A transformer on-line fault diagnosis and early warning method based on transformer transformation ratio
By collecting transformer voltage in real time and calculating the voltage ratio, and combining it with the ARIMA model for early warning, the problem of low accuracy in transformer fault diagnosis is solved, achieving rapid and accurate fault identification and early warning, and supporting intelligent operation and maintenance.
Patent Information
- Application Number
- CN202411660300.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-20
- Publication Date
- 2026-05-22
AI Technical Summary
Existing technologies for transformer fault diagnosis suffer from low accuracy, high cost, and inability to provide early warnings, especially for dry-type transformers, where accurate fault detection and early warning cannot be achieved through parameter identification.
By collecting the primary and secondary voltages of the transformer in real time, the voltage ratio is calculated and compared with the rated voltage ratio. The ratio is then predicted using the ARIMA model, and thresholds are set for early warning and fault identification.
It enables rapid and accurate identification and timely early warning of transformer faults, reduces costs, improves the efficiency and accuracy of fault detection, and supports unattended intelligent operation and maintenance.
Smart Images
Figure CN122072322A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of transformer fault detection technology, and in particular to a method for online fault diagnosis and early warning of transformers based on transformer turns ratio. Background Technology
[0002] Transformers are among the most critical pieces of equipment in a power grid. A failure in a transformer can cause significant economic losses and negative social impacts. Statistics show that online monitoring technology and condition-based maintenance can reduce annual transformer maintenance costs by 25% to 50%, and reduce average outage time by approximately 75%. Therefore, transformer condition monitoring and fault diagnosis have always been a key research focus for experts and scholars both domestically and internationally. Currently, most transformer fault diagnosis is achieved through dissolved gas analysis in oil. This method is effective in diagnosing overheating, discharge, and mechanical faults in transformers, but its accuracy remains insufficient, and it is only applicable to oil-fired transformers, not dry-type transformers.
[0003] In recent years, with the development of transformer condition-based maintenance technology, many scholars have conducted extensive research on parameter deviations in transformer faults. Studies have shown that winding deformation leads to changes in leakage inductance, and ground faults cause significant changes in both inductance and resistance. Therefore, many scholars both domestically and internationally have proposed new methods for establishing diagnostic criteria through parameter identification. For example, they use leakage inductance parameters in the transformer's T-type equivalent circuit as a criterion, and admittance parameters in the equivalent circuit to determine whether a transformer fault has occurred. Alternatively, they can use the reactance and resistance parameters in the transformer winding model to monitor and diagnose winding deformation. However, this method of impedance parameter identification based on the transformer's equivalent circuit is often unsatisfactory in practical applications. Due to missing data in the field and measuring instruments that do not meet actual requirements, the accuracy of parameter identification is low, and the computational cost is high. This makes it difficult to accurately and efficiently detect transformer faults and provides corresponding early warnings. Summary of the Invention
[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a transformer online fault diagnosis and early warning method based on transformer turns ratio, which can quickly and accurately identify the fault status of the transformer and provide timely early warning.
[0005] The objective of this invention can be achieved through the following technical solution: a method for online fault diagnosis and early warning of transformers based on transformer turns ratio, comprising the following steps:
[0006] S1. Real-time acquisition of the primary and secondary voltages of the transformer, and calculation of the corresponding real-time voltage turns ratio;
[0007] S2. Compare the real-time voltage ratio with the rated voltage ratio of the transformer to determine whether the transformer has a fault and the type of fault.
[0008] Based on the real-time voltage ratio, the turns ratio is predicted to obtain the predicted turns ratio value;
[0009] S3. Compare the predicted transformer ratio with the rated transformer ratio to determine whether to issue an early warning. If an early warning is issued, proceed to step S4; otherwise, return to step S1.
[0010] S4. Compare the predicted transformer ratio with the rated transformer ratio to determine whether the transformer has a fault and the type of fault.
[0011] Furthermore, step S1 specifically includes the following steps:
[0012] S11. Real-time acquisition of the primary and secondary voltages of the transformer, and storage of the acquired data in the database in array form according to the time series;
[0013] S12. Extract the latest primary-side voltage data and secondary-side voltage data from the database, calculate the current real-time voltage ratio, and store it in the database.
[0014] Furthermore, step S11 specifically involves using an intelligent measuring and control meter installed on the transformer to collect data on the primary and secondary voltages of the transformer.
[0015] Furthermore, the formula for calculating the real-time voltage transformation ratio in step S12 is as follows:
[0016]
[0017] Where K(k) is the transformer voltage ratio at the kth acquisition, U1(k) is the effective value of the transformer primary voltage at the kth acquisition, and U1(k) is the effective value of the transformer secondary voltage at the kth acquisition.
[0018] Furthermore, step S2 specifically includes the following steps:
[0019] S21. Determine the upper and lower threshold values based on the rated transformer ratio;
[0020] S22. Compare the real-time voltage ratio with the upper and lower threshold values to determine whether the transformer has a fault and the type of fault.
[0021] S23. Obtain historical real-time voltage ratio data from the database, combine it with the pre-trained ARIMA model to predict the voltage ratio, and output the predicted voltage ratio value for the set future time period.
[0022] Further, step S21 specifically involves calculating the upper and lower threshold values based on the allowable deviation of the rated tap voltage ratio, wherein the allowable deviation of the rated tap voltage ratio is ±0.5%.
[0023] Furthermore, the upper and lower threshold values in step S21 are specifically as follows:
[0024] K upper =K rated ×(1+0.005)
[0025] K lower =K rated ×(1-0.005)
[0026] Among them, K rated K is the rated transformation ratio of the transformer. upper K is the upper limit threshold. lower This is the lower threshold.
[0027] Furthermore, the specific process of step S22 is as follows:
[0028] If the real-time ratio K(k) exceeds the upper limit threshold K upper The initial assessment is that the DC resistance of the winding has increased, which may lead to winding overheating or other faults.
[0029] If the real-time ratio K(k) is less than the lower threshold K lower If the initial assessment is that an inter-turn short circuit has occurred in the winding, it may lead to winding damage or insulation failure.
[0030] Furthermore, in step S3, if the deviation between the predicted transformer ratio and the rated transformer ratio is within a first preset deviation range, a low-level early warning is issued.
[0031] If the deviation between the predicted value and the transformer's rated transformer ratio exceeds the second preset deviation range, an advanced early warning will be issued.
[0032] Furthermore, the first preset deviation range is between ±0.5% and ±1%.
[0033] The second preset deviation range is ±1%.
[0034] Compared with the prior art, the present invention has the following advantages:
[0035] This invention acquires real-time voltage data from the primary and secondary sides of a transformer and calculates the corresponding real-time voltage transformation ratio. By comparing this ratio with the transformer's rated transformation ratio, it determines whether a fault has occurred and the type of fault. Based on the real-time voltage transformation ratio, it performs transformation ratio prediction, obtaining a predicted value. This predicted value is then compared with the transformer's rated transformation ratio to determine if an early warning mechanism is triggered. If the early warning mechanism is activated, further fault diagnosis is performed on the transformer. This allows for direct fault identification using voltage parameters, enabling comprehensive, accurate, and rapid identification of transformer faults. Furthermore, the predictive algorithm implements the transformer's early warning function. The judgment criteria primarily rely on existing current parameters and voltage parameters from each side of the transformer. These parameters are readily available and do not significantly increase cost or difficulty, helping maintenance personnel quickly and accurately locate fault points and assess the extent of transformer damage. This has extremely high practical value in engineering applications. Attached Figure Description
[0036] Figure 1 This is a schematic diagram of the method flow of the present invention;
[0037] Figure 2 A logic framework for transformer fault diagnosis and early warning was built for this embodiment;
[0038] Figure 3 This is the equivalent circuit diagram of a transformer in a T-type configuration.
[0039] Figure 4 This is based on the transformer turns ratio principle. Detailed Implementation
[0040] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.
[0041] Example
[0042] like Figure 1 As shown, a method for online fault diagnosis and early warning of transformers based on transformer turns ratio includes the following steps:
[0043] S1. Real-time acquisition of the primary and secondary voltages of the transformer, and calculation of the corresponding real-time voltage turns ratio;
[0044] S2. Compare the real-time voltage ratio with the rated voltage ratio of the transformer to determine whether the transformer has a fault and the type of fault.
[0045] Based on the real-time voltage ratio, the turns ratio is predicted to obtain the predicted turns ratio value;
[0046] S3. Compare the predicted transformer ratio with the rated transformer ratio to determine whether to issue an early warning. If an early warning is issued, proceed to step S4; otherwise, return to step S1.
[0047] S4. Compare the predicted transformer ratio with the rated transformer ratio to determine whether the transformer has a fault and the type of fault.
[0048] This embodiment applies the above-described solution, such as Figure 2 As shown, the main contents include:
[0049] (1) Real-time acquisition of the voltage values required for protection criteria and storage in the IoTDB database. The transformer voltage ratio at the corresponding moment is calculated by acquiring the effective values of the primary and secondary voltages of the transformer and stored in the IoTDB database.
[0050] (2) Compare the calculated real-time voltage ratio with the normal ratio (i.e., the rated ratio marked on the transformer nameplate). If the real-time voltage ratio differs too much from the normal ratio, it is judged as a transformer fault and the differential main protection needs to be turned on.
[0051] (3) The diagnostic methods based on the transformer voltage ratio are as follows: 1. When the real-time ratio increases, it is initially judged that the DC resistance of the winding increases, and it is recommended to conduct a core inspection; 2. When the real-time ratio decreases, it is initially judged that an inter-turn short circuit has occurred in the winding, and it is recommended to conduct a detailed inspection of the transformer;
[0052] (4) The early warning method is as follows: using the ARIMA algorithm, the transformer ratio for the next 5 hours is predicted based on the historical transformer ratio in the IoTDB database; the predicted ratio value is compared with the early warning threshold. If the predicted ratio value differs too much from the threshold, it is determined that the transformer will fail in the future.
[0053] (5) If the early warning mechanism is triggered, the transformer will be further diagnosed. The diagnosis method is the same as step (3).
[0054] In step (1), information is collected from the transformer using a smart meter installed on the transformer. The real-time voltage values of the primary and secondary sides of the transformer are stored in the IoTDB database, denoted as U1 for the primary side voltage and U2 for the secondary side voltage. In the corresponding time series, data is collected once per minute, resulting in an array:
[0055] Primary voltage array U1: U1(k) represents the primary voltage value at the kth acquisition. The data is in array form, where each element corresponds to the effective voltage value acquired in one acquisition.
[0056] Secondary voltage array U2: U2(k) represents the secondary voltage value at the kth acquisition. The data is in array form and has the same structure as U1.
[0057] The transformer's turns ratio is calculated in real time using the primary and secondary voltage data. The transformer's turns ratio K is defined as the ratio of the primary electromotive force E1 to the secondary electromotive force E2, which is also numerically equal to the ratio of the number of turns in the primary winding N1 to the number of turns in the secondary winding N2.
[0058]
[0059] in accordance with Figure 3 The transformer T-type equivalent circuit in the diagram includes the resistance, reactance, and magnetizing branch of both the primary and secondary sides. Assume the primary voltage is... Secondary voltage is The load impedance is Z l ′ The corresponding current is and
[0060] Using frequency domain analysis to analyze the transformer, Kirchhoff's voltage law can be applied to obtain the following:
[0061] Primary voltage equation:
[0062]
[0063] in, It is the voltage drop across the excitation branch, and the secondary voltage equation is:
[0064]
[0065] Z l ′ It is the load impedance, and the transfer function G(s) is defined as the ratio of the secondary voltage to the primary voltage:
[0066]
[0067] Substituting the above primary and secondary voltage equations into the transfer function yields:
[0068]
[0069] Using the turns ratio relationship of an ideal transformer, assuming the primary current... and secondary current The relationship between them is:
[0070]
[0071] Substituting this into the transfer function, we get:
[0072]
[0073] As can be seen from the above transfer function expression, the gain of the transfer function is determined by the transformer turns ratio K, and the turns ratio K has a direct impact on the gain of the transfer function. Therefore, this scheme considers starting from the frequency domain analysis method, and can effectively diagnose transformer faults by monitoring changes in the turns ratio.
[0074] In this embodiment, the latest voltage data is extracted from the IoTDB database. The effective value of the primary voltage is recorded as U1(k) and the effective value of the secondary voltage is recorded as U2(k). The transformer voltage ratio K(k) is then calculated using the following formula:
[0075]
[0076] Store it in the IoTDB database to obtain the transformer turns ratio data arranged in time sequence.
[0077] In step (2), the real-time transformer turns ratio K is compared with the normal turns ratio K' on the transformer nameplate. In this embodiment, according to the 10kV transformer turns ratio test procedure: the allowable deviation of the rated tap voltage ratio is ±0.5%, and the voltage ratio of other taps should be within 1 / 10 of the transformer impedance voltage value (%), but should not exceed ±1%.
[0078] If the transformer turns ratio meets the following conditions:
[0079] K > 100.5% * K′ or K < 99.5% * K′
[0080] If this is detected, it indicates an abnormality in the transformer, and the differential protection will be activated.
[0081] In step (3), if the transformer turns ratio increases, it is preliminarily determined that it may be due to the increase in the DC resistance of the transformer.
[0082] like Figure 4 As shown, from the principle of transformer turns ratio, it can be seen that under normal circumstances:
[0083]
[0084] In case of abnormality, such as severe burnt-out of the tap changer contacts, a contact resistance R may be introduced into the contacts. Figure 4 As shown in section (b). During the transformer ratio test, due to the voltage drop across the contact resistance R, the actual voltage involved in the electromagnetic induction is U. 12 However, when the instrument actually calculates:
[0085]
[0086] Due to the voltage drop across resistor R, Therefore, K2 > K1, meaning the turns ratio has increased. At this point, a core inspection of the transformer is necessary.
[0087] In this embodiment, after obtaining the transformer voltage ratio K(k) at the current time point k, this ratio is compared with the rated ratio K on the transformer nameplate. rated Compare them.
[0088] Diagnostic rules: Based on the allowable deviation of ±0.5% for the rated tap voltage ratio, the upper and lower threshold values are set as follows:
[0089] K upper =K rated ×(1+0.005)
[0090] K lower =K rated ×(1-0.005)
[0091] If the transformation ratio increases, that is, the real-time transformation ratio K(k) exceeds the upper limit threshold K. upper If the initial assessment is that the DC resistance of the winding has increased, it may be due to overheating or other malfunctions. In this case, it is recommended to conduct a core inspection to confirm whether there is a problem with the winding.
[0092] If the transformation ratio decreases, that is, the real-time transformation ratio K(k) is less than the lower limit threshold K. lower If this is the case, the initial assessment is that an inter-turn short circuit has occurred in the winding, which may lead to winding damage or insulation failure. In this situation, it is recommended to conduct a detailed inspection of the transformer to ensure that the winding fault is detected and addressed promptly.
[0093] If the transformer turns ratio decreases, the most direct reason is that the number of turns N1 in the primary winding decreases or the number of turns N2 in the secondary winding increases. However, for a finished transformer, it is impossible to increase the number of turns. Therefore, the most likely reason is that the number of turns N1 in the primary winding decreases, that is, an inter-turn short circuit has occurred in the primary winding of the transformer.
[0094] When an alarm is triggered, maintenance personnel should promptly inspect and maintain the transformer based on the diagnostic results. For an increased turns ratio, it may be necessary to check if the DC resistance of the windings has increased abnormally; for a decreased turns ratio, it may be necessary to check for inter-turn short circuits in the windings, ensuring that faults are quickly identified and addressed.
[0095] In step (4), the ARIMA (AutoRegressive Integrated Moving Average) algorithm is used to predict the trend of the historical data of the transformer's real-time turns ratio K(t) in order to predict the changes in the turns ratio over a future period, thereby achieving early warning of potential transformer faults. The specific calculation steps are as follows:
[0096] 1) Data preparation
[0097] Extract historical transformer ratio data K(t) from the IoTDB database, where t represents time.
[0098] Differential processing is performed on the data to remove trend and seasonal components, thereby smoothing the data.
[0099] 2) Model building
[0100] The historical data of the variation ratio are modeled using an autoregressive component (AR), assuming a linear relationship between the current variation ratio K(t) and the variation ratio values of the previous few time points:
[0101] K(t)=φ1K(t-1)+φ2K(t-2)+…+φ p K(tp)+∈(t)
[0102] in, Here are the model parameters, and ε∈(t) represents white noise.
[0103] The moving average component (MA) is used to correct the forecast error to eliminate random fluctuations:
[0104] ∈(t)=θ1∈(t-1)+θ2∈(t-2)+…+θ q ∈(tq)
[0105] Where θ1, θ2...θ q This is the moving average coefficient.
[0106] 3) Model parameter estimation
[0107] The parameters of the ARIMA model are determined using the least squares method. and θ j Using these parameters, the transformation ratio K(t+n) at the future time t+n is predicted.
[0108] 4) Variation prediction
[0109] The transformer turns ratio is extracted from the IoTDB database and substituted into the model to obtain the predicted turns ratio value K(t+n).
[0110] In this embodiment, firstly, historical variation data K(t) over a past period is obtained from the IoTDB database, where t = 1, 2, ..., n represents historical time points. This data will be used as input to the ARIMA model to predict future variation.
[0111] Using an algorithm file named arima_prediction.py, the transformer turns ratio for the next 5 hours is predicted based on the ARIMA model, generating a prediction sequence K. p re(t+1),K pre(t+2),...,K p re(t+5), this prediction sequence reflects the trend of transformer turns ratio changes every hour in the future, and K is taken here. p re(t+5) represents the transformer turns ratio after 5 hours. The warning threshold is selected based on the transformer's nameplate rated turns ratio K. rated .
[0112] If the predicted value deviates slightly from the threshold range (e.g., the deviation is between ±0.5% and ±1%), a "low-level warning" will be issued to remind maintenance personnel to pay attention to the operating status of the transformer and to suggest arranging routine inspections.
[0113] If the predicted value deviates significantly from the threshold range (e.g., the deviation exceeds ±1%), an "advanced warning" will be issued, indicating to maintenance personnel that the transformer may be abnormal and suggesting in-depth condition monitoring or preventive maintenance to prevent the fault from developing further.
[0114] In step (5), the predicted turns ratio and the standard turns ratio are compared using the method in step (3) to analyze transformer faults and achieve early warning.
[0115] If the predicted ratio K pre (t+5) is greater than the upper threshold K upper The initial assessment is that the DC resistance of the windings has increased, which may cause overheating or other problems. It is recommended to perform a core inspection of the transformer to confirm whether there are any issues with the windings.
[0116] If the predicted ratio K pre (t+5) is less than the lower threshold K lower If this is the case, the initial assessment is that an inter-turn short circuit has occurred in the winding, which may lead to winding damage or insulation failure. In this situation, it is recommended to conduct a detailed inspection of the transformer to ensure that the winding fault is detected and addressed promptly.
[0117] If the predicted ratio exceeds the normal range, the system will trigger a fault alarm, notifying maintenance personnel to take necessary inspection and maintenance measures. Simultaneously, all relevant data, including the predicted value K, will be monitored. p The data, including re(t+5), real-time voltage data, and diagnostic results, will be stored in the IoTDB database for subsequent analysis and processing.
[0118] In summary, this solution, combining transformer mechanism analysis and historical data-driven approaches, can directly utilize existing on-site measuring equipment for online diagnosis and early warning of transformers. This solution integrates equipment electrical characteristic modeling and artificial intelligence technology based on big data mining, making it suitable for status monitoring and fault prediction of commonly used 10kV distribution transformers and distributed equipment in tunnel power systems. It supports unattended intelligent operation and maintenance, helps achieve predictable maintenance and lifespan management of equipment, and ensures the safe and reliable operation of tunnel power systems.
Claims
1. A method for online fault diagnosis and early warning of transformers based on transformer turns ratio, characterized in that, Includes the following steps: S1. Real-time acquisition of the primary and secondary voltages of the transformer, and calculation of the corresponding real-time voltage turns ratio; S2. Compare the real-time voltage ratio with the rated voltage ratio of the transformer to determine whether the transformer has a fault and the type of fault. Based on the real-time voltage ratio, the turns ratio is predicted to obtain the predicted turns ratio value; S3. Compare the predicted transformer ratio with the rated transformer ratio to determine whether to issue an early warning. If an early warning is issued, proceed to step S4; otherwise, return to step S1. S4. Compare the predicted transformer ratio with the rated transformer ratio to determine whether the transformer has a fault and the type of fault.
2. The online fault diagnosis and early warning method for transformers based on transformer turns ratio according to claim 1, characterized in that, Step S1 specifically includes the following steps: S11. Real-time acquisition of the primary and secondary voltages of the transformer, and storage of the acquired data in the database in array form according to the time series; S12. Extract the latest primary-side voltage data and secondary-side voltage data from the database, calculate the current real-time voltage ratio, and store it in the database.
3. The online fault diagnosis and early warning method for transformers based on transformer turns ratio according to claim 2, characterized in that, Specifically, step S11 involves using an intelligent measuring and control meter installed on the transformer to collect data on the primary and secondary voltages of the transformer.
4. The online fault diagnosis and early warning method for transformers based on transformer turns ratio according to claim 2, characterized in that, The formula for calculating the real-time voltage transformation ratio in step S12 is as follows: Where K(k) is the transformer voltage ratio at the kth acquisition, U1(k) is the effective value of the transformer primary voltage at the kth acquisition, and U1(k) is the effective value of the transformer secondary voltage at the kth acquisition.
5. The online fault diagnosis and early warning method for transformers based on transformer turns ratio according to claim 4, characterized in that, Step S2 specifically includes the following steps: S21. Determine the upper and lower threshold values based on the rated transformer ratio; S22. Compare the real-time voltage ratio with the upper and lower threshold values to determine whether the transformer has a fault and the type of fault. S23. Obtain historical real-time voltage ratio data from the database, combine it with the pre-trained ARIMA model to predict the voltage ratio, and output the predicted voltage ratio value for the set future time period.
6. The online fault diagnosis and early warning method for transformers based on transformer turns ratio according to claim 5, characterized in that, Specifically, step S21 involves calculating the upper and lower threshold values based on the allowable deviation of the rated tap voltage ratio, where the allowable deviation of the rated tap voltage ratio is ±0.5%.
7. The online fault diagnosis and early warning method for transformers based on transformer turns ratio according to claim 6, characterized in that, The upper and lower thresholds in step S21 are specifically as follows: K upper =K rated ×(1+0.005) K lower =K rated ×(1-0.005) Among them, K rated K is the rated transformation ratio of the transformer. upper K is the upper limit threshold. lower This is the lower threshold.
8. The online fault diagnosis and early warning method for transformers based on transformer turns ratio according to claim 7, characterized in that, The specific process of step S22 is as follows: If the real-time ratio K(k) exceeds the upper limit threshold K upper The initial assessment is that the DC resistance of the winding has increased, which may lead to winding overheating or other faults. If the real-time ratio K(k) is less than the lower threshold K lower If the initial assessment is that an inter-turn short circuit has occurred in the winding, it may lead to winding damage or insulation failure.
9. The online fault diagnosis and early warning method for transformers based on transformer turns ratio according to claim 1, characterized in that, In step S3, if the deviation between the predicted transformer ratio and the rated transformer ratio is within the first preset deviation range, a low-level early warning is issued. If the deviation between the predicted value and the transformer's rated transformer ratio exceeds the second preset deviation range, an advanced early warning will be issued.
10. The online fault diagnosis and early warning method for transformers based on transformer turns ratio according to claim 9, characterized in that, The first preset deviation range is between ±0.5% and ±1%. The second preset deviation range is ±1%.