Buoy anomaly detection method based on buoy model

Abstract

The application relates to the technical field of underwater equipment, in particular to a buoy anomaly detection method based on a buoy model, which comprises the following steps: constructing a buoy model of a profile drift buoy in a hovering stage, performing parameter configuration on the buoy model by using multi-source data through a least square method; obtaining current sea trial data returned by the profile drift buoy; predicting a first outer oil bag volume under a preset drift depth based on the buoy model, analyzing a second outer oil bag volume in the current sea trial data, determining whether an absolute difference value between the first outer oil bag volume and the second outer oil bag volume is smaller than or equal to a preset detection threshold value, if yes, parameter updating is not needed; otherwise, it is judged whether the buoy model has been updated, if the model has been updated, it is judged that the buoy is in an abnormal state; otherwise, a parameter updating step is executed according to a statistical distribution condition of the current sea trial data. Through the application, accurate identification of an operation state and dynamic updating of a parameter model are realized based on limited returned data.

Description

Technical Field

[0001] This application relates to the field of underwater equipment technology, and in particular to a buoy anomaly detection method based on a buoy model. Background Technology

[0002] In modern marine environmental monitoring, Argo buoys, as deep-sea profiling devices, have been widely used for long-term automated measurement of key ocean parameters such as temperature, salinity, and depth due to their low noise, low cost, long endurance, light weight, and ease of deployment. These buoys, through their diving and surfacing processes, enable vertical profiling observations from the surface to the deep sea, and are an important component of the Global Ocean Observing System (GOOS).

[0003] However, despite continuous advancements in structural design and measurement accuracy, Argo buoys, as single-use devices, are inherently limited by their inability to be recovered and maintained, making condition monitoring during operation particularly critical. Due to the complexity and variability of the marine environment, buoys can be affected by issues such as grounding, seal failure, or sensor malfunction, leading to decreased data acquisition accuracy or loss of observational functionality. Furthermore, Argo buoys typically rely on satellite communication for small-scale data transmission; under conditions of limited bandwidth and real-time performance, anomaly detection and condition diagnosis are challenging.

[0004] Current research has employed some methods to identify anomalies and update models using buoy operational data, but these methods still face significant challenges due to the dual challenges of drastic changes in the deep-sea environment and limited data availability. In particular, there is a lack of effective judgment criteria and quantitative analysis mechanisms for distinguishing between parameter drift and state anomalies.

[0005] Therefore, how to accurately identify the operational status of deep-sea profiling buoys and dynamically update their parameter models based on limited transmitted data, given the inability to physically recover and maintain them on-site, has become a key technical challenge that urgently needs to be overcome in the application of deep-sea profiling buoys. The existence of these problems not only restricts the accuracy of anomaly detection but also affects the reliability of deep-sea observation data and the depth of scientific research. Summary of the Invention

[0006] This application provides a buoy anomaly detection method based on a buoy model, including:

[0007] A buoy model of a profile drifting buoy in the hovering phase is constructed, and the parameters of the buoy model are configured using the least squares method with multi-source data.

[0008] The current sea trial data is periodically acquired via satellite communication from the profile drifting buoy.

[0009] predicting an outer oil bag volume at a preset drift depth based on the buoy model as a first outer oil bag volume and analyzing an outer oil bag volume in current sea trial data as a second outer oil bag volume determining an absolute difference between the first outer oil bag volume and the second outer oil bag volume whether the absolute difference is less than or equal to a preset detection threshold,

[0010] if yes, no parameter updating is needed;

[0011] otherwise, determining whether the buoy model has been updated,

[0012] if the model has been updated, determining that the buoy is in an abnormal state;

[0013] otherwise, performing a parameter updating step according to a statistical distribution of the current sea trial data.

[0014] Based on the above steps, the present application establishes a buoy model based on the stable behavior of a profile drift buoy in a hovering phase, and uses multi-source data for fitting and parameter setting; current sea trial data is periodically received through satellite communication, and a first outer oil bag volume is calculated by combining the buoy model; by comparing the difference between the second outer oil bag volume actually measured at present and the first outer oil bag volume, it is determined whether the detection threshold is exceeded to identify potential abnormalities; if the above absolute difference exceeds the detection threshold, it is determined whether to judge as abnormal according to a model update flag, or the parameter updating step is continued to be executed to update the model parameters based on statistical consistency.

[0015] In the above embodiment, the profile drift buoy determines the drift depth based on a preset control instruction, and when the depth value of the buoy within a certain period of time belongs to the drift depth range, it is determined that the buoy enters the hovering phase.

[0016] In some embodiments, the buoy model is represented as the following calculation model:

[0017] ,

[0018] wherein, is the mass of the profile drift buoy, is the acceleration of gravity, is the density of water at the operating depth is the volume of the glass sphere, is the volume of the outer oil bag, is the volume of other components of the buoy, is the compression coefficient of the glass sphere, is the compression coefficient of the outer oil bag, is the compression coefficient of other components of the buoy, is the compression coefficient of other components of the buoy, is a parameter error term, used to represent the error related to the aforementioned parameters such as density, glass sphere volume, outer oil bladder volume, other component volume, and corresponding compression coefficient.

[0019] In some embodiments, as the depth increases, the temperature decreases, resulting in an increase in density, and the same salinity affects the density, which also increases linearly with depth, causing water compression, resulting in an increase in density. The density can be simplified as an empirical function or fitted as a lookup table based on historical sea trial data.

[0020] In another embodiment, in the buoy model, the density of water as a function of depth , taking into account the coupling effects of temperature gradient, salinity distribution and hydrostatic pressure increase caused by depth change. By establishing a functional relationship between and , the buoyancy change can be accurately modeled, and the accuracy of buoyancy prediction of the buoy at different depths is enhanced, thereby improving the overall response characteristics of the model.

[0021] Based on the above buoy model, the embodiment of the present application is constructed based on the buoyancy balance principle, and the left side of the equation represents the gravity acting on the buoy, and the right side represents the buoyancy acting on the buoy at depth . The buoyancy part corresponds to the buoyancy acting on the glass sphere, the outer oil bladder and other structural components in the buoy in water. Since the material is compressible under water, its effective volume shrinks with increasing depth, so the volume compression coefficient is introduced in the buoyancy part.

[0022] By establishing a buoyancy calculation model that includes the compression characteristics of multiple components and the influence of water density change, the mechanical state of the profile drift buoy at different depths can be accurately modeled. The model also considers the non-rigid response characteristics of the main components of the buoy, making the calculation results closer to the actual operating state. The parameter error term provides flexible model adjustment capability, which is corrected by fitting methods such as least squares method, effectively reducing the model prediction error.

[0023] In some embodiments, the parameter error term is expressed as a calculation model determined as follows:

[0024] ,

[0025] wherein, , , is a first coefficient, a second coefficient, and a third coefficient.

[0026] Based on the above calculation model, the parameter error term of the buoy model in the hovering phase is expressed as a composite exponential function structure based on depth in the embodiment of the present application, and the depth As independent variables, the error trend with depth is described by three coefficients 、 、 The coefficients 、 、 are determined based on historical sea trial data fitting regression, and can be dynamically updated according to current sea trial data through a parameter updating step during operation to improve prediction accuracy.

[0027] In some embodiments, the multi-source data includes simulation results, laboratory data, and historical sea trial data, and the parameter configuration of the buoy model using the multi-source data specifically includes:

[0028] According to the simulation results, the laboratory data is used as the initial value (which can include the mass of the profile drift buoy, the volume of the buoy glass sphere, the volume of the buoy outer oil bag, and the volume of other components, the running speed, etc.) to determine the parameters of the buoy model;

[0029] The parameters of the buoy model are updated based on the historical sea trial data using the least square method.

[0030] In some embodiments, the parameter updating step further includes:

[0031] The statistical distribution of the current sea trial data is calculated, and it is determined whether the distribution follows the statistical distribution of the historical sea trial data;

[0032] If yes, the parameter error term is updated using the current sea trial data by the least square method, and the determination coefficient ,

[0033] It is further determined whether the determination coefficient is greater than 0.9, if yes, the parameter error term is updated again using the current sea trial data by the least square method, otherwise, the parameter error term is not updated;

[0034] Otherwise, the parameter error term is not updated using the current sea trial data.

[0035] Based on the above steps, the application dynamically updates the parameter error term in the buoy model based on the dual constraints of sea trial data distribution consistency determination and model fitting quality evaluation, to improve the fitting accuracy and environmental adaptability of the buoy model to the actual sea measurement data.

[0036] The least square method adopted by the embodiments of the present application can update the parameter error term under the condition of meeting the distribution consistency, and enhance the model's recognition ability to data quality according to the determination coefficient as a secondary update condition, effectively inhibit the influence of instantaneous abnormality or local noise on the parameter estimation of the parameter error term; setting the threshold of the determination coefficient helps to control the risk of overfitting while maintaining the accuracy of the model, making the model more robust in long-term operation, and also providing a quantitative standard for multi-round iteration optimization.

[0037] The details of one or more embodiments of the present application are presented in the following drawings and description to make other features, objects and advantages of the present application more clear and easy to understand. BRIEF DESCRIPTION OF DRAWINGS

[0038] The drawings described herein are used to provide further understanding of the present application, and form a part of the present application. The illustrative embodiments of the present application and their description serve to explain the present application, and do not constitute an improper limitation on the present application. In the drawings:

[0039] Figure 1 The flowchart of the buoy anomaly detection method of the embodiments of the present application;

[0040] Figure 2 The logic flowchart of the buoy anomaly detection of the embodiments of the present application;

[0041] Figure 3 The mapping relationship between the absolute difference and the pressure of the embodiments of the present application;

[0042] Figure 4 The variation of the absolute difference with the buoy running time line of the embodiments of the present application;

[0043] Figure 5 Another mapping relationship between the absolute difference and the pressure of the embodiments of the present application;

[0044] Figure 6 The evolution process of the prediction error of the embodiments of the present application. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical scheme and advantages of the present application more clear and understandable, the present application is described and explained below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and do not limit the present application. Based on the embodiments provided by the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0046] It is apparent that the drawings in the following description merely show some examples or embodiments of the present application, and for those skilled in the art, the present application can be applied to other similar situations without creative labor based on these drawings. In addition, it can be understood that although the efforts made in this development process can be complex and lengthy, some design, manufacture or production changes based on the technology disclosed in the present application are only routine technical means for those skilled in the art related to the disclosure of the present application, and should not be understood as insufficient disclosure of the present application.

[0047] In the present application, the phrase "embodiment" means that the specific features, structures or characteristics described in conjunction with the embodiment can be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment to other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described in the present application can be combined with other embodiments without conflict.

[0048] Unless otherwise defined, the technical terms or scientific terms involved in the present application should be understood as the usual meaning understood by those skilled in the art in the technical field to which the present application belongs. The terms "one", "a", "an", "the", and similar words involved in the present application do not represent quantity limitation, but can represent singular or plural. The terms "include", "contain", "have", and any variations thereof involved in the present application are intended to cover non-exclusive inclusion; for example, a process, method, system, product or device including a series of steps or modules (units) is not limited to the listed steps or units, but can also include steps or units not listed, or can also include other steps or units inherent to the process, method, product or device. The terms "connected", "connected", "coupled" and similar words involved in the present application are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The term "multiple" refers to two or more. The association between the associated objects is described by the term "and / or", which means that there can be three relationships, for example, "A and / or B" can mean that A exists alone, A and B exist together, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects. The terms "first", "second", "third", and the like involved in the present application are only to distinguish similar objects, and do not represent a specific order for the objects.

[0049] Figure 1 For the flowchart of the buoy anomaly detection method of the embodiments of the present application, referring to Figure 1 The buoy anomaly detection method based on the buoy model includes:

[0050] Step S1: constructing a model of the profile drift buoy in the hovering stage of the profile drift buoy, and configuring parameters of the model by least squares method using multi-source data.

[0051] Step S2: periodically obtaining current sea trial data returned by the profile drift buoy through satellite communication, the current sea trial data including real-time temperature, real-time salinity, real-time depth, historical oil discharge, etc.

[0052] Step S3: performing anomaly detection based on the model of the buoy and the current sea trial data.

[0053] Firstly, the buoy model as shown in the following formula (1) is constructed to simulate the motion state of the buoy,

[0054] (1)

[0055] wherein, m is the mass of the profile drift buoy, a is the acceleration of the profile drift buoy, g is the acceleration of gravity, p is the density of water at the running depth of the buoy, V is the volume of the glass sphere of the buoy, V is the volume of the outer oil bag of the buoy, V is the volume of other components of the buoy, the other components of the buoy being components of the buoy other than the glass sphere and the outer oil bag, for example but not limited to, the other components at least including: a protective shell, a sensor, an antenna, a counterweight, etc., β is the compression coefficient of the glass sphere, β is the compression coefficient of the outer oil bag, β is the compression coefficient of other components of the buoy, the other components of the buoy being components of the buoy other than the glass sphere and the outer oil bag, for example but not limited to, the other components at least including: a protective shell, a sensor, an antenna, a counterweight, etc., C is the drag coefficient of the profile drift buoy, v is the running speed of the profile drift buoy.

[0056] In actual application, an effective method for detecting the abnormal state of the buoy is to monitor the relationship between the buoyancy and the gravity. Therefore, the present application mainly focuses on the change of the motion state of the buoy during the hovering process of the buoy under water, at this time the speed and acceleration of the buoy are usually close to zero. In addition, the estimated volume and coefficient usually deviate from the true value. Considering these differences in the hovering stage, the buoy model with volume and parameter uncertainty can be represented as the following formula (2):

[0057]

[0058] wherein, , , , , ,​ , The relevant errors are respectively for density, glass sphere volume, outer oil bladder volume, other buoy component volume, glass sphere compressibility coefficient, outer oil bladder compressibility coefficient, and other buoy component compressibility coefficient.

[0059] However, in practical applications, accurately measuring these individual errors is often impractical. Therefore, in this embodiment, the aforementioned related errors are expressed as parameter error terms, resulting in the buoy model shown in the following expression:

[0060] ,

[0061] in, The mass of the profile drifting buoy, It is the acceleration due to gravity. For running depth The density of the water below, The volume of the glass sphere is... The volume of the outer oil sac. The volume refers to the volume of other components of the buoy. These other components are those excluding the glass bulb and external oil bladder. Examples, but not limited to, include at least: protective shell, sensors, antenna, and counterweights. The coefficient of compressibility of the glass sphere. The compression coefficient of the external oil bladder. The compression coefficient of other components of the buoy. This is the parameter error term, used to represent the errors related to the aforementioned parameters such as density, glass sphere volume, outer oil bladder volume, other component volumes, and corresponding compressibility coefficients.

[0062] In some embodiments, as depth increases, temperature decreases, leading to an increase in density. Similarly, salinity affects density, and as depth increases linearly, water compression also occurs, leading to an increase in density. Density can be simplified to an empirical function or fitted to a reference table based on historical sea trial data.

[0063] In another embodiment, in the buoy model, the density of water... As depth The function comprehensively considers the coupling effects of factors such as temperature gradient, salinity distribution, and hydrostatic pressure increase caused by depth changes. This is achieved by establishing... and The functional relationship between them can enable accurate modeling of buoyancy changes, enhance the accuracy of buoyancy prediction at different depths, and thus improve the overall response characteristics of the model.

[0064] Based on the above buoy model, the embodiments of this application are constructed based on the principle of buoyancy balance. The left side of the equation represents the gravity acting on the buoy, and the right side represents the buoy's position at depth. The buoyancy and error term received by the place. The buoyancy part corresponds to the buoyancy received by the volume of the glass ball, the outer oil bag and other structural parts in the buoy in the water. Due to the compressibility of the material under water, the effective volume shrinks with the increase of depth, so the volume compression coefficient is introduced in the buoyancy part.

[0065] By establishing a buoyancy calculation model containing multi-component compression characteristics and the influence of water density change, the mechanical state of the profile drift buoy at different depths can be accurately modeled. The model takes into account the non-rigid response characteristics of the main components of the buoy, making the calculation results closer to the actual running state. The parameter error term Flexible model adjustment capability is provided, which is fitted and corrected by least square method and the like, effectively reducing the model prediction error.

[0066] Specifically, the parameter error term is expressed as the following calculation model:

[0067]

[0068] Among them, , , The first coefficient, the second coefficient, and the third coefficient.

[0069] Based on the above calculation model, the parameter error term of the buoy model in the hovering stage is expressed as a composite exponential function structure based on depth, using depth As the independent variable, the change trend of the error with depth is described by three coefficients , , The coefficients , , Based on historical sea trial data fitting regression determination, and can be dynamically updated according to the current sea trial data through the parameter updating step during operation, to improve the prediction accuracy.

[0070] In some embodiments, the multi-source data includes simulation results, laboratory data, and historical sea trial data, and the parameter configuration of the buoy model using multi-source data specifically includes:

[0071] According to the simulation results, the laboratory data is used as the initial value (which can include the mass of the profile drift buoy, the volume of the buoy glass ball, the volume of the outer oil bag and other components, the running speed, etc.) to determine the parameters of the buoy model;

[0072] Based on the historical sea trial data, the least square method is used to update the parameters of the buoy model.

[0073] Figure 2 The floating buoy anomaly detection logic flowchart, please refer to​Figure 2 In the above embodiments, step S3 specifically comprises:

[0074] S301: predicting the outer oil bag volume at the preset drift depth based on the buoy model as a first outer oil bag volume , and analyzing the outer oil bag volume in the current sea trial data as a second outer oil bag volume ;

[0075] S302: determining whether the absolute difference between the first outer oil bag volume and the second outer oil bag volume is less than or equal to a preset detection threshold value,

[0076] If yes, no parameter updating is needed, and the process ends;

[0077] Otherwise, if the absolute difference between the first outer oil bag volume and the second outer oil bag volume is greater than the detection threshold value, step S303 is entered to determine whether the buoy model has been updated based on an update flag,

[0078] If the model has been updated, it is determined that the buoy is in an abnormal state;

[0079] Otherwise, parameter updating step S4 is performed according to the statistical distribution of the current sea trial data.

[0080] In the above embodiments, the cross-section drift buoy determines the drift depth based on a preset control instruction, and when the depth value of the buoy within a certain period of time belongs to the drift depth range, it is determined that the buoy enters the hovering stage.

[0081] Based on the above steps, the present application establishes a buoy model based on the stable behavior of the cross-section drift buoy in the hovering stage, and uses multi-source data for fitting and parameter setting; the current sea trial data is periodically received through satellite communication, the first outer oil bag volume is calculated combined with the buoy model, and the difference between the second outer oil bag volume actually measured at present and the first outer oil bag volume is compared to determine whether it exceeds the detection threshold value to identify potential abnormalities. If the above absolute difference exceeds the detection threshold value, it is determined whether to judge as abnormal according to the model update flag, or the parameter updating step is continued to be executed to update the model parameters based on statistical consistency.

[0082] In some embodiments, the update flag can be set as a Boolean variable or a state enumeration value. Taking the Boolean variable as an example, its typical definition is as follows:

[0083] flag_model_updated = True indicates that the model has been updated;

[0084] `flag_model_updated = False` indicates that the model has not been updated.

[0085] After the buoy model is built and initialized, the update flag is set to False by default. When the current sea trial data returned meets the statistical consistency distribution and the determination coefficient is greater than 0.9, the update flag is configured to True. If the buoy is judged to be abnormal, the flag can be reset to False to wait for the next valid update. When the buoy is in a stable state for more than a set time period or data batch number, the flag is reset to improve long-term adaptability.

[0086] The updated flag can be stored in the local cache of the profile drifting buoy and uploaded to the shore-based control center as status data along with other parameters via satellite communication for remote judgment of the model status; or the shore-based system can dynamically generate and maintain the flag through algorithm logic during the model fitting and data processing stages without the buoy's involvement.

[0087] In another embodiment, the historical oil discharge volume of this application embodiment can be calculated based on the volume of the outer oil bladder and a fixed oil discharge rate and time, where the oil discharge time is the time it takes for hydraulic oil to be discharged from the inner oil tank to the outer oil bladder. The calculation method for the historical oil discharge volume can be adjusted according to the characteristics of the specific hydraulic system by adjusting the parameter model, or can be supplemented by direct sensing measurement.

[0088] In some embodiments, parameter update step S4 further includes:

[0089] Calculate the statistical distribution of the current sea trial data and determine whether its distribution follows the statistical distribution of historical sea trial data;

[0090] If so, the parameter error term is updated using the current sea trial data and the least squares method, and the determination coefficient corresponding to the absolute difference is calculated. ,

[0091] Further determine the determination coefficient If the value is greater than 0.9, it means that the volume of the first outer oil bladder predicted by the buoy model is very close to the volume of the second outer oil bladder, indicating that the data is reliable. Then, the parameter error term is updated again using the least squares method with the current sea trial data, and the update flag is configured. Otherwise, it means that the data residual is dominant and the data is abnormal. The parameter error term is not updated.

[0092] Otherwise, do not update the parameter error term.

[0093] Based on the above steps, the embodiments of the present application dynamically update the parameter error term in the buoy model based on the dual constraints of sea trial data distribution consistency judgment and model fitting quality evaluation, to improve the fitting precision and environmental adaptability of the buoy model to actual sea measurement data.

[0094] The least squares method used in the embodiments of the present application can update the parameter error term under the condition of meeting the distribution consistency, and according to the determination coefficient as a secondary update condition, the model's ability to identify data quality is enhanced, and the influence of instantaneous abnormality or local noise on parameter estimation of the parameter error term is effectively suppressed; setting the threshold of the determination coefficient helps to control the risk of overfitting while maintaining the accuracy of the model, making the model more robust in long-term operation, and also providing a quantitative standard for multi-round iteration optimization.

[0095] In another embodiment, the threshold 0.9 of the determination coefficient can also be adjusted according to different environmental conditions or equipment performance requirements. For example, for deep sea observation tasks with high detection accuracy requirements, the threshold can be increased to 0.95; for environments with limited resources or high noise levels, the threshold can be appropriately reduced to 0.85 to improve the flexibility of the model.

[0096] In the above embodiments, the determination coefficient can be calculated by the following formula:

[0097]

[0098] wherein, SSR is the sum of squares of the absolute difference, SST is the sum of squares of the difference between the volume of the second outer oil bladder and the average volume of the plurality of outer oil bladders in the current sea trial data window, which can be determined by a certain time range.

[0099] The present application uses satellite communication to obtain the current sea trial data for updating the parameter error term of the buoy model, and uses the detection threshold as an auxiliary parameter to distinguish between parameter drift and state abnormality. The change of the volume of the outer oil bladder is used to monitor the relationship between buoyancy and gravity. If the absolute difference between the volume of the outer oil bladder calculated based on the buoy model and the volume of the outer oil bladder in the current sea trial data is greater than the detection threshold, and the buoy model has been updated, it is judged that the buoy is in an abnormal state. If the absolute difference is always lower than the detection threshold, the reason for the abnormal state of the buoy is attributed to parameter drift, and it is not identified as an abnormal state.

[0100] Figure 3 The mapping relationship between the absolute difference and the pressure is shown. Figure 3 The horizontal coordinate is the pressure, and the vertical coordinate is the numerical value of the absolute difference. The pressure is the depth of the associated function, the underwater pressure increases with the depth. The absolute difference (blue) is fitted with a power function (red solid line), and the shaded area represents the 99.7% confidence interval. The fitted curve effectively captures the non-linear relationship between the pressure and the absolute difference, and most of the absolute differences fall within the confidence interval, indicating that the estimation of the parametric error term is accurate. Accordingly, the present application uses the parametric error term as the term of the buoy model, which helps to improve the accuracy of the prediction results of the first outer oil bladder volume and reduce prediction errors.

[0101] Figure 4 The absolute difference is shown as a function of the buoy's running time. In the initial stage (first 30 data samples), the model update cannot be triggered due to the limited number of valid measurements returned in the current sea trial data; the value and the detection threshold are kept at their default settings (1 and 10, respectively). When the number of data samples returned in the sea trial data exceeds 30, the buoy model is updated for the first time (the update flag is set to 1), from the default value of 1 to 0.9909. As the buoy continues to run, the absolute difference gradually increases from about 1 mL to 16 mL, eventually exceeding the detection threshold and triggering a second update (the update flag is set to 2). During this period, from 0.9909 to 0.9876, and then the absolute difference decreases significantly. In the subsequent running stage, the absolute difference accumulates again and exceeds the threshold, triggering a third update (the update flag is set to 3). During this period, further decreases to 0.9807, and the absolute difference decreases significantly again. Overall, these results show that the proposed online updating strategy can effectively adjust the model parameters according to real-time running conditions, keeping the absolute difference within an acceptable range and enhancing the adaptability and robustness of the buoy model.

[0102] Based on another set of returned current sea trial data, the model is updated based on step S4 of the present application. Figure 5 The mapping relationship between the absolute difference and the pressure is shown, Figure 5 where the blue points represent the absolute difference, and the red shaded band represents the fitted curve and its confidence interval. According to the proposed updating strategy, the returned data is assigned different indicators: blue points represent the absolute difference points included in the model update (173 points); red "X" marks represent the absolute difference points that will cause the determination coefficient to be lower than 0.9, thereby reducing the model quality and being discarded; yellow squares represent the absolute difference points that exceed the "3-δ" threshold of the historical distribution, and these points are forced to be excluded as significantly abnormal (9 points).

[0103] Figure 6The evolution of the prediction error under this detection mechanism is shown. The blue curve represents the absolute difference, the dashed line represents the detection threshold, and the red dots represent the detected anomalies. A total of 7 update events are identified. At approximately the 31st data sample point (update flag = 1), when the cumulative number of data sample points exceeds 30, the model is updated for the first time, from 1.000 to 0.9918. Similarly, when the update flag is 2 to 7, the absolute difference exceeds the detection threshold, triggering subsequent parameter updates, and always keeping above 0.95. Between data sample points 66 and 72, an anomaly point appears, causing to drop below 0.9; therefore, these data are judged to be rejected in step S4, the parameter error term is not updated, and they are deleted from the historical database for updating. It is confirmed that the rejected data are indeed abnormal. Subsequently, between data sample points 141 and 149, a cluster of larger abnormal data appears, which deviates significantly from the historical distribution. These data sample points are forcibly excluded to prevent further model updates, again confirming their abnormal nature. Overall, these results show that the proposed online updating strategy not only enables dynamic model adaptation, but also effectively detects and isolates abnormal states, thereby improving the reliability and robustness of the buoy model.

[0104] It should be noted that the steps shown in the above flow or the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0105] The technical features of the above-described embodiments can be combined in any manner. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described, however, as long as the combination of the technical features does not contradict, it should be considered within the scope of the present disclosure.

[0106] The above-described embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the patent scope of the present application. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the patent protection scope of the present application should be subject to the appended claims.

Claims

1. A buoy anomaly detection method based on a buoy model, characterized in that, include: A buoy model of a profile drifting buoy in the hovering phase is constructed, and the parameters of the buoy model are configured using the least squares method with multi-source data. The current sea trial data transmitted back by the profile drifting buoy is obtained via satellite communication; Based on the buoy model, the volume of the outer oil bladder is predicted at a preset drift depth, and this volume is used as the first outer oil bladder volume. The actual volume of the outer oil bladder in the current sea trial data was analyzed and used as the volume of the second outer oil bladder. Determine the volume of the first outer oil bladder. and the volume of the second outer oil bladder Whether the absolute difference between them is less than or equal to the preset detection threshold. If so, then no parameter update is required; Otherwise, determine whether the buoy model has been updated. If the model has been updated, then the buoy is determined to be in an abnormal state; Otherwise, perform the parameter update step based on the statistical distribution of the current sea trial data.

2. The buoy anomaly detection method based on a buoy model according to claim 1, characterized in that, The buoy model is represented by the following calculation model: ， in, The mass of the profile drifting buoy, It is the acceleration due to gravity. For running depth The density of the water below, The volume of the glass sphere is... The volume of the outer oil sac. For the volume of other components of the buoy, The coefficient of compressibility of the glass sphere. The compression coefficient of the external oil bladder. The compression coefficient of other components of the buoy. This is the parameter error term.

3. The buoy anomaly detection method based on a buoy model according to claim 2, characterized in that, The parameter error term is determined by the following calculation model: ， in, , , These are the first coefficient, the second coefficient, and the third coefficient.

4. The buoy anomaly detection method based on a buoy model according to any one of claims 1 to 3, characterized in that, The parameter update step further includes: Calculate the statistical distribution of the current sea trial data and determine whether its distribution follows the statistical distribution of historical sea trial data; If so, the parameter error term is updated using the current sea trial data and the least squares method, and the determination coefficient corresponding to the absolute difference is calculated. , Further determine the determination coefficient If the value is greater than 0.9, then the parameter error term is updated again using the least squares method with the current sea trial data; otherwise, the parameter error term is not updated. Otherwise, do not use the current sea trial data to update the parameter error terms.

Images