Direct current magnetic bias on-line monitoring system based on multi-data fusion
The DC bias online monitoring system with multi-data fusion solves the limitations of single data monitoring in the power system, realizes comprehensive and accurate monitoring of the DC bias status of the transformer, eliminates monitoring blind spots, and promptly discovers potential problems.
Patent Information
- Application Number
- CN202510938130.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-08
- Publication Date
- 2025-09-16
- Estimated Expiration
- Not applicable · inactive patent
Smart Images
Figure CN120652196A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of DC bias magnetic online monitoring, in particular to a DC bias magnetic online monitoring system based on multi-data fusion. Background Art
[0002] In power system operation, DC magnetic bias has become a core issue threatening the safe and stable operation of transformers. When DC current enters the transformer through the grounded neutral point, it can cause problems such as half-cycle core saturation, excitation current distortion, increased vibration and noise, and local overheating. In severe cases, it can even cause equipment damage or system failure. With the expansion of power grids and the increasing integration of DC transmission projects, the causes of DC magnetic bias are becoming increasingly complex, further exacerbating the shortcomings of single-data monitoring. For example, in areas with multiple DC drop points, varying ground potential gradients can cause significant differences in the DC component experienced by different transformers. Relying solely on current monitoring can overlook the risk of local magnetic bias. Furthermore, factors such as transformer aging and harmonic pollution can alter the vibration and noise characteristics of magnetic bias, making it difficult to establish an accurate model for assessing the degree of magnetic bias using vibration monitoring alone. This limitation is particularly prominent in emerging power grid scenarios such as ultra-high voltage (UHV) and flexible DC power systems, necessitating the use of multi-data fusion technologies to overcome the bottleneck of single-data monitoring.
[0003] Based on this, in order to solve the above problems, the present invention provides a DC bias magnetic online monitoring system based on multi-data fusion. Summary of the Invention
[0004] In order to solve the problems existing in the above solutions, the present invention provides a DC bias magnetic online monitoring system based on multi-data fusion.
[0005] The purpose of the present invention can be achieved through the following technical solutions:
[0006] The DC bias magnetic online monitoring system based on multi-data fusion includes condition analysis module, monitoring module, and monitoring analysis module;
[0007] The condition analysis module is used to perform condition analysis on each transformer to obtain a transformer information map, which includes threshold data of each transformer and the transformer classification to which the corresponding transformer belongs.
[0008] Furthermore, condition analysis is performed on each transformer, including:
[0009] Obtain information about each transformer; set up a transformer distribution map based on the information about each transformer;
[0010] Identify various data types that require multiple data collection; set corresponding working condition items according to each data type;
[0011] Collect real-time data from each transformer according to the working condition items to obtain individual working condition data of the corresponding working condition items; integrate the individual working condition data corresponding to the transformer into working condition monitoring data, and add the working condition monitoring data to the transformer distribution map;
[0012] Classify each transformer according to the transformer distribution map to obtain several transformer classifications; mark the transformers in the transformer distribution map according to the transformer classifications;
[0013] Dynamically determine the threshold data of each transformer according to the transformer distribution map, and add the threshold data to the transformer distribution map; mark the current transformer distribution map as a transformer information map.
[0014] Furthermore, each transformer is classified according to the transformer distribution map, including:
[0015] Step SA1: setting a merging standard, where the merging standard is that the same threshold value can be set between corresponding transformers;
[0016] Identify the operating condition monitoring data of each transformer based on the transformer distribution map, evaluate the operating condition monitoring data between corresponding transformers based on the merging criteria, and obtain individual judgment results for each data type between corresponding transformers. The individual judgment results include whether the merging criteria are met or not met; perform binarization processing on each individual judgment result to obtain a classification vector between corresponding transformers;
[0017] Step SA2: judging whether the corresponding transformers belong to the same category based on the classification vector, classifying the transformers belonging to the same category into one category to obtain an initial classification;
[0018] Step SA3: Evaluate whether the corresponding initial classifications belong to the same classification, merge the initial classifications belonging to the same classification to obtain a new initial classification;
[0019] Step SA4: loop step SA3 until all initial classifications cannot be merged, and mark the remaining initial classifications as transformer classifications.
[0020] Further, when there are remaining transformers that are not in the transformer classification, a transformer classification is formed based on the transformers.
[0021] Furthermore, the threshold data of each transformer is dynamically determined based on the transformer distribution map, including:
[0022] According to the transformer distribution map, each transformer classification is identified in real time, and a transformer is selected from the transformer classification as a basic transformer; the basic transformer is analyzed to obtain threshold data of the basic transformer, and the threshold data is used as threshold data of each transformer in the transformer classification.
[0023] The monitoring module is used to perform real-time monitoring on each transformer and obtain multiple monitoring data of the transformer.
[0024] Furthermore, the transformer is monitored in real time, including:
[0025] Identify preset monitoring items, monitor the transformer in real time according to the monitoring items, and obtain single monitoring data of the monitoring items; identify the timestamp of the single monitoring data; integrate the individual monitoring data of the transformer according to the timestamp to obtain multiple monitoring data of the transformer.
[0026] The monitoring and analysis module is used to analyze multiple monitoring data of the corresponding transformer, identify threshold data of the corresponding transformer, analyze the multiple monitoring data according to the threshold data, and obtain a DC bias magnetic evaluation result of the corresponding transformer.
[0027] Furthermore, the DC bias magnetic evaluation results are analyzed in real time to obtain corresponding conflict evaluation results, which include conflicting results and normal results.
[0028] When the conflict assessment result is normal, no corresponding processing is performed;
[0029] When the conflict assessment result is a result conflict, determine the cause of the conflict and handle it accordingly.
[0030] Furthermore, the DC bias magnetic evaluation results are analyzed, including:
[0031] Establish a conflict identification model. The expression of the conflict identification model is:
[0032]
[0033] Where: AQ is the conflict material data, the output data is the conflict identification value CT(AQ), and the conflict identification value is 1 or 0;
[0034] Obtain corresponding conflict material data based on the DC bias magnetic evaluation results, analyze the conflict material data using the conflict identification model, and obtain corresponding conflict identification values;
[0035] When the conflict identification value is 1, the conflict evaluation result is a result conflict;
[0036] When the conflict identification value is 0, the conflict evaluation result is normal.
[0037] Furthermore, the determination of the causes of the conflict includes:
[0038] Mark the transformer whose conflict assessment result is a result conflict as a conflict transformer, and identify the transformer classification corresponding to the conflict transformer;
[0039] Identify multiple monitoring data and threshold data of the conflicting transformers, generate corresponding data retrieval formulas based on the multiple monitoring data and threshold data, retrieve historical monitoring data of each transformer in the transformer classification based on the data retrieval formula, and obtain conflict reference data; determine the conflict cause of the conflicting transformers based on the conflict reference data.
[0040] Furthermore, the historical monitoring data of each transformer in the transformer classification is retrieved according to the data retrieval formula, including:
[0041] Identify historical threshold data corresponding to historical monitoring data, perform feature extraction on the historical monitoring data and historical threshold data according to the data retrieval formula, obtain corresponding historical monitoring feature data, and calculate the similarity between the data retrieval formula and the historical monitoring feature data;
[0042] Obtain standard result data corresponding to historical monitoring feature data with a similarity greater than a threshold value X1, where the standard result data is composed of single standard results of each monitoring item; integrate the historical monitoring feature data and the standard result data into conflict reference data.
[0043] Compared with the prior art, the present invention has the following beneficial effects:
[0044] The present invention breaks through the limitations of traditional single data monitoring through multi-data fusion technology. In the operation of the power system, the DC bias magnetization phenomenon is affected by many factors, and single data monitoring cannot fully reflect the actual situation. For example, in areas with multiple DC landing points, the change in ground potential gradient makes the DC components borne by different transformers vary significantly. Relying solely on current monitoring may ignore the risk of local bias magnetization. The present invention integrates multi-source data such as current, vibration, and noise, and can capture the characteristic information of DC bias magnetization from different angles, realize comprehensive and accurate monitoring of the DC bias magnetization status of the transformer, effectively eliminate monitoring blind spots, and promptly discover potential DC bias magnetization problems. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0046] Figure 1 This is a principle block diagram of the present invention. DETAILED DESCRIPTION
[0047] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0048] like Figure 1 As shown, the DC bias magnetic online monitoring system based on multi-data fusion includes a condition analysis module, a monitoring module, and a monitoring analysis module;
[0049] The condition analysis module is used to perform condition analysis on each transformer that requires DC bias magnetic monitoring. That is, unless otherwise specified, the transformer described in the present invention is a transformer that requires DC bias magnetic monitoring; obtain equipment information of each transformer, such as transformer type, installation location, component information and other related information, and mark it as transformer information; set a transformer distribution map based on the information of each transformer, that is, distribute the transformers into the corresponding map according to the preset blocks, and supplement the corresponding transformer information;
[0050] Identify various data types that require subsequent multi-data collection, such as vibration, current, noise, temperature, and other data types; determine the operating conditions that have an impact on each data type; for example, when an oil-immersed transformer uses natural cooling (ONAN), the temperature increase causes the core resistivity to decrease, and the eddy current loss caused by DC bias magnetization increases, which may cause local overheating; while forced oil circulation (OFAF) can quickly dissipate heat, the oil flow noise may interfere with vibration monitoring; at high temperatures, the hysteresis loop of silicon steel sheets shifts, and the flux distortion caused by bias magnetization is more significant, but the vibration signal may produce additional components due to the thermal expansion effect; when the 5th and 7th harmonic currents in the power grid act together with DC bias magnetization, the core flux waveform is severely distorted, resulting in the complexity of the vibration harmonic components; for example, DC bias magnetization mainly produces even harmonics (2nd and 4th), while harmonic currents may be superimposed with odd harmonics, interfering with feature extraction. Harmonic current increases core loss, resulting in increased background noise (such as the frequency band below 100 Hz), which reduces the signal-to-noise ratio of the bias magnetic vibration signal. Based on the conclusions and common sense related to existing transformers, the operating condition items corresponding to each data type are determined.
[0051] Identify the existing working condition items, collect real-time data for each transformer based on the working condition items, and obtain the individual working condition data of the corresponding working condition items; integrate the individual working condition data corresponding to the transformer into working condition monitoring data, and add the working condition monitoring data to the transformer distribution map for real-time display. Some working condition items do not change data, so they can be collected when they are updated, such as cooling methods, cooling components, etc.
[0052] Each transformer is classified according to the transformer distribution map to obtain a plurality of transformer classifications; and corresponding markings are made in the transformer distribution map according to the transformer classifications.
[0053] Dynamically determine the threshold data of each transformer according to the transformer distribution map, such as vibration amplitude, total harmonic distortion rate THDi, harmonic vibration threshold, temperature and other thresholds; and add the obtained threshold data to the transformer distribution map.
[0054] In one embodiment, each transformer is classified according to a transformer distribution map, including:
[0055] Step SA1: setting a merging standard, where the merging standard is that the same threshold value can be set between corresponding transformers;
[0056] The operating condition monitoring data of each transformer is identified based on the transformer distribution map, and the operating condition monitoring data between the corresponding transformers are evaluated according to the merging standard to obtain the individual judgment results of the corresponding transformers on each data type. The individual judgment results include whether the merging standard is met or not met. Each individual judgment result is binarized, such as if the merging standard is met, the value is 1, and if the merging standard is not met, the value is 0, or other methods are used to assign corresponding values to the two results. A classification vector is formed according to the corresponding individual judgment results between the transformers, such as the values of the corresponding individual judgment results between the transformers are 1, 1, 1, 0, and 1 respectively.
[0057] Step SA2: judging whether the corresponding transformers belong to the same category according to the classification vector, classifying the transformers belonging to the same category into one category, and obtaining an initial classification, i.e., the classification corresponding to the two transformers;
[0058] Step SA3: Evaluate whether the corresponding initial classifications belong to the same classification, merge the initial classifications belonging to the same classification to obtain a new initial classification;
[0059] Step SA4: loop step SA3 until all initial classifications cannot be merged, and mark the remaining initial classifications as transformer classifications;
[0060] When there are remaining transformers that are not in the transformer classification, it means that they cannot be classified into the same category as other transformers, and a transformer classification is formed by the transformer.
[0061] In one embodiment, it is determined whether the corresponding transformers belong to the same category based on the classification vector, that is, if the values of the classification vector corresponding to each individual judgment result all meet the merging criteria, they are considered to belong to the same category.
[0062] Exemplarily, a single judgment result that satisfies the merging criteria is assigned a value of 1, and a single judgment result that does not satisfy the merging criteria is assigned a value of 0. When all elements in the classification vector are 1, they belong to the same classification.
[0063] In one embodiment, each transformer is classified according to a transformer distribution map, a merging standard is determined, and classification is performed based on the merging standard and existing technologies.
[0064] In one embodiment, the operating condition monitoring data between corresponding transformers are evaluated according to the merging standard, and judgment can be made based on existing technologies, such as using the historical data of the corresponding transformer under the operating condition to determine whether the same threshold data can be used, and then determining the single judgment result of the corresponding data type; intelligent models can also be established based on machine learning, deep learning algorithms, etc. for intelligent evaluation.
[0065] In one embodiment, the threshold data of each transformer is dynamically determined based on the transformer distribution map, and each transformer classification is identified in real time. An analysis is performed based on any transformer in the transformer classification to determine the threshold data of the transformer, and the threshold data is used as the threshold data of each transformer in the transformer classification; the threshold data includes the threshold value corresponding to each data type.
[0066] Threshold data for a single transformer can be set based on existing methods, such as manual setting; it can also be set based on machine learning, deep learning algorithms, etc. to establish an intelligent model for intelligent analysis; it can also be based on technologies such as digital twins to pre-establish a digital twin of the corresponding transformer, and set threshold data based on the digital twin simulation; at the same time, historical data of transformers under various historical conditions can be statistically analyzed to determine the threshold data of the transformer under various transformer classifications, and the staff can perform calibration and adjustment to form the threshold data of the corresponding transformer classification, and then perform corresponding matching; specifically, the threshold data of the corresponding transformer can be set according to a variety of methods.
[0067] That is, first dynamically analyze whether each transformer can be classified into one category, and then only the threshold data of one transformer in the category needs to be clarified to determine the threshold data of all transformers.
[0068] The monitoring module is used to monitor each transformer in real time and obtain multiple monitoring data of multi-data fusion, that is, the integrated data of the monitoring data of each monitoring device; such as the integrated data of vibration, current, noise, temperature and other data, which are integrated accordingly according to the corresponding timestamps.
[0069] That is, the corresponding monitoring items are determined according to the set monitoring equipment, real-time monitoring is performed according to the monitoring items, the corresponding single monitoring data are obtained, and the various single monitoring data are integrated according to the timestamps to obtain the corresponding multiple monitoring data.
[0070] The monitoring and analysis module is used to analyze multiple monitoring data, identify threshold data of the corresponding transformer, analyze the multiple monitoring data according to the threshold data, and obtain a DC bias magnetic evaluation result of the corresponding transformer.
[0071] In one embodiment, multiple monitoring data are analyzed according to threshold data. The analysis can be performed based on existing evaluation methods, such as establishing an intelligent model based on machine learning, deep learning algorithms, etc. to perform intelligent analysis to obtain corresponding DC bias magnetic evaluation results. Other evaluation methods can also be used, for example:
[0072] Assign weights to each monitoring item, such as:
[0073] Neutral point DC current: weight 0.4, threshold > 3A, risk level is high;
[0074] Harmonic content: weight 0.3, threshold THD>5%, risk level is medium;
[0075] Vibration and noise: weight 0.2, threshold >80dB, risk level is low;
[0076] Core temperature: weight 0.1, threshold temperature rise rate > 2°C / min, risk level is low;
[0077] Data cleaning and calibration:
[0078] Eliminate sensor noise, such as using a sliding average filter; correct for environmental interference, such as using a temperature compensation algorithm;
[0079] Mark data points exceeding the threshold using box plots or the 3σ criterion;
[0080] Single indicator over-limit judgment: If any monitoring item exceeds the threshold, the corresponding risk level will be directly triggered;
[0081] Multi-indicator joint evaluation: When multiple indicators are close to the threshold, a comprehensive judgment is made through weighted scoring;
[0082] Calculate the risk score: Risk score = ∑ (indicator weight × excess ratio);
[0083] For example: neutral point current 3.5A (16.7% over limit);
[0084] Harmonic content THD = 6% (exceeding the limit by 20%);
[0085] Vibration noise 75dB (not exceeded);
[0086] Risk score = 0.4 × 0.167 + 0.3 × 0.2 + 0.2 × 0 + 0.1 × 0 = 0.127
[0087] If the score is > 0.1 (custom threshold), it is determined to be a risk of magnetic bias.
[0088] In one embodiment, the DC bias magnetic evaluation result is analyzed in real time to determine whether the DC bias magnetic evaluation result has a result conflict phenomenon, and obtain a corresponding conflict evaluation result, the conflict evaluation result including a result conflict and a result normal;
[0089] When the conflict assessment result is normal, no corresponding processing is performed;
[0090] When the conflict assessment result is a result conflict, determine the cause of the conflict and handle it accordingly.
[0091] In one embodiment, whether the DC bias magnetic evaluation results have conflicting results can be determined based on existing methods, such as whether a single item exceeds the limit or multiple items are normal;
[0092] Example: The neutral point DC current exceeds the threshold (such as 3A), but the harmonic content and vibration noise are within the limits.
[0093] Multiple items exceed the limit but are logically contradictory;
[0094] Example: The neutral point current and harmonic content are both out of limits, but the core temperature does not rise significantly (which contradicts the causal chain of temperature rise caused by bias magnetization). The typical impact path of DC bias magnetization is: current ↑ → core saturation → harmonics ↑ → temperature rise ↑ → noise ↑;
[0095] Threshold boundary blur;
[0096] Example: The monitored value is close to the threshold (e.g., neutral point current = 2.9A, threshold 3A), making it difficult to directly determine whether it is abnormal.
[0097] Conflict between trend and instantaneous value;
[0098] Example: Momentary current exceeding limit but long-term trend stable, or vice versa.
[0099] Identify based on the above features and determine whether the results conflict.
[0100] In one embodiment, determining whether a DC bias magnetic evaluation result has a conflicting result phenomenon includes:
[0101] A conflict identification model is established. The conflict identification model is used to obtain the monitoring results of the transformer on various monitoring items and corresponding monitoring phenomena, such as current increase, harmonic increase, temperature increase, etc., based on the DC bias magnetic evaluation results; integrate them into conflict material data; analyze the conflict material data to determine whether the results corresponding to each monitoring item have a conflict situation, and use the corresponding historical data to establish a corresponding training set for training; the expression of the conflict identification model is:
[0102]
[0103] Where: AQ is the conflict material data. AQ meeting the result conflict requirement means that the corresponding monitoring results have conflicts, such as normal and over-limit conflicts. This is determined based on the comparison of single thresholds. The output data is the conflict identification value CT(AQ), which is 1 or 0.
[0104] Obtain corresponding conflict material data based on the DC bias magnetic evaluation results, analyze the conflict material data using the conflict identification model, and obtain corresponding conflict identification values;
[0105] When the conflict identification value is 1, the conflict evaluation result is a result conflict;
[0106] When the conflict identification value is 0, the conflict evaluation result is normal.
[0107] In one embodiment, when the conflict evaluation result is a result conflict, the cause of the conflict is determined, and the cause of the conflict can be determined based on an existing method.
[0108] For example, the individual items exceed the limit but the correlation is weak;
[0109] If the neutral point current exceeds the limit, but the harmonic content and vibration noise are normal, it is judged to be a suspected sensor false alarm or local interference.
[0110] Multiple items exceed the limit but are logically contradictory;
[0111] If both the neutral point current and harmonic content exceed the limit but the core temperature does not rise, it is determined that the core is not deeply saturated or there is a sensor error.
[0112] Threshold boundary blur;
[0113] If the monitored value is close to the threshold (e.g. within ±10%), it is determined that further verification is required (e.g. increasing the sampling frequency or redundant sensors).
[0114] In one embodiment, determining the cause of the conflict includes:
[0115] Obtain historical data on DC bias of the transformer, determine the correlation between various monitoring items when the transformer is DC biased based on the historical data, form an impact path, and determine the cause of the conflict based on the impact path.
[0116] For example, the monitoring data of a transformer is as follows:
[0117] Neutral point DC current: 3.8A (threshold 3A)
[0118] Harmonic content: THD = 4.2% (threshold 5%)
[0119] Vibration noise: 78dB (threshold 80dB)
[0120] Core temperature: 55°C (ambient temperature 30°C, temperature rise is normal)
[0121] Conflict Analysis:
[0122] Single item exceeds the limit: neutral point current exceeds the limit by 26.7%, but other indicators are normal.
[0123] Correlation verification:
[0124] DC bias magnetization usually causes a synchronous increase in harmonic content, but the current THD is only 4.2%, which is inconsistent with the current exceeding the limit.
[0125] Sensor calibration:
[0126] The backup sensor measured the neutral point current as 3.2A (still exceeding the limit), eliminating the sensor failure.
[0127] in conclusion:
[0128] Neutral point current is a true abnormal item, which may be caused by DC intrusion on the grid side;
[0129] The fact that the harmonic content does not exceed the limit may be related to the core not being deeply saturated, and continuous monitoring is required.
[0130] In one embodiment, determining the cause of the conflict includes:
[0131] Mark the transformer whose conflict assessment result is a result conflict as a conflict transformer, and identify the transformer classification corresponding to the conflict transformer;
[0132] Identify multiple monitoring data and threshold data of the conflicting transformer, and generate corresponding data retrieval formulas based on the multiple monitoring data and threshold data. For example, according to the above example, the corresponding characteristic data are distributed according to current, harmonic content, vibration noise, etc., to form a data retrieval formula of 3.8A (3A)-4.2% (5%)-78dB (80dB); retrieve the historical monitoring data of each transformer in the transformer classification according to the data retrieval formula to obtain conflict reference data; determine the conflict cause of the conflicting transformer based on the conflict reference data.
[0133] In one embodiment, the conflict cause of the conflict transformer is determined based on the conflict reference data, such as determining the abnormal cause in the current situation based on the conflict reference data, evaluating the possibility based on the proportion of the corresponding abnormal cause, and then determining the cause of the conflict; it is also possible to verify based on the manifestation of the abnormal cause and determine that it meets the abnormal cause as the cause of the conflict; it is also possible to establish an intelligent model based on intelligent algorithms such as machine learning and deep learning algorithms for intelligent analysis; specifically, other existing methods can also be used for analysis.
[0134] In one embodiment, historical monitoring data of each transformer in the transformer category is retrieved according to a data retrieval formula, including:
[0135] Identify the historical threshold data corresponding to the historical monitoring data, perform feature extraction on the historical monitoring data and the historical threshold data according to the data retrieval formula, and obtain the corresponding historical monitoring feature data, that is, perform feature extraction according to the data retrieval formula, and calculate the similarity between the data retrieval formula and the historical monitoring feature data, such as presetting the weight coefficient of each monitoring item, and then calculate the similarity based on an algorithm such as cosine similarity;
[0136] Obtain the standard result data corresponding to the historical monitoring feature data whose similarity is greater than the threshold X1. The standard result data consists of the single standard result of each monitoring item. The single standard result includes whether the result is abnormal and the cause of the abnormality.
[0137] Integrate historical monitoring feature data and standard result data into conflict reference data.
[0138] The above formulas are all calculated by removing dimensions and taking their numerical values. The formula is a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and preset thresholds in the formula are set by technicians in this field according to actual conditions or obtained by simulating a large amount of data.
[0139] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.
Claims
1. The DC bias magnetic online monitoring system based on multi-data fusion is characterized by: Including condition analysis module, monitoring module, monitoring and analysis module; The condition analysis module is used to perform condition analysis on each transformer to obtain a transformer information map, wherein the transformer information map includes threshold data of each transformer and the transformer classification to which the corresponding transformer belongs; The monitoring module is used to monitor each transformer in real time and obtain multiple monitoring data of the transformer; The monitoring and analysis module is used to analyze multiple monitoring data of the corresponding transformer, identify threshold data of the corresponding transformer, analyze the multiple monitoring data according to the threshold data, and obtain a DC bias magnetic evaluation result of the corresponding transformer.
2. The DC bias magnetic online monitoring system based on multi-data fusion according to claim 1 is characterized in that: Perform condition analysis on each transformer, including: Obtain information about each transformer; set up a transformer distribution map based on the information about each transformer; Identify various data types that require multiple data collection; set corresponding working condition items according to each data type; Collect real-time data from each transformer according to the working condition items to obtain individual working condition data of the corresponding working condition items; integrate the individual working condition data corresponding to the transformer into working condition monitoring data, and add the working condition monitoring data to the transformer distribution map; Classify each transformer according to the transformer distribution map to obtain several transformer classifications; mark the transformers in the transformer distribution map according to the transformer classifications; Dynamically determine the threshold data of each transformer according to the transformer distribution map, and add the threshold data to the transformer distribution map; mark the current transformer distribution map as a transformer information map.
3. The DC bias magnetic online monitoring system based on multi-data fusion according to claim 2 is characterized in that: Classify each transformer according to the transformer distribution diagram, including: Step SA1: setting a merging standard, where the merging standard is that the same threshold value can be set between corresponding transformers; Identify the operating condition monitoring data of each transformer based on the transformer distribution map, evaluate the operating condition monitoring data between corresponding transformers based on the merging criteria, and obtain individual judgment results for each data type between corresponding transformers. The individual judgment results include whether the merging criteria are met or not met; perform binarization processing on each individual judgment result to obtain a classification vector between corresponding transformers; Step SA2: judging whether the corresponding transformers belong to the same category based on the classification vector, classifying the transformers belonging to the same category into one category to obtain an initial classification; Step SA3: Evaluate whether the corresponding initial classifications belong to the same classification, merge the initial classifications belonging to the same classification to obtain a new initial classification; Step SA4: loop step SA3 until all initial classifications cannot be merged, and mark the remaining initial classifications as transformer classifications.
4. The DC bias magnetic online monitoring system based on multi-data fusion according to claim 3 is characterized in that: When there are remaining transformers that are not within the transformer category, a transformer category is formed based on the transformers.
5. The DC bias magnetic online monitoring system based on multi-data fusion according to claim 2 is characterized in that: Dynamically determine the threshold data for each transformer based on the transformer distribution map, including: According to the transformer distribution map, each transformer classification is identified in real time, and a transformer is selected from the transformer classification as a basic transformer; the basic transformer is analyzed to obtain threshold data of the basic transformer, and the threshold data is used as threshold data of each transformer in the transformer classification.
6. The DC bias magnetic online monitoring system based on multi-data fusion according to claim 1 is characterized in that: Real-time monitoring of transformers, including: Identify preset monitoring items, monitor the transformer in real time according to the monitoring items, and obtain single monitoring data of the monitoring items; identify the timestamp of the single monitoring data; integrate the individual monitoring data of the transformer according to the timestamp to obtain multiple monitoring data of the transformer.
7. The DC bias magnetic online monitoring system based on multi-data fusion according to claim 1 is characterized in that: Analyze the DC bias magnetic evaluation results in real time to obtain the corresponding conflict evaluation results, which include conflict results and normal results; When the conflict assessment result is normal, no corresponding processing is performed; When the conflict assessment result is a result conflict, determine the cause of the conflict and handle it accordingly.
8. The DC bias magnetic online monitoring system based on multi-data fusion according to claim 7 is characterized in that: Analyze the DC bias magnetic evaluation results, including: Establish a conflict identification model. The expression of the conflict identification model is: Where: AQ is the conflict material data, the output data is the conflict identification value CT(AQ), and the conflict identification value is 1 or 0; Obtain corresponding conflict material data based on the DC bias magnetic evaluation results, analyze the conflict material data using the conflict identification model, and obtain corresponding conflict identification values; When the conflict identification value is 1, the conflict evaluation result is a result conflict; When the conflict identification value is 0, the conflict evaluation result is normal.
9. The DC bias magnetic online monitoring system based on multi-data fusion according to claim 8 is characterized in that: Determination of the causes of the conflict, including: Mark the transformer whose conflict assessment result is a result conflict as a conflict transformer, and identify the transformer classification corresponding to the conflict transformer; Identify multiple monitoring data and threshold data of the conflicting transformers, generate corresponding data retrieval formulas based on the multiple monitoring data and threshold data, retrieve historical monitoring data of each transformer in the transformer classification based on the data retrieval formula, and obtain conflict reference data; determine the conflict cause of the conflicting transformers based on the conflict reference data.
10. The DC bias magnetic online monitoring system based on multi-data fusion according to claim 9 is characterized in that: Retrieve historical monitoring data of each transformer within the transformer category based on the data retrieval formula, including: Identify historical threshold data corresponding to historical monitoring data, perform feature extraction on the historical monitoring data and historical threshold data according to the data retrieval formula, obtain corresponding historical monitoring feature data, and calculate the similarity between the data retrieval formula and the historical monitoring feature data; Obtain standard result data corresponding to historical monitoring feature data with a similarity greater than a threshold value X1, where the standard result data is composed of single standard results of each monitoring item; integrate the historical monitoring feature data and the standard result data into conflict reference data.