A transformer oil level on-line monitoring method, system and device

Abstract

The application relates to the technical field of transformer oil level monitoring, in particular to a transformer oil level online monitoring method, system and device, the method comprising the following steps: collecting the oil level and oil temperature of the transformer oil pillow at each moment, and the inclination angle of the oil level collection equipment in each axial direction of a three-dimensional coordinate system at each moment; acquiring each abnormal oil level in the oil level data of the current moment and all moments before the current moment; determining the trend characteristic value of each abnormal oil level; acquiring the inclination characteristic value of each axial direction at the moment corresponding to each abnormal oil level; obtaining the non-noise confidence degree of each abnormal oil level, acquiring noise data in all abnormal oil levels, removing the noise of the noise data by using a digital filtering algorithm; and compensating and correcting the oil level at each moment after the noise removal treatment based on the inclination angle of all axial directions at each moment. Therefore, the authenticity and accuracy of the final oil level data monitoring are improved.

Description

Technical Field

[0001] This application relates to the field of transformer oil level monitoring technology, specifically to a method, system, and device for online monitoring of transformer oil level. Background Technology

[0002] As a key piece of equipment in the power system, the safe and stable operation of transformers plays a crucial supporting role in the reliability of the power grid. Transformer oil level, a key parameter reflecting the transformer's operating status, directly affects its heat dissipation, insulation, and protection mechanisms. Low oil levels lead to reduced cooling efficiency, increasing the risk of overheating and potentially causing equipment damage or fires. Therefore, online monitoring of transformer oil levels is necessary to ensure the transformer operates within a safe range and avoid safety hazards caused by abnormal oil levels.

[0003] Ultrasonic level sensors are commonly used non-contact level measurement devices, widely applied for monitoring liquid levels in various atmospheric pressure storage tanks and sealed containers. When using ultrasonic level sensors for online monitoring of transformer oil levels, the oil level data acquired is often affected by thermal noise from internal sensor components, shot noise, and external electromagnetic interference, introducing noise into the data. Therefore, current technologies typically use digital filtering algorithms to filter all acquired oil level data in real time to eliminate noise interference. However, real-time filtering of all acquired oil level data can easily lead to over-smoothing of non-noise data that reflects the true oil level change, thus affecting the accuracy of online transformer oil level monitoring. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide a method, system, and device for online monitoring of transformer oil levels. The specific technical solution adopted is as follows:

[0005] In a first aspect, embodiments of this application provide a method for online monitoring of transformer oil level, the method comprising the following steps:

[0006] The oil level and temperature in the transformer oil conservator at various times are collected, as well as the tilt angle of the oil level acquisition device in each axial direction of the three-dimensional coordinate system at each time.

[0007] Obtain the abnormal oil levels from the current time and all previous time data; analyze the monotonic change trend of oil temperature data at neighboring time points for each abnormal oil level, and determine the trend characteristic value of each abnormal oil level; analyze the difference between the tilt angle of each axial direction at the corresponding time point of each abnormal oil level and the tilt angle at neighboring time points, and determine the tilt characteristic value of each axial direction at the corresponding time point of each abnormal oil level; fuse the tilt characteristic values ​​of all axial directions at the corresponding time point of each abnormal oil level, and combine them with the trend characteristic value to obtain the non-noise confidence level of each abnormal oil level; obtain the noise data in all abnormal oil levels, and use a digital filtering algorithm to denoise the noise data;

[0008] Based on the tilt angles in all axial directions at each moment, the oil level at each moment after noise reduction is compensated and corrected.

[0009] In one embodiment, determining the trend characteristic value of each abnormal oil level includes:

[0010] The trend terms of oil temperature data at nearby times are obtained by using a time-series decomposition algorithm. The trend terms corresponding to all nearby times of each abnormal oil level are combined into a trend term sequence. The trend term sequence is tested to obtain the trend feature value of each abnormal oil level.

[0011] In one embodiment, the trend feature value is the absolute value of the test statistic obtained by performing a trend test on the trend term sequence using a trend test algorithm.

[0012] In one embodiment, determining the tilt feature value includes:

[0013] For the abnormal oil level at time i, the absolute values ​​of the differences between the tilt angles at time i and all its neighboring times are fused to determine the tilt characteristic values ​​for each axial direction at time i.

[0014] In one embodiment, the tilt characteristic value of each axial direction at the i-th time moment is the mean of the absolute values ​​of the differences between the tilt angles at the i-th time moment and all its neighboring times.

[0015] In one embodiment, the non-noise confidence level is the normalized result of the product of the mean of the tilt feature values ​​in all axial directions at the time corresponding to each abnormal oil level and the trend feature value.

[0016] In one embodiment, acquiring noise data from all abnormal oil levels includes: thresholding the non-noise confidence scores of all abnormal oil levels, and classifying abnormal oil levels with non-noise confidence scores less than the threshold as noise data.

[0017] In one embodiment, the compensation and correction of the oil level at each time point after noise reduction includes:

[0018] The oil level data after noise reduction and all non-noise oil level data are compensated and corrected respectively. Based on the tilt angle of all axial directions at each time, the oil level measurement error caused by the tilt of the oil level acquisition device at each time is calculated according to the trigonometric function relationship. The oil level data at all times is then compensated and corrected according to the calculated oil level measurement error.

[0019] Secondly, embodiments of this application also provide an online transformer oil level monitoring device, the device comprising:

[0020] An ultrasonic level sensor is used to collect the oil level in the transformer oil conservator at various times.

[0021] Temperature sensor, used to collect oil temperature in transformer oil conservator at various times;

[0022] Angle sensor is used to acquire the tilt angle of the ultrasonic liquid level sensor in each axial direction of the three-dimensional coordinate system.

[0023] The oil level filtering module is used to acquire abnormal oil levels from the current time and all previous time data; analyze the monotonic change trend of oil temperature data in the time series of each abnormal oil level, and determine the trend characteristic value of each abnormal oil level; analyze the difference between the tilt angle of each axial direction at the corresponding time of each abnormal oil level and the tilt angle of its neighboring time, and determine the tilt characteristic value of each axial direction at the corresponding time of each abnormal oil level; fuse the tilt characteristic values ​​of all axial directions at the corresponding time of each abnormal oil level, and combine them with the trend characteristic value to obtain the non-noise confidence of each abnormal oil level; acquire the noise data in all abnormal oil levels; and use a digital filtering algorithm to denoise the noise data.

[0024] The oil level correction module compensates and corrects the oil level at each moment after noise reduction, based on the tilt angle in all axial directions at each moment.

[0025] Thirdly, embodiments of this application also provide an online transformer oil level monitoring system, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0026] This application has at least the following beneficial effects:

[0027] This application collects oil level and temperature data from the transformer oil conservator at various times, as well as the tilt angle of the oil level acquisition device in each axial direction of the three-dimensional coordinate system at each time. Utilizing multi-dimensional data, it can more accurately identify abnormal oil levels, avoiding the limitations of traditional single-level oil level monitoring methods. It acquires abnormal oil levels from the current time and all previous time data; analyzes the monotonic change trend of oil temperature data in the time series of each abnormal oil level, determining the trend characteristic value of each abnormal oil level. The determination of the trend characteristic value effectively identifies false oil level changes caused by thermal expansion and contraction of the oil volume, avoiding the identification of non-noise oil level data as noise data, and helping to improve the accuracy and reliability of noise data identification.

[0028] Furthermore, the differences between the tilt angles of each axial direction at each abnormal oil level and the tilt angles at adjacent times are analyzed to determine the tilt characteristic values ​​of each axial direction at each abnormal oil level. Based on the three-dimensional axial angle difference analysis, instantaneous tilt interference caused by equipment vibration, foundation settlement, or external impact is eliminated, reducing the false alarm rate of noise oil level data identification caused by mechanical displacement, thereby improving the accuracy of noise oil level data identification. The tilt characteristic values ​​of all axial directions at each abnormal oil level are fused together with the trend characteristic values ​​to obtain the non-noise confidence of each abnormal oil level, and the noise data in all abnormal oil levels is obtained. The noise data is then denoised using a digital filtering algorithm. This improves the denoising effect of abnormal oil level data while avoiding the problem of excessive smoothing of non-noise interference oil level data. Based on the tilt angles of all axial directions at each time, the oil level at each time after denoising is compensated and corrected, improving the authenticity and accuracy of the final oil level data monitoring, avoiding equipment failures caused by abnormal oil levels, and improving the safety and stability of the transformer. Attached Figure Description

[0029] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A flowchart illustrating the steps of an online transformer oil level monitoring method according to one embodiment of this application;

[0031] Figure 2 Flowchart for determining non-noise confidence level. Detailed Implementation

[0032] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a transformer oil level online monitoring method, system, and apparatus proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0034] The following description, in conjunction with the accompanying drawings, details the specific scheme of the online transformer oil level monitoring method, system, and device provided in this application.

[0035] Please see Figure 1 The diagram illustrates a flowchart of an online transformer oil level monitoring method according to an embodiment of this application. The method includes the following steps:

[0036] S1 collects the oil level and temperature in the transformer oil conservator at various times, as well as the tilt angle of the oil level acquisition device in each axial direction of the three-dimensional coordinate system at each time.

[0037] This embodiment uses an ultrasonic level sensor to collect oil level data in the transformer oil conservator at various times, an angle sensor to collect tilt angle data of the ultrasonic level sensor in the X, Y, and Z axes of the three-dimensional coordinate system at various times, and a temperature sensor to collect oil temperature data in the transformer oil conservator at various times.

[0038] Among them, oil level data, tilt angle data, and oil temperature data are all collected synchronously, with the collection time interval set to 1 second. Implementers can set this according to the actual situation, and this embodiment does not impose any restrictions on it.

[0039] It should be noted that the three-dimensional coordinate system described in this embodiment is constructed with the magnet in the angle sensor as the origin, and the XOY plane of the three-dimensional coordinate system is parallel to the plane where the ultrasonic liquid level sensor is located.

[0040] In this embodiment, the collected oil level data is uniformly compensated and corrected at fixed time intervals. The fixed time interval is set to 3 minutes, but the implementer can set it according to actual conditions; this embodiment does not impose any restrictions. In another embodiment, the collected oil level data can be compensated and corrected in real time.

[0041] Using the Min-Max normalization method, all oil level data, oil temperature data, and tilt angle data along the X, Y, and Z axes collected within the fixed time period are normalized. The normalized data for each data type are then arranged in chronological order to form the oil level data time series A, oil temperature data time series B, X-axis tilt angle data time series C1, Y-axis tilt angle data time series C2, and Z-axis tilt angle data time series C3 within the fixed time period. The Min-Max normalization method is a well-known existing technology, and its specific process will not be elaborated upon. Implementers can choose other feasible normalization methods, and this embodiment does not impose any restrictions on this.

[0042] S2, obtain the abnormal oil levels in the oil level data of the current time and all previous times; analyze the monotonic change trend of the oil temperature data of the adjacent times of each abnormal oil level in the time series, and determine the trend characteristic value of each abnormal oil level.

[0043] Typically, noisy data in collected oil level data often exhibits more abnormal data distribution characteristics compared to normal oil level data due to its randomness. Therefore, in this embodiment, a time series anomaly detection algorithm is used to extract all abnormal data points in the oil level data time series A, which are denoted as abnormal oil levels. The set of all obtained abnormal data points is denoted as the abnormal data point set D of the oil level data time series A, which is used to characterize the set of all oil level data in the oil level data time series A that are affected by noise.

[0044] The time series anomaly detection algorithm described in this embodiment adopts the Isolation Forest algorithm, which is a well-known existing technology. Implementers can choose other feasible anomaly detection algorithms, such as the LOF anomaly detection algorithm, etc. This embodiment does not impose any restrictions on this.

[0045] However, when the ultrasonic waves emitted by the ultrasonic level sensor propagate through the oil in the transformer oil conservator, the propagation speed of the ultrasonic waves is usually affected by the temperature of the medium and will vary. Furthermore, the volume of oil in the transformer oil conservator will usually expand or contract with the change in oil temperature. Therefore, when the oil temperature in the transformer oil conservator shows a monotonous trend of increasing or decreasing, even if the oil level data collected at this time is not affected by noise, the oil level data will still show different data distribution characteristics compared with the oil level data collected at adjacent time points due to this monotonous trend of oil temperature. This makes it easy for time series anomaly detection algorithms to identify abnormal oil level data caused by noise interference.

[0046] Therefore, to avoid misclassifying the actual oil level data caused by changes in oil temperature within the transformer oil conservator as noise data and smoothing it during subsequent digital filtering, thus affecting the accuracy of the transformer oil level monitoring results, this embodiment makes the following processing:

[0047] Taking any abnormal oil level d in the abnormal data point set D as an example, let the collection time corresponding to the abnormal oil level d be the i-th time. The N times before the i-th time and the N times after the i-th time are taken as the neighboring times of the i-th time. In this embodiment, N=10. The implementer can set it according to the actual situation. This embodiment does not limit it.

[0048] The oil temperature data at time i and all its neighboring times are arranged in chronological order to form a local oil temperature sequence at time i. The trend term of each oil temperature in the local oil temperature sequence is extracted using the STL (Seasonal and Trend decomposition using Loess) time series decomposition algorithm to form a trend term sequence of the local oil temperature sequence. This reduces the impact of noise data in all oil temperature data collected within the local time period where the abnormal oil level d is located on the subsequent evaluation of whether all oil temperature data collected within the local time period where the abnormal oil level d is located has a monotonic data change trend. The STL time series decomposition algorithm is a well-known technology, and the specific process will not be described in detail.

[0049] The standardized Z-value of the trend term sequence of the local oil temperature sequence at time i is calculated using the Mann-Kendall trend test algorithm. The absolute value of the obtained standardized Z-value is used as the trend feature value of the abnormal oil level d, which is used to evaluate whether all oil temperature data collected within the local time period where the abnormal oil level d is located have a monotonic data change trend. The larger the trend feature value, the more monotonic the data change trend, and the less likely the abnormal oil level d is to be noise data. The Mann-Kendall trend test algorithm is a well-known technique, and the specific process will not be described in detail.

[0050] S3. Analyze the difference between the tilt angle of each axial direction at the time corresponding to each abnormal oil level and the tilt angle of its neighboring time, and determine the tilt characteristic value of each axial direction at the time corresponding to each abnormal oil level.

[0051] Because ultrasonic level sensors measure liquid level by emitting ultrasonic pulses and receiving the echo reflection signals from the liquid surface, during transformer operation, factors such as transformer foundation settlement and external impacts can cause the ultrasonic level sensor installed at the bottom of the transformer oil tank to tilt. When the ultrasonic level sensor suddenly tilts, the incident angle of the ultrasonic waves emitted by the sensor on the oil surface in the transformer oil tank changes abruptly. This causes the oil level data collected at this time to deviate significantly from the oil level data collected at adjacent times. Consequently, the oil level data collected at this time will also be identified by the time series anomaly detection algorithm as abnormal oil level data caused by noise interference.

[0052] Therefore, to avoid misclassifying the real oil level data caused by sudden tilting of the ultrasonic level sensor as noise data and smoothing it in the subsequent digital filtering process, thereby affecting the accuracy of the transformer oil level monitoring results, this embodiment makes the following processing:

[0053] Taking the abnormal oil level d at the i-th moment as an example, in the X-axis tilt angle data time series C1, the difference between the X-axis tilt angle at the i-th moment and the X-axis tilt angle at each of its neighboring moments is calculated, and based on the difference, the tilt characteristic value of the X-axis at the i-th moment is determined.

[0054] It should be noted that the difference represents the degree of difference between two variables, and can be calculated using methods such as the absolute value of the difference, the square of the difference, or the ratio.

[0055] In this embodiment, the absolute value of the difference between the X-axis tilt angle at time i and the X-axis tilt angle at each of its neighboring times is calculated. The mean of the absolute values ​​of the differences between the X-axis tilt angle at time i and the X-axis tilt angle at all its neighboring times is taken as the tilt characteristic value of the X-axis at time i. This is used to evaluate the degree of sudden X-axis tilt of the ultrasonic level sensor at the data acquisition time of the abnormal oil level d. The larger the tilt characteristic value of the X-axis, the greater the degree of sudden X-axis tilt.

[0056] Correspondingly, using the same calculation method as the tilt feature value of the X-axis at time i, in the Y-axis tilt angle data time series C2, the mean of the absolute values ​​of the differences between the Y-axis tilt angle at time i and the Y-axis tilt angle at all its neighboring times is calculated as the tilt feature value of the Y-axis at time i.

[0057] In the Z-axis tilt angle data time series C3, the mean of the absolute values ​​of the differences between the Z-axis tilt angle at time i and the Z-axis tilt angle at all its neighboring times is calculated as the tilt feature value of the Z-axis at time i.

[0058] S4. Integrate the tilt feature values ​​of all axial directions at the corresponding time of each abnormal oil level, and combine them with the trend feature values ​​to obtain the non-noise confidence of each abnormal oil level. Obtain the noise data in all abnormal oil levels, and use a digital filtering algorithm to denoise the noise data.

[0059] For the i-th time, the average value of the tilt characteristic values ​​along the X, Y, and Z axes is calculated to evaluate the degree of sudden tilting of the ultrasonic level sensor at the time of acquisition of the abnormal oil level d. The larger the average value, the greater the degree of sudden tilting, and the less likely the abnormal oil level d is to be noise data.

[0060] The normalized result of the product of the average value and the trend characteristic value at time i is calculated and used as the non-noise confidence score of the abnormal oil level d at time i. This score is used to assess whether the abnormal oil level d is non-noise data that reflects the actual oil level changes in the transformer oil conservator. The higher the non-noise confidence score, the more likely the abnormal oil level d is to be non-noise data. The normalization result is obtained using the Sigmoid function. The flowchart for determining the non-noise confidence score is as follows. Figure 2 As shown.

[0061] The non-noise confidence scores of all abnormal oil levels in the abnormal data point set D are used as inputs to the maximum inter-class variance algorithm, and the output is a segmentation threshold. The set of all abnormal oil levels in the abnormal data point set D with non-noise confidence scores less than the segmentation threshold is denoted as the noise data point set E, which is used to characterize all oil level data that need to be digitally filtered among all collected oil level data. The maximum inter-class variance algorithm is a well-known technique, and the specific process will not be described in detail.

[0062] The digital filtering algorithm is used to filter all abnormal oil levels in the set of noise data points E. In this embodiment, the digital filtering algorithm adopts the first-order lag filtering algorithm, which is a well-known existing technology. The implementer can choose other existing feasible digital filtering algorithms, such as median filtering algorithm, mean filtering algorithm, etc. This embodiment does not limit this.

[0063] S5, based on the tilt angles in all axial directions at each moment, compensate and correct the oil level at each moment after noise reduction processing.

[0064] In the oil level data time series A, the filtered oil level data from the noisy data point set E is replaced with its original oil level data to obtain the corrected oil level data time series. .

[0065] Based on the tilt angle data collected at various times along the X, Y, and Z axes, the corrected oil level data time series was calculated using trigonometric relationships. The oil level measurement error caused by tilt is included in the various oil level data, and the corrected oil level data time series is adjusted based on the calculated oil level measurement error. All oil level data are compensated and corrected to obtain the compensated and corrected oil level data at each time point, thus completing the online monitoring of the transformer oil level. The calculation of the oil level measurement error caused by tilting in the oil level data based on trigonometric functions is a well-known technique, and the specific process will not be elaborated further.

[0066] Based on the same inventive concept as the above method, this application also provides an online transformer oil level monitoring device, the device comprising:

[0067] An ultrasonic level sensor is used to collect the oil level in the transformer oil conservator at various times.

[0068] Temperature sensor, used to collect oil temperature in transformer oil conservator at various times;

[0069] Angle sensor is used to acquire the tilt angle of the ultrasonic liquid level sensor in each axial direction of the three-dimensional coordinate system.

[0070] The oil level filtering module is used to acquire abnormal oil levels from the current time and all previous time data; analyze the monotonic change trend of oil temperature data in the time series of each abnormal oil level, and determine the trend characteristic value of each abnormal oil level; analyze the difference between the tilt angle of each axial direction at the corresponding time of each abnormal oil level and the tilt angle of its neighboring time, and determine the tilt characteristic value of each axial direction at the corresponding time of each abnormal oil level; fuse the tilt characteristic values ​​of all axial directions at the corresponding time of each abnormal oil level, and combine them with the trend characteristic value to obtain the non-noise confidence of each abnormal oil level; acquire the noise data in all abnormal oil levels; and use a digital filtering algorithm to denoise the noise data.

[0071] The oil level correction module compensates and corrects the oil level at each moment after noise reduction, based on the tilt angle in all axial directions at each moment.

[0072] Based on the same inventive concept as the above method, this application embodiment also provides a transformer oil level online monitoring system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described transformer oil level online monitoring methods.

[0073] In summary, this application first collects the oil level and oil temperature in the transformer oil conservator at various times, as well as the tilt angle of the oil level acquisition device in each axial direction of the three-dimensional coordinate system at each time. By utilizing multi-dimensional data, abnormal oil levels can be identified more accurately, avoiding the limitations of traditional single oil level monitoring methods.

[0074] Secondly, obtain the abnormal oil levels from the current time and all previous time data; analyze the monotonic change trend of the oil temperature data in the time series of each abnormal oil level, and determine the trend characteristic value of each abnormal oil level; the determination of the trend characteristic value can effectively identify the false oil level changes caused by the thermal expansion and contraction of oil volume, avoid identifying non-noise oil level data as noise data, and help improve the accuracy and reliability of noise data identification.

[0075] Furthermore, the differences between the tilt angles of each axial direction at the time corresponding to each abnormal oil level and the tilt angles at adjacent times are analyzed to determine the tilt characteristic values ​​of each axial direction at the time corresponding to each abnormal oil level. Based on the three-dimensional axial angle difference analysis, the instantaneous tilt interference caused by equipment vibration, foundation settlement or external impact is eliminated, reducing the false alarm rate of noise oil level data identification caused by mechanical displacement, thereby improving the accuracy of noise oil level data identification.

[0076] Finally, by integrating the tilt feature values ​​of all axial directions at each abnormal oil level and combining them with the trend feature values, the non-noise confidence level of each abnormal oil level is obtained. Noise data from all abnormal oil levels is acquired, and a digital filtering algorithm is used to denoise the noise data. This improves the denoising effect of the abnormal oil level data and avoids the problem of excessive smoothing of non-noise interference oil level data. Based on the tilt angles of all axial directions at each time, the oil level at each time after denoising is compensated and corrected, improving the authenticity and accuracy of the final oil level data monitoring, avoiding equipment failures caused by abnormal oil levels, and improving the safety and stability of the transformer.

[0077] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0078] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0079] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A method for online monitoring of transformer oil level, characterized in that, The method includes the following steps: The oil level and temperature in the transformer oil conservator at various times are collected, as well as the tilt angle of the oil level acquisition device in each axial direction of the three-dimensional coordinate system at each time. Obtain the abnormal oil levels from the current time and all previous time data; analyze the monotonic change trend of oil temperature data at neighboring time points for each abnormal oil level, and determine the trend characteristic value of each abnormal oil level; analyze the difference between the tilt angle of each axial direction at the corresponding time point of each abnormal oil level and the tilt angle at neighboring time points, and determine the tilt characteristic value of each axial direction at the corresponding time point of each abnormal oil level; fuse the tilt characteristic values ​​of all axial directions at the corresponding time point of each abnormal oil level, and combine them with the trend characteristic value to obtain the non-noise confidence level of each abnormal oil level; obtain the noise data in all abnormal oil levels, and use a digital filtering algorithm to denoise the noise data; Based on the tilt angles in all axial directions at each moment, the oil level at each moment after noise reduction is compensated and corrected.

2. The online transformer oil level monitoring method as described in claim 1, characterized in that, The determination of the trend characteristic values ​​of each abnormal oil level includes: The trend terms of oil temperature data at nearby times are obtained by using a time-series decomposition algorithm. The trend terms corresponding to all nearby times of each abnormal oil level are combined into a trend term sequence. The trend term sequence is tested to obtain the trend feature value of each abnormal oil level.

3. The online transformer oil level monitoring method as described in claim 2, characterized in that, The trend feature value is the absolute value of the test statistic obtained by performing a trend test on the trend term sequence using a trend test algorithm.

4. The online transformer oil level monitoring method as described in claim 1, characterized in that, The determination of the tilt characteristic value includes: For the abnormal oil level at time i, the absolute values ​​of the differences between the tilt angles at time i and all its neighboring times are fused to determine the tilt characteristic values ​​for each axial direction at time i.

5. The online transformer oil level monitoring method as described in claim 4, characterized in that, The tilt characteristic value of each axial direction at time i is the mean of the absolute values ​​of the differences between the tilt angles at time i and all its neighboring times.

6. The online transformer oil level monitoring method as described in claim 1, characterized in that, The non-noise confidence level is the normalized result of the product of the mean of the tilt feature values ​​in all axial directions at the time corresponding to each abnormal oil level and the trend feature value.

7. The online transformer oil level monitoring method as described in claim 1, characterized in that, The step of acquiring noise data from all abnormal oil levels includes: thresholding the non-noise confidence scores of all abnormal oil levels, and classifying abnormal oil levels with non-noise confidence scores less than the threshold as noise data.

8. The online transformer oil level monitoring method as described in claim 1, characterized in that, The compensation and correction of the oil level at each time point after noise reduction includes: The oil level data after noise reduction and all non-noise oil level data are compensated and corrected respectively. Based on the tilt angle of all axial directions at each time, the oil level measurement error caused by the tilt of the oil level acquisition device at each time is calculated according to the trigonometric function relationship. The oil level data at all times is then compensated and corrected according to the calculated oil level measurement error.

9. A transformer oil level online monitoring device, applied to the transformer oil level online monitoring method described in claim 1, characterized in that, The device includes: An ultrasonic level sensor is used to collect the oil level in the transformer oil conservator at various times. Temperature sensor, used to collect oil temperature in transformer oil conservator at various times; Angle sensor is used to acquire the tilt angle of the ultrasonic liquid level sensor in each axial direction of the three-dimensional coordinate system. The oil level filtering module is used to acquire abnormal oil levels from the current time and all previous time data; analyze the monotonic change trend of oil temperature data in the time series of each abnormal oil level, and determine the trend characteristic value of each abnormal oil level; analyze the difference between the tilt angle of each axial direction at the corresponding time of each abnormal oil level and the tilt angle of its neighboring time, and determine the tilt characteristic value of each axial direction at the corresponding time of each abnormal oil level; fuse the tilt characteristic values ​​of all axial directions at the corresponding time of each abnormal oil level, and combine them with the trend characteristic value to obtain the non-noise confidence of each abnormal oil level; acquire the noise data in all abnormal oil levels; and use a digital filtering algorithm to denoise the noise data. The oil level correction module compensates and corrects the oil level at each moment after noise reduction, based on the tilt angle in all axial directions at each moment.

10. A transformer oil level online monitoring system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-8.

Images