An inverter anomaly detection method and system based on an industrial cloud platform

Abstract

The application provides an inverter abnormality detection method and system based on an industrial cloud platform, and relates to the technical field of inverter abnormality detection.The method comprises the following steps: collecting real-time operation data of each inverter and sending the data to an industrial cloud platform; constructing an inverter power prediction model; calculating predicted power through the inverter power prediction model according to the real-time operation data; calculating a power deviation value according to actual power and predicted power, and determining whether the deviation is greater than a preset error value in a continuous preset number of times; if yes, continuing to perform current waveform analysis; otherwise, returning to the initial detection; extracting the amplitude of a specific frequency component of current waveform data in the real-time operation data; calculating current harmonic indicators of input and output currents according to the components; determining whether any harmonic indicator exceeds a critical threshold; if yes, determining that the inverter is in an aging critical state, and replacing the aging device by using a differentiated online replacement strategy; otherwise, directly returning to the initial detection.

Description

Technical Field

[0001] This invention relates to the field of inverter anomaly detection technology, and in particular to an inverter anomaly detection method and system based on an industrial cloud platform. Background Technology

[0002] Inverters, as core components of power electronic systems, are widely used in photovoltaic power generation, wind power generation, electric vehicle drive, and industrial control. Their primary function is to convert direct current (DC) to alternating current (AC) to meet the power quality and frequency requirements of different application scenarios. Because inverters operate under high voltage, high frequency, and high power conditions, they are highly susceptible to factors such as temperature changes, electromagnetic interference, and load fluctuations, leading to various anomalies or faults, such as open circuits, short circuits, overheating, overload, and parameter drift. Therefore, anomaly detection technology for inverters has become a key research direction for ensuring the stable operation of power systems and the safe use of equipment.

[0003] Currently, inverter anomaly detection mainly employs two categories: diagnostic methods based on artificial intelligence models and analytical methods based on signal processing or model inference. The former typically utilizes deep learning or neural network algorithms to train a large number of known inverter fault samples, thereby establishing a fault identification model to achieve pattern classification and intelligent judgment of inverter operating status. The latter primarily extracts and analyzes features from the inverter's power or current signals, using methods such as spectrum analysis, wavelet transform, and parameter estimation to infer the inverter's operational health status or potential anomalies based on signal characteristic changes. Regarding inverter anomaly detection based on industrial cloud platforms, in recent years, with the gradual application of industrial cloud platforms in the power operation and maintenance field, some systems have begun to attempt to upload inverter operating data to the cloud to achieve centralized data management and remote monitoring across devices.

[0004] However, existing AI-based methods typically rely on a large number of fault samples for training, making it difficult to adapt to new fault modes that inverters encounter in complex environments. Furthermore, they suffer from poor real-time performance and limited operability. Traditional signal processing or model inference methods are mostly used for localized or offline analysis and have not fully leveraged the advantages of industrial cloud platforms in large-scale data fusion, online modeling, and real-time analysis. As a result, current remote monitoring of inverters still lacks efficient and real-time anomaly detection capabilities, which affects the system's operational stability and maintenance efficiency. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this invention is to provide an inverter anomaly detection method based on an industrial cloud platform. This method can solve the problems of existing artificial intelligence-based methods, which usually rely on a large number of fault samples for training, making it difficult to adapt to new fault modes that inverters may encounter in complex environments, and also have poor real-time performance and limited operability. Traditional signal processing or model inference methods are mostly used for localized or offline analysis and have not fully utilized the advantages of industrial cloud platforms in large-scale data fusion, online modeling, and real-time analysis. As a result, current remote inverter monitoring still lacks efficient and real-time anomaly detection capabilities, which affects the system's operational stability and maintenance efficiency.

[0006] A first aspect of this invention proposes an inverter anomaly detection method based on an industrial cloud platform, applied to an industrial cloud platform, wherein each distributed inverter is communicatively connected to the industrial cloud platform, comprising: S1: Collect real-time operating data of each inverter and send it to the industrial cloud platform; S2: Construct an inverter power prediction model; S3: Based on the real-time operating data, calculate the predicted power of the inverter using the inverter power prediction model; S4: Calculate the power deviation value based on the actual power in the real-time operating data and the predicted power, and determine whether the power deviation value for a consecutive preset number of times is greater than the preset error value; if yes, proceed to the next step; otherwise, return to S1; S5: Extract the amplitude of a specific component of the current waveform data from the real-time running data; S6: Calculate the input current harmonic index and the output current harmonic index based on the amplitude of the specific component; S7: Determine whether the input current harmonic index or the output current harmonic index is greater than the critical threshold; if so, determine that the inverter is in the critical aging state and proceed to S8; otherwise, return to S1 to continue collecting the inverter's operating data. S8: Output a replacement notification, use a differentiated online replacement strategy to replace the aging components of the inverter, and return to S1.

[0007] A second aspect of this invention provides an inverter anomaly detection system based on an industrial cloud platform, comprising: a processor and a memory; The memory stores programs or instructions that can run on the processor, and when the programs or instructions are executed by the processor, they implement the steps of the inverter anomaly detection method based on the industrial cloud platform of the first aspect.

[0008] A third aspect of the present invention provides a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implements the steps of the inverter anomaly detection method based on an industrial cloud platform as described in the first aspect.

[0009] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this embodiment of the invention, real-time operational data is collected and sent to an industrial cloud platform. A power prediction model is then constructed on the cloud platform to predict inverter power and combine this prediction with actual power for continuous deviation assessment. This enables the identification of potential anomalies without relying on a large number of historical fault samples, improving the method's adaptability to complex operating conditions and novel fault modes. Simultaneously, leveraging the cloud platform's advantages in large-scale real-time analysis and online modeling, current waveform data is analyzed. Waveform analysis and harmonic index extraction are triggered only when continuous power deviation anomalies occur, effectively reducing computational overhead and improving detection efficiency and system resource utilization. Furthermore, state determination based on harmonic thresholds, combined with a differentiated online replacement strategy, enables rapid identification and immediate processing of inverter aging states, significantly enhancing system stability, real-time detection, and the method's engineering feasibility. Attached Figure Description

[0010] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. Obviously, the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0011] Figure 1 This is a flowchart illustrating an inverter anomaly detection method based on an industrial cloud platform provided in an embodiment of the present invention.

[0012] Figure 2 This is a schematic diagram of the structure of an inverter anomaly detection system based on an industrial cloud platform provided in an embodiment of the present invention. Detailed Implementation

[0013] To enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0014] The inverter anomaly detection method based on an industrial cloud platform provided by the present invention will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.

[0015] Reference manual attached Figure 1 The diagram shows a flowchart of an inverter anomaly detection method based on an industrial cloud platform provided by an embodiment of the present invention.

[0016] This invention provides an inverter anomaly detection method based on an industrial cloud platform. The method is applied to an industrial cloud platform where each distributed inverter is communicatively connected to the platform. The method may include the following steps: S1: Collects real-time operating data from each inverter and sends it to the industrial cloud platform.

[0017] In one possible implementation, the real-time operating data specifically includes: current waveform data, inverter temperature, historical irradiance, historical power generation, input current, output current, total irradiance of the array plane, and actual power generation.

[0018] Specifically, current waveform data, inverter temperature, historical irradiance, historical power generation, input current, output current, total irradiance of the array plane, and actual power generation are all acquired in real time through the inverter's built-in voltage and current sensors, temperature sensors, irradiance meters, and their supporting data acquisition modules, and uploaded to the industrial cloud platform for unified management and analysis via the inverter's communication interface.

[0019] It should be noted that the real-time operational data encompasses multi-dimensional information reflecting the inverter's operating status and performance characteristics, including current waveform data, inverter temperature, historical irradiance, historical power generation, input current, output current, total irradiance of the array plane, and actual power generation. By collecting this multi-source operational data and sending it to the industrial cloud platform, a comprehensive characterization of the inverter's electrical characteristics, environmental changes, and historical operating patterns is achieved. This provides a sufficient data foundation for subsequent power prediction modeling, health status assessment, and anomaly detection, thereby ensuring the system's prediction accuracy, anomaly identification capability, and real-time operational reliability under complex operating conditions.

[0020] S2: Construct an inverter power prediction model.

[0021] The inverter power prediction model is a mathematical model built on the relationship between the total irradiance of the array plane and historical power generation data. It is used to predict in real time the power output of the photovoltaic inverter under current irradiance conditions. This model typically uses a linear regression method and dynamically updates the coefficients through a recursive least squares algorithm to achieve adaptation to environmental changes and component aging.

[0022] In one possible implementation, S2 specifically includes: S201: Obtain historical sample data and calculate the initial coefficients using the least squares method based on the historical sample data. in, This represents the initial coefficients, and N represents the total number of sampling points. This represents the total irradiance of the array plane at the i-th sampling point. This represents the actual power generation at the i-th sampling point.

[0023] Least squares is a classic mathematical optimization method used to find the optimal function model in data fitting that minimizes the sum of squared errors between observed and model predicted values. In regression analysis, it determines the optimal solution for model parameters by minimizing the sum of squared residuals (differences between actual and fitted values) for all sample points.

[0024] It should be noted that the historical sample data includes the total irradiance and actual power generation of the array plane from multiple historical data.

[0025] In this embodiment of the invention, the initial regression coefficients are calculated using the least squares method based on historical sample data, which can establish a reasonable starting point for the model. The least squares method minimizes the sum of squared errors between the predicted and actual power values, thereby obtaining an optimal fitting result based on historical data. This not only ensures that the model has strong predictive ability from the outset but also provides a robust foundation for subsequent recursive updates, improving the accuracy and reliability of the overall prediction model.

[0026] S202: Set the initial covariance matrix and forgetting factor.

[0027] It should be noted that setting the initial covariance to a large number can speed up the model's response to new data in the early stages and allow it to quickly adapt to the current operating conditions. The introduction of a forgetting factor, on the other hand, can reduce the influence of old data, enabling the model to maintain its adaptive capability under dynamic changes such as data drift or component aging.

[0028] Optionally, the covariance matrix P0 = 10000 (large number guarantees fast convergence), and the forgetting factor λ = 0.99.

[0029] S203: Based on the initial coefficients, the initial covariance matrix, and the forgetting factor, calculate the dynamic update coefficients and the covariance matrix used for the recursive dynamic update coefficients online recursively.

[0030] In one possible implementation, S203 specifically includes: S2031: Calculate the prior error: in, Let represent the prior error at time t. This represents the actual power generation at time t. This represents the dynamic update coefficient at time t-1. This represents the total irradiance of the array plane at time t.

[0031] S2032: Calculate the recursive gain matrix: in, This represents the recursive gain coefficient at time t. Let represent the covariance matrix at time t-1. This represents the forgetting factor.

[0032] S2033: Based on the initial coefficients, prior error, and recursive gain matrix, dynamically update the coefficients online recursively. in, This indicates that the coefficients are updated dynamically.

[0033] S2034: Calculate the covariance matrix online recursively based on the forgetting factor, the initial covariance matrix, and the gain matrix. in, Let represent the covariance matrix at time t.

[0034] In this embodiment of the invention, by updating the regression coefficients and covariance matrix in real time, the recursive least squares method can continuously optimize the power prediction model, ensuring it always reflects the latest environmental and equipment status. Unlike static models, the online update mechanism allows the system to dynamically adjust model parameters, improving the real-time performance and accuracy of inverter performance detection.

[0035] S204: Construct an inverter power prediction model based on the dynamically updated coefficients: in, This represents the predicted power generation at time t. This represents the dynamic update coefficient at time t. This represents the total irradiance of the array plane at time t.

[0036] Specifically, the system constructs an inverter power prediction model based on an industrial cloud platform to assess in real-time the power output of the inverter under current illumination conditions. The industrial cloud platform first acquires historical total irradiance and historical power generation data of the array plane uploaded by field devices, and calculates initial regression coefficients using the least squares method to establish a basic predictive model. Based on this, the platform sets an initial covariance matrix and a forgetting factor to enable the model to adapt to environmental changes and component aging. Subsequently, the industrial cloud platform dynamically updates the regression coefficients and covariance matrix using a recursive least squares algorithm based on real-time uploaded irradiance and power generation data, progressively optimizing the model parameters so that the prediction model continuously reflects the latest operating status of the inverter. Finally, the cloud platform uses the updated dynamic coefficients to calculate the real-time predicted power generation of the inverter, achieving data-driven, high-precision online power prediction.

[0037] S3: Based on real-time operating data, calculate the predicted power of the inverter using the inverter power prediction model.

[0038] S4: Based on the actual power and predicted power in the real-time operating data, calculate the power deviation value and determine whether the power deviation value for a consecutive preset number of times is greater than the preset error value. If yes, proceed to the next step. Otherwise, return to S1.

[0039] Optionally, the preset number of attempts is 3.

[0040] A sustained increase in power deviation is often an early sign of inverter performance degradation (such as MOSFET aging and deteriorating heat dissipation). By analyzing the trend of deviation values, abnormal trends can be detected before the equipment completely fails, enabling intelligent early warning and providing a time window for subsequent maintenance and component replacement.

[0041] Specifically, we designed a mechanism based on "power deviation judgment" to determine whether further anomaly analysis should be triggered. This part belongs to the "preliminary screening" stage of the entire method and is characterized by its lightweight, high efficiency, and intelligence. By comparing the actual power generation of the inverter with the theoretical power generation output by the prediction model in real time, the difference between them is calculated. If this deviation exceeds the pre-set error tolerance range in multiple consecutive samplings, the inverter is considered to have a potential anomaly and proceeds to the next stage of in-depth diagnosis. Otherwise, routine testing continues.

[0042] In this embodiment of the invention, by comparing the actual power generation with the predicted power in real time, calculating the power deviation, and determining whether it exceeds a set error threshold at multiple consecutive moments, a highly robust and sensitive intelligent anomaly screening mechanism is achieved. This model-based dynamic deviation judgment method can not only effectively identify early signs of equipment performance degradation or potential faults, but also suppress false alarms caused by light fluctuations, measurement noise, etc., thereby improving the accuracy of fault detection and the stability of system operation. In addition, the "continuous deviation judgment" strategy can also achieve trend tracking of equipment status, providing reliable data for subsequent refined diagnosis and intelligent maintenance, and is an important link connecting predictive analysis and fault location.

[0043] It should be noted that those skilled in the art can set the preset number of times and the preset error value according to actual needs, and this invention does not limit this.

[0044] S5: Extract the amplitude of a specific component of the current waveform data from the real-time running data.

[0045] Current waveform data refers to the continuous recording of the change of current in a circuit over a certain time range, reflecting the instantaneous characteristics and periodic behavior of the current. In power electronic systems such as photovoltaic inverters, current waveform data includes DC input current and AC output current, and can be used to analyze equipment operating status, power quality, and potential faults.

[0046] In one possible implementation, S5 specifically includes: S501: Collects DC input current and AC output current for multiple cycles in real-time operation data.

[0047] S502: Using Fast Fourier Transform, frequency domain analysis is performed on the DC input current and AC output current for each cycle to extract the amplitude of specific components of the input current and output current corresponding to each cycle.

[0048] The Fast Fourier Transform (FFT) is an efficient algorithm for calculating the Discrete Fourier Transform (DFT) and its inverse. It transforms a signal from the time domain to the frequency domain, thereby revealing the amplitude and phase characteristics of each frequency component in the signal. Compared to directly calculating the DFT, the FFT significantly reduces computational complexity, making it a core tool in signal processing, harmonic analysis, and spectral feature extraction.

[0049] In this embodiment of the invention, the current waveform in the time domain is converted into a frequency domain signal using Fast Fourier Transform (FFT), which clearly identifies the various frequency components and their amplitudes contained in the current. This is crucial for analyzing the power quality of inverters, determining whether harmonics are abnormal, and identifying aging or distortion problems at specific frequencies. The high efficiency of FFT also makes it suitable for real-time or embedded systems, improving system response speed and processing efficiency.

[0050] S503: By averaging, the amplitude of the specific component of the input current and the amplitude of the specific component of the output current in each cycle are denoised to obtain the amplitude of the specific component of the input current and the amplitude of the specific component of the output current.

[0051] S504: Combines the amplitude of a specific component of the denoised input current and the amplitude of a specific component of the denoised output current to form the amplitude of a specific component of the current waveform data.

[0052] In this embodiment of the invention, the amplitude of the specific frequency component of the current extracted from multiple cycles is averaged and denoised, which can effectively filter out high-frequency noise and random fluctuations introduced by environmental, power grid disturbances or sampling errors, thereby obtaining more stable and reliable feature values. This helps to improve the accuracy of subsequent harmonic index calculation and anomaly judgment, reduce the risk of misjudgment and missed judgment, and enhance the anti-interference capability of the system.

[0053] Specifically, to achieve high-quality acquisition of current waveforms and extract amplitude characteristics of key frequency components to provide data support for subsequent harmonic index analysis and anomaly detection, this step mainly includes the following three sub-steps: To ensure data temporal consistency, we synchronously sample the DC input current and AC output current, with the sampling time span covering multiple complete current cycles to obtain representative waveform data. The acquired current waveform is initially a time-domain signal that varies with time, but time-domain data cannot directly reveal which frequency components are included in the current. We use Fast Fourier Transform to convert it to the frequency domain, thus clearly showing the amplitude of key frequency components in the current. These frequencies correspond to the fundamental frequency of the power grid and its harmonics, effectively reflecting power quality and device status. To make the frequency characteristics more stable, we average the amplitude of the same frequency component extracted from multiple cycles. This method can reduce the noise impact caused by short-term fluctuations, external interference, or measurement errors, ultimately obtaining smoother and more reliable characteristic values.

[0054] It should be noted that during long-term operation of the inverter, power devices such as MOSFETs may experience performance degradation due to factors such as heat, aging, and environmental conditions, resulting in distortion of the current waveform. Traditional electrical parameter detection methods often fail to sensitively reflect this change. Therefore, we introduce a "current harmonic index" method based on frequency analysis. By analyzing the frequency domain characteristics of the input and output currents, we extract harmonic indices that reflect aging and distortion characteristics, providing data support for subsequent anomaly detection and equipment maintenance.

[0055] S6: Calculate the input current harmonic index and the output current harmonic index based on the amplitude of a specific component.

[0056] Among them, the input current harmonic index is a characteristic parameter that measures the relative intensity of harmonic components in the input current of the inverter. It is usually calculated by comparing the amplitude of the input current at 50Hz (fundamental wave) and 100Hz (secondary harmonic wave) frequency components, reflecting the degree of distortion of the input current waveform.

[0057] Among them, the output current harmonic index is a key parameter used to evaluate the quality of the output current waveform of a photovoltaic inverter. It is usually calculated by comparing the ratio of the amplitude of the 100Hz subharmonic to the amplitude of the 50Hz fundamental current, reflecting the relative intensity of the harmonic components in the output current.

[0058] In one possible implementation, the specific component amplitude specifically includes: the 50Hz component amplitude of the denoised input current, the 100Hz component amplitude of the denoised input current, the 50Hz component amplitude of the denoised output current, and the 100Hz component amplitude of the denoised output current. S6 specifically includes: S601: Calculate the input current harmonic index based on the amplitude of the 50Hz component and the amplitude of the 100Hz component of the denoised input current. in, This indicates the input current harmonics index. This indicates the amplitude of the 50Hz component of the noise-reducing input current. This indicates the amplitude of the 100Hz component of the noise-reducing input current.

[0059] It's important to note that aging of power devices (such as MOSFETs) can affect the conduction characteristics of the input current and introduce harmonics. Changes in harmonics directly reflect the health status of the device on the input side. Ideally, the input current should be "stable DC" on the DC input side. However, due to the switching action of the inverter's MOSFETs, a certain low-frequency ripple component is superimposed on the input current. The 100Hz component is the main interference component (because the 50Hz fundamental frequency of the power grid manifests as a 100Hz ripple on the DC side after rectification), while the 50Hz component is usually a secondary harmonic caused by aging of the switching devices and unstable drive. Therefore, the health status of the input side is described by the ratio of 50Hz / 100Hz: if the MOSFETs age, the 50Hz component increases significantly, causing this ratio to decrease, reflecting input distortion.

[0060] S602: Calculate the output current harmonic index based on the amplitude of the 50Hz component and the amplitude of the 100Hz component of the denoised output current. in, This indicates the output current harmonic distortion index. This indicates the amplitude of the 50Hz component of the noise-reducing output current. This indicates the amplitude of the 100Hz component of the noise-reducing output current.

[0061] It should be noted that on the AC output side, the current should be mainly based on the 50Hz fundamental frequency, which is the reference frequency of the power grid. The 50Hz component is the "normal component" of the output current, and the stronger it is, the healthier it is. The 100Hz component is the main low-order harmonic component, which often comes from the nonlinear effects inside the inverter (such as device aging and drive distortion). Therefore, the output side uses the ratio of 100Hz / 50Hz to measure the degree of distortion: when the MOSFET ages, the 100Hz component increases and the proportion rises, which can clearly reflect the waveform distortion.

[0062] In this embodiment of the invention, by analyzing the amplitude ratio of the denoised input current at the fundamental (50Hz) and subharmonic (100Hz) frequencies, an evaluation index for the input-side harmonic intensity is constructed, forming a simple yet efficient inverter health assessment mechanism. Compared to traditional full-spectrum analysis, this method only needs to focus on two frequency points, requires less computation, and is easy to implement in real time. Simultaneously, it possesses good sensitivity and trend-reflecting capabilities, enabling early detection of device aging and current distortion issues, thereby improving the targetedness and accuracy of diagnosis.

[0063] S7: Determine if the input current harmonic index or the output current harmonic index exceeds the critical threshold. If yes, determine that the inverter is in a critical aging state and proceed to S8. Otherwise, return to S1 to continue collecting the inverter's operating data.

[0064] It should be noted that during long-term operation of the inverter, internal power devices (such as MOSFETs) will gradually age due to time, thermal stress, and load fluctuations, manifesting as deterioration in conduction performance and distortion of the current waveform. To achieve intelligent health monitoring of the inverter, we have already extracted key frequency components from the current waveform and calculated the input and output current harmonic indices in previous steps. This step will compare these two indices with preset "critical thresholds" to determine whether the inverter is in a critical aging state, thus deciding whether to proceed to the next step of device replacement.

[0065] Specifically, the calculated input current harmonic index and output current harmonic index are compared with the preset critical threshold. If either condition is met, it indicates that the current waveform distortion has reached the aging critical level. The system determines that the inverter is in the aging critical state, and the process enters S8 to execute the device replacement strategy. If neither index exceeds the critical threshold, the inverter is determined to be in normal condition, and the process returns to S1 to continue detection.

[0066] In this embodiment of the invention, a critical threshold judgment mechanism based on harmonic indices is introduced, providing a simple, reliable, and automatically executable decision-making method for inverter aging detection. It not only achieves quantitative identification of the degree of distortion in the input and output current waveforms but also effectively improves the sensitivity to early aging problems of power devices. By setting a threshold for identifying critical states, it lays the foundation for subsequent graded maintenance and differentiated replacement strategies, and is a crucial step in realizing intelligent operation and maintenance and adaptive diagnosis of inverters.

[0067] In one possible implementation, the determination of the critical threshold specifically includes: S701: Determine the operating parameters of the inverter, including the input voltage and load resistance.

[0068] In this embodiment of the invention, the critical threshold should be calculated under actual operating conditions, rather than using a uniform, fixed threshold. Different inverters operate differently under different voltages and loads, and their sensitivity to changes in on-resistance also varies. By introducing "operating condition parameters," more targeted modeling can be achieved, improving the accuracy and applicability of detection, and ensuring that the threshold will not lead to misjudgment or missed judgment due to mismatch.

[0069] S702: Based on the aging characteristics of the power MOSFETs in the inverter, set the critical increment value of the on-resistance: in, This represents the critical increment value of the on-resistance. This represents the critical aging coefficient. This represents the initial on-resistance of the MOSFET.

[0070] Optionally, the critical aging factor is 5%.

[0071] For example, when using an STP8NM60 MOSFET, if its initial on-resistance is 0.9Ω, then the corresponding critical increment is: 0.05 × 0.9 = 45mΩ.

[0072] In this embodiment of the invention, based on the physical aging characteristics of MOSFETs, the judgment criteria are linked to the device's own characteristics (rather than empirical data), making the model more theoretically sound and engineeringly reasonable. Setting it to "a certain percentage of the initial on-resistance" allows for flexible adjustment based on the initial parameters of different devices, enhancing the method's portability and scalability.

[0073] S703: Based on operating condition parameters, circuit simulation software is used to model the aging process of the inverter and simulate the relationship between the critical increment of the on-resistance and the harmonic detection index. in, Indicates harmonic detection index, This represents any increment in on-resistance. Represents the slope parameter. This represents the harmonic detection index value under the initial state.

[0074] It should be noted that the initial harmonic detection index value is a constant, which is usually known. When determining the threshold, we often only compare whether the "exceeding portion" meets the standard. Therefore, we only focus on... The incremental portion resulting from device aging.

[0075] S704: Substitute the critical increment value of the on-resistance into the variation formula to calculate the critical threshold under the corresponding operating condition: in, This represents the critical threshold.

[0076] Specifically, first, we determine the inverter's operating parameters, including input voltage and load resistance. Different inverters respond differently to changes in device performance under different operating conditions; therefore, the critical threshold should be dynamically calculated based on actual operating conditions, rather than applying fixed values. This significantly improves the accuracy of judgment and avoids misjudgments or missed detections. Next, based on the aging characteristics of the power MOSFET, we set the critical increment value of its on-resistance. Subsequently, we use circuit simulation software to model the inverter's operation under aging conditions, obtaining the functional relationship between on-resistance change and harmonic parameters. Finally, we use the above relationship to calculate the critical harmonic threshold under the corresponding operating conditions.

[0077] In this embodiment of the invention, by combining the simulation results from the previous step with the set critical on-resistance increment, the critical judgment value to be used under the current operating condition is accurately obtained. This critical value can serve as the standard for subsequent judgment on whether to trigger a replacement operation, realizing closed-loop logic between diagnostic and maintenance strategies. This also avoids the adaptation error caused by using static "experience thresholds," making the entire system adaptive and engineering feasible.

[0078] S8: Output a replacement notification, use a differentiated online replacement strategy to replace the aging components of the inverter, and return to S1.

[0079] It should be noted that the differentiated online replacement strategy output by the industrial cloud platform is an intelligent maintenance method that selectively replaces aging power devices based on the specific fault characteristics and harmonic index trends of the inverter. This strategy determines the fault location based on different combinations of input and output current harmonic indices. For example, if only the input side is abnormal, the upper bridge arm MOSFET is replaced; if the output side or both are abnormal, the entire MOSFET group is replaced, achieving precise location and online switching of faulty components.

[0080] In one possible implementation, S8 specifically includes: S801: If the notification output by the industrial cloud platform indicates that the input current harmonic index is greater than the critical threshold, it is determined that the upper bridge arm MOSFET of the inverter is aging, and the process proceeds to S803.

[0081] S802: If the notification output by the industrial cloud platform shows that the output current harmonic index is greater than the critical threshold, or that both the input current harmonic index and the output current harmonic index are greater than the critical threshold, proceed to S804.

[0082] It should be noted that when only the output current harmonic indicators exceed the critical threshold, it is impossible to accurately determine whether the aging device is located in the upper or lower bridge arm, or whether multiple devices are degrading simultaneously. To ensure that the fault is completely eliminated and the system is restored to normal operation, a more prudent strategy is adopted at this time, namely, replacing all MOSFETs. This avoids overlooking potential fault points or causing repeated repairs, and ensures the overall reliability and operational stability of the inverter.

[0083] S803: The microcontroller controls the startup of the backup upper bridge arm MOSFET, sends a turn-on signal to the gate of the upper bridge arm MOSFET, waits for a short overlap time, sends a turn-off signal to the gate of the aging upper bridge arm MOSFET to cut off the current path of the aging transistor, and performs hot-plug physical replacement of the aging upper bridge arm. After the replacement is completed, the backup upper bridge arm MOSFET is turned off, the new MOSFET is connected to the upper bridge arm, and the system returns to S1 to continue detection.

[0084] In this embodiment of the invention, by first sending a turn-on signal to the gate of the spare MOSFET and then delaying the turn-off of the aging device, a brief conduction overlap is achieved, effectively avoiding current interruption or surge, and ensuring the safety and electrical continuity of the replacement process. The use of hot-swappable physical replacement not only improves maintenance efficiency but also significantly reduces system downtime. After replacement, the device is immediately connected to the detection system, forming a self-recovering closed-loop process, thereby realizing an efficient, safe, and intelligent device replacement mechanism for the inverter during operation.

[0085] S804: Replace all MOSFETs in the same way as S803, then return to S1 to continue testing.

[0086] In this embodiment of the invention, the differentiated online replacement strategy output by the industrial cloud platform can accurately determine the location of aging components in the inverter based on harmonic indicators, enabling targeted replacement of the upper bridge arm or the entire MOSFET group. This avoids unnecessary replacement of all components and reduces maintenance costs. Simultaneously, the replacement process employs microcontroller control and hot-swapping operations, supporting online replacement without system downtime, ensuring uninterrupted system operation, and improving maintenance efficiency and system reliability. This strategy ensures both thorough troubleshooting and flexible, intelligent operation and maintenance.

[0087] The inverter anomaly detection method based on an industrial cloud platform provided in this application can be executed by an inverter anomaly detection device based on an industrial cloud platform. This application example illustrates the inverter anomaly detection device based on an industrial cloud platform, using the execution of the inverter anomaly detection method based on an industrial cloud platform as an example.

[0088] Reference manual attached Figure 2The diagram shows a structural schematic of an inverter anomaly detection system based on an industrial cloud platform provided by an embodiment of the present invention.

[0089] This invention provides an inverter anomaly detection system 20 based on an industrial cloud platform, comprising: a processor 201 and a memory 202; The memory 202 stores programs or instructions that can run on the processor 201. When the program or instructions are executed by the processor 201, they implement the steps of the above-described inverter anomaly detection method based on the industrial cloud platform and achieve the same technical effect. To avoid repetition, the present invention will not repeat the above-described steps.

[0090] It should be understood that the processor 201 in this embodiment of the invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0091] It should also be understood that the memory 202 in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DR RAM).

[0092] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0093] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0094] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0095] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0096] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0097] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0098] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0099] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0100] This invention provides a readable storage medium comprising: storing a program or instructions on the readable storage medium, wherein when the program or instructions are executed by a processor, the program or instructions implement the steps of the above-described inverter anomaly detection method based on an industrial cloud platform, and can achieve the same technical effect. To avoid repetition, this invention will not elaborate further.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting inverter anomalies based on an industrial cloud platform, characterized in that, Applied to an industrial cloud platform, each distributed inverter is communicatively connected to the industrial cloud platform, including: S1: Collect real-time operating data of each inverter and send it to the industrial cloud platform; S2: Construct an inverter power prediction model; S2 specifically includes: S201: Obtain historical sample data, and calculate the initial coefficients based on the historical sample data using the least squares method; S202: Set the initial covariance matrix and forgetting factor; S203: Based on the initial coefficients, the initial covariance matrix, and the forgetting factor, calculate the dynamic update coefficients and the covariance matrix used to derive the dynamic update coefficients online recursively; S204: Construct the inverter power prediction model based on the dynamic update coefficients; S3: Based on the real-time operating data, calculate the predicted power of the inverter using the inverter power prediction model; S4: Calculate the power deviation value based on the actual power in the real-time operating data and the predicted power, and determine whether the power deviation value for a consecutive preset number of times is greater than the preset error value; if yes, proceed to the next step; otherwise, return to S1; S5: Extract the amplitude of a specific component of the current waveform data from the real-time running data; S6: Calculate the input current harmonic index and the output current harmonic index based on the amplitude of the specific component; S7: Determine whether the input current harmonic index or the output current harmonic index is greater than the critical threshold; if so, determine that the inverter is in the critical aging state and proceed to S8; otherwise, return to S1 to continue collecting the inverter's operating data. The method for determining the critical threshold specifically includes: S701: Determine the operating condition parameters of the inverter, wherein the operating condition parameters include the input voltage and the load resistance; S702: Set the critical increment value of the on-resistance based on the aging characteristics of the power MOSFET in the inverter; S703: Based on the operating condition parameters, use circuit simulation software to model the aging process of the inverter and simulate the relationship between the critical increment value of the on-resistance and the harmonic detection index. S704: Substitute the critical increment value of the on-resistance into the change relationship to calculate the critical threshold under the corresponding operating condition; S8: Output a replacement notification, use a differentiated online replacement strategy to replace the aging components of the inverter, and return to S1.

2. The inverter anomaly detection method based on an industrial cloud platform according to claim 1, characterized in that, The real-time operating data specifically includes: total irradiance of the array plane and actual power generation.

3. The inverter anomaly detection method based on an industrial cloud platform according to claim 1, characterized in that, S203 specifically includes: S2031: Calculate the prior error; S2032: Calculate the recursive gain matrix; S2033: Calculate the dynamically updated coefficients online recursively based on the initial coefficients, the prior error, and the recursive gain matrix; S2034: Calculate the covariance matrix online recursively based on the forgetting factor, the initial covariance matrix, and the gain matrix.

4. The inverter anomaly detection method based on an industrial cloud platform according to claim 1, characterized in that, The current waveform data includes DC input current and AC output current; S5 specifically includes: S501: Collect the DC input current and AC output current for multiple cycles in the real-time operating data; S502: Using Fast Fourier Transform, frequency domain analysis is performed on the DC input current and AC output current for each cycle to extract the amplitude of specific components of the input current and the amplitude of specific components of the output current corresponding to each cycle. S503: The amplitude of the specific component of the input current and the amplitude of the specific component of the output current in each cycle are denoised by averaging to obtain the amplitude of the specific component of the input current and the amplitude of the specific component of the output current. S504: Combine the amplitude of the specific component of the denoised input current and the amplitude of the specific component of the denoised output current to form the amplitude of the specific component of the current waveform data.

5. The inverter anomaly detection method based on an industrial cloud platform according to claim 1, characterized in that, The specific component amplitudes specifically include: the 50Hz component amplitude of the denoised input current, the 100Hz component amplitude of the denoised input current, the 50Hz component amplitude of the denoised output current, and the 100Hz component amplitude of the denoised output current; S6 specifically includes: S601: Calculate the input current harmonic index based on the amplitude of the 50Hz component of the denoised input current and the amplitude of the 100Hz component of the denoised input current; S602: Calculate the output current harmonic index based on the amplitude of the 50Hz component of the denoised output current and the amplitude of the 100Hz component of the denoised output current.

6. The inverter anomaly detection method based on an industrial cloud platform according to claim 1, characterized in that, S8 specifically includes: S801: If the input current harmonic index is greater than the critical threshold in the notification output by the industrial cloud platform, it is determined that the upper bridge arm MOSFET of the inverter is aging, and proceeds to S803. S802: If the notification output by the industrial cloud platform contains an output current harmonic index that is greater than the critical threshold, or if both the input current harmonic index and the output current harmonic index are greater than the critical threshold, proceed to S804. S803: The microcontroller starts the backup upper bridge arm MOSFET, sends a turn-on signal to the gate of the upper bridge arm MOSFET, waits for a short overlap time, sends a turn-off signal to the gate of the aged upper bridge arm MOSFET to cut off the current path of the aged transistor, and performs hot-plug physical replacement of the aged upper bridge arm. After the replacement is completed, the backup upper bridge arm MOSFET is turned off, the new MOSFET is connected to the upper bridge arm, and the process returns to S1 to continue detection. S804: Replace all MOSFETs in the same way as S803, then return to S1 to continue testing.

7. An inverter anomaly detection system based on an industrial cloud platform, characterized in that, include: Processor and memory; The memory stores programs or instructions that can run on the processor, and when the programs or instructions are executed by the processor, they implement the steps of the inverter anomaly detection method based on an industrial cloud platform as described in any one of claims 1 to 6.

8. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions, which, when executed by a processor, implement the steps of the inverter anomaly detection method based on an industrial cloud platform as described in any one of claims 1 to 6.

Images