Fishing-light complementary power station grounding system corrosion state real-time diagnosis method and system based on double-layer information fusion and intelligent prediction

Abstract

The invention discloses a fishery light complementary power station grounding system corrosion state real-time diagnosis method based on double-layer information fusion and intelligent prediction, and belongs to the field of power equipment state monitoring and predictive maintenance. Comprising the following steps: synchronously acquiring environment data of an environment where a grounding system is located, real-time corrosion rate data of a grounding body and grounding resistance data obtained by adopting a pulse type synchronous measurement technology; based on the collected data, calculating the real-time health degree SOH of the grounding system through a first-layer weighted fusion model; based on the historical SOH sequence, the future health trend of the grounding system is predicted through a long short-term memory (LSTM) network model, and the remaining life RUL of the grounding system is calculated. According to the invention, through anti-interference measurement and double-layer fusion analysis, real-time, multi-dimensional and accurate diagnosis and prospective prediction of the corrosion state of the fishing-light complementary power station grounding system are realized.

Description

Technical Field

[0001] This invention belongs to the field of power equipment condition monitoring and predictive maintenance technology, and in particular relates to a method and system for real-time diagnosis of corrosion status of grounding system of solar-fishery hybrid power station based on dual-layer information fusion and intelligent prediction. Background Technology

[0002] Grounding systems are a crucial component for ensuring the safe and stable operation of power facilities, especially important for solar-aquaculture power stations built in aquatic environments. Their grounding electrodes are constantly exposed to harsh environments such as high humidity and high salinity, making them susceptible to electrochemical corrosion, leading to deterioration of electrical performance and threatening the safety of the power station. Currently, the monitoring and maintenance of grounding systems mainly rely on the following existing technologies: First, regular manual inspections and excavation checks, using tools such as grounding resistance meters for measurement and visual inspection. While this method can detect existing problems, it is time-consuming, inefficient, and a reactive measure. Second, single-function online monitoring devices can automatically collect and upload specific parameters such as grounding resistance, avoiding the inconvenience of manual measurement. However, their monitoring dimensions are limited, typically focusing only on the electrical parameters themselves. Third, in the field of corrosion monitoring, there are electrochemical sensors based on principles such as the plate method (calculating the average corrosion rate by weighing) or the linear polarization resistance method, which can be used to obtain corrosion information. However, the former provides data with significant lag, and the latter is mostly used in laboratories or for monitoring isolated points.

[0003] However, the aforementioned existing technologies still have significant shortcomings when applied to the condition management of grounding systems in solar-aquaculture hybrid power plants. First, monitoring methods are passive and lagging, failing to achieve real-time, continuous perception of corrosion status, and lacking timely data support for operation and maintenance decisions. Second, the various monitoring technologies are fragmented, presenting a "data silo" state. Relying solely on isolated grounding resistance, environmental, or corrosion data makes it difficult to reveal the complete fault chain of "environment-driven corrosion process, corrosion process leading to electrical performance degradation," resulting in a one-sided diagnostic dimension and an inability to conduct accurate root cause analysis. Third, existing methods are entirely based on current or historical conditions, lacking the ability to predict corrosion trends and the remaining lifespan of the system. This forces operation and maintenance plans to rely on fixed cycles or post-event remediation, failing to achieve proactive predictive maintenance. Finally, in the complex electromagnetic environment of solar-aquaculture hybrid power plants (especially with DC components and stray currents), traditional grounding resistance measurement methods are susceptible to interference, leading to distortion of key basic data and affecting the reliability of any higher-level condition assessments. Therefore, how to achieve real-time, multi-dimensional, accurate diagnosis and proactive prediction of the corrosion status of grounding systems in solar-aquaculture hybrid power plants has become an urgent technical problem to be solved in this field. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a real-time diagnostic method and system for the corrosion status of a solar-fishery hybrid power station grounding system based on dual-layer information fusion and intelligent prediction, thereby resolving the issues present in the prior art.

[0005] To achieve the above objectives, in a first aspect, the present invention provides a real-time diagnostic method for the corrosion status of a solar-fishery hybrid power station grounding system based on dual-layer information fusion and intelligent prediction, comprising:

[0006] S1. Synchronously collect environmental data of the grounding system's environment, real-time corrosion rate data of the grounding body, and grounding resistance data; wherein, the grounding resistance data is obtained using pulse-type synchronous measurement technology;

[0007] S2. Based on the environmental data, real-time corrosion rate data, and grounding resistance data collected in step S1, the real-time health of the grounding system is calculated using the first-layer weighted fusion model.

[0008] S3. Based on the historical real-time health sequence, predict the future health trend sequence of the grounding system using a Long Short-Term Memory (LSTM) network model, and calculate the remaining lifetime of the grounding system.

[0009] Preferably, in step S1, the collected environmental data includes salt spray concentration, ambient temperature, and relative humidity; leakage current data of the grounding system is also collected.

[0010] Preferably, the pulsed synchronous measurement technology specifically involves injecting a current pulse signal of a specific frequency into the grounding grid and synchronously detecting the corresponding voltage response to calculate the grounding resistance.

[0011] Preferably, the calculation process of the first layer weighted fusion model in step S2 includes:

[0012] S21. Calculate the environmental corrosion index based on the environmental data;

[0013] S22. Calculate the Electrical Performance Index (EPI) based on the grounding resistance data and leakage current data;

[0014] S23. The real-time health status is calculated by weighting and fusing the environmental corrosivity index, the real-time corrosion rate, and the electrical performance index.

[0015] Preferably, in step S23, the weighted fusion model is:

[0016] SOH real−time =α*(1−ECI)+β*(1−V corr )+γ*EPI;

[0017] Where α, β, and γ are weighting coefficients, and α + β + γ = 1, V corr This represents the standardized real-time corrosion rate.

[0018] Preferably, in step S3, calculating the remaining lifetime of the grounding system specifically involves: determining the time point at which the system first falls below a preset failure threshold based on the future health trend curve predicted by the Long Short-Term Memory (LSTM) network model, and the time difference between this time point and the current time point is the remaining lifetime.

[0019] Preferably, the method further includes step S4, generating graded early warning information or predictive maintenance work orders based on the real-time health status and the remaining lifespan.

[0020] Secondly, the present invention also provides a real-time corrosion status diagnosis system for a solar-fishery hybrid power station grounding system based on dual-layer information fusion and intelligent prediction, used to implement the method described in the first aspect, including:

[0021] The field sensing layer includes an environmental sensor for collecting environmental data, a corrosion rate sensor for collecting real-time corrosion rate data, and an online grounding resistance monitor using pulse synchronous measurement technology.

[0022] The data transmission layer is used to transmit the data from the field perception layer to the cloud-based intelligent analysis platform via wireless communication.

[0023] The cloud-based intelligent analysis platform is configured to perform the calculation of the first-layer weighted fusion model and the prediction of the Long Short-Term Memory (LSTM) network model.

[0024] Preferably, the online grounding resistance monitoring instrument includes a main unit, a current electrode, and a voltage electrode; the current electrode and voltage electrode are arranged around the grounding grid, and the main unit is connected to the current electrode and voltage electrode through a measuring cable and is configured to perform the pulse synchronous measurement technology.

[0025] Preferably, the input of the Long Short-Term Memory (LSTM) network model is a time-series data window containing historical real-time health status, historical environmental corrosion index, and historical real-time corrosion rate, and the output is a health status prediction sequence for a future period of time.

[0026] Compared with the prior art, the present invention has the following advantages and technical effects:

[0027] This invention constructs a two-layer information fusion architecture that tightly integrates "real-time diagnosis (SOH calculation)" and "trend prediction (RUL calculation)." This architecture first fuses multi-source real-time data into a comprehensive health score (SOH), and then, based on reliable historical SOH sequences, uses a Long Short-Term Memory (LSTM) network model to predict future health trends and quantify remaining lifespan (RUL). This technical feature transforms the system output from simple data alarms to maintenance decision-making criteria with clear time windows, thereby achieving a fundamental shift in the operation and maintenance paradigm from passive response and periodic prevention to proactive prediction. This allows for the scientific formulation of maintenance plans, avoids over-maintenance or under-maintenance, and effectively optimizes the total lifecycle maintenance cost.

[0028] This invention simultaneously collects multi-dimensional data such as environmental conditions, corrosion rate, and grounding resistance, and calculates a unified health index (SOH) through a first-layer weighted fusion model. This technical feature breaks down data silos between different monitoring dimensions. Through the algorithm model, it inherently simulates the physical relationship between "harsh environment driving accelerated corrosion, and corrosion process leading to electrical performance degradation." This allows for the identification of different root causes, such as environmental corrosion, connection faults, or insulation degradation, from phenomena such as changes in resistance values. This provides a direct basis for judgment to take precise maintenance measures (such as anti-corrosion treatment, fastening connections, or insulation repair).

[0029] This invention employs a technical solution where the grounding resistance data is obtained using pulsed synchronous measurement technology. This addresses the complex electromagnetic environment (especially DC interference) of solar-fishery hybrid power plants by implementing source-level anti-interference measures. This technical feature ensures the accuracy of the acquisition of the core electrical parameter, grounding resistance, overcoming the shortcomings of traditional methods that are prone to distortion under strong interference. This fundamentally guarantees the reliability and credibility of all subsequent fusion calculations, condition diagnostics, and long-term predictions based on this data.

[0030] This invention constructs a complete closed loop from synchronous data acquisition by on-site sensors and data transmission to real-time fusion analysis in the cloud. This technology enables 24 / 7 uninterrupted status monitoring, instantly capturing transient and gradual information such as sudden changes in corrosion rates, increased environmental corrosivity, or early degradation of electrical performance. It completely changes the traditional passive mode that relies on periodic inspections and manual measurements, significantly shortening the time from the occurrence of an anomaly to its detection, and providing the possibility for timely intervention. Attached Figure Description

[0031] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0032] Figure 1 This is a flowchart illustrating the overall method of an embodiment of the present invention;

[0033] Figure 2 This is a flowchart of the real-time corrosion status diagnosis method for the grounding system of a solar-fishery hybrid power station according to an embodiment of the present invention.

[0034] Figure 3 This is a schematic diagram of a real-time corrosion diagnosis system for the grounding system of a solar-fishery hybrid power station, according to an embodiment of the present invention. Detailed Implementation

[0035] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0036] It should be noted that the steps shown in the flowchart in 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 may be executed in a different order than that shown here.

[0037] The technical terms used in the following embodiments will be explained first.

[0038] 1. Pulse-type synchronous measurement technology

[0039] Explanation: The core technology employed in the online grounding resistance monitoring instrument of this invention is that instead of continuously applying a measurement signal, it emits a brief, specific current pulse and simultaneously detects the corresponding voltage response.

[0040] This technology is analogous to using short, specific commands (pulses) in a noisy room (a complex electromagnetic interference environment) instead of continuously shouting, and focusing on listening to the corresponding echoes (synchronous detection). This effectively eliminates background noise (interference such as DC current) and enables accurate measurements. This is the cornerstone of ensuring data accuracy.

[0041] 2. Electrochemical corrosion rate sensor

[0042] Explanation: A sensor based on electrochemical principles (such as the linear polarization resistance method, LPR) can directly and in real time measure the instantaneous corrosion rate of metals, typically in millimeters per year (mm / a). It reflects the rate of corrosion "at this moment," rather than requiring a long time to obtain an average rate as in traditional plate methods.

[0043] 3. Salt spray concentration sensor

[0044] Explanation: Used to directly measure the concentration of tiny salt droplets (salt spray) in the air. Salt spray is one of the most significant environmental factors accelerating metal corrosion.

[0045] 4. Leakage current sensor

[0046] Explanation: A device worn on cables or grounding wires to detect minute currents that should not be present, leaking from lines or equipment into the ground. Excessive leakage current is an early sign of insulation deterioration or equipment failure.

[0047] 5. Two-layer information fusion

[0048] Explanation: The core architecture of this invention refers to a model that uses two levels and different strategies to comprehensively analyze multi-source data.

[0049] The first layer of fusion (real-time diagnostics) focuses on rapidly integrating multi-source data (environment, corrosion, electrical) at the current moment to derive a State of Health (SOH) score. It is a "snapshot" diagnostic.

[0050] The second layer of fusion (trend prediction) focuses on deep learning of historical time series data to identify patterns and predict future health trends. It's a "movie" prediction.

[0051] 6. Environmental Corrosion Index (ECI)

[0052] Explanation: A single quantitative index calculated by an algorithm that integrates multiple environmental parameters such as salt spray, humidity, and temperature. It represents the overall corrosiveness of an environment; the higher the value, the more severe the environment.

[0053] 7. Electrical Performance Index (EPI)

[0054] Explanation: A single quantitative index calculated by combining two key electrical parameters, grounding resistance and leakage current, using an algorithm. It represents the electrical safety status of a grounding system; the higher the value, the safer it is.

[0055] 8. State of Health (SOH)

[0056] Explanation: This is an abbreviation for State of Health. It is a comprehensive score between 0 and 1 calculated using a weighted fusion model, and is the final real-time diagnostic output of this invention. 1 represents a brand-new, perfect state, and 0 represents complete failure. It intuitively reflects the "health score" of the grounding system.

[0057] 9. LSTM model / Long Short-Term Memory network

[0058] Explanation: A special type of recurrent neural network, it is a model of deep learning. It excels at processing and predicting time series data. Internally, it has a complex "gate" structure that allows it to learn and remember long-term dependencies in the data.

[0059] Traditional models might only look at data from the last few days to predict tomorrow, while LSTM can remember and take into account the impact of a similar event from several months ago on the present. This is crucial for erosion, a process with strong "cumulative" and "historically correlated" characteristics.

[0060] 10. Remaining Lifetime (RUL)

[0061] Explanation: This is an abbreviation for Remaining Useful Life, a key metric in the field of predictive maintenance. In this invention, it specifically refers to the predicted time remaining from the current moment until the State of Health (SOH) of the grounding system drops to the failure threshold. For example, "remaining useful life of 180 days" provides maintenance personnel with a clear time window to plan maintenance.

[0062] Example 1

[0063] like Figure 1-2 As shown in the figure, this embodiment provides a real-time diagnostic method for the corrosion status of a solar-fishery hybrid power station grounding system based on dual-layer information fusion and intelligent prediction, including:

[0064] S1. Synchronously collect environmental data of the grounding system's environment, real-time corrosion rate data of the grounding body, and grounding resistance data; wherein, the grounding resistance data is obtained using pulse-type synchronous measurement technology;

[0065] Furthermore, in step S1, the collected environmental data includes salt spray concentration, ambient temperature, and relative humidity; leakage current data of the grounding system is also collected.

[0066] Furthermore, the pulsed synchronous measurement technology specifically involves injecting a current pulse signal of a specific frequency into the grounding grid and simultaneously detecting the corresponding voltage response to calculate the grounding resistance.

[0067] Specifically, pulse-based synchronous measurement technology injects a current pulse signal of a specific frequency and polarity into the grounding grid and simultaneously detects the corresponding voltage response. Its core advantage lies in its ability to effectively distinguish and separate the measured pulse signal from slowly changing DC interference and power frequency interference in the environment.

[0068] The "pulse-type" technology effectively suppresses interference from DC components and stray currents in solar-fishery hybrid power plants, ensuring the accuracy of grounding resistance measurement—a critical parameter—from the source. This provides a unique and reliable data guarantee for upper-level fusion analysis and intelligent prediction. This technology lays the foundation for the high reliability and credibility of the entire system's intelligent diagnostic results, ensuring the scientific nature of operation and maintenance decisions.

[0069] This embodiment fundamentally solves the problem of measurement inaccuracies in complex electromagnetic environments by adopting "pulse-type" synchronous measurement technology, providing a solid and reliable data foundation for upper-level intelligent diagnosis and prediction. This ensures the scientific nature, accuracy, and feasibility of all final diagnostic conclusions and operation and maintenance decisions, greatly improving the reliability of the entire power plant's intelligent operation and maintenance system.

[0070] Step S1 specifically includes:

[0071] S101: Synchronous data acquisition. The system controls a sensor network distributed throughout the power plant, synchronously acquiring the following data:

[0072] (1) Environmental data: collected by temperature, humidity and salt spray concentration sensors.

[0073] (2) Corrosion data: The instantaneous corrosion rate of the grounded flat steel is directly collected by an electrochemical corrosion rate sensor.

[0074] (3) Electrical data: The grounding resistance value after anti-interference is collected by the grounding resistance online monitoring instrument using the "pulse" synchronous measurement technology; the line insulation status data is collected by the leakage current sensor.

[0075] Data collection cycle: Automatically perform a full-site data collection once according to a preset strategy (such as every 5 minutes).

[0076] S102: Data cleaning and alignment;

[0077] Data cleaning: Algorithms such as interquartile range are applied to automatically identify and remove outliers caused by momentary sensor failures or communication interference.

[0078] Time alignment: Since the sampling frequencies of each sensor may be different, linear interpolation is used to unify all data onto the same timestamp sequence to ensure data consistency in the time dimension.

[0079] S3: Feature standardization.

[0080] The Z-Score normalization method is used to convert raw data of different dimensions and magnitudes (such as resistance in "ohms" and salt spray in "μg / m³") into dimensionless values.

[0081] The feature standardization formula is: ,in The original value, The sample mean. The standard deviation is denoted as .

[0082] The main purpose of feature standardization is to eliminate the influence of dimensions between features and ensure the fairness and accuracy of subsequent fusion models.

[0083] S2. Based on the environmental data, real-time corrosion rate data, and grounding resistance data collected in step S1, the real-time health of the grounding system is calculated using the first-layer weighted fusion model.

[0084] Furthermore, the calculation process of the first layer weighted fusion model in step S2 includes:

[0085] S21. Calculate the environmental corrosion index based on the environmental data;

[0086] S22. Calculate the Electrical Performance Index (EPI) based on the grounding resistance data and leakage current data;

[0087] S23. The real-time health status is calculated by weighting and fusing the environmental corrosivity index, the real-time corrosion rate, and the electrical performance index.

[0088] Specifically, the first layer of fusion analysis (real-time diagnosis) includes: the data fusion and preprocessing module receives the raw data and performs cleaning, alignment and standardization; the environmental corrosion index calculation unit calculates the ECI index based on the environmental data; the electrical performance index calculation unit calculates the EPI index based on the resistance and leakage current data; and the weighted fusion unit fuses the ECI, corrosion rate and EPI to output the real-time health status (SOH).

[0089] Step S2 involves fusing multi-source, heterogeneous real-time data into a single, intuitive, and comprehensive health indicator to achieve status diagnosis. Specifically, this includes:

[0090] S201: Calculation of Environmental Corrosion Index (ECI). Purpose: To quantify the instantaneous corrosion stress intensity of the environment on the grounding electrode.

[0091] The calculation formula is: ECI = w1 * C salt +w 2* RH+w 3* ;

[0092] C salt RH and T represent the standardized salt spray concentration, relative humidity, and temperature, respectively.

[0093] w1, w2, w3: weighting coefficients, which can be determined by principal component analysis (PCA) or by multiple linear regression with historical corrosion rate data.

[0094] S202: Calculation of Electrical Performance Index (EPI) Model. Purpose: To quantify the electrical safety status of a grounding system.

[0095] The calculation formula is: EPI=f(R) gnd ,I leak );

[0096] The formula function is a logic function, when the grounding resistance R gnd Excessive or leakage current I leak In abnormal situations, the EPI index drops sharply.

[0097] S203: Calculate Real-Time Health Status (SOH). Purpose: To obtain a comprehensive health score.

[0098] The formula for the weighted fusion model is:

[0099] SOH real−time =α*(1−ECI)+β*(1−V corr )+γ*EPI;

[0100] Among them, V corr The value represents the standardized real-time corrosion rate. α, β, and γ are the fusion weights (α+β+γ=1), determined using the AHP (Analytic Hierarchy Process) to reflect the contribution of different indicators to the overall health. SOH real−time Output a value between 0 and 1, where the closer to 1, the healthier the product.

[0101] S3. Based on the historical real-time health sequence, predict the future health trend sequence of the grounding system using a Long Short-Term Memory (LSTM) network model, and calculate the remaining lifetime of the grounding system.

[0102] Further, in step S3, calculating the remaining lifetime of the grounding system specifically involves: determining the time point at which the system first falls below a preset failure threshold based on the future health trend curve predicted by the Long Short-Term Memory (LSTM) network model, and the time difference between this time point and the current time point is the remaining lifetime.

[0103] Specifically, the second layer of fusion—LSTM-based time series prediction and remaining lifetime prediction—aims to predict the future health status and development trend of the grounding system by analyzing historical data sequences, and finally output a quantitative indicator with direct guiding significance—remaining lifetime (RUL).

[0104] Corrosion of grounding systems is a typical time-dependent process. The current corrosion state is not only influenced by today's environment, but also closely related to the environmental history and cumulative corrosion effects of the past few days, weeks, or even months. This dependence can be long-term.

[0105] Traditional neural networks or regression models struggle to capture such long-term dependencies. Long Short-Term Memory (LSTM) networks, as a special type of recurrent neural network (RNN), employ internal gating mechanisms (input gate, forget gate, output gate) that selectively memorize and forget information. They excel at learning and predicting long-term patterns and trends in time series, perfectly matching the physical characteristics of the corrosion process.

[0106] Detailed construction process of LSTM prediction model:

[0107] Step S301: Define the input features and the output target;

[0108] Input characteristics: In this embodiment, time series data strongly correlated with the corrosion process are selected as the model input. For a given prediction time point t, the model input is a multi-dimensional time series window containing data from the past N time steps:

[0109] X t =[vector1, vector2, ..., vector N ].

[0110] Each vector i Includes: vectors i =[SOH(t−i), ECI(t−i), V corr [(t−i)];

[0111] SOH(t−i): Overall health status at time t−i.

[0112] ECI(ti): Environmental corrosivity index at time t−i.

[0113] V corr (t−i): The normalized corrosion rate at time t−i.

[0114] Output objective: The model's task is to predict the SOH values ​​at the next M time steps, thereby depicting the future health trajectory.

[0115] Y t =[SOH(t+1),SOH(t+2),...,SOH(t+M)]

[0116] Step S302, LSTM model prediction;

[0117] The prepared sequence data is input into the trained Long Short-Term Memory (LSTM) network, and the workflow is as follows:

[0118] (1) Input sequence: The feature vectors of the past N time steps are input into the network in chronological order.

[0119] (2) Information transmission and memory:

[0120] The data X1 from the first time step is input into LSTM unit 1, which then computes a hidden state H1 and a cell state C1. C1 can be viewed as the unit's memory, carrying long-term information learned from the first step that may be useful for future predictions.

[0121] Next, X2, H1, and C1 are input into LSTM cell 2. The "forget gate" of cell 2 determines which information to discard from C1, and the "input gate" determines which new information from X2 and H1 to store in the cell state. Finally, H2 and the updated memory C2 are output.

[0122] This process is performed sequentially, with information at each step of the sequence being encoded, filtered, and then passed to the next step. This makes the hidden state H4 of the final LSTM unit 4 actually contain condensed information of the entire input sequence (from t-1 to tN).

[0123] (3) Output prediction: The hidden state H4 of the last LSTM unit is fed into a fully connected layer, which maps the high-level features to the output dimension required in this embodiment, that is, to generate the SOH prediction sequence for the next M time steps.

[0124] Step S303: Predict the SOH curve;

[0125] (1) Data preparation: Extract a large number of consecutive [X_t, Y_t] sample pairs from the historical database.

[0126] (2) Loss function: Defines the difference between the model's predicted value and the true value. Mean squared error (MSE) is typically used.

[0127] ;

[0128] in, For the model's true value, These are the model's predicted values.

[0129] (3) Training process: Using the backpropagation algorithm and optimizers such as Adam, all parameters (weights and biases) in the LSTM model are continuously adjusted with the goal of minimizing the value of the loss function MSE. This means that the model is constantly learning how to make the predicted SOH curve as close as possible to the historically actual SOH curve.

[0130] Step S304, calculate the remaining lifetime (RUL) from the predicted curve to the remaining lifetime (RUL).

[0131] Set a failure threshold: According to engineering safety specifications, a failure threshold for SOH is preset. (e.g., 0.3). When the SOH is below this value, the grounding system is considered unsafe and requires immediate replacement or overhaul.

[0132] RUL calculation: Apply the trained LSTM model to the current time point t. now Input the data from the most recent N days to obtain the predicted SOH curve for the next M days.

[0133] Compare this prediction curve with the failure threshold Compare them.

[0134] Remaining lifetime is defined as the period from the current time point t_now to the time point t_n when the predicted curve first crosses the failure threshold. failure The time elapsed.

[0135] RUL=t failure -t now ;

[0136] This embodiment, through learning from historical patterns, allows the model to extrapolate future performance degradation trajectories and present them in the intuitive and quantifiable form of "remaining lifespan." This enables managers to clearly know "how long a grounding electrode can still be used." This embodiment achieves an upgrade from "preventive maintenance" to "predictive maintenance," enabling the scientific formulation of repair and replacement plans, significantly reducing total lifecycle maintenance costs, and maximizing equipment utilization.

[0137] In this embodiment, the second-layer fusion analysis (trend prediction) includes: a time-series data processing unit organizes historical data to form training samples; an LSTM prediction model learns historical patterns and predicts future SOH trends; and a lifetime assessment unit calculates the remaining lifetime (RUL) based on the prediction curve.

[0138] S4. Generate graded early warning information or predictive maintenance work orders based on the real-time health status and the remaining lifespan.

[0139] Specifically, the diagnostic and predictive results are translated into concrete operational and maintenance action instructions, including:

[0140] S401, Real-time Status Alert:

[0141] The system is based on the real-time calculated SOH real−time Implement tiered early warning systems:

[0142] Healthy (SOH> 0.8): Normal condition.

[0143] Note (0.6 < SOH ≤ 0.8): The system records and prompts attention.

[0144] Warning (0.4 < SOH ≤ 0.6): An early warning is issued, and an inspection is recommended.

[0145] Hazard (SOH ≤ 0.4): An emergency alarm is issued, requiring immediate action.

[0146] S402, Predictive Maintenance Early Warning:

[0147] The system generates predictive maintenance work orders based on the predicted RUL value.

[0148] For example, when the predicted RUL is less than the preset safety period (e.g., 180 days), the system automatically pushes an alarm and suggestion: "Warning: The estimated remaining life of grounding electrode No. 3 in area A is 150 days. It is recommended to carry out anti-corrosion reinforcement treatment within 90 days."

[0149] S403, Decision Support and Visualization:

[0150] All diagnostic results, prediction curves, early warning information, and maintenance recommendations are centrally displayed on the cloud platform's visual interface.

[0151] Operations and maintenance personnel can view the data in real time through a web browser or mobile app, enabling data-driven scientific decision-making.

[0152] Example 2

[0153] like Figure 3 As shown, this embodiment provides a real-time corrosion status diagnosis system for a solar-fishery hybrid power station grounding system based on dual-layer information fusion and intelligent prediction, used to implement the method described in Embodiment 1, including:

[0154] The field sensing layer includes an environmental sensor for collecting environmental data, a corrosion rate sensor for collecting real-time corrosion rate data, and an online grounding resistance monitor using pulse synchronous measurement technology.

[0155] Furthermore, the online grounding resistance monitoring instrument includes a main unit, a current electrode, and a voltage electrode; the current electrode and voltage electrode are arranged around the grounding grid, and the main unit is connected to the current electrode and voltage electrode through a measuring cable and is configured to perform the pulse synchronous measurement technology.

[0156] Specifically, the components of the field sensing layer include: corrosion rate sensor, grounding resistance online monitoring instrument, environmental sensors (temperature and humidity sensor, salt spray concentration sensor), leakage current sensor and data acquisition unit.

[0157] A corrosion rate sensor is used to directly measure the instantaneous corrosion rate of grounded flat steel. The corrosion rate sensor is directly fixed to the surface of the grounded flat steel underwater or near the water surface via magnetic adsorption or a special clamp.

[0158] The core component of the online grounding resistance monitoring instrument must explicitly support "pulse-type" synchronous measurement technology, including the main unit and auxiliary measuring electrodes (current electrode and voltage electrode). The main unit of the online grounding resistance monitoring instrument is installed in the power plant operation and maintenance center. Its current electrode and voltage electrode are arranged in the soil or water surrounding the power plant grounding grid according to electrical regulations. The current electrode C is placed in the soil / water at a distance of 3-5 times the diagonal length of the grounding grid from its edge, and the voltage electrode P is placed in the middle between the current electrode and the grounding grid.

[0159] The environmental sensors include temperature and humidity sensors and salt spray concentration sensors. These sensors (temperature, humidity, and salt spray) are mounted on supports or towers that represent the atmospheric environment of the grounded electrode, avoiding direct sunlight. The temperature and humidity sensors are installed in the shaded area of ​​the power station supports or towers, 2-3 meters above the water surface. The salt spray concentration sensors are installed on supports at representative locations, avoiding direct sunlight and rain.

[0160] A leakage current sensor is used to monitor the leakage current from a line or device to ground. The leakage current sensor is mounted on the cable or grounding lead of a critical circuit.

[0161] The field sensing layer also includes a data acquisition unit, which collects, temporarily stores, and performs preliminary processing of data from each sensor. The data acquisition unit is installed in a waterproof junction box near the sensor concentration area.

[0162] The main unit of the grounding resistance online monitoring instrument is located in a standard cabinet in the power plant operation and maintenance center. It is connected to the remote measuring electrode through shielded twisted pair cable or optical fiber cable, and the power supply is taken from the uninterruptible power supply (UPS) system of the operation and maintenance center.

[0163] In this embodiment, the field sensor is connected to the power supply terminal of the data acquisition unit via a shielded cable; the power supply voltage is DC 12-24V, and it is powered by twisted pair cable or a separate power supply line.

[0164] The data transmission layer is used to transmit the data from the field perception layer to the cloud-based intelligent analysis platform via wireless communication.

[0165] Specifically, the data transmission layer components are industrial-grade wireless communication modules (such as 4G / 5G DTUs or LoRa modules), embedded in the data acquisition unit of the field sensing layer. The 4G / 5G industrial router is installed in the maintenance center cabinet, adjacent to the monitoring instrument host. Antennas are positioned on the roof of the maintenance center or in locations with good signal coverage.

[0166] The cloud-based intelligent analysis platform is configured to perform the calculation of the first-layer weighted fusion model and the prediction of the Long Short-Term Memory (LSTM) network model.

[0167] Furthermore, the input of the Long Short-Term Memory (LSTM) network model is a time-series data window containing historical real-time health status, historical environmental corrosion index, and historical real-time corrosion rate, and the output is a health status prediction sequence for a future period of time.

[0168] Specifically, the software modules deployed on the cloud server include a data fusion and preprocessing module, a two-layer information fusion analysis module, a status diagnosis and prediction module, an early warning and decision support module, a database, and a client interface. These software modules run on virtual servers or physical server clusters in the cloud.

[0169] A database is used to store historical data, model parameters, and diagnostic results.

[0170] Client-side interactive interfaces (such as web servers and app backends) are used to display results to users.

[0171] Multi-module collaborative decision-making includes: a status diagnosis and prediction module that generates health status assessment results; an early warning and decision support module that triggers early warnings of corresponding levels based on preset thresholds; a maintenance suggestion generation unit that outputs targeted maintenance plans based on diagnostic results and predicted trends; and a client interaction module that pushes the results to maintenance personnel in a visual format.

[0172] This embodiment can continuously upload the collected real-time data to the cloud-based intelligent analysis platform 24 / 7. Real-time data streams are the foundation for understanding system dynamics. Because the system can continuously sense changes in its state, it can detect abnormal signals in the initial stages of accelerated corrosion or deterioration of electrical performance, without waiting for the next scheduled inspection.

[0173] In this embodiment, the signal transmission connection includes:

[0174] Corrosion rate sensor → Data acquisition unit: 4-20mA analog signal line;

[0175] Environmental sensor → Data acquisition unit: RS-485 digital communication cable;

[0176] Leakage current sensor → Data acquisition unit: Modbus RTU communication protocol;

[0177] Measuring electrode (C / P) → Monitoring instrument main unit: Dedicated measuring cable with shielding layer.

[0178] In this embodiment, the power connection includes: monitoring instrument host → maintenance center UPS: AC 220V power line; data acquisition unit → field power: DC 24V industrial power supply.

[0179] In this embodiment, the data communication connection includes:

[0180] 1. Field layer communication:

[0181] A master-slave communication network is established between each sensor and the data acquisition unit;

[0182] Communication protocol: Modbus RTU over RS-485, baud rate 9600bps.

[0183] 2. Remote communication:

[0184] Data acquisition unit → Monitoring instrument host: Industrial Ethernet or fiber optic communication;

[0185] Monitoring unit host → Cloud platform: 4G / 5G wireless network, using TCP / IP protocol.

[0186] 3. Cloud-to-cloud layer communication:

[0187] The various software modules exchange data via RESTful APIs;

[0188] The database uses an SQL database connection;

[0189] The client interacts with the platform via HTTP / WebSocket.

[0190] Connection characteristics description: (1) Anti-interference design: all cables adopt double-layer shielding technology; (2) Waterproof and corrosion resistant: outdoor connections use IP67 protection level connectors; (3) Redundancy design: key communication links adopt dual-ring network redundant topology; (4) Security isolation: industrial firewalls are deployed between the field and the cloud.

[0191] This interconnection design ensures reliable system operation in complex electromagnetic environments, while also enabling real-time and accurate data transmission, providing a solid foundation for subsequent intelligent diagnostic analysis.

[0192] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A real-time diagnostic method for corrosion status of a solar-fishery hybrid power station grounding system based on dual-layer information fusion and intelligent prediction, characterized in that, Includes the following steps: S1. Synchronously collect environmental data of the grounding system's environment, real-time corrosion rate data of the grounding body, and grounding resistance data; wherein, the grounding resistance data is obtained using pulse-type synchronous measurement technology; S2. Based on the environmental data, real-time corrosion rate data, and grounding resistance data collected in step S1, the real-time health of the grounding system is calculated using the first-layer weighted fusion model. S3. Based on the historical real-time health sequence, predict the future health trend sequence of the grounding system using a Long Short-Term Memory (LSTM) network model, and calculate the remaining lifetime of the grounding system.

2. The method according to claim 1, characterized in that, In step S1, the collected environmental data includes salt spray concentration, ambient temperature, and relative humidity; leakage current data of the grounding system is also collected.

3. The method according to claim 1, characterized in that, The pulse-type synchronous measurement technology specifically involves injecting a current pulse signal of a specific frequency into the grounding grid and simultaneously detecting the corresponding voltage response to calculate the grounding resistance.

4. The method according to claim 2, characterized in that, The calculation process of the first layer weighted fusion model in step S2 includes: S21. Calculate the environmental corrosion index based on the environmental data; S22. Calculate the Electrical Performance Index (EPI) based on the grounding resistance data and leakage current data; S23. The real-time health status is calculated by weighting and fusing the environmental corrosivity index, the real-time corrosion rate, and the electrical performance index.

5. The method according to claim 4, characterized in that, In step S23, the weighted fusion model is: SOH real−time =α*(1−ECI)+β*(1−V corr )+γ*EPI; Where α, β, and γ are weighting coefficients, and α + β + γ = 1, V corr This represents the standardized real-time corrosion rate.

6. The method according to claim 1, characterized in that, In step S3, the calculation of the remaining lifetime of the grounding system is specifically as follows: based on the future health trend curve predicted by the Long Short-Term Memory (LSTM) network model, the time point at which it first falls below the preset failure threshold is determined, and the time difference between this time point and the current time point is the remaining lifetime.

7. The method according to claim 1, characterized in that, It also includes step S4, generating graded early warning information or predictive maintenance work orders based on the real-time health status and the remaining lifespan.

8. A real-time corrosion status diagnosis system for a solar-fishery hybrid power station grounding system based on dual-layer information fusion and intelligent prediction, used to implement the method described in any one of claims 1-7, characterized in that, include: The field sensing layer includes an environmental sensor for collecting environmental data, a corrosion rate sensor for collecting real-time corrosion rate data, and an online grounding resistance monitor using pulse synchronous measurement technology. The data transmission layer is used to transmit the data from the field perception layer to the cloud-based intelligent analysis platform via wireless communication. The cloud-based intelligent analysis platform is configured to perform the calculation of the first-layer weighted fusion model and the prediction of the Long Short-Term Memory (LSTM) network model.

9. The system according to claim 8, characterized in that, The online grounding resistance monitoring instrument includes a main unit, a current electrode, and a voltage electrode; the current electrode and voltage electrode are arranged around the grounding grid, and the main unit is connected to the current electrode and voltage electrode through a measuring cable and is configured to perform the pulse synchronous measurement technology.

10. The system according to claim 8, characterized in that, The input to the Long Short-Term Memory (LSTM) network model is a time-series data window containing historical real-time health status, historical environmental corrosion index, and historical real-time corrosion rate, and the output is a health status prediction sequence for a future period of time.

Images