Fault detection method for direct current converter transformer

Abstract

The invention discloses a fault detection method for a direct current converter transformer, and particularly relates to the technical field of power equipment detection, and the method comprises the steps: S01, obtaining real-time voltage data, current data and temperature data based on each data collection unit, and constructing real-time operation state data of the direct current converter transformer according to a Kalman filtering data fusion algorithm; s02, learning the running state data through a deep learning algorithm, and predicting a load loss estimated value in real time; and S03, controlling the DC converter transformer according to the real-time temperature data on the basis of the predicted load loss estimated value in combination with a self-adaptive control algorithm for model prediction control, and adjusting a management and control strategy in combination with the DC converter transformer to generate a self-adaptive control strategy. According to the invention, real-time voltage data, current data and temperature data of the DC converter transformer are acquired and monitored, so that the operation safety of the DC converter transformer is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment detection, and particularly relates to a fault detection method for a DC converter transformer. Background Art

[0002] A DC converter transformer is a special transformer used in a high-voltage direct current (HVDC) system, mainly for converting alternating current into direct current, or for achieving voltage conversion between different voltage levels in a DC power transmission system. In the power supply systems of municipal engineering or rail transit, the electrical equipment is of high importance, and failures can endanger the operation safety. At the same time, the power facilities are deployed relatively dispersedly, making inspection and maintenance inconvenient. Often, it is necessary for staff to conduct manual inspections of power facilities during outages. To solve the inconvenience of manual inspections, the operating data of the DC converter transformer is collected to obtain the operating state of the DC converter transformer, and then a health assessment is carried out.

[0003] For example, as disclosed (announced) in Publication No.: CN118033492A, Publication (Announcement) Date: May 14, 2024, a fault detection method, device and test system for a flexible DC converter transformer are provided. The method includes: receiving the condition information of the operating condition to be detected; obtaining the target voltage waveform data corresponding to the condition information from a preset voltage waveform database, where the preset voltage waveform database includes: the condition information of various operating conditions and their respective corresponding voltage waveform data; obtaining a voltage signal based on the target voltage waveform data, and sending the voltage signal to the flexible DC converter transformer to be detected after being amplified proportionally by a voltage amplifier, so as to complete the fault detection of the flexible DC converter transformer to be detected under the operating condition to be detected. This application can realize fault detection reflecting the actual operating conditions of the flexible DC converter transformer, improve the reliability of fault detection, and further improve the operating safety and reliability of the flexible DC system.

[0004] In the prior art including the above patent, the detection of the DC converter transformer mainly stays at the detection of voltage signals, lacking the detection of current signals. And the change of temperature during the operation of the DC converter transformer has a very serious impact on voltage and current, so it lacks comprehensive guiding significance. Summary of the Invention

[0005] The purpose of the present invention is to provide a fault detection method for a DC converter transformer to solve the above problems.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A fault detection method for a DC converter transformer, including:

[0007] S01. Obtain real-time voltage data, current data, and temperature data based on each data acquisition unit, and construct real-time operating state data of the DC converter transformer according to the Kalman filter data fusion algorithm;

[0008] S02. Learn the operating state data through a deep learning algorithm and predict the load loss estimate in real time;

[0009] S03. Based on the predicted load loss estimate and combined with the adaptive control algorithm using model predictive control, control the DC converter transformer according to the real-time temperature data, and combine the DC converter transformer adjustment and control strategy to generate an adaptive control strategy;

[0010] S04. Combine the operating state data and the predicted load loss estimate, use a machine learning algorithm to monitor the DC converter transformer in real time, and generate an intelligent protection and abnormal warning report for abnormal variable data.

[0011] Preferably, the construction of the real-time operating state data of the DC converter transformer in step S01 includes:

[0012] S11. Install voltage, current, and temperature sensors at key positions of the DC converter transformer using differential amplifier technology to obtain original monitoring data;

[0013] S12. Perform high-speed sampling on the original monitoring data and then perform digital processing, and obtain synchronous monitoring data through timestamps;

[0014] S13. Apply the Kalman filter algorithm to perform nonlinear estimation and noise removal on the synchronous monitoring data to obtain filtered monitoring data;

[0015] S13. Substitute into the model to integrate the filtered monitoring data to obtain the operating state data.

[0016] Preferably, the acquisition of voltage and current recorded in step S11 is performed separately based on the current access and outlet on the DC converter transformer.

[0017] Preferably, the learning of the operating state data through a deep learning algorithm in step S02 includes:

[0018] S21. Preprocess the operating state data using Z-score normalization;

[0019] S22. Based on the Adam optimizer and the mean squared error loss function, input the standardized state data for RNN model training, and use cross-validation to optimize hyperparameters and perform weight calculation to obtain a trained RNN model;

[0020] S23. Input the operation status data into the trained RNN model to obtain a real-time predicted load loss estimate.

[0021] Preferably, the formula for calculating weights by optimizing hyperparameters using cross-validation in step S22 includes:

[0022]

[0023] where M is a matrix, m represents the evaluation index correlation coefficient, and n is a constant;

[0024] Sum and transform the obtained preference relation matrix M row by row, and perform normalization processing to obtain the initial weight distribution corresponding to each evaluation index.

[0025] Preferably, in step S03, the adaptive control strategy is to optimize and integrate the initial weight and weight calibration value based on the cuckoo algorithm to obtain the total weight distribution value:

[0026] λ = α * λ 1 +(1 - α) * λ 2 ;

[0027] where λ 1 is the initial weight, λ 2 is the weight calibration value, taking 2, and α is the weighting coefficient.

[0028] Preferably, generating an anomaly warning report for the anomaly variable data in step S04 includes:

[0029] S41. Combine a decision tree and a support vector machine to build an ensemble learning model to improve the accuracy of anomaly detection and obtain an anomaly detection model;

[0030] S42. Use the anomaly detection model to analyze the operation status data and the predicted load loss estimate, and apply a Gaussian mixture model to label potential risks to obtain an anomaly warning report.

[0031] Preferably, generating intelligent protection for the anomaly variable data in step S04 includes uploading the anomaly warning report to a big data processing platform and actively giving feedback by selecting the best processing solution based on an expert knowledge base.

[0032] In the above technical solution, a fault detection method for a DC converter transformer provided by the present invention has the following beneficial effects: By collecting real-time voltage data, current data, and temperature data, and using an algorithm to give a real-time predicted load loss estimate, an adaptive control algorithm based on model predictive control is used based on this estimate to generate an adaptive control strategy, thereby coordinating the influence of the DC converter transformer temperature on voltage and current, and thus avoiding affecting the power supply stability of normal subways and subway stations.

[0033] Secondly, by predicting the load loss estimate in real time, machine learning algorithms can generate intelligent protection and anomaly warning reports for abnormal variable data, thereby notifying managers to perform manual troubleshooting in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0035] Figure 1 It is a flowchart provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0037] Embodiment 1

[0038] As Figure 1 shown, a fault detection method for a DC converter transformer includes:

[0039] S01. Based on each data acquisition unit, obtain real-time voltage data, current data, and temperature data, and construct real-time operation state data of the DC converter transformer according to the Kalman filter data fusion algorithm;

[0040] S02. Learn the operation state data through a deep learning algorithm and predict the load loss estimate in real time;

[0041] S03. Based on the predicted load loss estimate and combined with the adaptive control algorithm using model predictive control, control the DC converter transformer according to the real-time temperature data, and combine the DC converter transformer to adjust the control strategy to generate an adaptive control strategy;

[0042] S04. Combine the operating status data and the predicted load loss estimate, and use machine learning algorithms to monitor the DC converter transformer in real time, and generate intelligent protection and anomaly warning reports for abnormal variable data.

[0043] Specifically, through the current collector and voltage collector deployed at the wire interface of the DC converter transformer, the obtained voltage data and current data are transmitted. During the data transmission process, timestamps are set by fixed positions. For example, the first interface of the DC converter transformer wire interface is T1(00:00), and data transmission is carried out in this way successively.

[0044] Secondly, the load loss estimate usually refers to the energy loss caused by resistance or other factors when current flows through conductors, transformers or other electrical equipment in the power system. This loss is released in the form of heat, resulting in a reduction in system efficiency. And the so-called adaptive control algorithm can use the power formula P = I 2 R in Ohm's law, where P is the power loss (i.e., the load loss), I is the load current, and R is the resistance.

[0045] Through machine learning algorithms, real-time monitoring of the DC converter transformer is achieved. The methods for data processing by the machine learning algorithms in this implementation are as follows:

[0046] Data cleaning and preprocessing: Identify outliers in the data through algorithms, and then process or eliminate them. And use interpolation or other techniques to fill in the missing data.

[0047] Data normalization / standardization: Convert the data to a unified scale to improve the performance of the algorithm.

[0048] Feature selection: Select the features that are most helpful for model prediction from the original data.

[0049] Feature extraction: Generate new features to improve the expressive ability of the model.

[0050] Data augmentation: Improve the generalization ability of the model by generating more data samples.

[0051] Data transformation: Convert unstructured data into structured data for further analysis.

[0052] Model training and optimization: Select algorithms suitable for data characteristics and task requirements, such as regression, classification, clustering, etc. Optimize the hyperparameters of the algorithm to improve the performance of the model.

[0053] Prediction and decision-making: Use the trained model to predict new data.

[0054] Thus, the processing of abnormal variable data is realized and an abnormal warning report is generated, including:

[0055] S41. Combine a decision tree and a support vector machine to construct an integrated learning model to improve the accuracy of anomaly detection and obtain an anomaly detection model;

[0056] S42. Use the anomaly detection model to analyze the operation status data and the predicted load loss estimation value, and apply a Gaussian mixture model to mark potential risks to obtain an abnormal warning report.

[0057] In the above technology, through the real-time data acquisition of voltage data, current data and temperature data, the real-time predicted load loss estimation value is given by an algorithm. Based on this estimation value, an adaptive control algorithm of model predictive control is used to generate an adaptive control strategy, so as to coordinate the influence of the temperature of the DC converter transformer on the voltage and current, thereby avoiding affecting the power supply stability of normal subways and subway stations.

[0058] Secondly, through the real-time predicted load loss estimation value, machine learning algorithms can generate intelligent protection and abnormal warning reports for abnormal variable data, so as to notify the management personnel to perform manual troubleshooting in time.

[0059] Embodiment 2

[0060] Compared with Embodiment 1, this embodiment further proposes a method for constructing the operation status data of a real-time DC converter transformer, that is, data conversion is performed on the collected operation status data of the DC converter transformer. The detailed steps include:

[0061] S11. Install voltage, current and temperature sensors at key positions of the DC converter transformer using differential amplifier technology to obtain original monitoring data;

[0062] S12. After high-speed sampling of the original monitoring data, perform digital processing and obtain synchronous monitoring data through timestamps;

[0063] S13. Apply the Kalman filter algorithm to perform non-linear estimation and noise removal on the synchronous monitoring data to obtain filtered monitoring data;

[0064] S13. Substitute the model to integrate the filtered monitoring data to obtain the operation status data.

[0065] Furthermore, the voltage and current acquisition recorded in step S11 are respectively executed based on the current access and outlet on the DC converter transformer.

[0066] Specifically, real-time monitoring requires high-precision data acquisition. Especially in the power system, the working state of the transformer may change rapidly. High-speed sampling can ensure that fast-changing signals are captured.

[0067] Select an appropriate sampling rate, usually determined according to the changing frequency of the signal, to capture sufficiently detailed data. The sampling rate should be higher than the Nyquist frequency of the signal to avoid aliasing. Use an analog-to-digital converter (ADC) to convert the analog signal into a digital signal. Timestamp the collected data to ensure that the data from different sensors can be synchronously recorded.

[0068] Furthermore, in the above embodiments, the Kalman filter is an effective algorithm that can estimate and filter signals with noise. It is particularly effective in processing signals of dynamic systems and can provide more accurate state estimates. In this embodiment, a Kalman filter model is established to describe the dynamic change law of the system state. Describe how the sensor observes the system state, predict the state at the next time step according to the system model, combine the observed data with the predicted value, correct the state estimate, and update the state of the filter. Substitute the filtered data into the model integration. Through the model integration, the filtered monitoring data can be combined with the operation model of the system to obtain more comprehensive operation state information. Finally, input the filtered data into the operation state model for comprehensive analysis to generate an operation state report, including the health status, load condition, and potential risks of the transformer.

[0069] Embodiment III

[0070] Compared with Embodiment I, this embodiment further proposes to learn the operation state data through a deep learning algorithm, including:

[0071] S21. Preprocess the operation state data using Z-score normalization;

[0072] S22. Based on the Adam optimizer and the mean squared error loss function, input the standardized state data for RNN model training, and use cross-validation to optimize the hyperparameters and calculate the weights to obtain a trained RNN model;

[0073] S23. Input the operation state data into the trained RNN model to obtain a real-time predicted load loss estimate.

[0074] Furthermore, the formula for using cross-validation to optimize the hyperparameters and calculate the weights in step S22 includes:

[0075]

[0076] Among them, M is a matrix, m represents the evaluation index correlation coefficient, and n is a constant;

[0077] Sum and transform the obtained preference relation matrix M row by row, and perform normalization processing to obtain the initial weight distribution corresponding to each evaluation index.

[0078] Specifically, data preprocessing is one of the important pre - steps in the training of deep - learning models. By Z - score normalization, the operating - state data is converted into a standard normal distribution with a mean of 0 and a standard deviation of 1, which can effectively eliminate the dimensional differences between different features. This standardization process helps to improve the training efficiency and stability of the model, enabling the model to converge to the optimal solution faster during the training process. Preprocessing can also prevent specific features from having an excessive impact on model training, thereby improving the generalization ability of the model.

[0079] Furthermore, based on the Adam optimizer and the mean - squared - error loss function, the standardized state data is input for RNN model training, and cross - validation is used to optimize hyperparameters and calculate weights, obtaining a trained RNN model.

[0080] Moreover, the recursive neural network (RNN) in deep - learning algorithms is used to train the standardized operating - state data. RNN is good at processing time - series data, can capture the temporal features and dynamic changes in the data, and is suitable for tasks such as load - loss prediction. It can effectively handle the learning - rate adjustment problem in gradient descent, helping the model find the optimal solution faster. It is used for regression tasks to measure the difference between the predicted value and the actual value. By minimizing the loss function, the model can improve the accuracy of prediction. And the above - mentioned method evaluates the performance of the model under different hyperparameter settings, thereby selecting the best combination of hyperparameters. Through cross - validation, the over - fitting problem of the model can be avoided, ensuring the good performance of the model on new data.

[0081] Therefore, by using deep - learning algorithms to process the operating - state data, the prediction ability and accuracy of the device operating state can be significantly improved. The data - preprocessing step provides a more stable training basis for the model, while the model - training and optimization steps ensure the performance of the final model. Real - time prediction provides immediate feedback on the device state, helping to make effective decisions. Overall, this process helps to improve the reliability of the system, reduce maintenance costs, and optimize device management.

[0082] Embodiment 4

[0083] Compared with Embodiment 1, the further proposed adaptive control strategy is to optimize and integrate the initial weight and the weight calibration value based on the cuckoo algorithm to obtain the total weight distribution value:

[0084] λ = α * λ 1 +(1 - α)*λ 2 ;

[0085] where λ 1 is the initial weight, λ 2 is the weight calibration value, taking 2, and α is the weighting coefficient.

[0086] Specifically, the cuckoo algorithm is used to optimize the initial weights and weight calibration values of the control strategy. Set the initial weight values and the corresponding weight calibration values. These values are usually obtained based on the preliminary estimation or experience of the system. Then, by simulating the parasitic behavior of cuckoos, the weights are adjusted to gradually find a better weight configuration. Then, the weight values optimized by the cuckoo algorithm are integrated to obtain the final total weight allocation value. These optimized weight values can better adapt to the changes of the system and improve the control performance. Apply the optimized total weight allocation value to the adaptive control strategy to achieve more precise control.

[0087] In the above embodiment, by adjusting the control parameters in real time, the adaptive control strategy can maintain a high control accuracy under various operating conditions, thereby improving the overall performance of the system. Compared with the control system with fixed parameters, adaptive control can reduce the risk of over-adjustment and avoid the system from generating violent fluctuations due to inappropriate control parameters. Moreover, the adaptive control can adjust the control strategy according to the actual state of the system, thereby improving the stability of the system and reducing the risk of system instability.

[0088] Embodiment Five

[0089] Compared with Embodiment One, this embodiment further proposes to generate intelligent protection for abnormal variable data, including uploading the abnormal warning report to the big data processing platform and actively giving feedback by selecting the best processing solution based on the expert knowledge base.

[0090] Specifically, when the system detects abnormal variable data, it will automatically generate and upload an abnormal warning report to the big data processing platform. Through its powerful data processing capabilities, this platform can analyze and identify abnormal data patterns in real time and archive them as specific abnormal events. This step ensures that abnormal data can be recorded in a timely manner and included in the scope of further analysis. Based on the built-in algorithm, a detailed analysis of the abnormal data is carried out. The platform will not only search for potential patterns in the data but also compare these abnormal data with historical data to determine their possible causes and impacts. Then, the analysis results are transmitted to the expert knowledge base. The expert knowledge base contains the experience and professional knowledge of a large number of domain experts, which are used to formulate the best processing solution. The system will comprehensively consider the suggestions in the expert knowledge base, select the most appropriate processing solution, and give active feedback. The content of the feedback includes corrective suggestions for the abnormal data, preventive measures, and possible impact assessments.

[0091] In the above-mentioned technology, abnormal variable data can be quickly identified and processed, reducing the occurrence probability of potential risks. The real-time processing ability of the system ensures a rapid response to abnormal events, thus reducing the risk of business operation interruption. Through in-depth analysis of the big data platform and assisted decision-making of the expert knowledge base, the system can provide more accurate abnormal diagnosis and processing suggestions. This not only improves the accuracy of abnormal detection but also reduces the situations of false alarms and missed detections. The participation of the expert knowledge base makes the processing scheme more targeted and effective, ensuring the best decision-making can be made in complex situations, thus effectively protecting the stability and security of the system.

[0092] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program code.

[0093] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0094] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device realizes the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0095] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide for realizing the functions in Figure 1 one process or multiple processes and / or blocksFigure 1 Steps of functions specified in one or more boxes.

[0096] In the present invention, specific embodiments are used to elaborate on the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

[0097] The embodiments of the present application also provide a specific implementation manner of an electronic device capable of implementing all the steps in the method in the above embodiments. The electronic device specifically includes the following:

[0098] A processor, a memory, a communications interface, and a bus;

[0099] Wherein, the processor, the memory, and the communications interface complete mutual communication through the bus;

[0100] The processor is used to call the computer program in the memory, and when the processor executes the computer program, all the steps in the method in the above embodiments are implemented.

[0101] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all the steps in the method in the above embodiments. A computer program is stored on the computer-readable storage medium, and when the computer program is executed by the processor, all the steps in the method in the above embodiments are implemented.

[0102] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the hardware + program type of embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, reference can be made to the partial description of the method embodiments. Although the method operation steps are provided in the embodiments of this specification as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-creative means. The order of steps listed in the embodiments is only one way among many execution orders of the steps and does not represent the only execution order. When the actual device or terminal product is executed, it can be executed in the order of the method shown in the embodiments or the drawings or in parallel (for example, in an environment of parallel processors or multi-threaded processing, or even in a distributed data processing environment). The term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, product or device including a series of elements not only includes those elements but also includes other elements not expressly listed, or also includes elements inherent to such process, method, product or device. Without further limitation, it does not exclude the existence of additional identical or equivalent elements in the process, method, product or device including the said elements. For the convenience of description, when describing the above device, it is divided into various modules according to functions and described separately. Of course, when implementing the embodiments of this specification, the functions of each module can be implemented in the same or multiple software and / or hardware, or the modules implementing the same function can be realized by a combination of multiple sub-modules or sub-units, etc. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in the form of electricity, machinery or other forms. The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general computer, a special computer, an embedded processor or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks

[0103] Those skilled in the art should understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of completely hardware embodiments, completely software embodiments, or embodiments combining software and hardware aspects. Moreover, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code. Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. The key point of each embodiment is to illustrate the differences from other embodiments. In particular, for system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and reference can be made to the corresponding parts of the method embodiments for relevant details. In the description of this specification, the description of reference terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the embodiments of this specification.

[0104] In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples. The above is only the embodiments of the embodiments of this specification and is not used to limit the embodiments of this specification. For those skilled in the art, various changes and modifications can be made to the embodiments of this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the embodiments of this specification shall be included within the scope of the claims of the embodiments of this specification.

Claims

1. A fault detection method for a DC converter transformer, characterized in that: include: S01. Acquire real-time voltage data, current data and temperature data based on each data acquisition unit, and construct real-time operation status data of the DC converter transformer according to the Kalman filter data fusion algorithm; S02. Learning the operating status data through a deep learning algorithm, and predicting load loss estimation in real time; S03, based on the predicted load loss estimation and in combination with an adaptive control algorithm using model predictive control, controlling the DC converter transformer according to the real-time temperature data, and adjusting the management and control strategy of the DC converter transformer to generate an adaptive control strategy; S04. Combining the operating status data and the predicted load loss estimation, a machine learning algorithm is used to perform real-time monitoring of the DC converter transformer, and to generate intelligent protection and abnormal warning reports for abnormal variable data.

2. A DC converter transformer fault detection method according to claim 1, characterized in that: The step S01 of constructing real-time operation status data of the DC converter transformer includes: S11. Install voltage, current and temperature sensors at key locations of DC converter transformers using differential amplifier technology to obtain raw monitoring data; S12, performing digital processing on the original monitoring data after high-speed sampling, and obtaining synchronous monitoring data through time stamp; S13, applying a Kalman filter algorithm to perform nonlinear estimation and noise removal on the synchronous monitoring data to obtain filtered monitoring data; S13, substituting the filtered monitoring data into a model to integrate the filtered monitoring data to obtain operation status data.

3. A DC converter transformer fault detection method according to claim 1, characterized in that: The voltage and current collection recorded in step S11 is performed based on the current input and output of the DC converter transformer.

4. A DC converter transformer fault detection method according to claim 1, characterized in that: The step S02 includes learning the operating status data by using a deep learning algorithm, including: S21, preprocessing the operating status data using Z-score normalization; S22, based on the Adam optimizer and the mean square error loss function, input the standardized state data to perform RNN model training, and use cross-validation to optimize hyperparameters and perform weight calculation to obtain a trained RNN model; S23, inputting the operating status data into the trained RNN model to obtain a real-time predicted load loss estimate.

5. A DC converter transformer fault detection method according to claim 1, characterized in that: The formula for weight calculation using cross-validation to optimize hyperparameters in step S22 includes: Among them, M is a matrix, m represents the correlation coefficient of the evaluation index, and n is a constant; The obtained priority relationship matrix M is summed and transformed row by row, and normalized to obtain the initial weight distribution corresponding to each evaluation index.

6. A DC converter transformer fault detection method according to claim 1, characterized in that: The adaptive control strategy in step S03 is based on the cuckoo algorithm to optimize and integrate the initial weight and the weight calibration value to obtain the total weight distribution value: λ=a*λ 1 +(1-a)*λ 2 ; Among them, λ 1 is the initial weight, λ 2 is the weight calibration value, which is 2, and α is the weighting coefficient.

7. A DC converter transformer fault detection method according to claim 1, characterized in that: The step S04 of generating an abnormal warning report for abnormal variable data includes: S41. Combining decision tree and support vector machine to build an integrated learning model, improve the accuracy of anomaly detection, and obtain an anomaly detection model; S42: Analyze the operating status data and the predicted load loss estimate using the anomaly detection model, and apply a Gaussian mixture model to mark potential risks to obtain an anomaly warning report.

8. A DC converter transformer fault detection method according to claim 1, characterized in that: Generating intelligent protection for abnormal variable data in step S04 includes uploading the abnormal warning report to the big data processing platform, and actively providing feedback by selecting the best processing solution based on the expert knowledge base.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the fault detection method for the DC converter transformer according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the fault detection method for the DC converter transformer according to any one of claims 1 to 8 are implemented.

Images