Capacitor quasi-online state monitoring method and system for wind power converter

Abstract

The invention relates to the field of capacitor monitoring, and provides a capacitor quasi-online state monitoring method and system for a wind power converter, and the method comprises the steps: extracting a data slice comprising DC bus voltage, capacitor branch current, housing temperature and output power from a data buffer area; furthermore, output power data is utilized, intelligent screening is carried out through a quasi-steady-state slice trigger, and only effective data with the power average value large enough and the fluctuation small enough are reserved. And for the screened effective data slices, extracting the DC bus voltage and the capacitance branch current, and accurately identifying the voltage, current amplitude and phase information of the main harmonic through fast Fourier transform. Based on the key harmonic data, the complex impedance of the capacitor is calculated, and then the original ESR and capacitance estimation value are calculated. And finally, in combination with the capacitor shell temperature extracted from the same data slice, performing temperature compensation on the original parameters, and outputting a final result capable of truly reflecting the aging state of the capacitor.

Description

Technical Field

[0001] This application relates to the field of capacitor monitoring, and specifically to a method and system for quasi-online condition monitoring of capacitors in wind power converters. Background Technology

[0002] Wind power, as a clean and renewable energy source, occupies an increasingly important position in the global energy structure. In wind turbine generators, the wind power converter is a key device for efficiently converting wind energy into electrical energy and ensuring stable grid connection. The DC bus capacitor, as a core energy storage and filtering component in the wind power converter, directly affects the operating efficiency, reliability, and lifespan of the entire converter and even the wind turbine generator. During long-term operation, the equivalent series resistance of the DC bus capacitor gradually increases and its capacitance gradually decreases due to the influence of various factors such as temperature, ripple current, and voltage stress. This leads to increased DC bus voltage ripple, increased losses, and may even cause converter failure and shutdown, resulting in significant economic losses and maintenance costs. Therefore, effective condition monitoring of the capacitors in the wind power converter, timely detection of performance degradation trends, and predictive maintenance are crucial for ensuring the long-term stable operation of wind turbine generators.

[0003] However, effective condition monitoring of capacitors is not easy, and existing monitoring methods often have certain limitations. Traditional offline monitoring methods require shutting down the converter and removing the capacitors for measurement, which is time-consuming and labor-intensive, and cannot achieve real-time or near-real-time condition assessment, making it difficult to meet the needs of continuous operation in wind farms. While some online monitoring methods can acquire data in real time, they usually require complex sensor arrangements or significant modifications to existing circuits, increasing cost and system complexity. More importantly, for wind power converters, their operating conditions are affected by wind speed fluctuations, and parameters such as output power, voltage, and current are constantly changing dynamically. This makes it extremely difficult to accurately extract characteristic parameters of capacitors (such as ESR and capacitance values) under complex unsteady-state conditions. Most methods based on harmonic analysis or impedance measurement are highly dependent on the system being in a relatively stable operating state. Once wind speed or load changes drastically, the data will be filled with a large number of transient components and noise, leading to distorted monitoring results that cannot effectively reflect the true health status of the capacitors.

[0004] Therefore, an optimized capacitor monitoring scheme for wind power converters is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a method and system for quasi-online condition monitoring of capacitors in wind power converters.

[0006] According to one aspect of this application, a method for quasi-online condition monitoring of capacitors in a wind power converter is provided, comprising:

[0007] Extract the first wind power converter data slice from the data buffer. The first wind power converter data slice includes multiple sampling points, each of which includes DC bus voltage, capacitor branch current, capacitor casing temperature and converter output power.

[0008] The first wind power converter data slice is input into a quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice.

[0009] In response to the fact that the first wind power converter data slice is a quasi-steady-state data slice, adaptive harmonic identification is performed on the first wind power converter data slice to obtain the harmonic identification result;

[0010] Based on the harmonic identification results, the original capacitance estimation and ESR calculation are performed to obtain the original capacitance estimate and the original ESR estimate.

[0011] The measurement temperature is extracted from the data slice of the first wind power converter, and the original capacitance estimate and the original ESR estimate are temperature compensated based on the measurement temperature to obtain the temperature compensated ESR estimate and the temperature compensated capacitance estimate.

[0012] A health status assessment is performed based on the temperature-compensated ESR estimate and the temperature-compensated capacitance estimate to obtain the health status assessment result.

[0013] According to another aspect of this application, a quasi-online condition monitoring system for capacitors in wind power converters is provided, comprising:

[0014] The data slicing sampling module is used to extract the first wind power converter data slice from the data buffer. The first wind power converter data slice includes multiple sampling points, and each sampling point includes DC bus voltage, capacitor branch current, capacitor casing temperature and converter output power.

[0015] The data slice input module is used to input the first wind power converter data slice into the quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice.

[0016] The harmonic identification module is used to perform adaptive harmonic identification on the first wind power converter data slice, which is a quasi-steady-state data slice, to obtain the harmonic identification result.

[0017] The estimation calculation module is used to perform original capacitance estimation and ESR calculation on the harmonic identification results to obtain the original capacitance estimate and the original ESR estimate.

[0018] The temperature-compensated estimation module is used to extract the measurement temperature from the data slice of the first wind power converter, and perform temperature compensation on the original capacitance estimate and the original ESR estimate based on the measurement temperature to obtain the temperature-compensated ESR estimate and the temperature-compensated capacitance estimate.

[0019] The results evaluation module is used to evaluate the health status of the temperature-compensated ESR estimate and the temperature-compensated capacitance estimate to obtain the health status evaluation results.

[0020] Compared with existing technologies, this application provides a quasi-online condition monitoring method and system for capacitors in wind power converters. It extracts data slices from a data buffer, including DC bus voltage, capacitor branch current, casing temperature, and output power. Furthermore, using the output power data, a quasi-steady-state slice trigger performs intelligent filtering, retaining only valid data with sufficiently large average power values ​​and sufficiently small fluctuations, thus ensuring the reliability of subsequent analysis. For the filtered valid data slices, the system extracts the DC bus voltage and capacitor branch current, accurately identifying the voltage, current amplitude, and phase information of the main harmonics using Fast Fourier Transform. Based on these key harmonic data, the complex impedance of the capacitor is calculated, and then the original ESR and capacitance estimate are obtained. Finally, combined with the capacitor casing temperature extracted from the same data slice, temperature compensation is applied to the original parameters, outputting a calibrated final result that accurately reflects the aging state of the capacitor, providing reliable data support for predictive maintenance of wind turbine units. Attached Figure Description

[0021] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0022] Figure 1 This is a flowchart of a quasi-online condition monitoring method for capacitors in a wind power converter according to an embodiment of this application.

[0023] Figure 2 This is a schematic diagram of data flow in a quasi-online condition monitoring method for capacitors in a wind power converter according to an embodiment of this application.

[0024] Figure 3 This is a flowchart illustrating how the first wind power converter data slice is input into a quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice in the quasi-online state monitoring method for capacitors of a wind power converter according to an embodiment of this application.

[0025] Figure 4 This is a flowchart illustrating the adaptive harmonic identification of the first wind power converter data slice to obtain the harmonic identification result in the quasi-online state monitoring method for capacitors of a wind power converter according to an embodiment of this application.

[0026] Figure 5 This is a block diagram of a capacitor quasi-online condition monitoring system for wind power converters according to an embodiment of this application. Detailed Implementation

[0027] Various exemplary embodiments, features, and aspects of this application will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0028] To address the problems in the background technology, this application proposes a quasi-online condition monitoring method and system for capacitors in wind power converters. To solve the technical challenge of inaccurate capacitor condition monitoring under dynamic operating conditions of wind power converters, this method avoids continuous and invalid calculations. Instead, it introduces a quasi-steady-state slicing trigger and intelligently filters out the optimal data window where the system operates at higher power and with minimal fluctuations by analyzing the mean and standard deviation of the converter's output power. This design fundamentally avoids measurement distortion caused by data noise and transient effects during drastic dynamic changes, ensuring the validity of the analyzed data and solving the problem of accurate feature extraction under dynamic conditions in existing technologies. After capturing ideal data, the system accurately calculates the original ESR and capacitance value of the capacitor using adaptive harmonic identification technology. Furthermore, to eliminate interference from environmental factors, a temperature compensation mechanism is also integrated. By utilizing the measured capacitor casing temperature, the original estimated value is calibrated to a unified benchmark, thereby eliminating the apparent influence of temperature changes on the parameters and solving the problem of misjudgment caused by ignoring temperature in existing technologies. By combining two core design features—quasi-steady-state triggering and temperature compensation—the system effectively overcomes the two major challenges of dynamic operation and temperature sensitivity in wind power, enabling accurate assessment of the true health status of capacitors and providing a solid foundation for reliable predictive maintenance.

[0029] The technical solution of this application proposes a method and system for quasi-online condition monitoring of capacitors in wind power converters. Figure 1 This is a flowchart of a quasi-online condition monitoring method for capacitors in a wind power converter according to an embodiment of this application. Figure 2 This is a schematic diagram of data flow in a quasi-online condition monitoring method for capacitors in a wind power converter according to an embodiment of this application. Figure 1 and Figure 2As shown, the method for quasi-online state monitoring of capacitors in a wind power converter according to an embodiment of this application includes: S100, extracting a first wind power converter data slice from a data buffer, the first wind power converter data slice including multiple sampling points, each sampling point including DC bus voltage, capacitor branch current, capacitor casing temperature and converter output power; S200, inputting the first wind power converter data slice into a quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice; S300, responding to the first... The wind power converter data slice is a quasi-steady-state data slice. Adaptive harmonic identification is performed on the first wind power converter data slice to obtain harmonic identification results; S400, based on the harmonic identification results, the original capacitance estimation and ESR calculation are performed to obtain the original capacitance estimate and the original ESR estimate; S500, the measurement temperature is extracted from the first wind power converter data slice, and the original capacitance estimate and the original ESR estimate are temperature compensated based on the measurement temperature to obtain the temperature compensated ESR estimate and the temperature compensated capacitance estimate.

[0030] S600 performs a health status assessment based on temperature-compensated ESR and temperature-compensated capacitance estimates to obtain the health status assessment results.

[0031] Specifically, in step S100, a first wind power converter data slice is extracted from the data buffer. This first wind power converter data slice includes multiple sampling points, each including DC bus voltage, capacitor branch current, capacitor casing temperature, and converter output power. This can be understood as follows: to achieve quasi-online state monitoring of the capacitor, the first step is to extract a data slice containing key physical quantities from the data buffer. These quantities are indirectly calculated by analyzing their voltage and current responses in the circuit, while their performance is significantly affected by operating conditions and temperature. Therefore, all necessary information related to capacitor state calculation and operating condition judgment must be acquired synchronously. This provides a complete, synchronous, and time-aligned data foundation for subsequent quasi-steady-state judgment, parameter calculation, and temperature compensation. By packaging these physical quantities into a single data slice, it ensures that all analyses are based on data within the same time window, avoiding calculation errors introduced by data asynchrony. Ultimately, a structured raw dataset containing multi-dimensional information is obtained, providing standardized input for all subsequent algorithm modules and serving as the data source and foundation of the entire monitoring process.

[0032] More specifically, in one example of this application, a data acquisition and buffering system is first constructed. In the control system of the wind power converter, four key analog signals are synchronously acquired through a high-frequency sampling module: DC bus voltage, capacitor branch current, temperature sensor signal close to the capacitor casing, and the real-time output power of the converter. After analog-to-digital conversion, these signals form digital sampling points containing timestamps and the values ​​of the four physical quantities. These sampling points are continuously fed into a first-in-first-out (FIFO) data buffer in real time. This buffer is typically configured as a fixed-length queue; specifically, in one example of this application, it can store the sampling data from the most recent 1 or 2 seconds, continuously discarding the oldest data and storing the latest data in a rolling update manner.

[0033] Secondly, data slices are defined and extracted periodically. The system sets a fixed extraction cycle; specifically, in a specific example of this application, the extraction operation is performed once per second. At the trigger time of each extraction cycle, the system copies a complete data block from the current queue of the data buffer at once. This data block constitutes the first wind power converter data slice. This slice is a two-dimensional data matrix, where the rows represent continuous sampling points in time, and the columns correspond to the time series of four physical quantities: DC bus voltage, capacitor branch current, capacitor casing temperature, and converter output power, respectively. Specifically, in a specific example of this application, if the sampling frequency is 10kHz and a 1-second data slice is extracted, the slice will contain 10,000 sampling points, each containing information in these four dimensions. Alternatively, this implementation can also be achieved through a data preprocessing network model. This model is designed to receive the raw data stream from the sensor layer. The first layer of the model is a synchronization and alignment layer, responsible for timestamping the input signals to ensure data synchronization. The second layer is a sliding window layer that defines a fixed-size window (i.e., a data slice) on the aligned data stream and slides forward with a set step size. Whenever the window slides to a new position, the data within the window is packaged into a standardized data slice tensor. This tensor contains the time series of all necessary physical quantities within a predetermined time length and is directly supplied as the model output to the subsequent quasi-steady-state determination module. This model-based implementation allows for a more flexible and intelligent generation and supply of data slices.

[0034] Specifically, in step S200, the first wind power converter data slice is input into a quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice. It should be understood that in real-world scenarios where wind power and grid load fluctuate continuously, the system is in a dynamic or transient state most of the time. The capacitor voltage and current signals collected at this time contain a large amount of noise and non-periodic components. If used directly for parameter calculation without screening, the results will inevitably lead to huge deviations or even errors, failing to reflect the true health status of the capacitors. Therefore, before performing accurate parameter calculations, it is necessary to first determine the operating conditions at the time of data acquisition; this is a logical prerequisite for ensuring the accuracy of monitoring results. By analyzing the converter output power within the data slice—a key indicator that directly reflects the system's operational stability—the quasi-steady-state slice trigger aims to accurately identify and capture those quasi-steady-state time windows with stable system operation and high signal-to-noise ratio from the continuous data stream, thus filtering out the most ideal and reliable raw data for subsequent accurate calculations.

[0035] Figure 3 This is a flowchart illustrating how the first wind power converter data slice is input into a quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice in the quasi-online state monitoring method for capacitors in a wind power converter according to an embodiment of this application. Figure 3 As shown, according to an embodiment of this application, the method for monitoring the quasi-online state of capacitors in a wind power converter includes step S200: S210, extracting the time series of converter output power from the first wind power converter data slice; S220, extracting statistical features of the time series of converter output power, the statistical features including the average output power and the standard deviation of output power; S230, determining whether the first wind power converter data slice is a quasi-steady-state data slice by judging whether the statistical features meet the triggering conditions.

[0036] Specifically, in step S300, in response to the first wind power converter data slice being a quasi-steady-state data slice, adaptive harmonic identification is performed on the first wind power converter data slice to obtain harmonic identification results. It should be understood that the two core health indicators of a capacitor—equivalent series resistance and capacitance—cannot be directly measured; their numerical information is hidden in the relationship between the AC ripple current flowing through the capacitor and the ripple voltage across its terminals. In the actual operating scenario of a wind power converter, these ripple signals are not of a single frequency but are composed of multiple harmonic components superimposed. Therefore, in order to accurately calculate the impedance characteristics of the capacitor, it is necessary to first accurately separate and identify the key information of one or more dominant harmonic components from the complex mixed signal, namely the amplitude and phase of voltage and current, and then perform precise signal feature extraction based on high-quality data that has been confirmed as quasi-steady-state. The purpose of adaptive harmonic identification is to address the challenge of possible changes in the harmonic spectrum of the wind power converter at different operating points. Instead of simply analyzing a fixed frequency, it uses algorithms such as Fast Fourier Transform to perform spectral analysis on voltage and current time series, automatically searching for and locking onto the dominant harmonic frequency with the highest energy concentration and signal-to-noise ratio under the current operating conditions. Then, for this adaptively locked frequency, it precisely extracts the corresponding voltage amplitude, current amplitude, and the phase difference between them. This process aims to extract core frequency domain parameters from the original time-domain waveform that can be directly used for complex impedance calculations.

[0037] Figure 4 This is a flowchart illustrating the adaptive harmonic identification of the first wind power converter data slice to obtain harmonic identification results in the quasi-online state monitoring method for capacitors in a wind power converter according to an embodiment of this application. Figure 4 As shown, according to an embodiment of this application, the method for quasi-online status monitoring of capacitors in a wind power converter includes step S300 as follows: S310, extracting the time series of DC bus voltage and the time series of capacitor branch current from the first wind power converter data slice; S320, performing a fast Fourier transform on the time series of DC bus voltage and the time series of capacitor branch current to obtain a voltage spectrum and a current spectrum; S330, within a predetermined frequency search range, performing a peak search and main harmonic determination on the voltage spectrum to obtain the main harmonic frequency and the frequency index corresponding to the main harmonic frequency; S340, based on the frequency index corresponding to the main harmonic frequency, extracting the main harmonic voltage amplitude, main harmonic voltage phase, main harmonic current amplitude, and main harmonic current phase from the voltage spectrum and the current spectrum.

[0038] Specifically, in step S310, the time series of the DC bus voltage and the time series of the capacitor branch current are extracted from the first wind power converter data slice. It should be understood that the impedance of a capacitor, including ESR and capacitive reactance information, is determined by the voltage across its terminals and the current flowing through it. On the DC bus side of the wind power converter, the voltage across the capacitor is the DC bus voltage, and the current flowing through it is the capacitor branch current. These two physical quantities are indispensable and the most direct raw inputs for all subsequent electrical parameter calculations. The converter output power and capacitor casing temperature included in the data slice are used for operating condition judgment and temperature compensation, respectively, and do not directly participate in impedance calculation. After confirming that the current data slice is in an ideal quasi-steady state, the algorithm enters the signal processing stage. The goal of this stage is to extract key frequency domain features from the complex time-domain waveform. Therefore, it is necessary to accurately separate the DC bus voltage and capacitor branch current—two specific time series—from the data slice containing multiple physical quantities. The general data structure is transformed into a dedicated data format that can be directly used by signal processing algorithms (such as Fast Fourier Transform, FFT), ensuring the focus and efficiency of subsequent analysis. Specifically, in step S320, the time series of the DC bus voltage and the time series of the capacitor branch current are subjected to Fast Fourier Transform to obtain the voltage spectrum and current spectrum. It should be understood that although the original DC bus voltage and capacitor branch current time series contain all the information needed to calculate the capacitor state, this information exists in a mixed and implicit way in the complex time-domain waveform. In the actual operation scenario of wind power converters, only a complex signal superimposed with multiple frequency ripples can be observed, and it is impossible to directly separate the precise relationship between voltage and current at a specific harmonic frequency, which is precisely the core basis for calculating capacitor impedance. Therefore, it is necessary to transform the signal from the intuitive but information-mixed time domain to the characteristically clear frequency domain to reveal its internal frequency composition, and Fast Fourier Transform is the most computationally efficient standard algorithm to achieve this transformation. The two complex time-domain signals are decomposed into their respective frequency components to obtain the voltage spectrum and current spectrum. The system decouples and quantifies the harmonic information hidden in waveform fluctuations. Through FFT, the system can clearly see the energy distribution of voltage ripple and current ripple at different frequency points, that is, their respective amplitude and phase information. This provides an intuitive and analyzable frequency domain region for subsequently adaptively finding and locking the main harmonic frequency with the most concentrated energy and the highest signal-to-noise ratio, which is a necessary preparatory work for accurate harmonic identification.

[0039] Specifically, in step S330, within a predetermined frequency search range, peak search and main harmonic determination are performed on the voltage spectrum to obtain the main harmonic frequency and the frequency index corresponding to the main harmonic frequency. It is worth noting that the predetermined frequency search range here is 200Hz to 2000Hz. It should be understood that the voltage spectrum obtained after Fast Fourier Transform is a panoramic view containing information from DC to the Nyquist frequency. In the actual operation of wind power converters, not all frequency components have equal value for capacitor state assessment. The spectrum contains both key harmonics with concentrated energy generated by converter switching operations and a large amount of background noise and irrelevant low-amplitude harmonics. Using all harmonics indiscriminately or arbitrarily selecting a harmonic will introduce huge errors due to low signal-to-noise ratio. A reliable mechanism must be established to accurately find the main harmonic. A peak search is performed within a pre-defined frequency range that conforms to the physical characteristics of the converter (e.g., avoiding DC components and extremely high-frequency noise, focusing on the switching frequency and its harmonics) to find the harmonic component with the largest amplitude and most concentrated energy within that range. This harmonic is then used to calculate the capacitor impedance, ultimately outputting two key results: the true frequency value of the dominant harmonic and its position in the FFT result array, i.e., the frequency index.

[0040] More specifically, in a specific example of this application, the implementation of the peak search and main harmonic determination includes the following steps: First, setting a frequency search interval. This step defines an effective and reasonable range for the peak search to eliminate interference from irrelevant frequencies and improve computational efficiency. Based on the operating principle of wind power converters, the voltage ripple on their DC bus is mainly dominated by the switching frequency of the grid-side converter and its harmonics. Therefore, a frequency search interval is pre-set, with its lower limit slightly lower than the theoretical grid-side switching frequency (e.g., 0.8 times the switching frequency) and its upper limit slightly higher than the second harmonic of the switching frequency (e.g., 2.2 times the switching frequency). This ensures that the search range can completely cover the most likely main harmonic frequency while avoiding invalid searches in the DC component or higher-frequency noise regions. Second, executing the peak search algorithm. This step involves finding the point with the largest amplitude from discrete voltage spectrum data within the set interval. The system will traverse all voltage spectrum amplitude points within the frequency interval set in the first step. During this process, through iterative comparison, the spectrum point with the largest amplitude within the interval is found and recorded. The point with the largest amplitude is identified as the peak value. The third step is to determine the primary harmonic frequency and its index. This step converts the peak value information found in the second step into specific parameters required for subsequent calculations. The system records the frequency corresponding to the peak value found in the second step; this frequency is determined as the primary harmonic frequency used in this calculation. Simultaneously, the system records the position or index of this peak value within the entire spectrum array; this position or index is the frequency index corresponding to the primary harmonic frequency. This frequency index is crucial because it serves as the address pointer for accurately extracting the amplitude and phase of the corresponding harmonics from the voltage and current spectra, ensuring the frequency consistency of voltage and current analysis.

[0041] Specifically, in step S340, based on the frequency index corresponding to the primary harmonic frequency, the amplitude, phase, amplitude, and phase of the primary harmonic voltage, primary harmonic current, and primary harmonic current are extracted from the voltage spectrum and the current spectrum. It should be understood that while the previous steps successfully identified the most representative primary harmonic frequency with the highest signal-to-noise ratio and its corresponding frequency index from the complex spectrum, this index itself is merely a position pointer and has not yet extracted the core physical quantities required for calculation—namely, the detailed harmonic information of the voltage and current at that specific frequency. Therefore, in the technical solution of this application, this determined frequency index is further utilized as a precise navigation beacon to simultaneously perform addressing and data extraction in both the voltage spectrum and current spectrum data structures, thereby obtaining all four fundamental parameters necessary for calculating the complex impedance of the capacitor: the amplitude and phase of the primary harmonic voltage, and the amplitude and phase of the primary harmonic current. In this way, the results of the previous frequency search can be transformed into specific physical values ​​that can be used for quantitative calculation, providing a direct and complete data input for the final calculation of the equivalent series resistance and capacitance of the capacitor, which is a key bridge connecting harmonic location and parameter calculation.

[0042] Specifically, in step S400, based on the harmonic identification results, the original capacitance estimation and ESR calculation are performed to obtain the original capacitance estimate and the original ESR estimate. It should be understood that while the preceding steps have successfully extracted precise voltage, current amplitude, and phase information at the main harmonic frequency from the complex time-domain signal, these harmonic identification results are still discrete physical quantities and have not yet been integrated into core indicators that directly reflect the health status of the capacitor. Therefore, in the technical solution of this application, based on the fundamental laws of circuit theory, these extracted harmonic parameters are used to construct complex voltage and current phasors, and the complex impedance of the capacitor at the main harmonic frequency is calculated by phasor division. This transforms the discrete harmonic information into a comprehensive electrical parameter containing resistance and reactance components. In this way, the estimated value of the equivalent series current can be directly obtained by extracting the real part of the complex impedance, and the estimated value of the capacitance can be calculated through its imaginary part. Thus, for the first time, two key original indicators characterizing the current performance of the capacitor are obtained, providing core, uncalibrated reference data for subsequent temperature compensation and final health status assessment.

[0043] More specifically, in a specific example of this application, the implementation of the original capacitance estimation and ESR calculation includes the following steps: First, constructing a complex phasor. This step involves converting the scalar result output by the preceding harmonic identification module into a complex form conforming to AC circuit analysis specifications. Using the extracted main harmonic voltage amplitude (V... amp ) and the phase of the main harmonic voltage (V phase Construct the complex form of the voltage phasor V.phasor =V phase *(cos(V phase )+jsin(V phase Similarly, using the amplitude of the main harmonic current (I) amp ) and the phase of the main harmonic current (I) phase Construct the complex form of the current phasor I. phasor =I phase *(cos(I phase )+jsin(I phase ()). Where j is the imaginary unit. The second step is to calculate the complex impedance of the capacitor. This step applies the phasor form of Ohm's law to directly solve for the electrical impedance characteristics of the capacitor at the main harmonic frequency. By dividing the voltage phasor by the current phasor, the complex impedance Z of the capacitor is calculated. complex =V phasor / I phasor The calculation result Z complex It is a complex number of the form Z. complex =R + jX, where R is the real part of the complex impedance and X is the imaginary part of the complex impedance. The third step is to solve for the original ESR estimate and the original capacitance estimate. This step separates the specific physical parameters from the complex impedance. The real part R of the complex impedance directly corresponds to the equivalent series resistance of the capacitor; therefore, this real part is extracted as the original ESR estimate, i.e., the ESR. raw =Real(Z_complex). The imaginary part X of the complex impedance represents the capacitive reactance of the capacitor, which is related to the capacitance C and the dominant harmonic angular frequency ω (ω = 2πf). main f main The relationship between the main harmonic frequency and the capacitance is X = 1 / (ω*C). Based on this, the estimated value of the original capacitance, C, can be calculated from the imaginary part X through algebraic transformation. raw =-1 / (ω*Imag(Z) complex The complete solution process from harmonic identification results to raw ESR and capacitance estimates has been completed.

[0044] Specifically, in step S500, the measurement temperature is extracted from the first wind power converter data slice, and the original capacitance estimate and original ESR estimate are temperature-compensated based on the measurement temperature to obtain the temperature-compensated ESR estimate and temperature-compensated capacitance estimate. It should be understood that due to the physical characteristics of capacitors, their equivalent series resistance and capacitance are not constant but change significantly with fluctuations in operating temperature. This change is not a degradation of the capacitor itself but a normal physical response. Therefore, in the technical solution of this application, the capacitor casing temperature, which corresponds exactly to the parameter calculation time, is further extracted synchronously from the confirmed valid quasi-steady-state data slice. Using this real-time temperature value, the previously calculated original ESR and capacitance estimates are calibrated using a preset temperature compensation model (such as a two-dimensional lookup table) to remove and eliminate the temporary and reversible effects of temperature fluctuations on the calculation results. In this way, the measured values ​​can be corrected to a unified reference temperature, thereby obtaining more accurate temperature-compensated ESR and capacitance estimates that truly reflect the changes in the intrinsic health status of the capacitor caused by irreversible factors such as aging and loss, providing a reliable and comparable basis for the final health status assessment.

[0045] More specifically, in a specific example of this application, the implementation of the temperature compensation includes the following steps: First, extracting the measurement temperature. This step is to obtain the environmental variables necessary for parameter compensation. The system extracts capacitor casing temperature data from the same first wind power converter data slice used to calculate the original equivalent series resistance and capacitance values. Since all sampling points within the data slice are highly synchronized in time, the extracted temperature value accurately represents the actual operating temperature of the capacitor during electrical parameter measurement, and this temperature value is recorded as the measurement temperature. Second, establishing a temperature compensation model. This step is the core basis for constructing the relationship between parameters and temperature. This model is based on datasheets provided by the capacitor manufacturer or obtained through offline experimental calibration. The model describes the variation of the capacitor's equivalent series resistance and capacitance values ​​at different temperatures relative to a reference temperature (e.g., 25 degrees Celsius). This model is implemented as a two-dimensional lookup table. The lookup table uses temperature as the input index and outputs the corresponding equivalent series resistance temperature compensation coefficient and capacitance temperature compensation coefficient. For example, the temperature compensation coefficient for equivalent series resistance is defined as the ratio of the equivalent series resistance value at a certain temperature to the equivalent series resistance value at a reference temperature; the temperature compensation coefficient for capacitance is defined as the ratio of the capacitance value at a certain temperature to the capacitance value at a reference temperature. The third step is to perform temperature compensation calculations. This step calibrates the original calculated values ​​to the standard values ​​at the reference temperature. The system uses the measurement temperature obtained in the first step as input, and queries the two-dimensional lookup table established in the second step or calculates it through interpolation to obtain the corresponding compensation coefficients at that temperature, namely the temperature compensation coefficient for equivalent series resistance and the temperature compensation coefficient for capacitance. Subsequently, compensation rules are applied for calculation: the original estimated equivalent series resistance value is divided by the found temperature compensation coefficient for equivalent series resistance, and the result is the temperature-compensated estimated equivalent series resistance value; the original estimated capacitance value is divided by the found temperature compensation coefficient for capacitance, and the result is the temperature-compensated estimated capacitance value. Through this calculation, parameter estimates that have eliminated the influence of temperature and can be directly used for aging assessment can be obtained.

[0046] Specifically, in step S600, a health status assessment is performed based on the temperature-compensated ESR and capacitance estimates to obtain the health status assessment result. It should be understood that while the preceding steps have already obtained temperature-compensated ESR and capacitance estimates that accurately reflect the capacitor's intrinsic performance, these values ​​are merely quantified physical parameters and have not yet been transformed into qualitative health status conclusions that provide direct guidance for equipment maintenance personnel. Therefore, in the technical solution of this application, these precisely calibrated parameter values ​​are further compared with pre-set health status thresholds representing different stages of the capacitor's lifespan, from brand new to the end of its lifespan, thereby mapping the quantified parameter indicators to specific health levels (such as healthy, warning, or fault). This ultimately completes the closed loop from raw data acquisition to advanced decision support, transforming complex electrical parameter analysis results into intuitive and operable assessment conclusions of the wind power converter capacitor's health status, providing final and decisive technical support for predictive maintenance and avoiding unplanned downtime.

[0047] More specifically, in a specific example of this application, the implementation of the health status assessment includes the following steps: First, defining health status assessment rules. This step is based on datasheets provided by the capacitor manufacturer or aging failure standards determined through experimental testing. These standards define the changes in the temperature-compensated ESR and capacitance values ​​of the capacitor as a function of aging throughout its lifespan. Specifically, an initial ESR value and an initial capacitance value are set, representing the capacitor's new state. Simultaneously, an ESR failure threshold and a capacitance failure threshold are set, representing the end of the capacitor's lifespan. For example, ESR_EOL is defined as twice the initial value, and C_EOL is defined as 80% of the initial value. Furthermore, for more refined management, an ESR warning threshold and a capacitance warning threshold are further set; for example, the ESR warning threshold is defined as 1.5 times the initial value, and the capacitance warning threshold is defined as 85% of the initial value. Second, executing health status judgment logic. This step involves logically comparing the real-time acquired temperature-compensated ESR estimate and temperature-compensated capacitance estimate with the preset thresholds. The system will execute a multi-condition judgment logic. If the temperature-compensated ESR estimate is greater than the ESR failure threshold or less than the capacitor failure threshold, the health status assessment result is judged as faulty or requires immediate replacement. If the temperature-compensated ESR estimate is greater than the ESR warning threshold or less than the capacitor warning threshold, but has not yet reached the failure threshold, the health status assessment result is judged as warning or requires attention. If all parameters are within the normal range defined by the warning threshold, the health status assessment result is judged as healthy or normal. The third step is to output and record the assessment results. This step solidifies and presents the output of the judgment logic. The system records the generated health status assessment results (e.g., text labels for health, warning, and fault, or corresponding status codes) along with the current timestamp, the corresponding temperature-compensated ESR estimate, and the capacitor estimate in the log database, forming a complete health profile record. Simultaneously, this assessment result is sent to the wind turbine's local monitoring system or remote control center, and displayed to maintenance personnel through a human-machine interface in a graphical or list-based manner, thus completing a full health status assessment process.

[0048] Figure 5 This is a block diagram of a capacitor quasi-online condition monitoring system for wind power converters according to an embodiment of this application, such as... Figure 5As shown, the quasi-online state monitoring system 100 for capacitors of wind power converters according to an embodiment of this application includes: a data slice sampling module 110, used to extract a first wind power converter data slice from a data buffer, the first wind power converter data slice including multiple sampling points, each sampling point including DC bus voltage, capacitor branch current, capacitor casing temperature and converter output power; a data slice input module 120, used to input the first wind power converter data slice into a quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice; and a harmonic identification module 130, used to, if the first wind power converter data slice is a quasi-steady-state data slice, to determine whether the first wind power converter data slice is a quasi-steady-state data slice. The data slices of the power converter are subjected to adaptive harmonic identification to obtain harmonic identification results; the estimation calculation module 140 is used to perform original capacitance estimation and ESR calculation based on the harmonic identification results to obtain original capacitance estimates and original ESR estimates; the temperature-compensated estimation calculation module 150 is used to extract the measurement temperature from the first wind power converter data slice, and perform temperature compensation on the original capacitance estimates and original ESR estimates based on the measurement temperature to obtain temperature-compensated ESR estimates and temperature-compensated capacitance estimates; the result evaluation module 160 is used to perform health status evaluation on the temperature-compensated ESR estimates and temperature-compensated capacitance estimates to obtain health status evaluation results.

[0049] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for quasi-online condition monitoring of capacitors in wind power converters, characterized in that, include: Extract the first wind power converter data slice from the data buffer. The first wind power converter data slice includes multiple sampling points, each of which includes DC bus voltage, capacitor branch current, capacitor casing temperature and converter output power. The first wind power converter data slice is input into a quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice. In response to the fact that the first wind power converter data slice is a quasi-steady-state data slice, adaptive harmonic identification is performed on the first wind power converter data slice to obtain the harmonic identification result; Based on the harmonic identification results, the original capacitance estimation and ESR calculation are performed to obtain the original capacitance estimate and the original ESR estimate. The measurement temperature is extracted from the data slice of the first wind power converter, and the original capacitance estimate and the original ESR estimate are temperature compensated based on the measurement temperature to obtain the temperature compensated ESR estimate and the temperature compensated capacitance estimate. A health status assessment is performed based on the temperature-compensated ESR estimate and the temperature-compensated capacitance estimate to obtain the health status assessment result.

2. The method for quasi-online status monitoring of capacitors in wind power converters according to claim 1, characterized in that, The first wind power converter data slice is input into a quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice, including: Extract the time series of converter output power from the data slice of the first wind power converter; Extract the statistical features of the time series of the converter output power, including the average output power and the standard deviation of the output power; The determination of whether the first wind power converter data slice is a quasi-steady-state data slice is made by judging whether the statistical characteristics meet the triggering conditions.

3. The method for quasi-online status monitoring of capacitors in wind power converters according to claim 2, characterized in that, The triggering condition is that the average output power is greater than a first preset threshold and the standard deviation of the output power is less than a second preset threshold.

4. The method for quasi-online status monitoring of capacitors in wind power converters according to claim 1, characterized in that, In response to the fact that the first wind power converter data slice is a quasi-steady-state data slice, adaptive harmonic identification is performed on the first wind power converter data slice to obtain harmonic identification results, including: Extract the time series of DC bus voltage and capacitor branch current from the data slice of the first wind power converter; Fast Fourier transform is performed on the time series of the DC bus voltage and the time series of the capacitor branch current to obtain the voltage spectrum and the current spectrum; Within a predetermined frequency search range, peak search and main harmonic determination are performed on the voltage spectrum to obtain the main harmonic frequency and the frequency index corresponding to the main harmonic frequency. Based on the frequency index corresponding to the main harmonic frequency, the main harmonic voltage amplitude, main harmonic voltage phase, main harmonic current amplitude, and main harmonic current phase are extracted from the voltage spectrum and the current spectrum.

5. The method for quasi-online status monitoring of capacitors in wind power converters according to claim 4, characterized in that, The predetermined frequency search range is 200Hz to 2000Hz.

6. The method for quasi-online status monitoring of capacitors in wind power converters according to claim 5, characterized in that, Within a predetermined frequency search range, peak search and main harmonic determination are performed on the voltage spectrum to obtain the main harmonic frequency and the frequency index corresponding to the main harmonic frequency, including: Traverse all frequency indices of the voltage spectrum within the predetermined frequency search range, and extract the voltage amplitude corresponding to each frequency index from the voltage spectrum to obtain a sequence of voltage amplitudes; Find the largest voltage amplitude from the sequence of voltage amplitudes and record its frequency index; The frequency index corresponding to the largest voltage amplitude is converted into the actual frequency value to obtain the main harmonic frequency.

7. The method for quasi-online status monitoring of capacitors in wind power converters according to claim 4, characterized in that, Based on the harmonic identification results, the original capacitance is estimated and the ESR is calculated to obtain the original capacitance estimate and the original ESR estimate, including: Based on the magnitude and phase of the main harmonic voltage, construct complex voltage phasors and complex current phasors; Calculate the complex impedance of the capacitor based on the complex voltage phasor and the complex current phasor. Extract the real part of the complex impedance of the capacitor as the original ESR estimate; The original capacitance estimate is calculated based on the imaginary part of the complex impedance of the capacitor.

8. The method for quasi-online status monitoring of capacitors in wind power converters according to claim 7, characterized in that, The measurement temperature is extracted from the data slice of the first wind power converter, and the original capacitance estimate and original ESR estimate are temperature compensated based on the measurement temperature to obtain the temperature compensated ESR estimate and temperature compensated capacitance estimate, including: The capacitor casing temperature at the current sampling point is extracted from the data slice of the first wind power converter and used as the temperature at the time of measurement. Based on the comparison between the measured temperature and the reference temperature, the first temperature compensation coefficient and the second temperature compensation coefficient are obtained through a two-dimensional lookup table. Based on the first temperature compensation coefficient and the second temperature compensation coefficient, temperature compensation is performed on the original capacitance estimate and the original ESR estimate during the measurement to obtain the temperature-compensated ESR estimate and the temperature-compensated capacitance estimate, respectively.

9. A quasi-online condition monitoring system for capacitors in wind power converters, characterized in that, include: The data slicing sampling module is used to extract the first wind power converter data slice from the data buffer. The first wind power converter data slice includes multiple sampling points, and each sampling point includes DC bus voltage, capacitor branch current, capacitor casing temperature and converter output power. The data slice input module is used to input the first wind power converter data slice into the quasi-steady-state slice trigger to determine whether the first wind power converter data slice is a quasi-steady-state data slice. The harmonic identification module is used to perform adaptive harmonic identification on the first wind power converter data slice, which is a quasi-steady-state data slice, to obtain the harmonic identification result. The estimation calculation module is used to perform original capacitance estimation and ESR calculation on the harmonic identification results to obtain the original capacitance estimate and the original ESR estimate. The temperature-compensated estimation module is used to extract the measurement temperature from the data slice of the first wind power converter, and perform temperature compensation on the original capacitance estimate and the original ESR estimate based on the measurement temperature to obtain the temperature-compensated ESR estimate and the temperature-compensated capacitance estimate. The results evaluation module is used to evaluate the health status of the temperature-compensated ESR estimate and the temperature-compensated capacitance estimate to obtain the health status evaluation results.

Images