A method and system for online monitoring of DC support capacitor current in a flexible DC converter valve

Abstract

This application discloses a method and system for online monitoring of the DC support capacitor current in a flexible DC converter valve. The method includes: responding to an external sampling start command, controlling a Rogowski coil to collect operating data of a selected capacitor in the flexible DC converter valve; sequentially amplifying, filtering, and detecting outliers in the current waveform data to obtain current data to be corrected; correcting the current data to be corrected based on a pre-built neural network model to obtain current data to be analyzed; analyzing the current data to be analyzed based on big data analysis technology and machine learning algorithms; and sending a generated matching current control command to the selected capacitor based on the analysis results. The method provided in this application improves the monitoring accuracy of the capacitor.

Description

Technical Field

[0001] This application relates to the field of power monitoring technology, and in particular to a method for online monitoring of the current of the DC support capacitor of a flexible DC converter valve. Background Technology

[0002] In power systems, the DC support capacitor is one of the core components of the flexible DC converter valve. As a core device in flexible DC transmission projects, the converter valve is responsible for converting AC to DC power and controlling parameters such as voltage and current. The DC support capacitor plays a crucial supporting role in this process. As an important reactive power compensation device, the DC support capacitor plays a key role in improving the power factor of the power grid, reducing line losses, and enhancing voltage quality.

[0003] However, current changes are instantaneous. Traditional capacitor monitoring methods mainly rely on manual inspection and periodic testing. This method cannot accurately measure real-time current changes in capacitors, especially when the current is small or changes rapidly. Insufficient accuracy may lead to misjudgment of the capacitor's condition, thereby affecting the stability and safety of the flexible DC converter system. Summary of the Invention

[0004] This application provides a method and system for online monitoring of the DC support capacitor current of a flexible DC converter valve, in order to solve the technical problem of how to improve the accuracy of capacitor current monitoring.

[0005] To address the aforementioned technical problems, this application provides a method for online monitoring of the DC support capacitor current in a flexible DC converter valve, comprising:

[0006] In response to an external sampling start command, the Rogowski coil is controlled to collect operating data of a selected capacitor in the flexible DC converter valve, wherein the operating data includes at least the current waveform data of the terminal block of the selected capacitor;

[0007] The current waveform data is sequentially amplified, filtered, and outlier detected to obtain the current data to be corrected.

[0008] Based on a pre-built neural network model, the current data to be corrected is processed to obtain the current data to be analyzed. The correction process is configured to eliminate the fluctuation of current amplitude-frequency characteristics caused by instantaneous current disturbance.

[0009] The current data to be analyzed is analyzed based on big data analytics and machine learning algorithms;

[0010] Based on the analysis results, the generated matching current regulation command is sent to the selected capacitor.

[0011] As one preferred embodiment, the step of controlling the Rogowski coil to acquire operating data of a selected capacitor in the flexible DC converter valve in response to an external sampling start command includes:

[0012] The charging and discharging current of the selected capacitor is collected using a Rogowski coil, and the charging and discharging current is converted into a proportional induced electromotive force.

[0013] The induced electromotive force is digitally converted to obtain a digital signal;

[0014] The digital signal is sampled to obtain current waveform data.

[0015] As one preferred embodiment, the step of sequentially amplifying, filtering, and detecting outliers in the current waveform data to obtain the current data to be corrected includes:

[0016] The current waveform data is input into an operational amplifier, and the current waveform data is amplified to a preset factor by adjusting the feedback resistor of the operational amplifier.

[0017] The amplified current waveform data is input into a filter for filtering, and the filtered signal is output.

[0018] A statistical anomaly detection method is used to identify anomalies in the filtered current waveform data, obtain anomaly values, and then delete, replace, or mark the anomaly values ​​to obtain the current data to be corrected.

[0019] As one preferred embodiment, the current data to be corrected is processed based on a pre-built neural network model to obtain the current data to be analyzed, including:

[0020] Acquire historical current waveform data, wherein the historical current waveform data includes normal current waveform data and abnormal current waveform data;

[0021] An initial neural network model for correction is constructed, and the initial neural network model is trained based on the historical current waveform data to obtain a trained neural network model.

[0022] The current data to be corrected is input into the neural network model for processing to obtain the corrected current data to be analyzed.

[0023] As one preferred embodiment, the analysis of the current data to be analyzed based on big data analytics and machine learning algorithms includes:

[0024] The waveform feature data of the current waveform data is extracted, including peak data, valley data, average data, and fluctuation range data;

[0025] An initial waveform recognition model is constructed, and the initial waveform recognition model is trained using acquired historical waveform data and corresponding anomaly labels to obtain a trained waveform recognition model.

[0026] The waveform features are input into the waveform recognition model to classify the waveform features and obtain the operating state of the selected capacitor.

[0027] Another embodiment of this application provides an online monitoring system for the DC support capacitor current of a flexible DC converter valve, characterized in that it includes:

[0028] The acquisition module is used to control the Rogowski coil to acquire the operating data of a selected capacitor in the flexible DC converter valve in response to an external sampling start command, wherein the operating data includes at least the current waveform data of the terminal block of the selected capacitor;

[0029] The preprocessing module is used to sequentially amplify, filter, and detect outliers in the current waveform data to obtain the current data to be corrected.

[0030] A correction module is used to perform correction processing on the current data to be corrected based on a pre-built neural network model to obtain the current data to be analyzed, wherein the correction processing is configured to eliminate the fluctuation of current amplitude-frequency characteristics caused by instantaneous current disturbances.

[0031] The analysis module is used to analyze the current data to be analyzed based on big data analysis technology and machine learning algorithms;

[0032] The monitoring module is used to send the generated matching current regulation command to the selected capacitor based on the analysis results.

[0033] As one preferred embodiment, the acquisition module is specifically used for:

[0034] The charging and discharging current of the selected capacitor is collected using a Rogowski coil, and the charging and discharging current is converted into a proportional induced electromotive force.

[0035] The induced electromotive force is digitally converted to obtain a digital signal;

[0036] The digital signal is sampled to obtain current waveform data.

[0037] As one preferred embodiment, the preprocessing module is specifically used for:

[0038] The current waveform data is input into an operational amplifier, and the current waveform data is amplified to a preset factor by adjusting the feedback resistor of the operational amplifier.

[0039] The amplified current waveform data is input into a filter for filtering, and the filtered signal is output.

[0040] A statistical anomaly detection method is used to identify anomalies in the filtered current waveform data, obtain anomaly values, and then delete, replace, or mark the anomaly values ​​to obtain the current data to be corrected.

[0041] As one preferred embodiment, the correction module is specifically used for:

[0042] Acquire historical current waveform data, wherein the historical current waveform data includes normal current waveform data and abnormal current waveform data;

[0043] An initial neural network model for correction is constructed, and the initial neural network model is trained based on the historical current waveform data to obtain a trained neural network model.

[0044] The current data to be corrected is input into the neural network model for processing to obtain the corrected current data to be analyzed.

[0045] As one preferred embodiment, the analysis module is specifically used for:

[0046] The waveform feature data of the current waveform data is extracted, including peak data, valley data, average data, and fluctuation range data;

[0047] An initial waveform recognition model is constructed, and the initial waveform recognition model is trained using acquired historical waveform data and corresponding anomaly labels to obtain a trained waveform recognition model.

[0048] The waveform features are input into the waveform recognition model to classify the waveform features and obtain the operating state of the selected capacitor.

[0049] Compared to the prior art, the beneficial effects of the embodiments of this application are at least one of the following:

[0050] 1) This application uses Rogowski coils to collect operating data of selected capacitors in flexible DC converter valves, especially current waveform data of terminal posts. This method has high sensitivity and accuracy and can capture subtle features of current changes.

[0051] 2) This application amplifies, filters, and detects outliers in the collected current waveform data, which can effectively remove noise and interference, improve the signal-to-noise ratio of the data, and provide a high-quality data foundation for subsequent analysis.

[0052] 3) This application utilizes a pre-built neural network model to correct the current data, which can eliminate fluctuations in the current amplitude-frequency characteristics caused by instantaneous current disturbances, further improving the accuracy and reliability of the data. Simultaneously, this application analyzes the processed current data based on big data analytics, enabling the discovery of potential patterns and characteristics within the data, providing strong support for fault prediction and diagnosis. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the online monitoring method for the DC support capacitor current of a flexible DC converter valve in one embodiment of this application.

[0054] Figure 2 This is a schematic diagram of a current acquisition device in one embodiment of this application;

[0055] Figure 3 This is a schematic diagram of an online monitoring system for the DC support capacitor current of a flexible DC converter valve in one embodiment of this application;

[0056] Figure label:

[0057] The components include: 1. Acquisition circuit board; 2. Top cover plate; 3. Elastic fastening ring; 4. Insulating housing; 5. Temperature sensing device; and 6. Energy acquisition coil. Detailed Implementation

[0058] 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 some embodiments of this application, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of this application more thorough and comprehensive. 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.

[0059] In the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0060] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components. The terms "vertical," "horizontal," "left," "right," "upper," "lower," and similar expressions used herein are for illustrative purposes only and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting this application. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0061] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used in the specification of this application is for the purpose of describing specific embodiments only and is not intended to limit the application. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0062] One embodiment of this application provides a capacitor monitoring system; for details, please refer to [link to specific documentation]. Figure 1 , Figure 1 The diagram shown is a flowchart illustrating an online monitoring method for the DC support capacitor current of a flexible DC converter valve according to one embodiment of this application, which includes steps S1-S5:

[0063] S1: In response to an external sampling start command, control the Rogowski coil to collect the operating data of the selected capacitor in the flexible DC converter valve, wherein the operating data includes at least the current waveform data of the terminal block of the selected capacitor;

[0064] According to Faraday's law of electromagnetic induction, when current flows through a conductor, a magnetic field is generated around the conductor; conversely, when the magnetic field changes, an induced electromotive force (EMF) is generated in the conductor. The Rogowski coil operates based on this principle. When a capacitor charges and discharges, the change in current at its terminals induces an EMF in the Rogowski coil. This induced EMF is proportional to the current at the capacitor's terminals and can therefore be used to reflect changes in current.

[0065] Preferably, in one embodiment of this application, the step of controlling the Rogowski coil to collect operating data of a selected capacitor in the flexible DC converter valve in response to an external sampling start command includes:

[0066] The charging and discharging current of the selected capacitor is collected using a Rogowski coil, and the charging and discharging current is converted into a proportional induced electromotive force.

[0067] The induced electromotive force is digitally converted to obtain a digital signal;

[0068] The digital signal is sampled to obtain current waveform data.

[0069] To facilitate subsequent data processing and analysis, the induced electromotive force (EMF) needs to be converted into a digital signal. This process is typically achieved using an analog-to-digital converter (ADC). An ADC can convert continuous analog signals (such as induced EMF) into discrete digital signals. In this application, the ADC converts the induced EMF output from the Rogowski coil into a digital signal for subsequent sampling and processing.

[0070] Sampling is one of the key steps in digital signal acquisition. In this application, the digital signal output from the analog-to-digital converter is sampled using a sampler to obtain a series of discrete sampling points. These sampling points reflect the changes in the current at the capacitor terminals, i.e., the current waveform data. The selection of the sampling frequency should satisfy the Nyquist sampling theorem to avoid aliasing.

[0071] For details, see Figure 2 , Figure 2 This is a schematic diagram of a current acquisition device provided in an embodiment of this application. The insulating housing is the basic structure of the current acquisition device, made of high-quality insulating material to ensure that the internal circuitry is isolated from the external environment, preventing the intrusion of moisture, dust, and corrosive substances, thereby protecting the internal circuitry from damage. The top cover is responsible for sealing the current acquisition device, protecting the internal circuitry from physical damage. It is typically made of robust and durable materials, possessing sufficient strength and rigidity to withstand external shocks and vibrations. The housing dimensions of the sensing unit are adapted to the dimensions of the capacitor terminal insulator, ensuring a tight fit against the bottom of the insulator. This design not only improves measurement accuracy but also ensures the sensing unit remains stable under various operating conditions.

[0072] A Rogowski coil is a non-contact current sensor that measures the charging and discharging current of a capacitor without breaking the circuit. Its operating principle is based on Faraday's law of electromagnetic induction; when current flows through the coil, an induced electromotive force (EMF) proportional to the current is generated. The number of turns and dimensions of the Rogowski coil are designed according to the current measurement range and accuracy requirements. More turns result in a larger induced EMF, but also increase the coil's size and weight. Therefore, it is necessary to minimize the number of turns to optimize the coil structure while meeting measurement requirements. Furthermore, to convert the induced EMF into a voltage signal, this application uses two 1Ω high-precision resistors connected in parallel on the secondary side of the Rogowski coil. This design not only improves the signal-to-noise ratio but also effectively reduces measurement errors.

[0073] The elastic fastening ring ensures that the sensing unit can be securely fastened to the bottom of the capacitor terminal insulator, while providing a certain degree of deformation capability to adapt to insulators of different sizes. This design ensures that the sensing unit remains stable under various operating conditions.

[0074] The acquisition circuit board is the core component of the current acquisition device, integrating circuits for current acquisition, signal conditioning, and communication. These circuits work together to achieve digital acquisition and processing of the capacitor charging and discharging current waveform and the temperature of the bottom shell of the terminals. In one embodiment of this application, the acquisition circuit board includes an STM32 microcontroller.

[0075] S2: The current waveform data is sequentially amplified, filtered, and outlier detected to obtain the current data to be corrected;

[0076] Specifically, when the capacitor charges and discharges, current flows through the energy harvesting coil (Rogowski coil), generating an induced electromotive force proportional to the current according to Faraday's law of electromagnetic induction. This induced electromotive force is then amplified and filtered by the signal conditioning circuitry on the acquisition circuit board to remove noise and interference. The processed current signal is converted into a digital signal for subsequent analysis. The acquired current waveform data is processed by the microprocessor on the acquisition circuit board, including data compression and formatting steps, to reduce data transmission bandwidth and storage space requirements. Subsequently, this data is transmitted to the acquisition control unit via a wireless communication module or fiber optic communication interface for further analysis and processing.

[0077] Preferably, in one embodiment of this application, the step of sequentially amplifying, filtering, and detecting outliers on the current waveform data to obtain the current data to be corrected includes:

[0078] The current waveform data is input into an operational amplifier, and the current waveform data is amplified to a preset factor by adjusting the feedback resistor of the operational amplifier.

[0079] The amplified current waveform data is input into a filter for filtering, and the filtered signal is output.

[0080] A statistical anomaly detection method is used to identify anomalies in the filtered current waveform data, obtain anomaly values, and then delete, replace, or mark the anomaly values ​​to obtain the current data to be corrected.

[0081] In this step, the AD627 operational amplifier is used to construct the amplification circuit. The AD627 is a high-precision, low-noise operational amplifier suitable for amplifying weak signals. The amplification factor is adjusted according to the signal strength and processing requirements. If the signal is weak, a higher amplification factor is needed; if the signal is strong, the amplification factor can be appropriately reduced to avoid signal distortion. Precise control of the amplification factor can be achieved by adjusting the values ​​of the feedback resistor and the input resistor in the amplifier circuit.

[0082] Furthermore, current signals may contain high-frequency noise and interference components, which can affect the accuracy and stability of the signal. Therefore, filtering circuits are needed to remove these components and improve signal quality. In this module, an RC filter circuit is used to filter the amplified voltage signal. The design of the filter circuit needs to consider the balance between the signal's frequency characteristics and the filtering effect. If the filtering effect is too strong, it may filter out useful signal components; if the filtering effect is too weak, it will not effectively remove noise and interference components. Therefore, appropriate filter components and parameters need to be selected according to the signal's frequency characteristics and processing requirements.

[0083] The voltage signal, after amplification and filtering, needs to be converted into a digital signal for subsequent digital signal processing and analysis. In this module, a high-precision analog-to-digital converter (ADC) is used to sample and quantize the voltage signal. The ADC's sampling rate and resolution are selected based on signal characteristics and processing requirements. A higher sampling rate captures more signal details; higher resolution represents a more precise signal. The digitized signal can then undergo further processing such as filtering, smoothing, and feature extraction to more accurately reflect the capacitor's current state. These processing results provide strong support for subsequent fault diagnosis, early warning, and condition assessment.

[0084] In the current waveform data upload mode, the STM32 microcontroller samples the current signal at a frequency of 50kHz using its internal ADC for 20ms, obtaining a total of 1000 data points. This waveform data is divided into 40 packets, each containing approximately 25 data points. Then, the 40 packets are sequentially transmitted to the data aggregation unit via the LoRa communication subunit of the data receiving unit. The transmission time for each packet is approximately 75ms, with a total transmission time of approximately 3 seconds.

[0085] S3: Based on a pre-built neural network model, the current data to be corrected is processed to obtain the current data to be analyzed, wherein the correction process is configured to eliminate the fluctuation of current amplitude-frequency characteristics caused by instantaneous current disturbances.

[0086] Preferably, in one embodiment of this application, the step of correcting the current data to be corrected based on a pre-built neural network model to obtain the current data to be analyzed includes:

[0087] Acquire historical current waveform data, wherein the historical current waveform data includes normal current waveform data and abnormal current waveform data;

[0088] An initial neural network model for correction is constructed, and the initial neural network model is trained based on the historical current waveform data to obtain a trained neural network model.

[0089] The current data to be corrected is input into the neural network model for processing to obtain the corrected current data to be analyzed.

[0090] In this step, historical data containing both normal and abnormal current waveforms is collected. This data should cover various current variation scenarios, including current fluctuations under normal operating conditions and abnormal current fluctuations caused by transient disturbances. The aim is to ensure the diversity and representativeness of the data so that a neural network model with strong generalization ability can be trained.

[0091] Choose a suitable neural network model for processing time series data, such as a recurrent neural network (RNN) or its variants (e.g., LSTM, GRU), or a convolutional neural network (CNN) combined with one-dimensional convolutional layers to process the current waveform data. Train the initial neural network model using historical current waveform data. During training, the model will learn how to identify and correct abnormal fluctuations in the current data. Continuously adjust the model's weights and bias parameters using optimization methods such as backpropagation and gradient descent to minimize prediction error.

[0092] Perform necessary post-processing on the corrected current waveform data, such as inverse normalization, to restore the original dimensions of the data. Use a validation dataset (if available) to verify the accuracy and reliability of the corrected current waveform data. The correction effect can be evaluated by calculating error indices before and after correction (such as mean square error (MSE), correlation coefficient, etc.).

[0093] S4: Analyze the current data to be analyzed based on big data analysis technology and machine learning algorithms;

[0094] Preferably, in one embodiment of this application, the analysis of the current data to be analyzed based on big data analysis technology and machine learning algorithms includes:

[0095] The waveform feature data of the current waveform data is extracted, including peak data, valley data, average data, and fluctuation range data;

[0096] An initial waveform recognition model is constructed, and the initial waveform recognition model is trained using acquired historical waveform data and corresponding anomaly labels to obtain a trained waveform recognition model.

[0097] The waveform features are input into the waveform recognition model to classify the waveform features and obtain the operating state of the selected capacitor.

[0098] Specifically, raw current waveform data is acquired from data sources (such as sensors, data acquisition systems, etc.). This data is usually presented in time series form, containing information on how the current changes over time. The raw current waveform data is processed to extract key waveform feature data. This feature data is used for subsequent analysis and identification. Common waveform features include peak data (the maximum value in the current waveform), valley data (the minimum value in the current waveform), average data (the arithmetic mean of the current waveform), and fluctuation range data (the difference between the maximum and minimum values ​​in the current waveform).

[0099] Based on the characteristics of the current waveform data and the analysis requirements, a suitable machine learning algorithm is selected to construct an initial waveform recognition model. Commonly used machine learning algorithms include decision trees, support vector machines (SVM), and neural networks. These algorithms can output classification results based on the input feature data (such as peak values, valley values, average values, and fluctuation ranges).

[0100] The initial waveform recognition model is trained using acquired historical waveform data and corresponding anomaly labels. The historical waveform data should contain both normal and abnormal current waveform samples, with anomaly labels indicating whether each sample belongs to an abnormal category. During training, the model learns how to identify the category (normal or abnormal) of current waveforms based on the input feature data. By adjusting the model's parameters and structure, its performance can be optimized, improving classification accuracy and generalization ability.

[0101] The extracted waveform feature data (peak value, valley value, average value, and fluctuation range) is input into the trained waveform recognition model. The waveform recognition model will classify the current waveform based on the input feature data and output the classification result (normal or abnormal). If the classification result is abnormal, further analysis of the cause and extent of the abnormality may be needed to take appropriate measures.

[0102] In embodiments of this application, the classification results can also be used to determine the operating status of a selected capacitor. If the current waveform is identified as abnormal, it may indicate that the capacitor has problems such as failure or performance degradation. By periodically or in real-time monitoring and analyzing the current waveform data, potential problems with the capacitor can be detected in a timely manner, thereby avoiding the occurrence of failures or reducing their impact.

[0103] S5: Based on the analysis results, the generated matching current regulation command is sent to the selected capacitor.

[0104] Specifically, the analysis results of the current data are interpreted to clarify the characteristics, anomalies, and possible causes of the current waveform. The response of the capacitor to current fluctuations is analyzed to determine if there is performance degradation, malfunction, or parameter adjustments required. Based on the analysis results, the current parameters requiring adjustment, such as current magnitude, frequency, and phase, are determined. The logic for generating current control commands is designed to ensure that the commands accurately reflect the requirements of the analysis results. The generated current control commands are verified to ensure they comply with the capacitor's control requirements and safety specifications. The commands are optimized based on actual conditions to improve the accuracy and efficiency of control. The generated current control commands are sent to the selected capacitor through appropriate communication protocols and interfaces. The transmission of commands is ensured to be free from interference and loss, maintaining their integrity and accuracy. Upon receiving the commands, the capacitor adjusts its operating state according to the requirements. Current changes in the capacitor during the adjustment process are monitored to ensure that it adjusts according to the command requirements.

[0105] Another embodiment of this application provides an online monitoring system for the DC support capacitor current of a flexible DC converter valve. For details, please refer to [link to relevant documentation]. Figure 3 , Figure 3 The diagram shown is a schematic of an online monitoring system for the DC support capacitor current of a flexible DC converter valve according to one embodiment of this application. It includes: a data acquisition module 11, a preprocessing module 12, a calibration module 13, an analysis module 14, and a monitoring module 15.

[0106] The acquisition module 11 is used to control the Rogowski coil to acquire the operating data of a selected capacitor in the flexible DC converter valve in response to an external sampling start command, wherein the operating data includes at least the current waveform data of the terminal of the selected capacitor.

[0107] Preprocessing module 12 is used to sequentially amplify, filter and detect outliers in the current waveform data to obtain the current data to be corrected;

[0108] The correction module 13 is used to perform correction processing on the current data to be corrected based on a pre-built neural network model to obtain the current data to be analyzed, wherein the correction processing is configured to eliminate the fluctuation of current amplitude-frequency characteristics caused by instantaneous current disturbances.

[0109] Analysis module 14 is used to analyze the current data to be analyzed based on big data analysis technology and machine learning algorithms;

[0110] The monitoring module 15 is used to send the generated current regulation command that matches the analysis results to the selected capacitor.

[0111] Preferably, in one embodiment of this application, the acquisition module is specifically used for:

[0112] The charging and discharging current of the selected capacitor is collected using a Rogowski coil, and the charging and discharging current is converted into a proportional induced electromotive force.

[0113] The induced electromotive force is digitally converted to obtain a digital signal;

[0114] The digital signal is sampled to obtain current waveform data.

[0115] Preferably, in one embodiment of this application, the preprocessing module is specifically used for:

[0116] The current waveform data is input into an operational amplifier, and the current waveform data is amplified to a preset factor by adjusting the feedback resistor of the operational amplifier.

[0117] The amplified current waveform data is input into a filter for filtering, and the filtered signal is output.

[0118] A statistical anomaly detection method is used to identify anomalies in the filtered current waveform data, obtain anomaly values, and then delete, replace, or mark the anomaly values ​​to obtain the current data to be corrected.

[0119] Preferably, in one embodiment of this application, the correction module is specifically used for:

[0120] Acquire historical current waveform data, wherein the historical current waveform data includes normal current waveform data and abnormal current waveform data;

[0121] An initial neural network model for correction is constructed, and the initial neural network model is trained based on the historical current waveform data to obtain a trained neural network model.

[0122] The current data to be corrected is input into the neural network model for processing to obtain the corrected current data to be analyzed.

[0123] Preferably, in one embodiment of this application, the analysis module is specifically used for:

[0124] The waveform feature data of the current waveform data is extracted, including peak data, valley data, average data, and fluctuation range data;

[0125] An initial waveform recognition model is constructed, and the initial waveform recognition model is trained using acquired historical waveform data and corresponding anomaly labels to obtain a trained waveform recognition model.

[0126] The waveform features are input into the waveform recognition model to classify the waveform features and obtain the operating state of the selected capacitor.

[0127] Compared to the prior art, the beneficial effects of the embodiments of this application are at least one of the following:

[0128] 1) This application uses Rogowski coils to collect operating data of selected capacitors in flexible DC converter valves, especially current waveform data of terminal posts. This method has high sensitivity and accuracy and can capture subtle features of current changes.

[0129] 2) This application amplifies, filters, and detects outliers in the collected current waveform data, which can effectively remove noise and interference, improve the signal-to-noise ratio of the data, and provide a high-quality data foundation for subsequent analysis.

[0130] 3) This application utilizes a pre-built neural network model to correct the current data, which can eliminate fluctuations in the current amplitude-frequency characteristics caused by instantaneous current disturbances, further improving the accuracy and reliability of the data. Simultaneously, this application analyzes the processed current data based on big data analytics, enabling the discovery of potential patterns and characteristics within the data, providing strong support for fault prediction and diagnosis.

[0131] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for online monitoring of DC support capacitor current of a LCC valve, characterized in that, The method comprises the following steps: in response to an external sampling start instruction, controlling a Rogowski coil to collect operation data of a selected capacitor in a HVDC converter valve, wherein the operation data at least includes current waveform data of a terminal column of the selected capacitor; amplifying, filtering and detecting outliers of the current waveform data in sequence to obtain to-be-corrected current data; based on a pre-constructed neural network model, correcting the to-be-corrected current data to obtain to-be-analyzed current data, wherein the correction processing is configured to eliminate current amplitude-frequency characteristic fluctuations caused by current transient disorder; based on big data analysis technology and machine learning algorithms, analyzing the to-be-analyzed current data; according to the analysis result, sending a generated current regulation instruction matched therewith to the selected capacitor.

2. The HVDC valve DC grading capacitor current on-line monitoring method of claim 1, wherein, The method of controlling a Rogowski coil to collect operation data of a selected capacitor in a HVDC converter valve in response to an external sampling start instruction comprises the following steps: based on the Rogowski coil, collecting charge and discharge currents of the selected capacitor, and converting the charge and discharge currents into proportional induced electromotive forces; digitally converting the induced electromotive forces to obtain digital signals; sampling the digital signals to obtain current waveform data.

3. The DC support capacitor current on-line monitoring method of the LCC valve according to claim 1, characterized in that, The method of amplifying, filtering and detecting outliers of the current waveform data in sequence to obtain to-be-corrected current data comprises the following steps: inputting the current waveform data into an operational amplifier, and amplifying the current waveform data to a preset multiple by adjusting the feedback resistance of the operational amplifier; inputting the amplified current waveform data into a filter for filtering processing, and outputting filtered signals; based on a statistical outlier detection method, identifying outliers of the filtered current waveform data, deleting, replacing or marking the outliers to obtain the to-be-corrected current data.

4. The HVDC valve DC blocking capacitor current on-line monitoring method of claim 1, wherein, The method of correcting the to-be-corrected current data based on a pre-constructed neural network model to obtain to-be-analyzed current data comprises the following steps: obtaining historical current waveform data, wherein the historical current waveform data includes normal current waveform data and abnormal current waveform data; constructing an initial neural network model for correction, training the initial neural network model based on the historical current waveform data to obtain a trained neural network model; inputting the to-be-corrected current data into the neural network model for processing to obtain corrected to-be-analyzed current data.

5. The HVDC valve DC blocking capacitor current on-line monitoring method of claim 1, wherein, The method of analyzing the to-be-analyzed current data based on big data analysis technology and machine learning algorithms comprises the following steps: extracting waveform feature data of the current waveform data, wherein the waveform feature data includes peak value data, valley value data, average value data and fluctuation range data; constructing an initial waveform recognition model, training the initial waveform recognition model based on obtained historical waveform data and corresponding abnormal labels to obtain a trained waveform recognition model; inputting the waveform features into the waveform recognition model to classify the waveform features to obtain a working state of the selected capacitor.

6. A DC support capacitor current on-line monitoring system for a LCC, characterized in that, The method comprises the following steps: The acquisition module is configured to, in response to an external sampling start instruction, control a Rogowski coil to collect operation data of a selected capacitor in a LCC valve, wherein the operation data at least includes current waveform data of a terminal post of the selected capacitor; The preprocessing module is configured to sequentially perform amplification, filtering and outlier detection processing on the current waveform data to obtain to-be-corrected current data; The correction module is configured to correct the to-be-corrected current data based on a pre-constructed neural network model to obtain to-be-analyzed current data, wherein the correction processing is configured to eliminate current amplitude-frequency characteristic fluctuations caused by current transient disorder; The analysis module is configured to analyze the to-be-analyzed current data based on big data analysis technology and a machine learning algorithm; The monitoring module is configured to send generated current regulation instructions matched with the analysis result to the selected capacitor.

7. The HVDC valve DC blocking capacitor current on-line monitoring system of claim 6, wherein, The acquisition module is specifically configured to: collect charge-discharge current of the selected capacitor based on the Rogowski coil, and convert the charge-discharge current into proportional induced electromotive force; perform digital conversion on the induced electromotive force to obtain a digital signal; sample the digital signal to obtain current waveform data.

8. The HVDC valve DC blocking capacitor current on-line monitoring system of claim 6, wherein, The preprocessing module is specifically configured to: input the current waveform data into an operational amplifier, amplify the current waveform data to a preset multiple by adjusting a feedback resistor of the operational amplifier; input the amplified current waveform data into a filter for filtering processing, and output filtered signals; perform abnormality identification on the filtered current waveform data based on a statistical abnormality detection method to obtain abnormal values, and perform deletion, replacement or marking processing on the abnormal values to obtain the to-be-corrected current data.

9. The HVDC valve DC blocking capacitor current on-line monitoring system of claim 6, wherein, The correction module is specifically configured to: obtain historical current waveform data, wherein the historical current waveform data includes normal current waveform data and abnormal current waveform data; construct an initial neural network model for correction, train the initial neural network model based on the historical current waveform data to obtain a trained neural network model; input the to-be-corrected current data into the neural network model for processing to obtain corrected to-be-analyzed current data.

10. The HVDC valve DC blocking capacitor current on-line monitoring system of claim 6, wherein, The analysis module is specifically configured to: extract waveform feature data of the current waveform data, wherein the waveform feature data includes peak value data, valley value data, average value data and fluctuation range data; construct an initial waveform identification model, train the initial waveform identification model based on obtained historical waveform data and corresponding abnormal labels to obtain a trained waveform identification model; input the waveform feature into the waveform identification model to classify the waveform feature to obtain a working state of the selected capacitor.

Images