Modular redundant offshore wind power cathodic protection pcb control system

The modular and redundant PCB control system enables the identification and dynamic monitoring of abnormal potential areas in the offshore wind power cathodic protection system. This solves the problems of erroneous switching and insufficient early warning timeliness in the existing cathodic protection system, improves the system's reliability and fault self-healing capability, and ensures the structural safety of the wind turbine foundation.

CN123063784APending Publication Date: 2026-11-10SUZHOU XIBEIYOU HYDROGEN ENERGY TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611252871.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-18
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

The existing control schemes for offshore wind power cathodic protection systems lack redundant design at the PCB level for the specific needs of cathodic protection, such as potential distribution monitoring and regional protection. This makes the control system prone to erroneous switching and control output interruption during faults, unable to effectively identify abnormal potential areas, and with insufficient early warning timeliness, thus affecting the structural safety of the wind turbine foundation.

Method used

The PCB control system, which adopts a modular redundancy design, uses a data acquisition unit, a status evaluation unit, a connectivity expansion unit, a failure determination unit, a feature generation unit, and a redundancy switching decision unit to identify and dynamically monitor abnormal potential regions. Based on the spatial distribution characteristics and changing trends of abnormal regions, it makes redundancy switching decisions to ensure a seamless transfer of control.

Benefits of technology

It improves the service reliability and fault self-healing capability of the offshore wind power cathodic protection system, reduces the probability of false switching, ensures the stable operation of the cathodic protection system in harsh marine environments, and provides technical support for the development of deep-sea wind power.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123063784A_ABST
    Figure CN123063784A_ABST
Patent Text Reader

Abstract

The application belongs to the technical field of wind power protection, and particularly relates to a modularized redundant offshore wind power cathodic protection PCB control system. Cathodic protection potential distribution data of offshore wind power foundation structure is acquired, the data is fed into a cathodic protection state evaluation model, and an abnormal area identification result of the protection potential is output. Connected domain expansion processing is performed on the abnormal area, and whether the expanded area data meets a protection failure early warning trigger condition is determined. A feature set is introduced into a protection system early warning decision model, and a modularized redundant switching control instruction is output to control the switching between work and standby of multiple PCB control modules. The application can make a reliable switching decision based on abnormal potential space evolution characteristics, effectively avoid misoperation caused by single threshold judgment, and through standby sampling and output isolation design of the hot standby module, seamless transfer of control right is realized, and the service reliability and fault self-healing ability of the cathodic protection system in a harsh marine environment are significantly improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of wind power protection technology, specifically relating to a modular redundant offshore wind power cathodic protection PCB control system. Background Technology

[0002] Offshore wind power facilities operate in a marine environment characterized by high salt spray, high humidity, strong waves and currents, and complex electrochemical corrosion. Corrosion protection of their steel foundation structures, such as monopiles and jacket structures, is crucial for ensuring the safe operation of wind turbines. Currently, cathodic protection systems, including sacrificial anode and impressed current methods, are widely used in corrosion protection engineering for offshore wind turbine foundations. By applying a protective current to the steel structure or connecting a more negatively charged anode material, the potential of the structure is maintained within a thermodynamically stable region, thereby inhibiting electrochemical corrosion reactions. However, with the increasing distance of wind farms from shore, greater water depth, and the increasing size of foundation structures, the reliability and maintainability of cathodic protection systems face severe challenges. In existing engineering practices, the control of cathodic protection systems typically uses a single PCB control board to acquire potential signals, adjust the protective potential in a closed loop, and manage the anode current output. If this control board malfunctions due to moisture condensation, salt spray corrosion, component aging, or transient overvoltage, it will directly lead to uncontrolled protective potential, potentially causing localized overprotection hydrogen evolution embrittlement or underprotection accelerated corrosion of the foundation structure, seriously threatening the structural safety of the wind turbine foundation.

[0003] In the existing technology, some offshore wind power projects have tried to adopt dual-machine hot standby or cold standby schemes to improve the fault tolerance of the cathodic protection control system. However, most of these schemes are based on the external redundancy architecture of general industrial controllers and do not carry out dedicated PCB-level redundancy design for the unique potential distribution monitoring and regional protection requirements of cathodic protection.

[0004] Specifically, conventional redundancy schemes often rely on host computers or external switching relays to switch between primary and backup boards. The switching decision logic is overly simplistic, typically based on a single trigger condition such as power supply status or communication timeout. This lacks the ability to regionally identify and trend-judge abnormalities in the spatial distribution of cathodic protection potentials, making it susceptible to erroneous switching due to instantaneous potential fluctuations or local measurement noise, causing unnecessary system disturbances. Furthermore, existing schemes lack refined timing management during backup board startup, often resulting in backup boards being forcibly connected to the control loop before completing internal parameter initialization or clock synchronization. This leads to control output interruptions or potential surges during switching, exacerbating the transient instability of the protection system. In addition, most existing systems' signal processing units lack the ability to dynamically track and analyze the morphological evolution of abnormal potential regions. They cannot predict protection failure risks in advance based on the spatial expansion trend of abnormal regions, only triggering alarms when the overall protection potential deviates significantly from the threshold. This insufficient early warning time leaves maintenance personnel with extremely limited response windows. Summary of the Invention

[0005] To address the aforementioned issues, there is an urgent need for a modular redundancy offshore wind power cathodic protection control system at the PCB control level. This system should be able to autonomously identify the spatial distribution characteristics of abnormal potential areas from multi-channel potential distribution data, accurately determine the precursory states of protection failure by analyzing regional connectivity trends, and then, when necessary, smoothly switch between the working module and the backup module in an orderly start-up sequence. Simultaneously, the system's redundancy switching decisions should be based on the dynamic evolution information of abnormal areas, rather than simple logic triggered by a single threshold, thereby reducing the probability of erroneous switching and improving the reliability and rationality of switching commands. By establishing a high-speed interconnect bus between PCB control modules, configuring independent power management and communication interfaces, and enabling the backup module to continuously sample without outputting control signals in standby mode, it is expected that seamless transfer of control can be achieved while ensuring continuous potential monitoring. This will significantly improve the service reliability and fault self-healing capability of the offshore wind power cathodic protection system, providing solid technical support for deep-sea wind power development.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] A modular redundant PCB control system for cathodic protection of offshore wind power consists of a data acquisition unit, a status assessment unit, a connectivity expansion unit, a failure determination unit, a feature generation unit, and a redundancy switching decision unit.

[0008] The data acquisition unit is connected to the cathodic protection potential sensor of the offshore wind power foundation structure to collect potential distribution data.

[0009] The status assessment unit receives the potential distribution data, feeds the potential distribution data into the cathodic protection status assessment model, and obtains the identification results of the protection potential abnormal area.

[0010] The connected domain expansion unit performs connected domain expansion processing on the abnormal potential region identification result of the protection potential abnormal region to generate expanded abnormal region data.

[0011] The failure determination unit determines whether the protection failure early warning triggering conditions are met based on the expanded abnormal area data.

[0012] When the failure determination unit determines that the result is consistent, the feature generation unit performs signal slicing and temporal semantic parsing on the identification result of the abnormal protection potential region to generate a system early warning candidate feature set.

[0013] The redundancy switching decision unit imports the system early warning candidate feature set into the protection system early warning decision model and outputs modular redundancy switching control commands.

[0014] Furthermore, the abnormal potential region connected domain expansion process performed by the connected domain expansion unit adopts a four-neighbor connected domain algorithm to expand the neighborhood of each abnormal pixel in the protection potential abnormal region identification result, merges the adjacent abnormal pixels into the same connected region, and expands the outer boundary of each connected region outward based on a preset expansion step size to form an expanded abnormal region.

[0015] The expansion step size is dynamically determined by the connected region expansion unit based on the potential gradient magnitude of the boundary pixels of the current connected region. When the boundary gradient magnitude exceeds a set threshold, a smaller expansion step size is used, and when the boundary gradient magnitude is lower than a set threshold, a larger expansion step size is used.

[0016] After completing the independent expansion of each connected region, the connected domain expansion unit performs a merging operation on the overlapping expanded regions, merging the overlapping parts into an overall abnormal region.

[0017] When merging overlapping regions, the connected domain expansion unit uses the geometric center of the overlapping region as the center of the merged region and recalculates the outer boundary of the merged region.

[0018] Furthermore, the failure determination unit determines whether the protection failure early warning triggering conditions are met based on the expanded abnormal area data, specifically as follows:

[0019] Calculate the area ratio of the expanded abnormal region in the potential distribution data, and calculate the rate of change of the area ratio relative to the previous sampling time. When the area ratio exceeds a predetermined threshold and the rate of change shows an increasing trend, it is determined to meet the protection failure early warning triggering condition.

[0020] The determination of the growth trend is based on the sign of the rate of change at multiple consecutive sampling times. If the rate of change at three consecutive sampling times is positive, it is considered to be a growth trend.

[0021] The predetermined threshold is dynamically adjusted by the failure determination unit based on the baseline area ratio obtained during the initial system calibration. The baseline area ratio is the statistical average of the abnormal area ratio in the unprotected failure state of the system.

[0022] In the calculation of the rate of change, if the area of ​​the abnormal region at the previous sampling time is zero, the rate of change is based on the area ratio at the current sampling time.

[0023] Furthermore, the signal slicing processing performed by the feature generation unit divides the protection potential abnormal region identification results into multiple consecutive time slices according to the time dimension. Each time slice corresponds to an abnormal region identification result sequence within a fixed time window. The duration of the fixed time window is determined according to the sampling period of the potential distribution data. Each time slice contains abnormal region identification results at multiple sampling times.

[0024] The temporal semantic parsing process performed by the feature generation unit extracts the spatial location change trajectory, geometric morphology evolution features, center point displacement vector, total area change curve, and number change curve of the abnormal region within each time slice for the abnormal region identification result sequence.

[0025] The spatial position change trajectory is obtained by recording the coordinate sequence of the centroid of the abnormal region at each sampling moment. The geometric morphological evolution features are obtained by calculating the aspect ratio of the bounding rectangle of the abnormal region, the perimeter of the region, and the compactness of the region at each sampling moment. The displacement vector is obtained by the difference in centroid coordinates between adjacent sampling moments. The area change curve is obtained by counting the total number of effective level pixels in the abnormal region at each sampling moment. The quantity change curve is obtained by counting the number of independent connected regions at each sampling moment.

[0026] The feature generation unit concatenates the extraction results of all time slices in chronological order to form the system early warning candidate feature set, which consists of anomaly region location sequence, morphology sequence, displacement vector sequence, area change sequence, and quantity change sequence.

[0027] After generating the system early warning candidate feature set, the feature generation unit transmits the system early warning candidate feature set to the redundancy switching decision unit in a structured data format, wherein the structured data format is a set of key-value pairs.

[0028] Furthermore, the modular redundancy switching control command consists of a switching enable signal for switching from the main control PCB module to the backup PCB module and the startup timing configuration parameters of the backup PCB module.

[0029] The switching enable signal is a high-level active digital pulse signal, used to trigger the output shutdown of the current working module and the output turn-on of the standby module.

[0030] The startup timing configuration parameters include the power supply establishment delay of the backup module, the clock synchronization calibration delay, and the control parameter loading delay. The redundancy switching decision unit outputs multiple control sub-signals in sequence according to the startup timing configuration parameters to complete the startup process of the backup module in sequence.

[0031] The startup timing configuration parameters of the backup module also include the self-test status waiting time and the output stabilization waiting time of the backup module.

[0032] Furthermore, the system has multiple redundant PCB control modules. Each PCB control module is independently configured with a cathodic protection potential signal acquisition channel and a potential control output channel. The switching control command output by the redundancy switching decision unit is used to select one of the PCB control modules as the current working module, and the remaining PCB control modules as hot standby modules.

[0033] Each PCB control module has an independent power management submodule and a communication interface submodule. The PCB control modules are interconnected through a high-speed backplane bus. The hot standby module continuously receives potential distribution data but does not output control signals in standby mode. When the redundancy switching decision unit outputs a switching control command, it simultaneously sends a shutdown command to the currently working module and an activation command to the selected standby module. The activation command contains the initial operating parameters of the standby module.

[0034] The hot standby module maintains a high impedance output port in standby mode to avoid interfering with the output of the currently working module.

[0035] Furthermore, the data acquisition unit is connected to the potential signal acquisition channel of each PCB control module through a synchronous serial interface, and the status evaluation unit obtains the potential distribution data from the data acquisition unit through a high-speed data bus;

[0036] The condition assessment unit has built-in the cathodic protection condition assessment model, which outputs abnormal area identification results based on potential distribution data. The abnormal area identification results are represented in the form of a binary image.

[0037] The connected component expansion unit receives the binarized image and performs the connected component expansion process on the connected regions marked with effective level to generate an expanded binarized image.

[0038] The failure determination unit receives the expanded binarized image, counts the total number of pixels with valid level markers, and determines whether the protection failure warning trigger condition is met based on the ratio of the total number of pixels to the total number of pixels in the binarized image and the gradient of the ratio over time.

[0039] The feature generation unit is activated only when the failure determination unit outputs a positive determination result. The feature generation unit obtains the original protection potential abnormal region identification result from the state evaluation unit through the data bus, and performs the signal slicing and timing semantic parsing on the identification result.

[0040] After receiving the system early warning candidate feature set, the redundancy switching decision unit calls the protection system early warning decision model, which outputs a switching priority sequence corresponding to multiple redundant PCB control modules.

[0041] The modular redundancy switching control command determines the target switching module based on the switching priority sequence and generates the corresponding switching control signal.

[0042] Furthermore, the early warning decision model of the protection system is a rule-based reasoning model based on decision trees;

[0043] The redundancy switching decision unit matches the system early warning candidate feature set with the early warning rule, and the redundancy switching action corresponding to the successfully matched early warning rule is the modular redundancy switching control command.

[0044] The matching process involves the redundancy switching decision unit sequentially comparing each sequence value in the system early warning candidate feature set with the feature conditions of each early warning rule. When all sequence values ​​meet the feature conditions of a certain early warning rule, the early warning rule is determined to be a successful match.

[0045] The warning rules include a default rule. When no rule is matched successfully, the redundancy switching decision unit outputs a switching control command pointing to the highest priority backup module.

[0046] Furthermore, when the failure determination unit determines that the expanded abnormal area data does not meet the protection failure warning triggering conditions, it maintains the working state of the currently working PCB control module, and both the feature generation unit and the redundancy switching decision unit remain in standby state, without performing signal slicing, timing semantic parsing and redundancy switching control command output operations.

[0047] In the standby state, the feature generation unit stops slicing and parsing the identification results of the abnormal protection potential region, and the redundancy switching decision unit stops calling the protection system early warning decision model and generating switching control commands.

[0048] The failure determination unit repeatedly performs the determination operation in each sampling period until the determination result meets the triggering condition. Then, it sends a start enable signal to the feature generation unit. After receiving the start enable signal, the feature generation unit starts to perform signal slicing and temporal semantic parsing.

[0049] When the failure determination unit calculates the ratio and change gradient in each sampling period, it only relies on the expanded abnormal area data in the current sampling period and does not depend on historical data caching.

[0050] Furthermore, the data acquisition unit consists of multiple potential acquisition channels, which take turns acquiring the potential signals of each sensor in a time-division multiplexing manner, and arrange the acquired potential signals according to the sensor position coordinates to form the potential distribution data;

[0051] The multiple potential acquisition channels are configured on different PCB control modules of the system. Each PCB control module is provided with at least one potential acquisition channel. The time-division multiplexing method is executed by the channel selection controller inside the data acquisition unit. The channel selection controller selects each potential acquisition channel in sequence according to a fixed channel polling order. It completes one potential signal sampling in each selection cycle and stores the sampled value in the buffer unit corresponding to the sensor coordinate. The data in all buffer units together form the potential distribution data.

[0052] The polling order of the channel selection controller is arranged according to the physical location of each sensor, so that the collected potential distribution data can be directly mapped to the spatial coordinate grid of the offshore wind power foundation.

[0053] This invention belongs to the field of wind power protection technology, specifically relating to a modular redundant PCB control system for offshore wind power cathodic protection. It acquires cathodic protection potential distribution data of the offshore wind power foundation structure, feeds this data into a cathodic protection status assessment model, and outputs the identification results of abnormal protection potential areas. Connectivity expansion processing is performed on the abnormal areas, and the expanded area data is used to determine whether the protection failure early warning triggering conditions are met. The feature set is imported into the protection system early warning decision model, and modular redundant switching control commands are output to control the working and standby switching between multiple PCB control modules. This invention can make reliable switching decisions based on the spatial evolution characteristics of abnormal potentials, effectively avoiding malfunctions caused by single threshold judgments. Simultaneously, through the standby sampling and output isolation design of the hot standby module, seamless transfer of control is achieved, significantly improving the service reliability and fault self-healing capability of the cathodic protection system in harsh marine environments. Attached Figure Description

[0054] Figure 1 This is a system structure diagram of a modular redundant offshore wind power cathodic protection PCB control system claimed in an embodiment of the present invention;

[0055] Figure 2 This is a flowchart illustrating the operation of a modular redundant offshore wind power cathodic protection PCB control system claimed in an embodiment of the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0057] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of those features. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications in the embodiments of this application, such as up, down, left, right, front, back, etc., are only used to explain the relative positional relationships and movements between components in a specific orientation as shown in the accompanying drawings. If the specific orientation changes, the directional indications will change accordingly. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0058] References to embodiments herein mean that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0059] According to the first embodiment of the present invention, referring to Figure 1 and Figure 2 This invention claims protection for a modular redundant offshore wind power cathodic protection PCB control system, which consists of a data acquisition unit, a status assessment unit, a connectivity expansion unit, a failure determination unit, a feature generation unit, and a redundancy switching decision unit.

[0060] The data acquisition unit is connected to the cathodic protection potential sensor of the offshore wind power foundation to collect potential distribution data;

[0061] The condition assessment unit receives potential distribution data, feeds the potential distribution data into the cathodic protection condition assessment model, and obtains the identification results of abnormal protection potential areas.

[0062] The connectivity expansion unit performs connectivity expansion processing on the abnormal potential region identification results to generate expanded abnormal region data;

[0063] The failure determination unit determines whether the protection failure early warning triggering conditions are met based on the expanded abnormal area data.

[0064] When the failure determination unit determines that the result is satisfactory, the feature generation unit performs signal slicing and temporal semantic parsing on the identification result of the abnormal protection potential area to generate a system early warning candidate feature set.

[0065] The redundancy switching decision unit imports the system early warning candidate feature set into the protection system early warning decision model and outputs modular redundancy switching control commands.

[0066] In this embodiment, the system is deployed on a booster station platform of an offshore wind farm. Its data acquisition unit is connected via an RS-485 bus to 12 silver / silver chloride reference electrodes installed on the outside of the steel piles of the wind turbine foundation. Each electrode forms a potential measurement loop with the steel structure. The data acquisition unit continuously acquires the potential value of each electrode position at a sampling rate of 10 times per second, and arranges these potential values ​​into an 8-row × 12-column potential distribution matrix according to the arrangement of the electrodes on the circumference. Each element in the matrix corresponds to the potential amplitude of a spatial position, in millivolts.

[0067] The status assessment unit reads the potential distribution matrix from the data acquisition unit via a parallel data bus and inputs the matrix into the cathodic protection status assessment model. The model has pre-stored a normal potential range threshold, such as -850mV to -1050mV. The model compares each potential value in the matrix with the normal threshold range one by one, marks the potential locations outside the range as abnormal, and outputs a binary label matrix with the same size as the input matrix. Abnormal locations are marked as logic 1, and normal locations are marked as logic 0. This is the result of the protection potential abnormal area identification. After acquiring the binarized marker matrix, the connected component expansion unit performs a four-neighbor connectivity check on the logical 1 region. This check verifies whether the four adjacent pixels above, below, left, and right of each logical 1 pixel are all logical 1 pixels. Connected logical 1 pixels are merged into the same region, and the outer boundary of each region is extended outward pixel by pixel. The extension distance is dynamically adjusted based on the potential gradient value of the boundary pixel of the region. If the potential change rate at the boundary is greater than 5 millivolts per millimeter, it extends by 2 pixels; if it is less than 5 millivolts per millimeter, it extends by 5 pixels. After the extension is completed, if different regions overlap, the overlapping parts are merged into a whole region to obtain the expanded abnormal region data. The failure determination unit receives the expanded abnormal region data, calculates the percentage of the total number of logical 1 pixels in the matrix, and records the percentage values ​​at three consecutive sampling times. If the current percentage exceeds 15% and the percentage at the next time step is greater than the previous time step, it is determined that the protection failure warning trigger condition is met.

[0068] The feature generation unit is activated only after the triggering condition is met. It obtains the original abnormal region identification result, i.e., the unexpanded binary matrix, from the state evaluation unit. It divides the matrix into slices for every 50 consecutive sampling times in chronological order, with each slice covering a duration of 5 seconds. It tracks the location of the abnormal region at each time point within each slice, records the time series of the centroid coordinates of each abnormal region, and the time series of the number of abnormal pixels in each region. It combines these sequences from all slices into a system early warning candidate feature set. The redundancy switching decision unit receives this feature set, compares it with the decision rules stored internally, determines the target PCB module number to which it needs to switch, such as module 2, and generates a corresponding switching control command. This command is output in the form of a digital pulse with a pulse width of 20 milliseconds.

[0069] Furthermore, the abnormal potential region connected domain expansion process performed by the connected domain expansion unit adopts a four-neighbor connected domain algorithm to expand the neighborhood of each abnormal pixel in the protection potential abnormal region identification result, merges the adjacent abnormal pixels into the same connected region, and expands the outer boundary of each connected region outward based on a preset expansion step size to form an expanded abnormal region.

[0070] The expansion step size is dynamically determined by the connected region expansion unit based on the potential gradient magnitude of the boundary pixels of the current connected region. When the boundary gradient magnitude exceeds a set threshold, a smaller expansion step size is used, and when the boundary gradient magnitude is lower than a set threshold, a larger expansion step size is used.

[0071] After completing the independent expansion of each connected region, the connected domain expansion unit performs a merging operation on the overlapping expanded regions, merging the overlapping parts into an overall abnormal region.

[0072] When merging overlapping regions, the connected domain expansion unit uses the geometric center of the overlapping region as the center of the merged region and recalculates the outer boundary of the merged region.

[0073] In this embodiment, when the connected component expansion unit executes the four-neighbor connectivity algorithm, the specific operation process is as follows: First, scan each pixel of the entire binary labeling matrix. When encountering a logical 1 pixel that has not yet been labeled by any region, use this pixel as a seed to recursively check its neighboring pixels in the left, top, right, and bottom directions. Assign all recursively reachable logical 1 pixels to the same connected region number until the region can no longer be expanded. Then continue scanning the next unlabeled logical 1 pixel until all abnormal pixels are numbered. For each connected region, calculate its outer boundary pixel set, i.e. Boundary pixels with at least one four-neighbor pixel set to logic 0 are identified. For each boundary pixel, the potential value at the corresponding position in the original potential distribution matrix and the potential value of its adjacent external neighbors are read. The potential difference is calculated, and this difference is divided by the pixel spacing, which is fixed at 0.5 meters, to obtain the gradient magnitude. If the gradient magnitude is higher than 8 millivolts per meter, the expansion step size of the boundary pixel is set to 1 pixel. If it is between 4 and 8 millivolts per meter, the step size is 2 pixels. If it is lower than 4 millivolts per meter, the step size is 3 pixels. For each boundary pixel, the corresponding step size is extended outward along its normal direction, and the newly added pixels after the extension are marked as logic 1.

[0074] After all boundary pixels have been expanded, check whether all regions overlap: if two or more regions share the same pixel coordinates after expansion, merge the numbers of these regions, and merge the shared pixels and all the original pixels of the two regions into a new single region. The geometric center of the new region is obtained by calculating the arithmetic mean of the coordinates of all pixels in the merged region, and this center is used as a reference point for subsequent judgment of changes in the region's position. At the same time, recalculate the bounding rectangle of the new region, and define the new boundary of the new region with the coordinates of the upper left and lower right corners of the rectangle.

[0075] Furthermore, the failure determination unit determines whether the protection failure early warning triggering conditions are met based on the expanded abnormal area data, specifically as follows:

[0076] Calculate the area ratio of the expanded abnormal region in the potential distribution data, and calculate the rate of change of the area ratio relative to the previous sampling time. When the area ratio exceeds a predetermined threshold and the rate of change shows an increasing trend, it is determined to meet the protection failure early warning triggering condition.

[0077] The determination of the growth trend is based on the sign of the rate of change at multiple consecutive sampling times. If the rate of change at three consecutive sampling times is positive, it is considered to be a growth trend.

[0078] The predetermined threshold is dynamically adjusted by the failure determination unit based on the baseline area ratio obtained during the initial system calibration. The baseline area ratio is the statistical average of the abnormal area ratio in the unprotected failure state of the system.

[0079] In the calculation of the rate of change, if the area of ​​the abnormal region at the previous sampling time is zero, the rate of change is based on the area ratio at the current sampling time.

[0080] In this embodiment, the failure determination unit receives an augmented binarized image once every 0.1 seconds during each sampling period. Let the current sampling time be t, the total number of pixels in the image be N = 968 × 12, and the number of logical 1 pixels be N1(t); calculate the area ratio R(t) = N1(t) / N. Meanwhile, during the initialization phase, after the system is powered on for the first time and runs for 10 minutes, it continuously records the area ratio at 100 sampling times and calculates the average value as the baseline area ratio R_base. In this embodiment, the measured R_base is 3.2%. Then, each time the failure judgment unit receives a new frame of image, it calculates the difference between the current R(t) and the baseline value R_base. If the difference is greater than 12 percentage points, i.e., R(t) > 15.2%, then the rate of change ΔR(t) = R(t) - R(t-1) is further calculated, where R(t-1) is the area ratio 0.1 seconds before the previous sampling time. If the abnormal area at the previous sampling time is zero, i.e., R(t-1) = 0, then ΔR(t) is directly defined as R(t) itself. Next, the failure determination unit maintains a change rate sign cache of length 3 to store the signs of ΔR(t-2), ΔR(t-1), and ΔR(t) as positive, negative, or zero, respectively. If the current ΔR(t) > 0 and the previous two change rates are also greater than zero, that is, the change rates are positive for three consecutive time points, and the current R(t) > 15.2%, then it is determined that the protection failure warning trigger condition is met. If the change rate sign is non-positive, the cache is reset and it is determined that the condition is not met. This determination logic only relies on the data of the current time point and the previous two time points and does not involve the historical storage of longer time periods.

[0081] Furthermore, the signal slicing processing performed by the feature generation unit divides the protection potential abnormal region identification results into multiple consecutive time slices according to the time dimension. Each time slice corresponds to an abnormal region identification result sequence within a fixed time window. The duration of the fixed time window is determined according to the sampling period of the potential distribution data. Each time slice contains abnormal region identification results at multiple sampling times.

[0082] The temporal semantic parsing process performed by the feature generation unit extracts the spatial location change trajectory, geometric morphology evolution features, center point displacement vector, total area change curve, and number change curve of the abnormal region within each time slice for the abnormal region identification result sequence.

[0083] The spatial position change trajectory is obtained by recording the coordinate sequence of the centroid of the abnormal region at each sampling moment. The geometric morphological evolution features are obtained by calculating the aspect ratio of the bounding rectangle of the abnormal region, the perimeter of the region, and the compactness of the region at each sampling moment. The displacement vector is obtained by the difference in centroid coordinates between adjacent sampling moments. The area change curve is obtained by counting the total number of effective level pixels in the abnormal region at each sampling moment. The quantity change curve is obtained by counting the number of independent connected regions at each sampling moment.

[0084] The feature generation unit concatenates the extraction results of all time slices in chronological order to form the system early warning candidate feature set, which consists of anomaly region location sequence, morphology sequence, displacement vector sequence, area change sequence, and quantity change sequence.

[0085] After generating the system early warning candidate feature set, the feature generation unit transmits the system early warning candidate feature set to the redundancy switching decision unit in a structured data format, wherein the structured data format is a set of key-value pairs.

[0086] In this embodiment, the feature generation unit is activated only after the failure determination unit outputs a positive signal.

[0087] After startup, the feature generation unit obtains the raw anomaly region identification results for the most recent 300 sampling times (30 seconds) from the state evaluation unit, with each time point corresponding to a binarized matrix. The signal slicing process divides the 300 matrices into 6 consecutive time slices in chronological order, with each slice containing 50 matrices and corresponding to a duration of 5 seconds.

[0088] For each of the 50 matrices in a slice, the feature generation unit performs the following parsing operations: First, for each matrix, extract the centroid coordinates x, y of all connected regions. The centroid coordinates are calculated as the average of the horizontal and vertical coordinates of all logical 1 pixels in the region, rounded to the nearest pixel. Then, arrange the centroid coordinates of the 50 times within the same slice in chronological order to form a spatial position change trajectory. For example, the centroid is (3,5) at time 1, (4,5) at time 2, and up to (7,8) at time 50. Record the coordinate sequence of this trajectory. At the same time, calculate the aspect ratio of the bounding rectangle of each region at each time, i.e., the width of the rectangle divided by the height in pixels, and the perimeter of the region by counting the number of boundary pixels. The region compactness is defined as 4π × area / perimeter squared, and the area is directly taken as the number of logical 1 pixels. Record these geometric parameters in chronological order as a morphological sequence. Next, calculate the difference vectors Δx and Δy between the centroid coordinates of adjacent time points. For example, subtracting the centroid from the centroid at time 1 from the centroid at time 2 results in (1,0). Record all difference vectors to form a displacement vector sequence. Simultaneously, count the total number of logical 1 pixels in all abnormal regions at each time point as an area change sequence. For example, the total number of abnormal pixels at time 1 is 120, at time 2 it is 135, ... at time 50 it is 210. Count the number of independently connected regions at each time point as a quantity change sequence. For example, there are 3 regions at time 1, 4 at time 2, ... at time 50 it is 6.

[0089] The five sequences mentioned above are concatenated in chronological order to form a candidate feature set for system early warning. This feature set is stored in key-value pairs, where the keys are location trajectory, morphological sequence, displacement vector, area sequence, and quantity sequence, and the corresponding values ​​are the corresponding time series arrays. The feature generation unit transmits this key-value pair set to the redundancy switching decision unit through the shared memory of the system's internal data exchange area.

[0090] Furthermore, the modular redundancy switching control command consists of a switching enable signal for switching from the main control PCB module to the backup PCB module and the startup timing configuration parameters of the backup PCB module.

[0091] The switching enable signal is a high-level active digital pulse signal, used to trigger the output shutdown of the current working module and the output turn-on of the standby module.

[0092] The startup timing configuration parameters include the power supply establishment delay of the backup module, the clock synchronization calibration delay, and the control parameter loading delay. The redundancy switching decision unit outputs multiple control sub-signals in sequence according to the startup timing configuration parameters to complete the startup process of the backup module in sequence.

[0093] The startup timing configuration parameters of the backup module also include the self-test status waiting time and the output stabilization waiting time of the backup module.

[0094] In this embodiment, the switching control command output by the redundancy switching decision unit includes a high-level active digital pulse signal switching enable signal and a set of startup timing configuration parameters. The digital pulse signal has a voltage of 5 volts and a duration of 50 milliseconds. The rising edge triggers the output relay of the currently working PCB module to disconnect, and the falling edge triggers the output relay of the standby PCB module to engage. The startup timing configuration parameters include four delay values: a power supply setup delay of 200 milliseconds to wait for the standby module's power regulator to output a stable voltage; a clock synchronization calibration delay of 150 milliseconds to lock the phase of the standby module's internal oscillator with the system's master clock; a control parameter loading delay of 100 milliseconds to allow the standby module to read control parameters such as the protection potential setting value and PID coefficient from the non-volatile memory; a self-test state waiting time of 300 milliseconds to allow the standby module to complete internal analog channel self-tests and communication link detection; and an output stabilization waiting time of 250 milliseconds to ensure that the standby module's drive circuit reaches a stable output state before output enable.

[0095] The redundancy switching decision unit sends out five sub-control signals in the above order. Each sub-signal is a timer trigger pulse with a corresponding delay duration. Specifically, it first sends out a power supply setup delay pulse, then automatically sends out a clock synchronization calibration delay pulse after 200 milliseconds, then sends out a parameter loading delay pulse after 150 milliseconds, and so on. Finally, after the output stabilizes and the waiting time ends, it pulls the switching enable signal high to complete the entire switching process.

[0096] Furthermore, the system has multiple redundant PCB control modules. Each PCB control module is independently configured with a cathodic protection potential signal acquisition channel and a potential control output channel. The switching control command output by the redundancy switching decision unit is used to select one of the PCB control modules as the current working module, and the remaining PCB control modules as hot standby modules.

[0097] Each PCB control module has an independent power management submodule and a communication interface submodule. The PCB control modules are interconnected through a high-speed backplane bus. The hot standby module continuously receives potential distribution data but does not output control signals in standby mode. When the redundancy switching decision unit outputs a switching control command, it simultaneously sends a shutdown command to the currently working module and an activation command to the selected standby module. The activation command contains the initial operating parameters of the standby module.

[0098] The hot standby module maintains a high impedance output port in standby mode to avoid interfering with the output of the currently working module.

[0099] In this embodiment, the system is configured with a total of three PCB control modules, numbered A, B, and C. Each module independently has six potential signal acquisition channels corresponding to six sensors and six potential control output channels corresponding to six auxiliary anodes. Module A serves as the currently operating module, while modules B and C serve as hot standby modules. Each module is equipped with a separate DC-DC power conversion submodule, with a 48V DC bus input and ±5V and ±12V analog power outputs. Additionally, each module has a CAN communication interface submodule for data exchange with the host computer.

[0100] The three modules are interconnected via a high-speed backplane bus using LVDS differential signaling at a transmission rate of 100 megabits per second. This backplane simultaneously transmits potential distribution data and module status information. In standby mode, hot standby modules B and C continuously acquire potential signals and update their internal buffers through their potential acquisition channels. However, the analog output terminals of their control output channels are configured in a high-impedance state (tri-state output pins), not driving any voltage to prevent current conflicts with module A's output. When the redundancy switching decision unit outputs a switching command selecting module B as the new working module, this command simultaneously sends a shutdown command to module A via the backplane bus, forcing module A's output channel into a high-impedance state, and sends an activation command to module B. The activation command includes initial operating parameters, such as a protection potential target value of -950mV, a current limit of 5 amps, and a filter coefficient of 0.8. After receiving the activation command, module B configures its control register according to these parameters and takes over the output after completing the startup sequence. During standby, the hot standby module's output port remains in a high-impedance state until it receives an activation command or switching command, ensuring that only one module drives the anode output at any given time.

[0101] Furthermore, the data acquisition unit is connected to the potential signal acquisition channel of each PCB control module through a synchronous serial interface, and the status evaluation unit obtains the potential distribution data from the data acquisition unit through a high-speed data bus;

[0102] The condition assessment unit has built-in the cathodic protection condition assessment model, which outputs abnormal area identification results based on potential distribution data. The abnormal area identification results are represented in the form of a binary image.

[0103] The connected component expansion unit receives the binarized image and performs the connected component expansion process on the connected regions marked with effective level to generate an expanded binarized image.

[0104] The failure determination unit receives the expanded binarized image, counts the total number of pixels with valid level markers, and determines whether the protection failure warning trigger condition is met based on the ratio of the total number of pixels to the total number of pixels in the binarized image and the gradient of the ratio over time.

[0105] The feature generation unit is activated only when the failure determination unit outputs a positive determination result. The feature generation unit obtains the original protection potential abnormal region identification result from the state evaluation unit through the data bus, and performs the signal slicing and timing semantic parsing on the identification result.

[0106] After receiving the system early warning candidate feature set, the redundancy switching decision unit calls the protection system early warning decision model, which outputs a switching priority sequence corresponding to multiple redundant PCB control modules.

[0107] The modular redundancy switching control command determines the target switching module based on the switching priority sequence and generates the corresponding switching control signal.

[0108] In this embodiment, the data acquisition unit is connected to the acquisition channels on the three PCB control modules via a synchronous serial interface SPI with a clock frequency of 2 MHz. However, currently only the acquisition channel of the working module uploads data to the data acquisition unit, and the acquisition data of the backup module is only used for internal self-testing. The status evaluation unit obtains the potential distribution data updated every 0.1 seconds from the data acquisition unit through the high-speed data bus PCIEx1 link. This data is an 8×12 floating-point matrix.

[0109] The cathodic protection status assessment model built into the status assessment unit is a threshold comparator. This comparator compares each potential value with the pre-stored upper and lower limits of -850mV and -1050mV to generate a binary image matrix, in which abnormal potentials exceeding the range are marked as logic 1 and normal potentials are marked as logic 0. After receiving the binary image, the connected component expansion unit uses the eight-neighbor connectivity algorithm, that is, it checks the eight neighboring pixels around each logic 1 pixel and classifies any pixel with logic 1 in any of the eight directions into the same connected region, thereby achieving clustering of abnormal regions.

[0110] After clustering, the diagonal length of the bounding rectangle of each connected region is calculated in pixels. For example, if the bounding rectangle of a region is 8 pixels wide and 6 pixels high, then the diagonal length is 10 pixels. The expansion radius is determined based on this diagonal length: if the diagonal length is less than 15 pixels, the expansion radius is 2 pixels; if it is between 15 and 30 pixels, the expansion radius is 4 pixels; if it is greater than 30 pixels, the expansion radius is 6 pixels. Then, this radius is used to expand all boundary pixels of each region outwards at equal intervals, generating an expanded binarized image. The failure determination unit receives this expanded image, counts the total number of logical 1 pixels M, and calculates the ratio P = M / 96 to the total number of pixels 96. Simultaneously calculate the difference between the current ratio P(t) and the previous ratio P(t-1), divide it by the sampling interval of 0.1 seconds to obtain the change gradient G(t) in units of per second; the preset normal ratio range is 0% to 12% determined based on historical normal data. If P(t) > 12% and G(t) > 0, it is determined that the protection failure warning trigger condition is met.

[0111] The feature generation unit only starts after the failure determination unit outputs a positive decision. It obtains the original, unexpanded binary recognition result from the state evaluation unit via the data bus and performs slicing and semantic parsing. After receiving the feature set, the redundancy switching decision unit calls its internal decision model, which is a rule base based on a lookup table. This model outputs a switching priority sequence, for example, [B,C,A], indicating that the first choice is to switch to B, the second choice is C, and the last choice is A. Based on this sequence, the redundancy switching decision unit generates a switching control signal, which includes the gating address of the target module (e.g., binary 01 represents module B) and the switching execution time, which is set to the current system clock plus a fixed 200-millisecond delay window to wait for the current cycle to complete.

[0112] Furthermore, the early warning decision model of the protection system is a rule-based reasoning model based on decision trees;

[0113] The redundancy switching decision unit matches the system early warning candidate feature set with the early warning rule, and the redundancy switching action corresponding to the successfully matched early warning rule is the modular redundancy switching control command.

[0114] The matching process involves the redundancy switching decision unit sequentially comparing each sequence value in the system early warning candidate feature set with the feature conditions of each early warning rule. When all sequence values ​​meet the feature conditions of a certain early warning rule, the early warning rule is determined to be a successful match.

[0115] The warning rules include a default rule. When no rule is matched successfully, the redundancy switching decision unit outputs a switching control command pointing to the highest priority backup module.

[0116] In this embodiment, the protection system's early warning decision model is a rule base based on a decision tree. This rule base is stored in the internal flash memory of the redundancy switching decision unit and contains a total of 7 early warning rules. Each rule has a feature condition part and an action part. The feature condition part sets constraint ranges for 5 types of feature sequences: position trajectory, shape sequence, displacement vector, area sequence, and quantity sequence.

[0117] For example, the feature conditions for Rule 1 are: the cumulative centroid displacement in the position trajectory exceeds 15 pixels, the average aspect ratio of the bounding rectangle in the shape sequence is greater than 2.5, the vector direction in the displacement vector sequence points northeast three times consecutively, the area in the area sequence at the last moment is more than twice the area at the beginning moment, and the number of connected regions in the quantity sequence increases from 3 to more than 5. The action part specifies switching to module B, and the switching method is immediate switching; the feature conditions for Rule 2 are that the area sequence growth rate exceeds 30% per minute and the quantity sequence remains constant, and the action specifies switching to module C; after receiving the system warning candidate feature set, the redundant switching decision unit sequentially compares each sequence value in the feature set with the feature conditions of each rule.

[0118] For example, firstly, the total distance of the centroid coordinate change in the position trajectory sequence is calculated by summing the absolute values ​​of the displacements at each moment. If it is greater than 15 pixels, then the first condition is met. Next, the average of all aspect ratios in the shape sequence is checked; if it is 2.7, then the condition is met. Then, the displacement vector sequence is checked; if the displacement direction angles for three consecutive times are between 30° and 60° in the northeast direction, then the condition is met. Next, the first and last values ​​of the area sequence are compared; if the last value is 2.3 times the first value, then the condition is met. Finally, the quantity sequence is checked; if the initial value is 3 and the final value is 6, then the condition is met. When all conditions are met, rule 1 is considered a successful match, and the redundancy switching decision unit immediately executes the action part, outputting the instruction to switch to module B. If rule 1 is not completely matched, rule 2 is checked, and so on. If none of the rules are matched, the default rule is triggered. The action part of this default rule specifies switching to the highest priority backup module, i.e., module B. The priority is determined by the system configuration.

[0119] Furthermore, when the failure determination unit determines that the expanded abnormal area data does not meet the protection failure warning triggering conditions, it maintains the working state of the currently working PCB control module, and both the feature generation unit and the redundancy switching decision unit remain in standby state, without performing signal slicing, timing semantic parsing and redundancy switching control command output operations.

[0120] In the standby state, the feature generation unit stops slicing and parsing the identification results of the abnormal protection potential region, and the redundancy switching decision unit stops calling the protection system early warning decision model and generating switching control commands.

[0121] The failure determination unit repeatedly performs the determination operation in each sampling period until the determination result meets the triggering condition. Then, it sends a start enable signal to the feature generation unit. After receiving the start enable signal, the feature generation unit starts to perform signal slicing and temporal semantic parsing.

[0122] When the failure determination unit calculates the ratio and change gradient in each sampling period, it only relies on the expanded abnormal area data in the current sampling period and does not depend on historical data caching.

[0123] In this embodiment, the failure determination unit calculates the ratio P(t) and the gradient G(t) every 0.1 seconds in each sampling period. If P(t) ≤ 12% or G(t) ≤ 0, it is determined that the triggering condition is not met. At this time, the failure determination unit does not send a start enable signal to the feature generation unit, the feature generation unit remains in standby state, its internal clock gate is turned off, and it does not perform slicing or parsing operations on any potential identification results. At the same time, the redundancy switching decision unit also remains in standby state, its output port remains at a low level, and it does not generate any switching command.

[0124] In standby mode, the data input buffer of the feature generation unit does not receive new recognition results, and its processor core suspends feature extraction tasks, maintaining only communication monitoring functions. The redundancy switching decision unit stops calling the decision model, and its internal rule matching engine is in sleep mode. The failure determination unit performs independent determination in each sampling period, without relying on cached data from previous periods. That is, each time the current expanded binarized image is reread, P(t) and G(t) are calculated. G(t) requires P(t-1) from the previous moment, but only the P value from the previous moment is retained, and earlier data is not retained. Once P(t) > 12% and G(t) > 0 in a certain sampling period, the failure determination unit immediately generates a positive pulse enable signal with a width of 5 milliseconds, which is transmitted to the interrupt pin of the feature generation unit. After receiving the interrupt, the feature generation unit exits the standby state, begins to acquire the recognition results of the most recent 300 sampling moments, and performs signal slicing and temporal semantic parsing. If the failure determination unit determines that it does not meet the requirements again in a subsequent sampling period, it disables the enable signal again, but does not interrupt the ongoing feature generation process. If it has already started, it continues to complete the process.

[0125] Furthermore, the data acquisition unit consists of multiple potential acquisition channels, which take turns acquiring the potential signals of each sensor in a time-division multiplexing manner, and arrange the acquired potential signals according to the sensor position coordinates to form the potential distribution data;

[0126] The multiple potential acquisition channels are configured on different PCB control modules of the system. Each PCB control module is provided with at least one potential acquisition channel. The time-division multiplexing method is executed by the channel selection controller inside the data acquisition unit. The channel selection controller selects each potential acquisition channel in sequence according to a fixed channel polling order. It completes one potential signal sampling in each selection cycle and stores the sampled value in the buffer unit corresponding to the sensor coordinate. The data in all buffer units together form the potential distribution data.

[0127] The polling order of the channel selection controller is arranged according to the physical location of each sensor, so that the collected potential distribution data can be directly mapped to the spatial coordinate grid of the offshore wind power foundation.

[0128] In this embodiment, the data acquisition unit has 12 potential acquisition channels, each corresponding to a silver / silver chloride reference electrode, numbered S1 to S12. These electrodes are evenly distributed along the circumference of the wind turbine foundation steel piles, with each electrode spaced 30° apart. The data acquisition unit is equipped with a channel selection controller, which is a loop counter that sequentially selects channels S1, S2, ..., S12 in a fixed polling order, and then returns to S1.

[0129] After each channel is selected, the sampling and holding circuit remains stable for 1 millisecond, then performs analog-to-digital conversion at a resolution of 16 bits, with a conversion time of 0.5 milliseconds. Therefore, the sampling period for each channel is 1.5 milliseconds. The total time for polling all 12 channels is 18 milliseconds, corresponding to a frame rate of approximately 55.6 Hz. However, the overall system sampling rate is set to 10 Hz, so a complete frame of data is actually output every 100 milliseconds, with the controller buffering the most recent round of data. Each PCB control module is configured with 4 potential acquisition channels, totaling 12 channels across 3 modules. Module A is responsible for S1-S4, Module B for S5-S8, and Module C for S9-S12. The polling order of the channel selection controller is arranged according to the physical location of the electrodes, i.e., starting from due north and proceeding clockwise. This ensures that the acquired potential values, after being stored in this order, can be directly mapped to a spatial coordinate grid. For example, position 1 is mapped to angle 0°, position 2 to 30°, and so on.

[0130] The controller writes the sampled values ​​into a 12×1 buffer array, with array indices corresponding to sensor numbers. This array is then expanded into an 8×12 potential distribution matrix, where 8 rows are virtual rows (only one row of data is actually valid). However, to ensure consistency with subsequent processing, the data is copied to all rows, forming a uniform distribution matrix. This matrix is ​​then transmitted to the state assessment unit via the data bus. When a working module switches, the new module takes over its assigned acquisition channel, while the channel selection controller maintains the same polling order and timing to ensure that the spatial mapping of the potential distribution data remains consistent.

[0131] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, 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 system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0132] 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 units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0133] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A modular redundant PCB control system for offshore wind power cathodic protection, characterized in that, It consists of a data acquisition unit, a status assessment unit, a connected component expansion unit, a failure determination unit, a feature generation unit, and a redundancy switching decision unit; The data acquisition unit is connected to the cathodic protection potential sensor of the offshore wind power foundation structure to collect potential distribution data. The status assessment unit receives the potential distribution data, feeds the potential distribution data into the cathodic protection status assessment model, and obtains the identification results of the protection potential abnormal area. The connected domain expansion unit performs connected domain expansion processing on the abnormal potential region identification result of the protection potential abnormal region to generate expanded abnormal region data. The failure determination unit determines whether the protection failure early warning triggering conditions are met based on the expanded abnormal area data. When the failure determination unit determines that the result is consistent, the feature generation unit performs signal slicing and temporal semantic parsing on the identification result of the abnormal protection potential region to generate a system early warning candidate feature set. The redundancy switching decision unit imports the system early warning candidate feature set into the protection system early warning decision model and outputs modular redundancy switching control commands.

2. The system according to claim 1, characterized in that, The abnormal potential region connection domain expansion process performed by the connection domain expansion unit adopts the four-neighbor connection algorithm to expand the neighborhood of each abnormal pixel in the protection potential abnormal region identification result, merges the adjacent abnormal pixels into the same connection region, and expands the outer boundary of each connection region outward based on the preset expansion step size to form an expanded abnormal region. The expansion step size is dynamically determined by the connected region expansion unit based on the potential gradient magnitude of the boundary pixels of the current connected region. When the boundary gradient magnitude exceeds a set threshold, a smaller expansion step size is used, and when the boundary gradient magnitude is lower than a set threshold, a larger expansion step size is used. After completing the independent expansion of each connected region, the connected domain expansion unit performs a merging operation on the overlapping expanded regions, merging the overlapping parts into an overall abnormal region. When merging overlapping regions, the connected domain expansion unit uses the geometric center of the overlapping region as the center of the merged region and recalculates the outer boundary of the merged region.

3. The system according to claim 2, characterized in that, The failure determination unit determines whether the protection failure early warning triggering conditions are met based on the expanded abnormal area data, specifically: Calculate the area ratio of the expanded abnormal region in the potential distribution data, and calculate the rate of change of the area ratio relative to the previous sampling time. When the area ratio exceeds a predetermined threshold and the rate of change shows an increasing trend, it is determined to meet the protection failure early warning triggering condition. The determination of the growth trend is based on the sign of the rate of change at multiple consecutive sampling times. If the rate of change at three consecutive sampling times is positive, it is considered to be a growth trend. The predetermined threshold is dynamically adjusted by the failure determination unit based on the baseline area ratio obtained during the initial system calibration. The baseline area ratio is the statistical average of the abnormal area ratio in the unprotected failure state of the system. In the calculation of the rate of change, if the area of ​​the abnormal region at the previous sampling time is zero, the rate of change is based on the area ratio at the current sampling time.

4. The system according to claim 3, characterized in that, The signal slicing process performed by the feature generation unit divides the protection potential abnormal region identification results into multiple consecutive time slices according to the time dimension. Each time slice corresponds to a sequence of abnormal region identification results within a fixed time window. The duration of the fixed time window is determined according to the sampling period of the potential distribution data. Each time slice contains abnormal region identification results at multiple sampling times. The temporal semantic parsing process performed by the feature generation unit extracts the spatial location change trajectory, geometric morphology evolution features, center point displacement vector, total area change curve, and number change curve of the abnormal region within each time slice for the abnormal region identification result sequence. The spatial position change trajectory is obtained by recording the coordinate sequence of the centroid of the abnormal region at each sampling moment. The geometric morphological evolution features are obtained by calculating the aspect ratio of the bounding rectangle of the abnormal region, the perimeter of the region, and the compactness of the region at each sampling moment. The displacement vector is obtained by the difference in centroid coordinates between adjacent sampling moments. The area change curve is obtained by counting the total number of effective level pixels in the abnormal region at each sampling moment. The quantity change curve is obtained by counting the number of independent connected regions at each sampling moment. The feature generation unit concatenates the extraction results of all time slices in chronological order to form the system early warning candidate feature set, which consists of anomaly region location sequence, morphology sequence, displacement vector sequence, area change sequence, and quantity change sequence. After generating the system early warning candidate feature set, the feature generation unit transmits the system early warning candidate feature set to the redundancy switching decision unit in a structured data format, wherein the structured data format is a set of key-value pairs.

5. The system according to claim 4, characterized in that, The modular redundancy switching control command consists of the switching enable signal for switching from the main control PCB module to the backup PCB module and the startup timing configuration parameters of the backup PCB module. The switching enable signal is a high-level active digital pulse signal, used to trigger the output shutdown of the current working module and the output turn-on of the standby module. The startup timing configuration parameters include the power supply establishment delay of the backup module, the clock synchronization calibration delay, and the control parameter loading delay. The redundancy switching decision unit outputs multiple control sub-signals in sequence according to the startup timing configuration parameters to complete the startup process of the backup module in sequence. The startup timing configuration parameters of the backup module also include the self-test status waiting time and the output stabilization waiting time of the backup module.

6. The system according to claim 5, characterized in that, The system has multiple redundant PCB control modules. Each PCB control module is independently configured with a cathodic protection potential signal acquisition channel and a potential control output channel. The switching control command output by the redundancy switching decision unit is used to select one of the PCB control modules as the current working module, and the remaining PCB control modules as hot standby modules. Each PCB control module has an independent power management submodule and a communication interface submodule. The PCB control modules are interconnected through a high-speed backplane bus. The hot standby module continuously receives potential distribution data but does not output control signals in standby mode. When the redundancy switching decision unit outputs a switching control command, it simultaneously sends a shutdown command to the currently working module and an activation command to the selected standby module. The activation command contains the initial operating parameters of the standby module. The hot standby module maintains a high impedance output port in standby mode to avoid interfering with the output of the currently working module.

7. The system according to claim 6, characterized in that, The data acquisition unit is connected to the potential signal acquisition channel of each PCB control module through a synchronous serial interface, and the status evaluation unit obtains the potential distribution data from the data acquisition unit through a high-speed data bus. The condition assessment unit has built-in the cathodic protection condition assessment model, which outputs abnormal area identification results based on potential distribution data. The abnormal area identification results are represented in the form of a binary image. The connected component expansion unit receives the binarized image and performs the connected component expansion process on the connected regions marked with effective level to generate an expanded binarized image. The failure determination unit receives the expanded binarized image, counts the total number of pixels with valid level markers, and determines whether the protection failure warning trigger condition is met based on the ratio of the total number of pixels to the total number of pixels in the binarized image and the gradient of the ratio over time. The feature generation unit is activated only when the failure determination unit outputs a positive determination result. The feature generation unit obtains the original protection potential abnormal region identification result from the state evaluation unit through the data bus, and performs the signal slicing and timing semantic parsing on the identification result. After receiving the system early warning candidate feature set, the redundancy switching decision unit calls the protection system early warning decision model, which outputs a switching priority sequence corresponding to multiple redundant PCB control modules. The modular redundancy switching control command determines the target switching module based on the switching priority sequence and generates the corresponding switching control signal.

8. The system according to claim 7, characterized in that, The protection system's early warning decision model is a rule-based reasoning model based on decision trees; The redundancy switching decision unit matches the system early warning candidate feature set with the early warning rule, and the redundancy switching action corresponding to the successfully matched early warning rule is the modular redundancy switching control command. The matching process involves the redundancy switching decision unit sequentially comparing each sequence value in the system early warning candidate feature set with the feature conditions of each early warning rule. When all sequence values ​​meet the feature conditions of a certain early warning rule, the early warning rule is determined to be a successful match. The warning rules include a default rule. When no rule is matched successfully, the redundancy switching decision unit outputs a switching control command pointing to the highest priority backup module.

9. The system according to claim 8, characterized in that, When the failure determination unit determines that the expanded abnormal area data does not meet the protection failure warning triggering conditions, it maintains the working state of the currently working PCB control module, and both the feature generation unit and the redundancy switching decision unit remain in standby state, without performing signal slicing, timing semantic parsing and redundancy switching control command output operations. In the standby state, the feature generation unit stops slicing and parsing the identification results of the abnormal protection potential region, and the redundancy switching decision unit stops calling the protection system early warning decision model and generating switching control commands. The failure determination unit repeatedly performs the determination operation in each sampling period until the determination result meets the triggering condition. Then, it sends a start enable signal to the feature generation unit. After receiving the start enable signal, the feature generation unit starts to perform signal slicing and temporal semantic parsing. When the failure determination unit calculates the ratio and change gradient in each sampling period, it only relies on the expanded abnormal area data in the current sampling period and does not depend on historical data caching.

10. The system according to claim 9, characterized in that, The data acquisition unit consists of multiple potential acquisition channels. The potential acquisition channels acquire the potential signals of each sensor in turn in a time-division multiplexing manner, and arrange the acquired potential signals according to the sensor position coordinates to form the potential distribution data. The multiple potential acquisition channels are configured on different PCB control modules of the system. Each PCB control module is provided with at least one potential acquisition channel. The time-division multiplexing method is executed by the channel selection controller inside the data acquisition unit. The channel selection controller selects each potential acquisition channel in sequence according to a fixed channel polling order. It completes one potential signal sampling in each selection cycle and stores the sampled value in the buffer unit corresponding to the sensor coordinate. The data in all buffer units together form the potential distribution data. The polling order of the channel selection controller is arranged according to the physical location of each sensor, so that the collected potential distribution data can be directly mapped to the spatial coordinate grid of the offshore wind power foundation.