Submarine pipeline anomaly condition real-time detection method, device, equipment and medium

Abstract

This invention primarily applies to the field of artificial intelligence technology. It discloses a method, apparatus, device, and medium for real-time detection of anomalies in subsea pipelines. The method includes: monitoring the subsea pipeline using a sensor array to obtain a sensor dataset; determining the mean offset of the data during monitoring based on the sensor dataset; updating preset relaxation parameters according to the mean offset; determining the cumulative sum of sensor data based on the updated relaxation parameters and the current offset, and using this as the updated data change trend; and determining an anomaly in the subsea pipeline and issuing an alarm when the updated data change trend falls outside a preset range. This application effectively improves detection sensitivity and accuracy, and reduces the probability of false alarms or missed alarms when detecting anomalies in subsea pipelines.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to a method, apparatus, equipment, and medium for real-time detection of anomalies in submarine pipelines. Background Technology

[0002] Submarine pipelines, as a critical infrastructure for global energy transportation, are widely used for the transoceanic transmission of oil and natural gas. Buried beneath the seabed for extended periods, these pipelines face complex marine environmental challenges, including seawater corrosion, ocean current impacts, and deep-sea pressure. These factors can lead to pipeline fatigue, cracking, corrosion, damage, or rupture. Timely detection of anomalies is crucial for protecting submarine pipelines. Traditional detection techniques compare sensor signals from the submarine pipeline to one or more fixed thresholds to identify anomalies. However, the complex and ever-changing seabed environment causes continuous variations in sensor signals, making false alarms or missed detections common with traditional techniques. Summary of the Invention

[0003] This invention provides a method, apparatus, equipment, and medium for real-time detection of anomalies in subsea pipelines. When detecting anomalies in subsea pipelines, it can effectively improve detection sensitivity and accuracy, and reduce the probability of false alarms or missed alarms.

[0004] This invention provides a method for real-time detection of anomalies in subsea pipelines, the method comprising:

[0005] Sensor datasets were obtained by monitoring subsea pipelines using a cluster of sensors.

[0006] Based on the sensor dataset, determine the mean offset of the data during the monitoring of the subsea pipeline;

[0007] The preset relaxation parameters are updated based on the mean offset, wherein the relaxation parameters are parameters used to adjust the monitoring sensitivity when monitoring the subsea pipeline;

[0008] Based on the updated relaxation parameters and the current offset, the cumulative sum of sensor data is determined and used as the updated data change trend, wherein the current offset is the difference between the current sensor signal and the mean of the sensor data;

[0009] When the updated data change trend is outside the preset value range, it is determined that the subsea pipeline has an abnormal situation and an alarm is issued.

[0010] Furthermore, the step of determining an abnormal situation in the subsea pipeline and issuing an alarm when the updated data change trend is outside the preset value range includes:

[0011] If the cumulative sum is a positive cumulative sum, then the positive cumulative sum is compared with a preset first alarm threshold. When the positive cumulative sum is greater than the first alarm threshold, it is determined that the subsea pipeline has an abnormal situation and an alarm is issued.

[0012] If the cumulative sum is negative, the negative cumulative sum is compared with a preset second alarm threshold. When the negative cumulative sum is less than the second alarm threshold, an abnormal situation is determined to have occurred in the subsea pipeline and an alarm is issued.

[0013] Furthermore, the method also includes:

[0014] When an anomaly is detected in the subsea pipeline, an anomaly sensor dataset is acquired.

[0015] A time-domain feature matrix is ​​generated based on the time-domain features of the aforementioned abnormal sensor dataset;

[0016] After the time-domain feature matrix is ​​input into a preset neural network, the neural network identifies the abnormal situation and outputs the category of the abnormal situation.

[0017] Furthermore, when an anomaly is determined to occur in the subsea pipeline, acquiring the anomaly sensor dataset includes:

[0018] When an abnormality is detected in the subsea pipeline, the current time point is recorded as the first time point and new sensor signals are continuously acquired until no abnormality is detected in the subsea pipeline, at which point the second time point is recorded.

[0019] Each new sensor signal acquired during the time period between the first time point and the second time point is regarded as an abnormal signal;

[0020] The abnormal sensor dataset is generated based on multiple abnormal signals.

[0021] Furthermore, the method also includes:

[0022] Obtain a historical abnormal event dataset, wherein each sample in the historical abnormal event dataset has a category label;

[0023] Based on the temporal features of each sample, a sample feature matrix is ​​generated, where each row of the sample feature matrix represents a sensor and each column of the sample feature matrix represents a temporal feature.

[0024] The sample feature matrix and each category label are input into a preset Bayesian convolutional neural network model to train the Bayesian convolutional neural network model.

[0025] When training the Bayesian convolutional neural network model, the Bayesian convolutional neural network model is updated using a preset loss function until a fully trained Bayesian convolutional neural network model is obtained.

[0026] The trained Bayesian convolutional neural network model is used as the neural network.

[0027] Furthermore, after inputting the time-domain feature matrix into a preset neural network, the neural network identifies the abnormal situation and outputs the category of the abnormal situation, including:

[0028] The temporal feature matrix is ​​input into the neural network, which includes at least a convolutional layer, a fully connected layer, and an output layer.

[0029] After the temporal features of the temporal feature matrix are convolved through the convolutional layer, multiple temporal feature vectors are obtained.

[0030] After the multiple time feature vectors are input to the output layer through the fully connected layer, the output layer maps each time feature vector to a target abnormal event category among multiple abnormal event categories to obtain the classification result of the abnormal situation and the confidence interval of the classification result.

[0031] Furthermore, the sensor dataset obtained by monitoring the subsea pipeline using a sensor array includes:

[0032] Acquire the raw signal detected by each sensor in the sensor group;

[0033] Each of the original signals is arranged to generate an original signal sequence;

[0034] Based on the mean and standard deviation of the signal when no abnormal changes are detected, the original signal sequence is converted into a standardized signal sequence with a standard normal distribution and used as the preprocessed sensor dataset.

[0035] The present invention also provides a real-time detection device for abnormal conditions in subsea pipelines, the device comprising:

[0036] The data acquisition module is used to monitor the subsea pipeline using a group of sensors and obtain sensor datasets.

[0037] The first calculation module is used to determine the mean offset of the data during the monitoring of the subsea pipeline based on the sensor dataset.

[0038] An update module is used to update preset relaxation parameters based on the mean offset, wherein the relaxation parameters are parameters used to adjust the monitoring sensitivity when monitoring the subsea pipeline;

[0039] The second calculation module is used to determine the cumulative sum of sensor data based on the updated relaxation parameters and the current offset, and to use it as the updated data change trend, wherein the current offset is the difference between the current sensor signal and the mean of the sensor data;

[0040] The output module is used to determine that the subsea pipeline has an abnormal situation and issue an alarm when the trend of the updated data changes outside the preset value range.

[0041] The present invention also provides an electronic device, the electronic device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the real-time detection method for abnormal conditions of subsea pipelines as described in any of the preceding claims.

[0042] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the real-time detection method for abnormal conditions of subsea pipelines as described in any of the preceding claims.

[0043] The present invention has at least the following beneficial effects:

[0044] In this technical solution, firstly, the mean offset of the sensor dataset is calculated, which helps determine whether an anomaly has occurred in the subsea pipeline. Next, based on the calculated mean offset, preset relaxation parameters are dynamically updated. The relaxation parameters are key to adjusting monitoring sensitivity; dynamically adjusting them effectively improves detection sensitivity, reduces the probability of false alarms and missed alarms, and thus improves detection accuracy. Then, based on the updated relaxation parameters and the current offset, the cumulative sum of the sensor data is calculated, reflecting the trend of data change. Finally, this updated data change trend is monitored. If this trend exceeds the preset normal range, an anomaly is determined in the subsea pipeline, and an alarm is immediately issued. This intelligent monitoring and early warning mechanism facilitates timely detection and handling of anomalies in the subsea pipeline, ensuring its safe operation. Attached Figure Description

[0045] The accompanying drawings are provided to further understand the technical solutions of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the technical solutions of the present invention, and do not constitute a limitation on the technical solutions of the present invention.

[0046] Figure 1 This is a flowchart of the steps involved in the real-time detection method for abnormal situations in subsea pipelines;

[0047] Figure 2 This is a flowchart of step S101 in the real-time detection method for abnormal situations in submarine pipelines;

[0048] Figure 3 This is another step in the real-time detection method for abnormal conditions in subsea pipelines;

[0049] Figure 4 This is a flowchart of step S301 in the real-time detection method for abnormal situations in submarine pipelines;

[0050] Figure 5 This is a flowchart of step S303 in the real-time detection method for abnormal situations in submarine pipelines;

[0051] Figure 6 This is another step in the real-time detection method for abnormal conditions in subsea pipelines;

[0052] Figure 7 This is a schematic diagram of the structure of a real-time detection device for abnormal conditions in submarine pipelines;

[0053] Figure 8 This is a schematic diagram of the structure of an electronic device. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0055] Researchers have found that current methods for detecting anomalies in subsea pipelines rely on fixed-threshold monitoring. These methods use one or more fixed thresholds to detect abnormal changes. The traditional CUSUM (Cumulative Sum Control Chart) method calculates the cumulative sum of sensor signals and compares it to the set thresholds to detect deviations from the normal range, thus identifying anomalies. However, due to threshold limitations, this method has relatively fixed sensitivity and is only suitable for detecting simple signal changes, not complex and variable signals in the seabed environment. Seabed environmental factors, such as ocean currents, cause continuous changes in sensor signals from subsea pipelines, making false alarms or missed alarms likely when using traditional CUSUM for anomaly detection. Furthermore, after detecting an anomaly in the subsea pipeline, the weights of the traditional neural network used to identify the anomaly category are fixed values ​​obtained during training. The model calculates forward propagation using these fixed weights, outputting a single classification result, which cannot address the uncertainty of the outcome. To address these issues, this application provides the following embodiments to solve the problems.

[0056] Please refer to Figure 1 , Figure 1 This is a flowchart of the steps involved in the real-time detection method for abnormal situations in subsea pipelines.

[0057] This embodiment provides a method for real-time detection of abnormal situations in subsea pipelines, including:

[0058] S101. Sensor datasets are obtained by monitoring the subsea pipeline using a sensor array.

[0059] S102. Based on the sensor dataset, determine the mean offset of the data during the monitoring of the subsea pipeline.

[0060] S103. Update the preset relaxation parameters according to the mean offset, where the relaxation parameters are used to adjust the monitoring sensitivity when monitoring the subsea pipeline.

[0061] S104. Based on the updated relaxation parameters and the current offset, determine the cumulative sum of the sensor data and use it as the updated data change trend, where the current offset is the difference between the current sensor signal and the mean of the sensor data.

[0062] S105. When the updated data trend is outside the preset value range, an abnormal situation is determined to have occurred in the submarine pipeline and an alarm is issued.

[0063] In step S101 of some embodiments, a sensor dataset is obtained by collecting sensor monitoring data on temperature, strain, and vibration in the subsea pipeline. It should be noted that the sensors in the sensor group are distributed at different locations within the subsea pipeline; therefore, sensor signals from different locations within the subsea pipeline can be collected using these sensors.

[0064] In step S102 of some embodiments, mean offset estimation based on EWMA and adaptive signal offset estimation based on EWMA are used. By dynamically smoothing the current signal and historical signals, the threshold, sensitivity and relaxation value in the detection process are adaptively adjusted, which can better adapt to the continuous changes in signals caused by factors such as ocean current oscillation in the marine environment.

[0065] In some embodiments, a method for real-time detection of anomalies in subsea pipelines may involve: preprocessing and analyzing the collected monitoring data using an online monitoring and diagnostic system based on ACUSUM; if abnormal changes are detected, extracting the signal region exhibiting the abnormal change and sending an alarm. It should be noted that ACUSUM uses EWMA-based adaptive signal offset estimation, which dynamically smooths the current signal against historical signals to adaptively adjust the sensitivity and relaxation value during the detection process.

[0066] Understandably, the relaxation parameter determines the sensitivity to changes in sensor data. If the mean shift is large, the relaxation parameter may need to be adjusted to make it more sensitive in order to detect anomalies more quickly.

[0067] Optionally, the relaxation parameter can be dynamically adjusted based on the magnitude of the mean offset. A rule can be set, such as reducing the relaxation parameter by 10% for every 1% deviation from a threshold.

[0068] Understandably, a cumulative sum can be calculated based on the deviation between the collected sensor signal and the average sensor data. For example, if the collected sensor signal value is higher than the average sensor data, this deviation is added to the positive cumulative sum; conversely, it is added to the negative cumulative sum. This allows for the monitoring of rising or falling trends in electricity consumption. By continuously calculating the cumulative sum, abnormal changes in sensor data can be detected. A continuously increasing positive cumulative sum may indicate an abnormal rise in sensor data; a continuously increasing negative cumulative sum may indicate an abnormal fall in sensor data.

[0069] Please refer to Figure 2 , Figure 2 This is a flowchart of step S101 in the real-time detection method for abnormal situations in submarine pipelines.

[0070] In some embodiments, step S101 includes:

[0071] S201. Obtain the raw signal detected by each sensor in the sensor group.

[0072] S202. Arrange each original signal to generate an original signal sequence.

[0073] S203. Based on the mean and standard deviation of the signal when no abnormal changes are detected, the original signal sequence is converted into a standardized signal sequence with a standard normal distribution and used as the preprocessed sensor dataset.

[0074] Optionally, the original signal sequence or the normalized signal sequence includes multiple elements, the number of which is equal to the number of sensors, and each element represents a set of signals acquired by a sensor.

[0075] Understandably, because temperature, strain, and vibration sensors are deployed at different locations on the subsea pipeline, the signal scales of sensors at different depths will differ, necessitating signal preprocessing. Standardization operations are performed on the signal sequences to eliminate these scale differences between the different sensors.

[0076] In some embodiments, step S105 includes:

[0077] If the cumulative sum is positive, it is compared with a preset first alarm threshold. If the positive cumulative sum is greater than the first alarm threshold, an abnormal situation is determined to have occurred in the subsea pipeline and an alarm is issued. If the cumulative sum is negative, it is compared with a preset second alarm threshold. If the negative cumulative sum is less than the second alarm threshold, an abnormal situation is determined to have occurred in the subsea pipeline and an alarm is issued.

[0078] Please refer to Figure 3 , Figure 3 This is another step in the real-time detection method for abnormal conditions in submarine pipelines.

[0079] In some embodiments, the real-time detection method for anomalies in subsea pipelines further includes:

[0080] S301. When an abnormal situation is determined to occur in the subsea pipeline, obtain the abnormal sensor data set.

[0081] S302. Generate a time-domain feature matrix based on the time-domain features of the abnormal sensor dataset.

[0082] S303. After inputting the time-domain feature matrix into the preset neural network, the neural network identifies the abnormal situation and outputs the category of the abnormal situation.

[0083] In some embodiments, the categories of anomalies that can be identified include submarine earthquakes, anchor strikes, anchor pulls, and swaying caused by ocean currents.

[0084] In some embodiments, the time-domain characteristics may be the standard deviation, peak-to-peak value, or root mean square value of the sensor signal.

[0085] Please refer to Figure 4 , Figure 4 This is a flowchart of step S301 in the real-time detection method for abnormal situations in submarine pipelines.

[0086] In some embodiments, step S301 includes:

[0087] S401. When an abnormal situation is detected in the subsea pipeline, the current time point is recorded as the first time point and new sensor signals are continuously acquired until no abnormal situation is detected in the subsea pipeline, at which point the second time point is recorded.

[0088] S402. Each new sensor signal acquired during the time period between the first time point and the second time point is regarded as an abnormal signal.

[0089] S403. Generate an abnormal sensor dataset based on multiple abnormal signals.

[0090] In one specific embodiment, determining an abnormal situation in the subsea pipeline and simultaneously recording the time point before the abnormal change occurs are both part of the process. The moment in time And continue monitoring; if within the set time... No further abnormal changes were detected within the area; this time point was recorded. time interval All sensor signals within the dataset are extracted as an anomaly sensor dataset. During the step of generating a time-domain feature matrix based on the time-domain features of the anomaly sensor dataset, time intervals are extracted. The standard deviation, peak-to-peak value, root mean square value, and other time-domain characteristics of all sensor signals are included in a time-domain feature matrix, which is:

[0091]

[0092] in, The number of features.

[0093] Please refer to Figure 5 , Figure 5 This is a flowchart of step S303 in the real-time detection method for abnormal situations in submarine pipelines.

[0094] In some embodiments, step S303 includes:

[0095] S501. Input the temporal feature matrix into the neural network, which includes at least a convolutional layer, a fully connected layer, and an output layer.

[0096] S502. After convolving the temporal features of the temporal feature matrix through a convolutional layer, multiple temporal feature vectors are obtained.

[0097] S503. After multiple time feature vectors are input to the output layer through a fully connected layer, the output layer maps each time feature vector to the target abnormal event category among multiple abnormal event categories to obtain the classification result of the abnormal situation and the confidence interval of the classification result.

[0098] In one specific embodiment, the trained neural network includes two convolutional layers, two fully connected layers, and one output layer. The first convolutional layer uses a 3x3 convolutional kernel for convolution operations, where the weights of the convolutional kernel are... It incorporates Bayes' theorem and weights Modeling a probability distribution from fixed values, where the weights... The initial values ​​are set to a normal distribution. The second convolutional layer uses a 2x2 convolutional kernel. The flattened temporal feature vector passes through two fully connected layers before entering the output layer. The output layer uses the Softmax activation function to map the temporal feature vector to different abnormal event categories. After two convolutional and pooling layers, the temporal feature vector is combined and used as the input to the fully connected layer.

[0099] Exemplarily, a specific embodiment illustrates how the output layer maps each temporal feature vector to a target anomalous event category among multiple anomalous event categories to obtain the classification result of the anomalous situation and the confidence interval of the classification result. In this embodiment, assume there are 3 sensors, each collecting 100 data points in 1 second, forming a feature vector containing 300 elements. This vector is the temporal feature vector, which captures the dynamic changes of the machine within that second. These temporal feature vectors are input into a fully connected layer of a neural network. The role of the fully connected layer is to transform these high-dimensional data points into a smaller, more easily processed representation. For example, the fully connected layer might transform the 300-element vector into a vector containing 10 elements, which is a compressed representation of the original data, retaining the most important information. Next, this 10-element vector is passed to the output layer. The task of the output layer is to map these feature vectors to different anomalous event categories. Suppose we have 5 possible anomalous events. The output layer will have a softmax layer containing 5 neurons, each neuron corresponding to an anomalous event category. The softmax layer calculates the probability of each category and outputs a probability distribution, indicating which category the model believes the current sensor data most likely belongs to. For example, if the model believes the current abnormal state is most likely overheating, then the neuron corresponding to overheating will output the highest probability. Furthermore, this embodiment also provides a confidence interval for the classification result. This confidence interval is obtained through uncertainty estimation of the model. A narrow confidence interval indicates a highly certain classification result; a wide confidence interval indicates a less certain classification result, which may require more data or human intervention for confirmation.

[0100] Understandably, this embodiment uses a Bayesian convolutional neural network model to transform weight parameter estimation into probability distribution estimation. By sampling the weight distribution, the uncertainty of the classification results is quantified. The model output includes not only the abnormal event category results but also the confidence interval of the results, enabling staff to make more cautious maintenance decisions based on the confidence interval of the results.

[0101] Please refer to Figure 6 , Figure 6 This is another step in the real-time detection method for abnormal conditions in submarine pipelines.

[0102] In some embodiments, the real-time detection method for anomalies in subsea pipelines further includes:

[0103] S601. Obtain the historical abnormal event dataset. Each sample in the historical abnormal event dataset has a category label.

[0104] S602. Based on the temporal features of each sample, generate a sample feature matrix. Each row of the sample feature matrix represents a sensor, and each column of the sample feature matrix represents a temporal feature.

[0105] S603. Input the sample feature matrix and each category label into the preset Bayesian convolutional neural network model to train the Bayesian convolutional neural network model.

[0106] S604. When training a Bayesian convolutional neural network model, the model is updated using a preset loss function until a fully trained Bayesian convolutional neural network model is obtained.

[0107] S605. Use the trained Bayesian convolutional neural network model as a neural network.

[0108] In one specific embodiment, the process of training a Bayesian convolutional neural network model is as follows:

[0109] The Bayesian convolutional neural network (CNN) model was trained using an existing dataset of subsea pipeline anomaly events (including anchor damage and natural events). Each data sample has multiple temporal features (such as peak-to-peak value and standard deviation), and each data sample is assigned a class label. The temporal features of the anomaly event dataset are combined into a feature matrix, which, along with the class labels, serves as the input to the Bayesian CNN model. The feature matrix has a 6x6 dimension, with rows representing one temperature sensor, one vibration sensor, and four strain sensors, and columns representing peak-to-peak value, variance, root mean square value, mean, standard deviation, and skewness. The Bayesian CNN quantifies the uncertainty of the model's classification of noisy data by sampling the weight distribution. The model's output classification result includes variance, which represents the model's confidence interval for that classification result. The loss function is calculated and used for backpropagation of the model, finally yielding the trained Bayesian CNN model. The loss function formula is as follows:

[0110]

[0111] The loss function combines KL divergence and log-likelihood, where... For approximate posterior distribution, For the prior distribution, Let be the likelihood function. The number of samples.

[0112] Understandably, when using a Bayesian convolutional neural network model, the model models the probability distribution of the weights, samples different weights from the current weight distribution, generates a series of prediction results, and estimates the uncertainty of the prediction by the mean and variance of these results.

[0113] The technical solution of this application also provides a specific embodiment of a method for real-time detection of abnormal conditions in subsea pipelines using an ACUSUM-based online monitoring and diagnostic system.

[0114] In this embodiment, before performing anomaly detection on the monitored data signals, the initial parameters of the system are first set:

[0115]

[0116] in The initial values ​​for the Exponentially Weighted Moving Average (EWMA) operator. and These are the initial values ​​for the upward and downward ACUSUM statistics, respectively.

[0117] After initializing the system parameters, extract the monitoring data signal sequence. ,in This represents the number of sensors.

[0118] The signal is converted into a dimensionless standard normal distribution to obtain a standardized sequence. The standardized formula is:

[0119]

[0120] in, and These represent the mean and standard deviation of the signal when there are no abnormal changes.

[0121] EWMA-based mean shift estimation of the standardized signal The EWMA operator formula for estimating the mean shift is as follows:

[0122]

[0123] in It is a smoothing constant used to control the weight of the latest data.

[0124] based on The signal mean offset is used to update the relaxation parameters of ACUSUM. Relaxation parameters Calculated using the following formula:

[0125]

[0126] in This is the preset minimum offset. This is an estimate of the mean offset of the signal at the current moment.

[0127] Based on the signal offset, the cumulative sums for the positive and negative directions are calculated separately to detect abnormal upward or downward trends in the signal.

[0128] The positive ACUSUM statistic is as follows:

[0129]

[0130] The negative ACUSUM statistic is as follows:

[0131]

[0132] Positive or negative accumulation exceeding the set alarm threshold When this happens, the system will issue an abnormal alarm, where the threshold is... The value is set based on historical data. Specific triggering conditions are as follows:

[0133]

[0134] Continue monitoring the signal during the period in which the abnormal signal was detected.

[0135] The system analyzes the data of regions exhibiting abnormal changes, extracts matrices, and then imports the extracted feature matrices into an anomaly event recognition system based on a Bayesian convolutional neural network to identify the anomaly event, determine its category, and send the corresponding diagnostic results.

[0136] It is understood that in this embodiment, the ACUSUM-based online monitoring and diagnostic system and the Bayesian convolutional neural network-based subsea pipeline anomaly detection and classification system utilize the online monitoring and diagnostic system to monitor changes in subsea pipeline sensor signals in real time, promptly detect abnormal signal changes and issue anomaly alarms, and simultaneously extract features from the abnormal change regions as input to the subsea pipeline anomaly detection and classification system. Finally, the system obtains the anomaly event category result and its confidence interval as the anomaly event alarm result. This embodiment overcomes the shortcomings of fixed-threshold-based monitoring techniques in detecting abnormal changes in complex and ever-changing marine environmental signals, and the uncertainty of results that traditional neural networks cannot provide.

[0137] In any of the above embodiments, firstly, by analyzing the sensor dataset, the mean offset of the data is calculated, which helps determine whether an anomaly has occurred in the subsea pipeline. Next, based on the calculated mean offset, the system dynamically updates preset relaxation parameters. Relaxation parameters are key to adjusting monitoring sensitivity; by dynamically adjusting the relaxation parameters, detection sensitivity can be effectively improved, and the probability of false alarms and missed alarms can be reduced, thereby improving detection accuracy. Then, based on the updated relaxation parameters and the current offset, the cumulative sum of sensor data is calculated, reflecting the trend of data change. Finally, this updated data change trend is monitored. If this trend exceeds the preset normal value range, the system will determine that an anomaly has occurred in the subsea pipeline and immediately issue an alarm. This intelligent monitoring and early warning mechanism helps to promptly detect and handle anomalies in the subsea pipeline, ensuring its safe operation.

[0138] Please refer to Figure 7 , Figure 7 This is a schematic diagram of the structure of a real-time detection device for abnormal conditions in submarine pipelines.

[0139] This embodiment also provides a real-time detection device for abnormal conditions in subsea pipelines, including:

[0140] The data acquisition module 701 is used to monitor the subsea pipeline through a sensor array to obtain a sensor dataset;

[0141] The first calculation module 702 is used to determine the mean offset of data during the monitoring of the subsea pipeline based on the sensor dataset;

[0142] The update module 703 is used to update the preset relaxation parameters according to the mean offset. The relaxation parameters are used to adjust the monitoring sensitivity when monitoring the subsea pipeline.

[0143] The second calculation module 704 is used to determine the cumulative sum of sensor data based on the updated relaxation parameters and the current offset, and to use it as the updated data change trend, wherein the current offset is the difference between the current sensor signal and the mean of the sensor data.

[0144] The output module 705 is used to determine that there is an abnormality in the submarine pipeline and issue an alarm when the trend of the updated data changes outside the preset value range.

[0145] It will be understood by those skilled in the art that all or some of the steps and apparatuses in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. As is known to those skilled in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0146] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0147] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement any of the above-mentioned methods for real-time detection of abnormal situations in submarine pipelines.

[0148] refer to Figure 8 , Figure 8 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0149] The processor 801 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0150] The memory 802 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 802 can store operating devices and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 802 and called and executed by the processor 801 using the real-time detection method for abnormal conditions of subsea pipelines according to the embodiments of this application.

[0151] The 803 input / output interface is used to implement information input and output.

[0152] The communication interface 804 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0153] Bus 805 transmits information between various components of the device (e.g., processor 801, memory 802, input / output interface 803, and communication interface 804);

[0154] The processor 801, memory 802, input / output interface 803, and communication interface 804 are connected to each other within the device via bus 805.

[0155] It is understood that the content of the above method embodiments is applicable to the embodiments of this electronic device. The specific functions implemented by the embodiments of this electronic device are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0156] This application also provides a computer-readable storage medium storing a processor-executable program, which, when executed by a processor, is used to implement the real-time detection method for subsea pipeline anomalies as described in any of the above specific embodiments.

[0157] This application also discloses a computer program product, including a computer program or computer instructions, which are stored in a computer-readable storage medium. The processor of the computer device reads the computer program or computer instructions from the computer-readable storage medium and executes the computer program or computer instructions, causing the computer device to perform the real-time detection method for subsea pipeline anomalies as described in any of the preceding embodiments.

[0158] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0159] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices. It should be understood that in this application, “at least one” means one or more, and “more than one” means two or more.

[0160] In the several embodiments provided in this application, it should be understood that the disclosed apparatus, devices, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0161] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0162] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0163] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0164] Although the description of this application has been quite detailed and particularly focused on several of the described embodiments, it is not intended to limit itself to any of these details or embodiments or any particular embodiment. Rather, it should be considered as effectively covering the intended scope of this application by referring to the appended claims and taking into account the prior art, which provides for a broad possible interpretation of these claims. Furthermore, the foregoing description of this application with respect to embodiments foreseeable by the inventors is intended to provide a useful description, and non-substantial modifications to this application that have not yet been foreseen may still represent equivalent modifications.

Claims

1. A method for real-time detection of anomalies in subsea pipelines, characterized in that, The method includes: Sensor datasets were obtained by monitoring subsea pipelines using a cluster of sensors. Acquire the raw signal detected by each sensor in the sensor group; Each of the original signals is arranged to generate an original signal sequence; Based on the mean and standard deviation of the signal when no abnormal changes are detected, the original signal sequence is converted into a standardized signal sequence with a standard normal distribution and used as the preprocessed sensor dataset. Based on the sensor dataset, the mean offset of the data during the monitoring of the subsea pipeline is determined, wherein the mean offset is obtained by performing mean offset estimation based on exponential weighted moving average on the standardized signal of the standardized signal sequence. Based on the magnitude of the mean offset, a preset relaxation parameter is dynamically adjusted, wherein the relaxation parameter is a parameter used to adjust the monitoring sensitivity when monitoring the subsea pipeline; Based on the updated relaxation parameters and the current offset, the cumulative sum of sensor data is determined and used as the updated data change trend, wherein the current offset is the difference between the current sensor signal and the mean of the sensor data; If the cumulative sum is a positive cumulative sum, then the positive cumulative sum is compared with a preset first alarm threshold. When the positive cumulative sum is greater than the first alarm threshold, it is determined that the subsea pipeline has an abnormal situation and an alarm is issued. If the cumulative sum is negative, the negative cumulative sum is compared with a preset second alarm threshold. When the negative cumulative sum is less than the second alarm threshold, an abnormal situation is determined to have occurred in the subsea pipeline and an alarm is issued.

2. The method for real-time detection of abnormal conditions in subsea pipelines according to claim 1, characterized in that, The method further includes: When an anomaly is detected in the subsea pipeline, an anomaly sensor dataset is acquired. A time-domain feature matrix is ​​generated based on the time-domain features of the aforementioned abnormal sensor dataset; After the time-domain feature matrix is ​​input into a preset neural network, the neural network identifies the abnormal situation and outputs the category of the abnormal situation.

3. The method for real-time detection of abnormal conditions in subsea pipelines according to claim 2, characterized in that, When an anomaly is detected in the subsea pipeline, the abnormal sensor dataset is obtained, including: When an abnormality is detected in the subsea pipeline, the current time point is recorded as the first time point and new sensor signals are continuously acquired until no abnormality is detected in the subsea pipeline, at which point the second time point is recorded. Each new sensor signal acquired during the time period between the first time point and the second time point is regarded as an abnormal signal; The abnormal sensor dataset is generated based on multiple abnormal signals.

4. The method for real-time detection of abnormal conditions in subsea pipelines according to claim 2, characterized in that, The method further includes: Obtain a historical abnormal event dataset, wherein each sample in the historical abnormal event dataset has a category label; Based on the temporal features of each sample, a sample feature matrix is ​​generated, where each row of the sample feature matrix represents a sensor and each column of the sample feature matrix represents a temporal feature. The sample feature matrix and each category label are input into a preset Bayesian convolutional neural network model to train the Bayesian convolutional neural network model. When training the Bayesian convolutional neural network model, the Bayesian convolutional neural network model is updated using a preset loss function until a fully trained Bayesian convolutional neural network model is obtained. The trained Bayesian convolutional neural network model is used as the neural network.

5. The method for real-time detection of abnormal conditions in subsea pipelines according to claim 2, characterized in that, After inputting the time-domain feature matrix into a preset neural network, the neural network identifies the anomaly and outputs the category of the anomaly, including: The temporal feature matrix is ​​input into the neural network, which includes at least a convolutional layer, a fully connected layer, and an output layer. After the temporal features of the temporal feature matrix are convolved through the convolutional layer, multiple temporal feature vectors are obtained. After the multiple time feature vectors are input to the output layer through the fully connected layer, the output layer maps each time feature vector to a target abnormal event category among multiple abnormal event categories to obtain the classification result of the abnormal situation and the confidence interval of the classification result.

6. A real-time detection device for abnormal conditions in subsea pipelines, characterized in that, The device includes: The data acquisition module is used to monitor the subsea pipeline using a sensor array to obtain sensor datasets. Acquire the raw signal detected by each sensor in the sensor group; Each of the original signals is arranged to generate an original signal sequence; Based on the mean and standard deviation of the signal when no abnormal changes are detected, the original signal sequence is converted into a standardized signal sequence with a standard normal distribution and used as the preprocessed sensor dataset. The first calculation module is used to determine the mean offset of data during the monitoring of the subsea pipeline based on the sensor dataset, wherein the mean offset is obtained by performing mean offset estimation based on exponential weighted moving average on the standardized signal of the standardized signal sequence. The update module is used to dynamically adjust the preset relaxation parameters according to the magnitude of the mean offset, wherein the relaxation parameters are parameters used to adjust the monitoring sensitivity when monitoring the subsea pipeline; The second calculation module is used to determine the cumulative sum of sensor data based on the updated relaxation parameters and the current offset, and to use it as the updated data change trend, wherein the current offset is the difference between the current sensor signal and the mean of the sensor data; The output module is used to compare the positive cumulative sum with a preset first alarm threshold if the cumulative sum is a positive cumulative sum, and to determine that the subsea pipeline has an abnormal situation and issue an alarm when the positive cumulative sum is greater than the first alarm threshold. If the cumulative sum is negative, the negative cumulative sum is compared with a preset second alarm threshold. When the negative cumulative sum is less than the second alarm threshold, an abnormal situation is determined to have occurred in the subsea pipeline and an alarm is issued.

7. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the real-time detection method for abnormal conditions of submarine pipelines as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the real-time detection method for abnormal conditions of submarine pipelines as described in any one of claims 1 to 5.

Images