Offshore wellhead multi-sensor detection method and device

Abstract

The present application discloses an offshore wellhead multi-sensor detection method and device, which relates to the technical field of acoustic emission sensors. The method includes: acquiring multi-sensor data; wherein the multi-sensor data comes from acoustic emission sensors at multiple different installation positions; optimizing the multi-sensor data to obtain an optimized signal; extracting the time domain features and frequency domain features of the optimized signal, and obtaining the key signal therein through feature fusion; determining the time difference between the key signals, and using a positioning algorithm to determine the source position of the key signal based on the time difference; classifying the key signals through a machine training algorithm to obtain the classification category of the key signals; obtaining the classification category and source position of each key signal according to the source position and classification category of the key signal. The method can accurately detect, identify and classify different types of acoustic emission events, and has high positioning accuracy.

Description

Technical Field

[0001] The present application relates to the technical field of acoustic emission sensors, and in particular to a multi-sensor detection method and device for an offshore wellhead. Background Art

[0002] With the vigorous development of the new generation of electronic information technology, the Internet of Things technology has gradually become the core driving force for the society to move towards the Internet of Everything. In the process of target detection, industrial application, natural resource exploitation and weather forecasting, through the integration of various sensor resources through the Internet of Things platform, big data analysis and artificial intelligence algorithm and other technologies, a multi-source data fusion system can be built to deeply explore the value behind the data, provide targeted solutions for all walks of life, and promote the development of society in a more intelligent direction.

[0003] In this development context, traditional technologies usually use multiple sensors and collaborate using ultrasonic detection, X-ray detection, and electromagnetic detection. Although the data of multiple sensors are fused and reasoned through methods such as Bayesian theorem, Dempster-Shafer evidence theory, or production rules, the reliability of data collection and processing has been significantly improved compared to single-sensor systems, how to efficiently fuse data from multiple sensors is still under exploration.

[0004] In addition, the existing technology will divide the structure according to the sensor fusion method, and send the raw data obtained by each sensor directly to the central processor for fusion processing. The data processing accuracy is high, but this method has very high requirements on the processor, high economic cost, but low reliability. After data fusion, the target detection point needs to be accurately located. Usually, multiple sensors are used to establish an overdetermined set of equations, and then the least squares method is applied to solve them to eliminate the influence of random errors. However, different positioning points may have different positioning errors. In practical applications, especially for underground detection, which is dangerous and requires accurate positioning, the effect is poor and it is difficult to achieve accurate positioning. Summary of the invention

[0005] The embodiments of the present application solve the problems of low detection accuracy and inaccurate positioning of the positioning method based on multiple acoustic emission sensors in the prior art by providing an offshore wellhead multi-sensor detection method and device. The Gray Wolf algorithm is introduced to establish a Gray Wolf optimization model to process multi-sensor data, achieve preliminary denoising, and after dimensionality reduction processing of the signal, the optimized signals of multiple sensors at different locations are predicted through state equations and observation equations, and further processed using adaptive weighting functions, and then combined with machine training algorithms to determine the classification category of the source location, thereby improving the accuracy of damage positioning. It achieves accurate identification and classification of different types of acoustic emission events and provides alarm feedback, providing early warning information and maintenance suggestions so that relevant personnel can handle them reasonably and improve response speed and reliability.

[0006] In a first aspect, an embodiment of the present application provides an offshore wellhead multi-sensor detection method, comprising: acquiring multi-sensor data; wherein the multi-sensor data comes from acoustic emission sensors at multiple different installation positions; optimizing the multi-sensor data to obtain an optimized signal; extracting the time domain features and frequency domain features of the optimized signal, and obtaining the key signal therein through feature fusion; determining the time difference between the key signals, and determining the source position of the key signal based on the time difference using a positioning algorithm; classifying the key signals through a machine training algorithm to obtain the classification category of the key signals; and obtaining the classification category of each key signal and its source position according to the source position and the classification category of the key signal.

[0007] In combination with the first aspect, in a possible implementation, the optimizing the multi-sensor data to obtain an optimized signal comprises: inputting the multi-sensor data into a gray wolf optimization model; initializing wolf pack parameters, and randomly generating positions of individuals in the wolf pack, and executing a decomposition step until a preset termination condition is reached to obtain an optimal position; wherein the wolf pack parameters comprise the number of individuals in the wolf pack, a convergence parameter, and the number of iterations; the decomposition step comprises: defining a fitness function, and using the fitness function to calculate the fitness value of the position of each individual; updating the positions of individuals in the wolf pack according to the fitness value; executing the decomposition step based on the updated positions of individuals in the wolf pack; using the optimal position as a decomposition parameter, and using the decomposition parameter to decompose and reconstruct the multi-sensor data to obtain the optimized signal.

[0008] In combination with the first aspect, in a possible implementation method, the extracting of time domain features and frequency domain features of the optimized signal includes: defining the state equation and observation equation of the optimized signal based on the Kalman filtering algorithm, and iteratively performing an update step until the time domain features and / or the frequency domain features are traversed; the update step is as follows: setting the estimated value and uncertainty of the state estimate at the current moment; using the state equation to predict the state value and uncertainty of the state estimate at the next moment based on the state value and uncertainty of the state estimate at the current moment; updating the state value and uncertainty of the state estimate at the next moment through the observation equation and the time domain features or the frequency domain features; using the state value and uncertainty of the state estimate at the next moment as the state value and uncertainty of the state estimate at the current moment, and performing the update step.

[0009] In combination with the first aspect, in a possible implementation manner, determining the time difference between the key signals includes:

[0010] An adaptive weighting function is defined, and the time difference between the key signals is calculated by the adaptive weighting function; wherein the adaptive weighting function is as follows:

[0011] ;in,

[0012] ;

[0013] In the formula, represents the adaptive weighting function, represents the weight of adaptive weighting, represents the adjustment parameter, represents the signal-to-noise ratio, Indicates the preset signal-to-noise ratio, Key signal indicating input With key signal The coherence function of Key signal indicating input The autopower spectrum function of Key signal indicating input The autopower spectrum function of Key signal indicating input With key signal The cross power spectrum function of The weight parameter representing the key signal of the input.

[0014] In combination with the first aspect, in a possible implementation manner, the calculating the time difference between the key signals by the adaptive weighting function further includes: performing weighted processing on the adaptive weighting function to obtain a generalized cross-correlation function, and taking the time delay corresponding to the maximum value thereof as the time difference, as follows:

[0015] ;

[0016] In the formula, represents the delay variable, Indicates the time difference, express The corresponding input when the maximum value is obtained, Indicates the current time. and They represent different key input signals respectively.

[0017] In combination with the first aspect, in a possible implementation method, the use of a positioning algorithm to determine the source position of the key signal based on the time difference includes: using a positioning algorithm based on the time difference to determine the source position of the key signal; wherein the positioning algorithm includes: determining the signal distance between the key signal and each acoustic emission sensor according to the time difference; determining the signal direction of the key signal according to the signal distance and the geometric layout between each acoustic emission sensor; and determining the source position using a triangulation algorithm based on the signal distance and the signal direction.

[0018] In combination with the first aspect, in a possible implementation manner, after obtaining the optimized signal, the method further includes: performing dimensionality reduction processing on the optimized signal.

[0019] In the second aspect, an embodiment of the present application provides an offshore wellhead multi-sensor detection device, including: a data acquisition module, used to obtain multi-sensor data; wherein the multi-sensor data comes from acoustic emission sensors at multiple different installation positions; a data processing module, used to optimize the multi-sensor data to obtain an optimized signal; a feature extraction module, used to extract the time domain features and frequency domain features of the optimized signal, and obtain the key signal therein through feature fusion; a signal processing module, used to determine the time difference between the key signals, and use a positioning algorithm to determine the source position of the key signal based on the time difference; a classification module, used to classify the key signals through a machine training algorithm to obtain the classification category of the key signals; an event positioning module, used to obtain the classification category of each of the key signals and its source position according to the source position and the classification category of the key signal.

[0020] In a third aspect, an embodiment of the present application provides a device for executing a multi-sensor detection method for an offshore wellhead, the device comprising: a processor; a memory for storing processor executable instructions; when the processor executes the executable instructions, it implements the method described in the first aspect or any possible implementation method of the first aspect.

[0021] In a fourth aspect, an embodiment of the present application provides a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium includes a medium for storing a computer program or instructions, which, when executed, enables the method described in the first aspect or any possible implementation method of the first aspect to be implemented.

[0022] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0023] The embodiment of the present application optimizes the multi-sensor data by adopting an offshore wellhead multi-sensor detection method and device, thereby effectively removing the noise of the original multi-sensor data. The key time domain features and frequency domain features are extracted from each optimized signal, and the key signal is obtained by feature fusion. The key signal is processed using an adaptive weighting function, the time difference of the arrival time of each acoustic emission sensor signal is calculated, and the time difference is classified, and finally the source position of the key signal of different categories is accurately determined. The final positioning accuracy and the judgment of the damage event are improved, while also reducing the burden on the central processing unit and improving the detection speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0025] Figure 1 A flow chart of an offshore wellhead multi-sensor detection method provided in an embodiment of the present application;

[0026] Figure 2 A flowchart of processing optimization signals provided in an embodiment of the present application;

[0027] Figure 3 A schematic diagram of the structure of an offshore wellhead multi-sensor detection device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0029] The following describes some of the techniques involved in the embodiments of the present application to facilitate understanding, and they should be considered as merely exemplary. Therefore, it should be appreciated by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, some descriptions of well-known functions and structures are omitted in the following description.

[0030] Figure 1 It is a flow chart of the offshore wellhead multi-sensor detection method provided in an embodiment of the present application, including steps 101 to 106. Figure 1 This is only an execution order shown in the embodiment of the present application, and does not represent the only execution order of the offshore wellhead multi-sensor detection method. In the case that the final result can be achieved, Figure 1 The steps shown may be performed in parallel or in reverse, as follows.

[0031] Step 101: Acquire multi-sensor data. The multi-sensor data comes from acoustic emission sensors at multiple different installation positions. In the embodiment of the present application, multiple acoustic emission sensors are arranged for detecting offshore oil and gas wells, and the installation positions of these acoustic emission sensors can meet the detection requirements of different detection areas and orientations. The multi-sensor data is obtained by synchronously recording data with a high sampling rate data acquisition card to ensure the consistency of data in each channel.

[0032] Step 102: Optimize the multi-sensor data to obtain an optimized signal. In the embodiment of the present application, the gray wolf optimization model is selected for data processing, and the gray wolf optimization model is used to optimize the parameters of the variational mode decomposition to adapt it to different signal characteristics, thereby effectively removing noise from the multi-sensor data. Figure 2 As shown, multi-sensor data is input into the gray wolf optimization model; wolf pack parameters are initialized, and positions of individuals in the wolf pack are randomly generated, and decomposition steps are performed until the preset termination condition is reached to obtain the optimal position; wherein the wolf pack parameters include the number of individuals in the wolf pack, convergence parameters and number of iterations;

[0033] The decomposition steps include:

[0034] Define a fitness function, and use the fitness function to calculate the fitness value of each individual's position; update the position of the individual in the wolf pack according to the fitness value; perform a decomposition step based on the updated position of the individual in the wolf pack; use the optimal position as a decomposition parameter, and use the decomposition parameter to decompose and reconstruct the multi-sensor data to obtain an optimized signal.

[0035] Specifically, the individuals (wolves) in the wolf pack are divided into four types, namely the optimal solution, the second-best solution, the third-best solution and the remaining solutions. The position of each wolf is determined by Encircling. For example, the number of individuals in the wolf pack is set to 20, and the current position of each wolf is represented by a two-dimensional vector express, and The convergence parameter ranges of are [3,10] and [1000,3000] respectively. During initialization, the position of each wolf is randomly generated within the parameter range:

[0036] The updated positions of each wolf are as follows:

[0037] ;in,

[0038] , .

[0039] In the formula, represents the updated position of each wolf, Indicates the current position of the wolf. Indicates the location of the optimal solution, suboptimal solution, or third optimal solution in the wolf pack. represents the coefficient vector that determines the direction and distance of movement of each wolf during the search process, represents the random coefficient vector that controls each wolf to move to the optimal solution, and represents a random vector, Represents the convergence factor, which decreases linearly with the number of iterations. Each wolf represents the parameters of the current variational mode decomposition. By updating the parameters of the current variational mode decomposition, it approaches the position of the optimal solution, the second-best solution, or the third-best solution to find the global optimal solution, that is, the parameters corresponding to the optimal solution of the variational mode decomposition.

[0040] According to the minimum envelope entropy, the effective component is updated, and the positions of the wolves in the optimal solution, the second optimal solution, and the third optimal solution are updated; in order to evaluate the position of each wolf, that is, each group The quality of the vector, the fitness function is defined as follows:

[0041] ;

[0042] In the formula, The fitness function is defined as the mean square error between the denoised multi-sensor data and the original multi-sensor data. Represents the original multi-sensor data sampling points, Represents the denoised multi-sensor data sampling points, represents the number of individuals in the wolf pack, Indicates Among them, the wolf pack level division is updated according to the value of the fitness function, and the positions of the optimal solution, the second-best solution and the third-best solution are updated, and then some individuals are eliminated and / or new individuals are generated.

[0043] If the current number of iterations reaches the number of iterations set in the preset termination condition, the parameters corresponding to the variational mode decomposition are output. Exemplarily, the number of iterations is set to 10 times.

[0044] The adaptive degree function is used to find the parameters corresponding to the optimal solution of the variational modal decomposition of the multi-sensor data, and then it is used as the decomposition parameter to denoise and reconstruct the optimized signal. In the embodiment of the present application, the gray wolf optimization model is used to adaptively find the optimal decomposition parameters, and each mode is appropriately processed, such as selecting some low-frequency signals to reconstruct the optimized signal. Without the need for manual adjustment, the key features of the multi-sensor data are retained while removing the noise in the multi-sensor data, simplifying the amount of calculation in the subsequent steps.

[0045] Step 103: extract the time domain features and frequency domain features of the optimized signal, and obtain the key signal therein by feature fusion. In the embodiment of the present application, the time domain features include: one or more of the maximum value, minimum value, peak value, mean value, variance, standard deviation, kurtosis, skewness, root mean square value, waveform factor, peak factor and impulse factor. The frequency domain features include: one or more of the centroid frequency, mean square frequency, root mean square frequency, frequency variance, frequency standard deviation, power spectrum entropy, energy entropy and singular spectrum entropy.

[0046] The optimization signal is processed by principal component analysis method to reduce the dimension, as follows:

[0047] ;in,

[0048] ,

[0049] ,

[0050] ;

[0051] In the formula, represents the variance, express The orthogonal characteristics of the dimension, Indicates An optimization signal, represents the average value of all optimized signals, represents the mean vector, represents the total number of optimized signals, represents the calculation of the standardized covariance matrix, represents transpose. Dimensional features are mapped to In terms of the characteristics of the dimension, , The dimension is a newly constructed orthogonal feature, which is the main component, different from the traditional method. Subtract from dimension feature Dimensional features.

[0052] In the embodiment of the present application, PCA (principal component analysis method) is selected to perform dimensionality reduction processing on the optimization signal, and the distance between the optimization signals is defined by variance, and the data is projected along the maximum direction, which is easier to distinguish. Each feature is standardized so that its mean is 0 and its variance is 1, which helps to eliminate the dimensional influence between different features.

[0053] In the embodiment of the present application, in order to ensure that the selected principal component retains useful information as much as possible, the optimized signal is subjected to dimensionality reduction processing. The principal component contribution rate is introduced to measure the amount of information contained in the principal component, and a threshold of the contribution rate is set. By calculating the principal component load matrix, the weight of each original time domain feature and / or frequency domain feature on the principal component is obtained, and the feature importance is calculated and sorted, effectively retaining the maximum information of the optimized signal.

[0054] In an embodiment of the present application, the state equation and observation equation of the optimized signal are defined based on the Kalman filtering algorithm, and the update step is iteratively performed until the time domain characteristics and / or frequency domain characteristics are traversed; the update step is as follows: setting the estimated value and uncertainty of the state estimate at the current moment; using the state equation to predict the state value and uncertainty of the state estimate at the next moment based on the state value and uncertainty of the state estimate at the current moment; updating the state value and uncertainty of the state estimate at the next moment through the observation equation and the time domain characteristics or the frequency domain characteristics; using the state value and uncertainty of the state estimate at the next moment as the state value and uncertainty of the state estimate at the current moment.

[0055] Exemplarily, the state equation of the optimized signal is as follows:

[0056] ;

[0057] In the formula, The state equation is The state vector at time, express time, express The next moment of the moment ( The state transition matrix of the system from The state of the moment is transferred to The state evolution of the moment, express The state vector at time, express The control input matrix at the moment describes the influence of the external control quantity on the system. express The external control quantity at the moment, express The noise of the prediction process at that moment.

[0058] The state equation is given by The state at the moment and the external control quantity, prediction process noise and other control variables are used to predict the next moment ( The state of the system at a certain moment in time describes how the system state changes over time.

[0059] The observation equation of the optimized signal is as follows:

[0060] ;

[0061] In the formula, represents the observation equation, Moment The observed values ​​of the acoustic emission sensors, express time, Indicates An acoustic emission sensor, Indicated in Moment The measurement matrix of the acoustic emission sensors is used to map the system state to the measurement space. represents the state equation, Indicated in Moment The measurement noise of each sensor is Indicates Sensors, Represents the total number of acoustic emission sensors.

[0062] All optimization signals are fused into a comprehensive observation matrix, as follows:

[0063] ;

[0064] In the formula, express The observation matrix at time, express The observation matrix at time, Indicated in Moment The observed values ​​of the acoustic emission sensors, Indicated in Moment The measurement matrix of the acoustic emission sensors is used to map the system state to the measurement space. Indicates Sensors, Represents the total number of acoustic emission sensors.

[0065] Among them, the state equation is linked to the observation equation through a comprehensive observation matrix, the error between the two is corrected, and the accuracy of the state estimation at the next moment is improved.

[0066] The prediction of the state value and uncertainty of the state estimate at the next moment is as follows:

[0067] ,

[0068] ;

[0069] In the formula, Indicated in moment to the next moment ( The state prediction at time Indicated in The state prediction of the current moment. express The state transfer matrix at time , represents the transposed matrix, express The control input matrix at the moment describes the influence of the external control quantity on the system. express The external control quantity at the moment, Indicated in The covariance matrix of the state estimate at the next moment, Indicated in The covariance matrix of the state estimate at the current moment, express The covariance of the moments.

[0070] Among them, the state value of the state estimate at the next moment is predicted based on the current state and control variables, and the covariance matrix is ​​used to represent the uncertainty of the system.

[0071] The measurement noise covariance matrix is ​​expanded according to the comprehensive observation matrix as follows:

[0072] ;

[0073] In the formula, represents the covariance matrix of the measurement noise, Indicated in Moment The covariance matrix of the acoustic emission sensor measurement noise is, It means taking the diagonal matrix, Represents the total number of acoustic emission sensors.

[0074] Update the state value and uncertainty of the state estimate at the next moment, as follows:

[0075] ,

[0076] ,

[0077] ;

[0078] In the formula, express The Kalman gain at time t, Indicated in The covariance matrix of the state estimate at the next moment, express The observation matrix at time, represents the transposed matrix, represents the inverse matrix, represents the covariance matrix of the measurement noise, Indicated in The state update at each moment corrects the prediction error. Indicated in moment to the next moment ( The state prediction at time express The observation matrix at time, Indicated in The covariance matrix after the moment is corrected, indicating the uncertainty after the update process, Represents the identity matrix.

[0079] Among them, the estimated value of the prediction stage is updated by the current measurement value, the predicted state covariance matrix and the measurement covariance matrix are used to calculate the Kalman gain, and the state estimate is updated according to the actual measurement data. With the introduction and update of the optimization signal, the uncertainty of the state estimate will change. The covariance matrix can accurately reflect this change and reduce the impact of measurement errors.

[0080] The Kalman filter algorithm is used to process the optimized signals from different acoustic emission sensors to obtain the key signal, so that the accuracy and reliability of the key signal are improved. State estimation And state transition prediction, predict the state and covariance at the next moment. The update step combines the sensor measurement matrix , correct the predicted state estimate. By iteratively predicting and updating, the state estimate is continuously optimized, and important key information in the time domain features and / or frequency domain features can be retained.

[0081] Step 104: Determine the time difference between the key signals, and use the positioning algorithm to determine the source position of the key signal based on the time difference. In the embodiment of the present application, a phase weighting function (SCOT) is combined with a maximum likelihood weighting function (ML) to form an adaptive weighting function. Specifically,

[0082] ;in,

[0083] ;

[0084] ;

[0085] ;

[0086] In the formula, represents the adaptive weighting function, represents the weight of adaptive weighting, represents the adjustment parameter, represents the signal-to-noise ratio, Indicates the preset signal-to-noise ratio, represents the phase weighting function, which enhances the delay estimation capability by combining amplitude and phase information in the frequency domain. represents the maximum likelihood weighted function, which can adjust the weight value according to the signal-to-noise ratio, so as to better suppress the interference of noise. and Respectively represent the key input signals With key signal The autopower spectrum function of Key signal indicating input With key signal The cross power spectrum function of The weight parameter representing the key signal of the input.

[0087] Among them, when Higher than When , the adaptive coefficient is close to 1, This means that most of the weight is provided by the smooth coherent transformation. When is higher, the noise component in the key signal is less, and the weight of the phase weighting function can more accurately reflect the time-frequency characteristics of the key signal. Therefore, the weight of the phase weighting function can improve the accuracy of time delay estimation. Lower than When , the adaptive coefficient is close to 0, , which means that most of the weights of the adaptive weighting function are provided by the maximum likelihood weighting function. When it is low, there are more noise components in the key signal, and the weight of the phase weighting function may be greatly affected by the noise, resulting in inaccurate time delay estimation. At this time, the weight of the maximum likelihood weighting function can suppress noise more effectively in a noisy environment.

[0088] The traditional maximum likelihood weighting function requires prior knowledge of statistical characteristics when applied, which has certain limitations. The embodiment of the present application combines the characteristics of the phase weighting function with the maximum likelihood weighting function in different noise environments, enhances the robustness of the adaptive weighting function in a noise environment, and improves the accuracy of delay estimation.

[0089] The adaptive weighting function is weighted to obtain the generalized cross-correlation function, and the delay corresponding to the maximum value is taken as the time difference, as follows:

[0090] ;

[0091] ;

[0092] In the formula, represents the generalized cross-correlation function, represents the weight of the generalized cross-correlation function, represents the adjustment parameter, Key signal indicating input With key signal The modulus of the squared coherence function, Key signal indicating input The autopower spectrum function of Key signal indicating input With key signal The cross power spectrum function of Key signal indicating input With key signals The cross power spectrum function, represents the carrier signal, The weight parameter representing the key signal of the input.

[0093] Using Fourier transform, the cross-correlation problem in the time domain can be converted to the frequency domain. Through frequency domain processing, especially in the face of complex environments such as noise and multipath effects, the robustness of cross-correlation can be enhanced, as follows:

[0094] ;

[0095] In the formula, represents the generalized cross-correlation function in the frequency domain, represents the delay variable, represents the inverse Fourier transform, represents the frequency domain weighting function, Key signal indicating input The complex conjugate of the Fourier transform of Key signal indicating input The Fourier transform of The weight parameter representing the key signal of the input.

[0096] ;

[0097] In the formula, represents the delay variable, Indicates the time difference, express The corresponding input when the maximum value is obtained, Indicates the current time. and They represent different key input signals respectively.

[0098] Among them, the key signal through the input Relative to the key signal The time difference between the two is calculated at different time offsets The generalized cross-correlation function The peak position corresponds to , which is the time difference between the two key signals.

[0099] For example, assuming there are two acoustic emission sensors, the signals received by them are Fourier transformed to obtain characteristic signals and , using the generalized cross-correlation formula and combining it with the adaptive weighting function, the generalized cross-correlation function in this application is calculated. Then the cross-correlation function in the frequency domain is converted back to the time domain through the inverse Fourier transform, and the time delay corresponding to the maximum value is found, which is the time difference . Assume that the key signal and The time difference between , the generalized cross-correlation function is The key signal is determined when the maximum value is obtained at any time. Relative to The corresponding time delay is 5 time units.

[0100] A positioning algorithm is used based on the time difference to determine the source position of the key signal; wherein the positioning algorithm includes: determining the signal distance between the key signal and each acoustic emission sensor according to the time difference; determining the signal direction of the key signal according to the signal distance and the geometric layout between each acoustic emission sensor; and determining its source position according to the signal distance and the signal direction using a triangulation algorithm.

[0101] Specifically, different positioning models can be combined according to actual conditions to determine the source position of the signal source of different key signals for precise positioning. For example, when the wellhead device or the detection area is in a two-dimensional plane, a plane model can be selected for positioning, as follows:

[0102] ;

[0103] In the formula, and Respectively represent the signal source of the key signal (i.e. multi-sensor data) to different acoustic emission sensors and The distance Indicates delay, Indicates the speed of sound.

[0104] Among them, it is assumed that the position coordinates of multiple acoustic emission sensors are , by taking the time delay of each pair of acoustic emission sensors Combined with the installation position of the acoustic emission sensor, the signal source of the key signal to the acoustic emission sensor can be calculated. and By combining multiple sets of equations, we can obtain the specific position of the signal source of the key signal in the two-dimensional plane.

[0105] For example, when the wellhead device is in a cylindrical or cylindrical structure, the acoustic emission sensor can be installed on the surface of the wellhead housing or a similar cylinder for detection, as follows:

[0106] ;

[0107] In the formula, and Acoustic Emission Sensor and The location coordinates of and The polar coordinates representing the source location of the critical signal, Acoustic Emission Sensor and The delay, Indicates the speed of sound.

[0108] Among them, it is assumed that the polar coordinate position of each acoustic emission sensor is , , , and combine multiple sets of equations to obtain the specific position of the signal source in the cylinder.

[0109] Those skilled in the art may select different positioning models according to actual conditions, and make full use of the time delay information and the geometric relationship between the source position and the acoustic emission sensor to achieve accurate positioning of the source position of the key signal.

[0110] Step 105: Classify the key signal through the machine training algorithm to obtain the classification category of the key signal. In the embodiment of the present application, the key signal obtained is classified by using a support vector machine (SVM). Whether the support vector machine can efficiently classify the signal often depends on the selection of the kernel function. It is necessary to consider not only the current signal characteristics, but also the subsequent parameter adjustment, overfitting and other practical applications. In the embodiment of the present application, a Gaussian radial basis kernel function is selected to process the key signal and map it to an infinite-dimensional space, which can better capture the complex patterns and nonlinear relationships in the data. Specifically as follows,

[0111] ;

[0112] In the formula, represents the Gaussian radial basis kernel function, represents the natural exponential function, Represents the width parameter, which is used to control the kernel function. and The feature vector representing the key signal input to the support vector machine.

[0113] The kernel function is selected according to the key signal and the classification model is constructed based on it, as follows:

[0114] ;in,

[0115] ,

[0116] .

[0117] In the formula, and represents the Lagrange multiplier, and The feature vector representing the key signal input to the support vector machine, and Indicates The first The feature vector label of the key signal, represents the regularization parameter, which is used to control the tolerance of the classification model to misclassification. Indicates the total number of key signals, represents the Gaussian radial basis kernel function.

[0118] Construct a decision function and evaluate the training results as follows:

[0119] ;in,

[0120] ;

[0121] .

[0122] In the formula, represents the decision function, Represents a custom signal, represents the Lagrange multiplier, and Indicates The first The labels of the feature vectors of the key signals, Indicates The characteristic vector of the key signal, represents the Gaussian radial basis kernel function, represents the bias term, represents the predicted label, Represents a trained classification model.

[0123] The trained classification model is used to classify the key signals to obtain the classification categories of the key signals.

[0124] Step 106: The classification category and source position of each key signal are obtained according to the source position and classification category of the key signal. In the embodiment of the present application, by combining the source position of the key signal with the classification category of the support vector machine, the damage type corresponding to each source position is determined, and the target event is accurately detected by multiple acoustic emission sensors. The classification categories are divided into abnormal signals and normal signals. Abnormal signals are key signals corresponding to the preset boundary value of the delay. The source position where such key signals are located may have groove cracks, seawater erosion corrosion, pressure fluctuations or temperature changes. According to data processing and analysis, the early warning system is triggered based on the detection results of the abnormal signal to provide early warning information and maintenance suggestions. The alarm mechanism can be implemented by communicating with the operator and sending a notification to the operator. The notification method can be a sound alarm, a visual prompt, a text message or an email notification, so that the operator can take corresponding measures in time. Through real-time monitoring and alarming, abnormal conditions can be quickly warned and the operator can be reminded. This helps to ensure the safety and stability of the operation and take appropriate maintenance or repair measures in time to reduce potential risks and losses.

[0125] Although the present application provides method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-creative labor. The order of steps listed in this embodiment is only one way of executing the order of many steps and does not represent the only execution order. When the actual device or client product is executed, it can be executed in sequence or in parallel according to the method shown in this embodiment or the accompanying drawings (for example, in a parallel processor or multi-threaded processing environment).

[0126] like Figure 3 As shown, the embodiment of the present application also provides an offshore wellhead multi-sensor detection device 300. The device includes: a data acquisition module 301, a data processing module 302, a feature extraction module 303, a signal processing module 304, a classification module 305 and an event positioning module 306, as follows.

[0127] The data acquisition module 301 is used to acquire multi-sensor data; wherein the multi-sensor data comes from acoustic emission sensors at multiple different installation positions.

[0128] The data processing module 302 is used to optimize the multi-sensor data to obtain an optimized signal.

[0129] The feature extraction module 303 is used to extract the time domain features and frequency domain features of the optimization signal, and obtain the key signal therein through feature fusion.

[0130] The signal processing module 304 is used to determine the time difference between the key signals and determine the source position of the key signal based on the time difference using a positioning algorithm.

[0131] The classification module 305 is used to classify the key signal through a machine training algorithm to obtain a classification category of the key signal.

[0132] The event location module 306 is used to obtain the classification category and source location of each key signal according to the source location and classification category of the key signal.

[0133] Some modules in the apparatus described in the present application can be described in the general context of computer executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0134] The devices or modules described in the above application embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described in various modules according to their functions. When implementing the embodiments of the present application, the functions of each module can be implemented in the same or multiple software and / or hardware. Of course, the module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0135] The methods, devices or modules described in this application can be implemented in the form of computer-readable program codes. The controller can be implemented in any appropriate manner. For example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program codes (such as software or firmware) that can be executed by the (micro)processor, logic gates, switches, application-specific integrated circuits (English: Application Specific Integrated Circuit; Abbreviation: ASIC), programmable logic controllers and embedded microcontrollers. Examples of controllers include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program codes, the controller can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers and embedded microcontrollers by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included in it for implementing various functions can also be regarded as structures within the hardware component. Or even, the means for realizing various functions may be regarded as both a software module for realizing the method and a structure within a hardware component.

[0136] The embodiment of the present application also provides a device for the offshore wellhead multi-sensor detection method, the device comprising: a processor; a memory for storing processor executable instructions; when the processor executes the executable instructions, the method described in the embodiment of the present application is implemented.

[0137] The embodiments of the present application also provide a non-volatile computer-readable storage medium on which a computer program or instruction is stored. When the computer program or instruction is executed, the method described in the embodiments of the present application is implemented.

[0138] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist independently, or two or more modules may be integrated into one module.

[0139] The above storage media include but are not limited to random access memory (RAM), read-only memory (ROM), cache, hard disk drive (HDD) or memory card. The memory can be used to store computer program instructions.

[0140] It can be seen from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus necessary hardware. Based on this understanding, the technical solution of the present application can be essentially or partly reflected in the prior art in the form of a software product, or it can be reflected in the implementation process of data migration. The computer software product can be stored in a storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a mobile terminal, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0141] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. All or part of this application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, mobile communication terminals, multi-processor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc.

[0142] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit the present application. Although the present application has been described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present application.

Claims

1. An offshore wellhead multi-sensor detection method, characterized in that: include: Acquire multi-sensor data; The multi-sensor data is from acoustic emission sensors at multiple different installation locations; Optimizing the multi-sensor data to obtain an optimized signal; Extracting the time domain features and frequency domain features of the optimized signal, and obtaining the key signal therein by feature fusion; Determining the time difference between the key signals, and determining the source position of the key signal based on the time difference using a positioning algorithm; Classifying the key signal by a machine training algorithm to obtain a classification category of the key signal; The classification category and the source position of each of the key signals are obtained according to the source position and the classification category of the key signal.

2. The method according to claim 1, characterized in that The step of optimizing the multi-sensor data to obtain an optimized signal includes: inputting the multi-sensor data into a gray wolf optimization model; Initialize wolf pack parameters, randomly generate positions of individuals in the wolf pack, and execute decomposition steps until a preset termination condition is reached to obtain the optimal position; wherein the wolf pack parameters include the number of individuals in the wolf pack, convergence parameters, and number of iterations; The decomposition step comprises: A fitness function is defined, and the fitness value of each individual position is calculated using the fitness function; updating the positions of individuals in the wolf pack according to the fitness values; The decomposition step is performed based on the updated positions of the individuals in the wolf pack; The optimal position is used as a decomposition parameter, and the multi-sensor data is decomposed and reconstructed using the decomposition parameter to obtain the optimized signal.

3. The method according to claim 1, characterized in that The extracting of the time domain features and the frequency domain features of the optimized signal comprises: Based on the Kalman filter algorithm, a state equation and an observation equation of the optimization signal are defined, and an updating step is iteratively performed until the time domain features and / or the frequency domain features are traversed; The update steps are as follows: Set the estimated value and uncertainty of the state estimate at the current moment; Using the state equation, the state value and uncertainty of the state estimate at the next moment are predicted based on the state value and uncertainty of the state estimate at the current moment; Update the state value and uncertainty of the state estimation at the next moment by using the observation equation and the time domain characteristics or the frequency domain characteristics; The updating step is performed by taking the state value and uncertainty of the state estimation at the next moment as the state value and uncertainty of the state estimation at the current moment.

4. The method according to claim 1, characterized in that The determining the time difference between the key signals comprises: An adaptive weighting function is defined, and the time difference between the key signals is calculated by the adaptive weighting function; wherein the adaptive weighting function is as follows: ;in, ； In the formula, represents the adaptive weighting function, represents the weight of adaptive weighting, represents the adjustment parameter, represents the signal-to-noise ratio, Indicates the preset signal-to-noise ratio, Key signal indicating input With key signal The coherence function of Key signal indicating input The autopower spectrum function of Key signal indicating input The autopower spectrum function of Key signal indicating input With key signal The cross power spectrum function of The weight parameter representing the key signal of the input.

5. The method according to claim 4, characterized in that The calculating the time difference between the key signals by the adaptive weighting function comprises: The adaptive weighting function is weighted to obtain a generalized cross-correlation function, and the time delay corresponding to the maximum value thereof is used as the time difference, as follows: ； In the formula, represents the delay variable, Indicates the time difference, express The corresponding input when the maximum value is obtained, Indicates the current time. and They represent different key input signals respectively.

6. The method according to claim 1, characterized in that The method of determining the source position of the key signal based on the time difference by using a positioning algorithm includes: Determine the source position of the key signal using a positioning algorithm based on the time difference; wherein the positioning algorithm includes: Determine the signal distance between the key signal and each acoustic emission sensor according to the time difference; Determining the signal direction of the key signal according to the signal distance and the geometric layout between each acoustic emission sensor; The source position is determined according to the signal distance and the signal direction using a triangulation algorithm.

7. The method according to claim 1, characterized in that After obtaining the optimized signal, the method further includes: A dimension reduction process is performed on the optimized signal.

8. Offshore wellhead multi-sensor detection device, characterized in that: include: Data acquisition module, used to obtain multi-sensor data; The multi-sensor data is from acoustic emission sensors at multiple different installation locations; A data processing module, used for optimizing the multi-sensor data to obtain an optimized signal; A feature extraction module is used to extract the time domain features and frequency domain features of the optimized signal, and obtain the key signal therein by feature fusion; A signal processing module, used to determine the time difference between the key signals, and determine the source position of the key signal based on the time difference using a positioning algorithm; A classification module, used for classifying the key signal through a machine training algorithm to obtain a classification category of the key signal; The event location module is used to obtain the classification category and the source position of each key signal according to the source position and the classification category of the key signal.

9. A device for performing an offshore wellhead multi-sensor detection method, characterized in that: include: processor; a memory for storing processor-executable instructions; When the processor executes the executable instructions, the method according to any one of claims 1 to 7 is implemented.

10. A non-volatile computer-readable storage medium, characterized in that: The device comprises a computer program or an instruction for storing the computer program or the instruction, which, when executed, enables the method according to any one of claims 1 to 7 to be implemented.

Images