculation formula of Sigma point weight is: .

[0060] Where: is the weight coefficient for predicting the update of state variables, The weight coefficients updated for state variable covariance prediction need to be normalized. l, a, b are all unscented transformation parameters, and , n is the dimension of the state vector, α It is usually a positive number not greater than 1.

[0061] Use sigma point weights to predict point sets. .

[0062] The covariance matrices of the state variable prediction values ​​and errors are: .

[0063] is the system noise autocorrelation covariance matrix. Perform Cholesky calculation on the factors of its predicted mean and state variable covariance: .

[0064] Where, qr {·} is qr decomposition, cholupdate {⋅} is the first-order update of Cholesky decomposition.

[0065] Calculate the system observation value based on the predicted value of the point set and perform measurement update: .

[0066] The covariance of the measurements is: .

[0067] Where, represents the 1:2n measurement update value obtained during the untraceable transformation process, represents the 0th measurement update value obtained during the traceless transformation, is the observation noise autocorrelation covariance matrix. The filter gain is: .

[0068] Filter result update: .

[0069] Since the calculation process will produce rounding errors, the matrix positive definiteness is difficult to guarantee, so we propose ( r is an upper triangular matrix, q is an orthogonal matrix), then have to: At the same time: Similarly, the following can be applied: Transformed into .

[0070] Calculate the sampling update particles, and update the sampling density function of each particle state by the posterior probability density function calculated by unscented transformation. N particles are extracted as follows: .

[0071] In the formula 0:k-1 express 0 arrive k-1 Sampling time, 1:k Indicates the 1 arrive k Sampling time.

[0072] New sample updates weights: Normalize it to get: Resampling: .

[0073] Take here If resampling is required, resample while ensuring the diversity of particles, appropriately reduce the number of particles with low weights, and update the weights. .

[0074] Output: .

[0075] In the formula is the estimated value of the state variable after unscented sampling filtering, is its error covariance matrix.

[0076] The pipeline defect prediction and assessment based on the AKF and SR-UPF disclosed in this invention combines the adaptive Kalman filter (AKF) and the square root unscented particle filter (SR-UPF) to achieve dynamic correction and optimization of the corrosion and crack growth mechanism model and stochastic process modeling. The VB-UKF effectively addresses the parameter uncertainty and noise interference issues of the mechanism model, significantly improving prediction accuracy and the model's adaptability. Simultaneously, the SR-UPF optimizes the state estimation of the stochastic process, addresses particle degradation, reduces computational complexity, and enhances modeling efficiency and real-time performance. When predicting defect growth, the mechanism model, stochastic modeling, and filtering techniques are combined to provide more accurate defect growth trend prediction and reliability assessment. Ultimately, this achieves precise monitoring of the health of engineering structures and early warning of failure risks, providing a scientific basis for engineering maintenance and decision-making, improving system reliability and safety, and extending the service life of structures and equipment.

[0077] refer to Figure 2The present invention also discloses a pipeline defect prediction and evaluation system based on AKF and SR-UPF, the system comprising: The data acquisition module 110 is used to acquire the observation data of the pipeline; The defect prediction module 120 is configured to obtain the growth pattern of pipeline defects through a preset mechanism model based on the observation data; The disturbance setting module 130 is used to introduce random disturbances in the defect growth process based on the growth law of pipeline defects through a preset random process model to generate pipeline defect prediction results; The mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, and the random process model optimizes particle weights and distribution through square root unscented particle filtering.

[0078] The mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, specifically including: The mechanism model represents the defect growth state evolution and observation process through linear state equations and observation equations; Based on the linear state equation and observation equation, the state value is predicted and updated through adaptive Kalman filtering, and the parameter uncertainty and noise distribution of the mechanism model are dynamically estimated and corrected to obtain a corrected mechanism model.

[0079] Based on the linear state equation and observation equation, the state value is predicted and updated through adaptive Kalman filtering, specifically including: Based on the linear state equation and observation equation, the current state is predicted by adaptive Kalman filtering to obtain the state value; The state value is compared with the latest observation data, and the state value is corrected and updated according to the comparison result.

[0080] The random process model uses square root unscented particle filtering to optimize particle weights and distribution, including: The random process model uses the posterior probability density function obtained by unscented transformation as the prior probability density function of the particle filter through the square root unscented particle filter, resamples the particle filter, and optimizes the particle weight and distribution.

[0081] The random process model uses the square root unscented particle filter to take the posterior probability density function obtained by unscented transformation as the prior probability density function of the particle filter, and resamples the particle filter, including: Initialize filtering conditions and generate particle swarm; Obtain sigma points by performing unscented transformation based on particle swarm; Perform point set prediction based on sigma point weights, generate predicted values, calculate observed values ​​based on predicted values, and perform measurement updates; Based on the measurement update results, the sampling update particles are calculated and the weights are updated. After resampling, the state variable estimation value and the error covariance matrix after unscented sampling filtering calculation are output.

[0082] The calculation process of the sigma point weight is: .

[0083] Where: is the weight coefficient for predicting the update of state variables, is the weight coefficient for the updated state variable covariance prediction, l, a, b are all unscented transformation parameters, and , n is the dimension of the state vector, α It is usually a positive number not greater than 1.

[0084] The pipeline defect prediction and assessment system based on AKF and SR-UPF provided by the present invention uses adaptive Kalman filtering to process the uncertainty and noise interference of the mechanism model parameters, dynamically adjusts the weight and noise covariance matrix, and enables the mechanism model to be adjusted in real time according to the observed data, thereby improving the accuracy and adaptability of the prediction. The square root unscented particle filter is used to filter and correct the state estimation of the random process. Through weight optimization and particle compression, the particle degradation problem is effectively solved, the computational complexity is reduced, and the filtering efficiency and modeling accuracy are improved.

[0085] Figure 3 An example of a physical structure diagram of an electronic device is shown below. Figure 3 As shown, the electronic device may include: a processor 310, a communications interface 320, a memory 330, and a communications bus 340, wherein the processor 310, the communications interface 320, and the memory 330 communicate with each other via the communications bus 340. The processor 310 may call logic instructions in the memory 330 to execute a pipeline defect prediction and assessment method based on the AKF and SR-UPF. The method includes: obtaining pipeline observation data; obtaining the growth pattern of pipeline defects using a preset mechanism model based on the observation data; introducing random perturbations in the defect growth process using a preset random process model based on the pipeline defect growth pattern to generate pipeline defect prediction results; wherein the mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution using an adaptive Kalman filter, and the random process model optimizes particle weights and distribution using a square root unscented particle filter.

[0086] Furthermore, the logic instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0087] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a pipeline defect prediction and assessment method based on AKF and SR-UPF provided by the above methods. The method includes: obtaining pipeline observation data; obtaining the growth law of pipeline defects through a preset mechanism model based on the observation data; introducing random disturbances in the defect growth process through a preset random process model based on the growth law of pipeline defects to generate pipeline defect prediction results; wherein, the mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, and the random process model optimizes particle weights and distribution through square root unscented particle filtering.

[0088] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute a pipeline defect prediction and assessment method based on AKF and SR-UPF provided by the above-mentioned methods, the method comprising: obtaining observation data of the pipeline; obtaining the growth law of pipeline defects through a preset mechanism model based on the observation data; introducing random disturbances in the defect growth process through a preset random process model based on the growth law of pipeline defects to generate pipeline defect prediction results; wherein, the mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, and the random process model optimizes particle weights and distribution through square root unscented particle filtering.

[0089] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0090] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A pipeline defect prediction and assessment method based on AKF and SR-UPF, characterized in that: include: Obtain observation data of the pipeline; Obtaining the growth law of pipeline defects through a preset mechanism model based on the observation data; Based on the growth law of pipeline defects, random disturbances in the defect growth process are introduced through a preset random process model to generate pipeline defect prediction results; The mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, and the random process model optimizes particle weights and distribution through square root unscented particle filtering.

2. The pipeline defect prediction and assessment method based on AKF and SR-UPF according to claim 1 is characterized in that: The mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, specifically including: The mechanism model represents the defect growth state evolution and observation process through linear state equations and observation equations; Based on the linear state equation and observation equation, the state value is predicted and updated through adaptive Kalman filtering, and the parameter uncertainty and noise distribution of the mechanism model are dynamically estimated and corrected to obtain a corrected mechanism model.

3. The pipeline defect prediction and assessment method based on AKF and SR-UPF according to claim 2 is characterized in that: The prediction and update of the state value based on the linear state equation and the observation equation by adaptive Kalman filtering specifically includes: Based on the linear state equation and observation equation, the current state is predicted by adaptive Kalman filtering to obtain the state value; The state value is compared with the latest observation data, and the state value is corrected and updated according to the comparison result.

4. The pipeline defect prediction and assessment method based on AKF and SR-UPF according to claim 1 is characterized in that: The random process model optimizes particle weights and distribution through square root unscented particle filtering, specifically including: The random process model uses the posterior probability density function obtained by unscented transformation as the prior probability density function of the particle filter through the square root unscented particle filter, resamples the particle filter, and optimizes the particle weight and distribution.

5. The pipeline defect prediction and assessment method based on AKF and SR-UPF according to claim 4 is characterized in that: The random process model uses the posterior probability density function obtained by the unscented transformation as the prior probability density function of the particle filter through the square root unscented particle filter, and resamples the particle filter, including: Initialize filtering conditions and generate particle swarm; Obtain sigma points by performing unscented transformation based on particle swarm; Perform point set prediction based on sigma point weights, generate predicted values, calculate observed values ​​based on predicted values, and perform measurement updates; Based on the measurement update results, the sampling update particles are calculated and the weights are updated. After resampling, the state variable estimation value and the error covariance matrix after unscented sampling filtering calculation are output.

6. The pipeline defect prediction and assessment method based on AKF and SR-UPF according to claim 5 is characterized in that: The calculation process of the sigma point weight is: Where: is the weight coefficient for predicting the update of state variables, is the weight coefficient for the updated state variable covariance prediction, λ, α, β are all unscented transformation parameters, and , n is the dimension of the state vector, α It is usually a positive number not greater than 1.

7. A pipeline defect prediction and assessment system based on AKF and SR-UPF, characterized by: The system comprises: Data acquisition module, used to obtain pipeline observation data; A defect prediction module, configured to obtain the growth law of pipeline defects through a preset mechanism model based on the observation data; The disturbance setting module is used to introduce random disturbances in the defect growth process based on the growth law of pipeline defects through a preset random process model to generate pipeline defect prediction results; The mechanism model dynamically estimates and corrects parameter uncertainty and noise distribution through adaptive Kalman filtering, and the random process model optimizes particle weights and distribution through square root unscented particle filtering.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the pipeline defect prediction and assessment method based on AKF and SR-UPF as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the pipeline defect prediction and assessment method based on AKF and SR-UPF as claimed in any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the pipeline defect prediction and assessment method based on AKF and SR-UPF as claimed in any one of claims 1 to 6 is implemented.

Images