Electrical equipment maintenance analysis system and method based on Internet of Things

Abstract

The invention discloses a power equipment maintenance analysis system and method based on the Internet of Things, and relates to the technical field of power equipment state monitoring, and the method comprises the steps: collecting a historical infrared thermal image sequence of a transformer in a normal operation state, and extracting temperature data of a sleeve and a winding region to form a historical temperature sequence; coherence coefficients between the sequences are calculated through power spectrum density estimation and amplitude square coherence analysis, and a historical coherence coefficient set is established; determining a dynamic threshold interval through statistical analysis; infrared thermal images are collected in real time, a real-time coherence coefficient is calculated, and when the coherence coefficient exceeds a threshold value, automatic early warning is carried out; the system comprises four core modules, namely a historical data acquisition and processing module, a historical coherence analysis module, a real-time monitoring and analysis module and an early warning judgment and output module. According to the method, a coherence analysis technology is applied, a dynamic early warning mechanism is established by monitoring correlation characteristics of temperature changes of a transformer bushing and a winding area, and accurate recognition and intelligent early warning of early faults of the transformer are achieved.

Description

Technical Field

[0001] This invention relates to the field of power equipment condition monitoring technology, specifically a power equipment maintenance and analysis system and method based on the Internet of Things. Background Technology

[0002] As a core component of the power transmission and distribution system, the operational reliability of power transformers directly affects the safe and stable operation of the entire power grid. With the rapid advancement of smart grid construction, higher demands are being placed on the accuracy of power equipment condition monitoring and the timeliness of fault early warning.

[0003] Currently, transformer condition monitoring mainly relies on periodic inspections, preventative tests, and online monitoring technologies. Among these, infrared thermography, as a non-contact temperature measurement method, has been widely used in detecting overheating defects in power equipment. Traditional infrared diagnostic methods are mainly based on absolute temperature threshold methods, relative temperature difference judgment methods, or thermal image feature analysis. These methods determine faults by setting fixed temperature limits or by comparing temperature differences between different phases or between similar equipment. However, these methods have significant limitations: First, transformer operating temperature is affected by a combination of factors such as load current, ambient temperature, and cooling conditions, making it difficult to adapt to complex operating conditions using fixed threshold methods, which can easily lead to false alarms or missed alarms. Second, most existing methods focus on local temperature characteristics or the spatial distribution of the temperature field, failing to fully consider the thermal dynamic coupling relationship between different transformer components, especially the coordinated change characteristics between closely thermally related components such as bushings and windings.

[0004] Existing diagnostic methods based on infrared thermography are mostly limited to static analysis of temperature scalars or spatial gradients of the temperature field. These methods fail to effectively utilize the temporal characteristics of the dynamic evolution of temperature signals over time, and even more so, they fail to quantify the mutual influence of thermal dynamic processes between different components within a transformer from a system correlation perspective. Essentially, they diagnose through "isolated snapshots," lacking a deep characterization of the core physical processes of heat transfer and thermal balance between components. This results in insufficient sensitivity in identifying latent faults, and warning thresholds are usually determined empirically, exhibiting poor adaptability.

[0005] In recent years, with the development of IoT technology, transformer condition monitoring has been able to acquire continuous and large amounts of infrared thermal imaging time-series data, providing a data foundation for in-depth research on the dynamic temperature characteristics of transformers. However, how to extract effective fault features from this time-series data, especially how to quantitatively analyze the coordinated change patterns among related components, remains a challenge for current technology.

[0006] Therefore, there is an urgent need for a new monitoring method that can deeply analyze the temperature correlation characteristics between key transformer components and adaptively establish dynamic early warning thresholds, so as to achieve accurate early warning and precise maintenance of transformers for early faults. Summary of the Invention

[0007] The purpose of this invention is to provide a power equipment maintenance and analysis system and method based on the Internet of Things to solve the problems raised in the prior art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a power equipment maintenance and analysis system and method based on the Internet of Things. The methods include: Step S1: Collect historical infrared thermograms of the transformer within several consecutive time windows under normal operating conditions to form several historical infrared thermogram sequences; obtain historical infrared thermograms of the bushing area and the winding area from the historical infrared thermograms to form a region pair; extract historical temperature data of the bushing area and the winding area from the historical infrared thermogram sequences to form historical temperature sequences of the bushing area and the winding area. By establishing historical temperature baseline data under normal operating conditions, a reliable reference is provided for subsequent coherence analysis, ensuring the accuracy of fault diagnosis. This zone pair design can capture the thermal correlation characteristics between key transformer components.

[0009] Step S2: Calculate the coherence coefficient between the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region within several time windows to form a set of historical coherence coefficients; perform statistical analysis on the set of historical coherence coefficients to determine the normal fluctuation range, which serves as the dynamic threshold interval for coherence anomalies of the region pair. By using coherence analysis to quantify the correlation between bushing and winding temperature changes, early fault characteristics that are difficult to detect using traditional methods can be captured.

[0010] Step S3: Acquire real-time infrared thermal images of the transformer within a continuous time window to form a real-time infrared thermal image sequence; extract the real-time temperature sequences of the bushing region and winding region from the real-time infrared thermal image sequence, and calculate the real-time coherence coefficient; Real-time monitoring uses the same technical approach as historical analysis to ensure consistency in feature extraction and facilitate effective comparison with historical benchmarks.

[0011] Step S4: Compare the real-time coherence coefficient with the coherence anomaly dynamic threshold. When the real-time coherence coefficient exceeds the coherence anomaly dynamic threshold range, issue an early warning.

[0012] By comparing real-time features with historical benchmarks, data-driven intelligent fault diagnosis can be achieved, enabling timely detection of abnormal equipment conditions.

[0013] Furthermore, step S1 includes: Step S1-1: Acquire historical infrared thermal images of the transformer under normal operating conditions for n consecutive time windows, where the length of each time window is t; thus forming a set of n historical infrared thermal image sequences, denoted as . ,in, This represents the historical infrared thermal image sequence for the first time window. This represents the historical infrared thermal image sequence for the second time window. This represents the historical infrared thermal image sequence for the i-th time window. The sequence represents the historical infrared thermal images of the nth time window; the historical infrared thermal image sequence of the ith time window contains m historical infrared thermal images arranged in chronological order. By adopting a continuous time window data acquisition method, the continuity and integrity of historical data are ensured, providing data support for establishing accurate dynamic thresholds.

[0014] Step S1-2: Obtain the historical infrared thermal image of the bushing area and the historical infrared thermal image of the winding area from the historical infrared thermal image, and define them as a region pair; Defining the thermally coupled bushing and winding as a pair of associated regions effectively captures the coordinated temperature changes between them. This pairing provides a clear analytical object for subsequent coherence analysis.

[0015] Steps S1-3: Traverse the set of historical infrared thermal image sequences The historical infrared thermal image sequence of the i-th time window Historical infrared thermal images in the database, and perform the following operations on the historical infrared thermal images: Step S1-3-1: Calculate the average temperature of all pixels within the sleeve region to obtain the temperature value of the sleeve region; extract the historical infrared thermal image sequence of the i-th time window. The casing region temperature values ​​at all times are arranged in chronological order to form the historical temperature sequence of the casing region for the i-th time window, denoted as . ,in This represents the temperature of the first historical infrared thermal image of the casing region in the i-th time window. This represents the temperature of the second historical infrared thermal image of the casing area in the i-th time window. This represents the temperature of the m-th historical infrared thermal image of the casing area in the i-th time window; Step S1-3-2: Calculate the average temperature of all pixels within the winding region to obtain the temperature value of the winding region; then process the historical infrared thermal image sequence of the i-th time window. The winding region temperature values ​​at all times are arranged in chronological order to form the historical temperature sequence of the winding region for the i-th time window, denoted as . ,in This represents the temperature of the first historical infrared thermal image of the winding region in the i-th time window. This represents the temperature of the second historical infrared thermal image of the winding region in the i-th time window. This represents the temperature of the m-th historical infrared thermal image of the winding region in the i-th time window.

[0016] By calculating the regional average temperature and constructing a time series, the spatial temperature distribution is transformed into a time-series signal, facilitating analysis using signal processing techniques. This data processing method can effectively extract the dynamic characteristics of temperature changes.

[0017] Furthermore, step S2 includes: Step S2-1: For the historical temperature sequence of the casing region in the i-th time window Historical temperature sequence of winding region Perform power spectral density estimation to obtain the sequence Power spectral density estimation ,sequence Power spectral density estimation and the historical temperature sequence of the casing area Historical temperature sequence of winding region Cross-power spectral density estimation ; By converting the time-domain temperature signal to the frequency domain through power spectral density estimation, the frequency characteristics of temperature changes can be revealed.

[0018] Step S2-2: Based on the power spectral density estimate obtained in step S2-1, the amplitude squared coherence method is used to calculate the historical temperature sequence of the casing region in the i-th time window. Historical temperature sequence of the winding region in the i-th time window The coherence coefficient is calculated using the following formula: ; The amplitude squared coherence coefficient can accurately quantify the degree of linear correlation between two temperature sequences at different frequencies, reflecting the strength of thermal dynamic coupling between components.

[0019] Step S2-3: Calculate the coherence coefficients at several frequency points in the frequency domain. The average value is used as the single historical coherence coefficient for the i-th time window. ; By averaging multiple frequency points, a single index representing the coherence characteristics of the entire frequency band is obtained, simplifying the complexity of subsequent statistical analysis.

[0020] Step S2-4: Traverse the set of historical infrared thermal image sequences For all n historical infrared thermal image sequences, repeat step S2-1 to obtain a set of historical coherence coefficients consisting of n individual historical coherence coefficients. ,in This represents the single historical coherence coefficient for the first time window. This represents the single historical coherence coefficient for the second time window. This represents the single historical coherence coefficient for the i-th time window. Represents the single historical coherence coefficient for the nth time window; A set of historical coherence coefficients is constructed to form a reference benchmark for the normal operating status of the equipment, providing a data foundation for setting dynamic thresholds. This multi-time-window statistical analysis can cover various normal operating conditions.

[0021] Step S2-5: For the set of historical coherence coefficients Perform statistical analysis and calculate its mean. and standard deviation and with the mean Centered on the target, and with a width of k times the standard deviation, the dynamic threshold range for coherence anomalies is determined as follows: , where k is a preset constant.

[0022] Setting a dynamic threshold range based on statistical principles can adapt to the coherence fluctuation range during normal device operation. This data-driven threshold setting method is more adaptable to complex and ever-changing actual operating environments than a fixed threshold.

[0023] Furthermore, step S3 includes: Step S3-1: Acquire real-time infrared thermal images of the transformer within the current time window, where the time window length is t; thus forming a real-time infrared thermal image sequence. The real-time infrared thermal imaging sequence It contains m real-time infrared thermal images arranged in chronological order; Use the same time window specification as historical data to ensure data consistency between real-time monitoring and historical analysis.

[0024] Step S3-2: Traverse the real-time infrared thermal image sequence The real-time infrared thermal image in the image is used to perform the following operations: Step S3-2-1: Calculate the average temperature of all pixels within the sleeve region to obtain the temperature value of the sleeve region; then, use the infrared thermal image sequence of the current time window... The casing area temperature values ​​at all times are arranged in chronological order to form the real-time temperature sequence of the casing area for the current time window, denoted as . ,in This indicates the temperature of the first real-time infrared thermal image of the casing area within the current time window. This indicates the temperature of the second real-time infrared thermal image of the casing area within the current time window. This indicates the temperature of the m-th real-time infrared thermal image of the casing area in the current time window; Step S3-2-2: Calculate the average temperature of all pixels within the winding region to obtain the temperature value of the winding region; then, use the infrared thermal image sequence of the current time window... The winding region temperature values ​​at all times are arranged in chronological order to form the real-time temperature sequence of the winding region for the current time window, denoted as . ,in This indicates the temperature of the first real-time infrared thermal image of the winding region within the current time window. This indicates the temperature of the second real-time infrared thermal image of the winding region within the current time window. This indicates the temperature of the m-th winding region in the current time window using real-time infrared thermal imaging. The real-time temperature sequence extraction uses the same algorithm as the historical data processing to ensure consistency in feature calculation.

[0025] Step S3-3: Calculate the real-time temperature sequence of the casing region for the current time window using the same amplitude squared coherence method as in step S2-1. Real-time temperature sequence of the winding region in the current time window The coherence coefficient between them; calculate the average value of the coherence coefficients at several frequency points in the frequency domain, and take the average value as the real-time coherence coefficient, denoted as . .

[0026] Maintain complete consistency between real-time coherence calculation and historical analysis methods to ensure the comparability of feature metrics.

[0027] Furthermore, step S4 includes: Step S4-1: Calculate the real-time coherence coefficient obtained in step S3. The coherence anomaly dynamic threshold range determined in step S2 Compare; The coherence coefficient calculated in real time is compared with the normal fluctuation range obtained by learning. Step S4-2: When the real-time coherence coefficient Exceeding the dynamic threshold range of coherence anomalies When the transformer is found to be in an abnormal state, an early warning is issued.

[0028] Once a coherence characteristic is detected to exceed the normal fluctuation range, an early warning mechanism is immediately triggered to enable early detection and timely handling of faults.

[0029] The system includes: a historical data acquisition and processing module, a historical coherence analysis module, a real-time monitoring and analysis module, and an early warning judgment and output module; The historical data acquisition and processing module is used to acquire historical infrared thermograms of the transformer within several consecutive time windows under normal operating conditions, forming several historical infrared thermogram sequences; to obtain historical infrared thermograms of the bushing area and the winding area from the historical infrared thermograms, forming a region pair; and to extract historical temperature data of the bushing area and the winding area from the historical infrared thermogram sequences, forming historical temperature sequences of the bushing area and historical temperature sequences of the winding area. The historical coherence analysis module is used to calculate the coherence coefficient between the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region within several time windows, forming a set of historical coherence coefficients; statistical analysis is performed on the set of historical coherence coefficients to determine the normal fluctuation range, which serves as the dynamic threshold interval for coherence anomalies of the region pair. The real-time monitoring and analysis module is used to acquire real-time infrared thermal images of the transformer within a continuous time window, forming a real-time infrared thermal image sequence; extract the real-time temperature sequences of the bushing area and winding area from the real-time infrared thermal image sequence, and calculate their coherence coefficient; The early warning judgment and output module is used to compare the real-time coherence coefficient with the coherence anomaly threshold. When the real-time coherence coefficient exceeds the dynamic threshold range of coherence anomaly, an early warning is issued.

[0030] The historical data acquisition and processing module includes: The sequence generation unit is used to acquire historical infrared thermal images of the transformer under normal operating conditions for n consecutive time windows, where the length of the time window is t; thereby forming a set of n historical infrared thermal image sequences. A region pair definition unit is used to obtain the historical infrared thermal image of the bushing region and the historical infrared thermal image of the winding region from the historical infrared thermal image, and define them as a region pair. The temperature sequence generation unit is used to traverse the historical infrared thermal image sequence of each time window in the historical infrared thermal image sequence set, calculate the average temperature of all pixels in the sleeve region and the winding region, and form the historical temperature sequence of the sleeve region and the historical temperature sequence of the winding region, respectively.

[0031] The historical coherence analysis module includes: The power spectrum estimation unit is used to estimate the power spectral density of the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region for each time window, and obtain the corresponding power spectral density estimate and cross-power spectral density estimate. The coherence calculation unit, based on power spectral density estimation, uses the amplitude squared coherence method to calculate the coherence coefficient between the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region. The coefficient averaging unit is used to calculate the average value of the coherence coefficients at several frequency points in the frequency domain, and this average value is used as the single historical coherence coefficient for the corresponding time window. The coefficient set generation unit is used to traverse all historical infrared thermal image sequences and generate a historical coherence coefficient set consisting of multiple single historical coherence coefficients. The threshold determination unit is used to perform statistical analysis on the historical coherence coefficient set, calculate its mean and standard deviation, and determine the dynamic threshold range of coherence anomalies.

[0032] The real-time monitoring and analysis module includes: The real-time sequence generation unit is used to acquire real-time infrared thermal images of the transformer within the current time window and form a real-time infrared thermal image sequence. The real-time temperature extraction unit is used to traverse the real-time infrared thermal image sequence, calculate the average temperature of all pixels in the sleeve area and the winding area, and form the real-time temperature sequence of the sleeve area and the real-time temperature sequence of the winding area respectively. The real-time coherence calculation unit uses the same amplitude squared coherence method as the historical coherence analysis to calculate the coherence coefficient between the real-time temperature sequence of the bushing region and the real-time temperature sequence of the winding region, and calculates the average value of the coherence coefficient at several frequency points in the frequency domain, and uses this average value as the real-time coherence coefficient.

[0033] The early warning judgment and output module includes: The comparison and judgment unit is used to compare the real-time coherence coefficient with the dynamic threshold range of coherence anomalies. The early warning information generation unit determines that the transformer is in an abnormal state and issues an early warning when the real-time coherence coefficient exceeds the dynamic threshold range of coherence anomaly.

[0034] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention overcomes the limitations of traditional static threshold methods. By establishing a dynamic threshold range based on statistical analysis of historical normal conditions, it can adapt to different operating conditions and environmental conditions of transformers, effectively avoiding false alarms and missed alarms caused by load fluctuations, changes in ambient temperature, and other factors. More importantly, by analyzing the coherence characteristics of bushing and winding temperature sequences in the frequency domain, it can capture subtle changes in the thermal dynamic correlation characteristics between the two components even before the absolute temperature values ​​show obvious anomalies, thereby achieving a more sensitive and advanced early warning of latent faults in transformers.

[0035] 2. This invention innovatively applies coherence analysis from the field of signal processing to transformer infrared thermography analysis for the first time. By defining the bushing and winding as a "region pair" and quantifying the temporal synergy of their temperature fluctuations, it is possible to deeply explore the inherent thermal coupling relationship and fault propagation patterns between components. This correlation analysis method not only provides a new dimension for fault diagnosis but also significantly improves the confidence and reliability of fault determination through a "cross-validation" mechanism.

[0036] 3. This invention combines Internet of Things (IoT) technology with advanced signal processing methods to construct a complete closed-loop intelligent analysis system. This system can automatically complete the entire process from data acquisition, temperature sequence extraction, coherence calculation, dynamic threshold updates to intelligent early warning, greatly reducing manual intervention and improving operation and maintenance efficiency. This integrated solution provides reliable technical support for the intelligent upgrade of power equipment from "periodic maintenance" to "condition-based maintenance," and has significant engineering application value. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the structure of an Internet of Things-based power equipment maintenance and analysis system according to the present invention; Figure 2 This is a flowchart illustrating an IoT-based power equipment maintenance and analysis method according to the present invention. Figure 3 This is a flowchart illustrating the historical coherence analysis module of the power equipment maintenance analysis method based on the Internet of Things of the present invention. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] Example: Figure 1 , Figure 2 and Figure 3 As shown, the present invention provides a technical solution: a power equipment maintenance and analysis system and method based on the Internet of Things. The methods include: Step S1: Collect historical infrared thermograms of the transformer within several consecutive time windows under normal operating conditions to form several historical infrared thermogram sequences; obtain historical infrared thermograms of the bushing area and the winding area from the historical infrared thermograms to form a region pair; extract historical temperature data of the bushing area and the winding area from the historical infrared thermogram sequences to form historical temperature sequences of the bushing area and the winding area. Step S2: Calculate the coherence coefficient between the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region within several time windows to form a set of historical coherence coefficients; perform statistical analysis on the set of historical coherence coefficients to determine the normal fluctuation range, which serves as the dynamic threshold interval for coherence anomalies of the region pair. Step S3: Acquire real-time infrared thermal images of the transformer within a continuous time window to form a real-time infrared thermal image sequence; extract the real-time temperature sequences of the bushing region and winding region from the real-time infrared thermal image sequence, and calculate the real-time coherence coefficient; Step S4: Compare the real-time coherence coefficient with the coherence anomaly dynamic threshold. When the real-time coherence coefficient exceeds the coherence anomaly dynamic threshold range, issue an early warning.

[0040] Furthermore, step S1 includes: Step S1-1: Collect historical infrared thermal images of the transformer under normal operating conditions for n=30 consecutive time windows, with each time window length t=24 hours, thus forming a set of 30 historical infrared thermal image sequences. , where each sequence A historical infrared thermal image sequence representing a time window, containing 1440 historical infrared thermal images arranged in chronological order; Step S1-2: Obtain the historical infrared thermal image of the bushing area and the historical infrared thermal image of the winding area from the historical infrared thermal image using an image recognition algorithm, and define them as a pair of regions. The coordinate range of the bushing area is (100,150) to (300,350), containing 5000 pixels, and the coordinate range of the winding area is (400,200) to (600,400), containing 8000 pixels. Steps S1-3: Traverse the set of historical infrared thermal image sequences The historical infrared thermal image sequence of the 5th time window in the series For each historical infrared thermal image in the database, perform the following processing operations: Step S1-3-1: Calculate the average temperature of all pixels within the sheath area to obtain the temperature value of the sheath area, and then use the historical infrared thermal image sequence of the 5th time window. The casing area temperature values ​​at all times are arranged in chronological order to form the historical temperature sequence of the casing area for the 5th time window, denoted as . (A total of 1440 temperature values), of which the first 5 temperature values ​​are as follows: ℃ ℃ ℃ ℃ ℃, the last temperature value ℃; Step S1-3-2: Similarly, calculate the average temperature of all pixels within the winding region to obtain the temperature value of the winding region, and then use the historical infrared thermal image sequence of the 5th time window. The winding region temperature values ​​at all times are arranged in chronological order to form the historical temperature sequence of the winding region for the 5th time window, denoted as . (A total of 1440 temperature values), of which the first 5 temperature values ​​are as follows: ℃ ℃ ℃ ℃ ℃, the last temperature value ℃.

[0041] Furthermore, step S2 includes: Step S2-1: Analyze the historical temperature sequence of the casing region for the 5th time window. Historical temperature sequence of winding region Perform power spectral density estimation to obtain the sequence Power spectral density estimation ,sequence Power spectral density estimation And the cross-power spectral density estimation of the two. Specifically, the Welch method is used: the 1440 data points are divided into 9 segments, each with 256 data points and an overlap of 128 points; a Hanning window function is applied to each segment for windowing; an FFT is performed on each windowed segment to obtain its frequency domain representation; finally, the power spectral density is estimated using the following formula: ; ; Where w=9 represents the number of segments; the frequency points are obtained as follows: sampling frequency Hz, FFT point count N=256, frequency resolution Hz, frequency vector ,Right now Hz, Hz, Hz,..., Hz. (Based on frequency point) Hz, the frequency of each segment is calculated. The values ​​are [0.25, 0.27, 0.26, 0.28, 0.24, 0.29, 0.23, 0.25, 0.26], then... ; each paragraph The values ​​are [0.31, 0.33, 0.32, 0.34, 0.30, 0.35, 0.29, 0.31, 0.32], then... ; each paragraph The values ​​are [0.25, 0.27, 0.26, 0.28, 0.24, 0.29, 0.23, 0.25, 0.26], then... ; Step S2-2: Based on the power spectral density estimate obtained in step S2-1, the amplitude squared coherence method is used to calculate the historical temperature sequence of the casing region in the fifth time window. Historical temperature sequence of winding region The coherence coefficient is calculated using the following formula: ; by frequency point Hz, ; Step S2-3: Select frequency points in the frequency range of 0.001-0.1Hz in the frequency domain and calculate the coherence coefficients of these frequency points. The average value is determined by selecting 128 frequency points from the frequency vector that satisfy the condition 0.001Hz ≤ f ≤ 0.1Hz. The specific calculation process involves calculating the coherence coefficient values ​​at each of the 128 frequency points as [0.81, 0.78, 0.83, 0.77, 0.82, ..., 0.84], and then calculating the average value. This average value is used as the single historical coherence coefficient for the fifth time window. ; Step S2-4: Traverse the set of historical infrared thermal image sequences For all 30 historical infrared thermogram sequences, repeat steps S2-1 to S2-3 to obtain a set of historical coherence coefficients consisting of 30 individual historical coherence coefficients. ; Step S2-5: For the set of historical coherence coefficients Statistical analysis was performed using formulas. Calculate the mean Using formula Calculate the standard deviation Centered on the mean μ, and with k = 2 standard deviations as the width, the dynamic threshold interval for coherence anomalies is determined as follows: .

[0042] Furthermore, step S3 includes: Step S3-1: Acquire real-time infrared thermal images of the transformer within the current time window, with the time window length maintained at t=24 hours, to form a real-time infrared thermal image sequence. The sequence contains 1,440 real-time infrared thermal images arranged in chronological order; Step S3-2: Traverse the real-time infrared thermal image sequence The real-time infrared thermal images in the image are processed by performing the following operations on each real-time infrared thermal image: Step S3-2-1: Calculate the average temperature of all pixels within the sheath area to obtain the temperature value of the sheath area, and then use the infrared thermal image sequence of the current time window. The casing area temperature values ​​at all times are arranged in chronological order to form the real-time temperature sequence of the casing area for the current time window, denoted as . (A total of 1440 temperature values), of which the first 5 temperature values ​​are as follows: ℃ ℃ ℃ ℃ ℃, the last temperature value ; Step S3-2-2: Calculate the average temperature of all pixels within the winding region to obtain the temperature value of the winding region, and then use the infrared thermal image sequence of the current time window. The winding region temperature values ​​at all times are arranged in chronological order to form the real-time temperature sequence of the winding region for the current time window, denoted as . (A total of 1440 temperature values), of which the first 5 temperature values ​​are as follows: ℃ ℃ ℃ ℃ ℃, the last temperature value ℃; Step S3-3: Calculate the real-time temperature sequence of the casing region for the current time window using the same amplitude squared coherence method as in step S2-1. Real-time temperature sequence of winding region The coherence coefficient between frequencies is calculated by averaging the coherence coefficients at 128 frequency points within the same frequency range of 0.001-0.1Hz. Specifically, the coherence coefficient values ​​[0.54, 0.55, 0.53, 0.56, 0.54, ..., 0.55] are calculated at each of these 128 frequency points, and the average value is then taken. This average value is used as the real-time coherence coefficient. .

[0043] Furthermore, step S4 includes: Step S4-1: Calculate the real-time coherence coefficient obtained in step S3. Compare with the coherence anomaly dynamic threshold range [0.67, 0.91] determined in step S2; Step S4-2: Due to the real-time coherence coefficient If the value exceeds the lower limit of the dynamic threshold range [0.67, 0.91] for coherence anomalies, the transformer is deemed to be in an abnormal state, and an early warning is issued.

[0044] The system includes: a historical data acquisition and processing module, a historical coherence analysis module, a real-time monitoring and analysis module, and an early warning judgment and output module; The historical data acquisition and processing module is used to acquire historical infrared thermograms of the transformer within several consecutive time windows under normal operating conditions, forming several historical infrared thermogram sequences; to obtain historical infrared thermograms of the bushing area and the winding area from the historical infrared thermograms, forming a region pair; and to extract historical temperature data of the bushing area and the winding area from the historical infrared thermogram sequences, forming historical temperature sequences of the bushing area and historical temperature sequences of the winding area. The historical coherence analysis module is used to calculate the coherence coefficient between the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region within several time windows, forming a set of historical coherence coefficients; statistical analysis is performed on the set of historical coherence coefficients to determine the normal fluctuation range, which serves as the dynamic threshold interval for coherence anomalies of the region pair. The real-time monitoring and analysis module is used to acquire real-time infrared thermal images of the transformer within a continuous time window, forming a real-time infrared thermal image sequence; extract the real-time temperature sequences of the bushing area and winding area from the real-time infrared thermal image sequence, and calculate their coherence coefficient; The early warning judgment and output module is used to compare the real-time coherence coefficient with the coherence anomaly threshold. When the real-time coherence coefficient exceeds the dynamic threshold range of coherence anomaly, an early warning is issued.

[0045] The historical data acquisition and processing module includes: The sequence generation unit is used to acquire historical infrared thermal images of the transformer under normal operating conditions for n consecutive time windows, where the length of the time window is t; thereby forming a set of n historical infrared thermal image sequences. A region pair definition unit is used to obtain the historical infrared thermal image of the bushing region and the historical infrared thermal image of the winding region from the historical infrared thermal image, and define them as a region pair. The temperature sequence generation unit is used to traverse the historical infrared thermal image sequence of each time window in the historical infrared thermal image sequence set, calculate the average temperature of all pixels in the sleeve region and the winding region, and form the historical temperature sequence of the sleeve region and the historical temperature sequence of the winding region, respectively.

[0046] The historical coherence analysis module includes: The power spectrum estimation unit is used to estimate the power spectral density of the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region for each time window, and obtain the corresponding power spectral density estimate and cross-power spectral density estimate. The coherence calculation unit, based on power spectral density estimation, uses the amplitude squared coherence method to calculate the coherence coefficient between the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region. The coefficient averaging unit is used to calculate the average value of the coherence coefficients at several frequency points in the frequency domain, and this average value is used as the single historical coherence coefficient for the corresponding time window. The coefficient set generation unit is used to traverse all historical infrared thermal image sequences and generate a historical coherence coefficient set consisting of multiple single historical coherence coefficients. The threshold determination unit is used to perform statistical analysis on the historical coherence coefficient set, calculate its mean and standard deviation, and determine the dynamic threshold range of coherence anomalies.

[0047] The real-time monitoring and analysis module includes: The real-time sequence generation unit is used to acquire real-time infrared thermal images of the transformer within the current time window and form a real-time infrared thermal image sequence. The real-time temperature extraction unit is used to traverse the real-time infrared thermal image sequence, calculate the average temperature of all pixels in the sleeve area and the winding area, and form the real-time temperature sequence of the sleeve area and the real-time temperature sequence of the winding area respectively. The real-time coherence calculation unit uses the same amplitude squared coherence method as the historical coherence analysis to calculate the coherence coefficient between the real-time temperature sequence of the bushing region and the real-time temperature sequence of the winding region, and calculates the average value of the coherence coefficient at several frequency points in the frequency domain, and uses this average value as the real-time coherence coefficient.

[0048] The early warning judgment and output module includes: The comparison and judgment unit is used to compare the real-time coherence coefficient with the dynamic threshold range of coherence anomalies. The early warning information generation unit determines that the transformer is in an abnormal state and issues an early warning when the real-time coherence coefficient exceeds the dynamic threshold range of coherence anomaly.

[0049] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A power equipment maintenance and analysis method based on the Internet of Things, characterized in that: Includes the following steps: Step S1: Collect historical infrared thermograms of the transformer within several consecutive time windows under normal operating conditions to form several historical infrared thermogram sequences; obtain historical infrared thermograms of the bushing area and the winding area from the historical infrared thermograms to form a region pair; extract historical temperature data of the bushing area and the winding area from the historical infrared thermogram sequences to form historical temperature sequences of the bushing area and the winding area. Step S2: Calculate the coherence coefficient between the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region within several time windows to form a set of historical coherence coefficients; perform statistical analysis on the set of historical coherence coefficients to determine the normal fluctuation range, which serves as the dynamic threshold interval for coherence anomalies of the region pair. Step S3: Acquire real-time infrared thermal images of the transformer within a continuous time window to form a real-time infrared thermal image sequence; extract the real-time temperature sequences of the bushing region and winding region from the real-time infrared thermal image sequence, and calculate the real-time coherence coefficient; Step S4: Compare the real-time coherence coefficient with the coherence anomaly dynamic threshold. When the real-time coherence coefficient exceeds the coherence anomaly dynamic threshold range, issue an early warning.

2. The method for power equipment maintenance and analysis based on the Internet of Things according to claim 1, characterized in that: Step S1 includes: Step S1-1: Acquire historical infrared thermal images of the transformer under normal operating conditions for n consecutive time windows, where the length of each time window is t; thus forming a set of n historical infrared thermal image sequences, denoted as . ,in, This represents the historical infrared thermal image sequence for the first time window. This represents the historical infrared thermal image sequence for the second time window. This represents the historical infrared thermal image sequence for the i-th time window. The sequence represents the historical infrared thermal images of the nth time window; the historical infrared thermal image sequence of the ith time window contains m historical infrared thermal images arranged in chronological order. Step S1-2: Obtain the historical infrared thermal image of the bushing area and the historical infrared thermal image of the winding area from the historical infrared thermal image, and define them as a region pair; Steps S1-3: Traverse the set of historical infrared thermal image sequences The historical infrared thermal image sequence of the i-th time window Historical infrared thermal images in the database, and perform the following operations on the historical infrared thermal images: Step S1-3-1: Calculate the average temperature of all pixels within the sleeve region to obtain the temperature value of the sleeve region; extract the historical infrared thermal image sequence of the i-th time window. The casing region temperature values ​​at all times are arranged in chronological order to form the historical temperature sequence of the casing region for the i-th time window, denoted as . ,in This represents the temperature of the first historical infrared thermal image of the casing region in the i-th time window. This represents the temperature of the second historical infrared thermal image of the casing area in the i-th time window. This represents the temperature of the m-th historical infrared thermal image of the casing area in the i-th time window; Step S1-3-2: Calculate the average temperature of all pixels within the winding region to obtain the temperature value of the winding region; then process the historical infrared thermal image sequence of the i-th time window. The winding region temperature values ​​at all times are arranged in chronological order to form the historical temperature sequence of the winding region for the i-th time window, denoted as . ,in This represents the temperature of the first historical infrared thermal image of the winding region in the i-th time window. This represents the temperature of the second historical infrared thermal image of the winding region in the i-th time window. This represents the temperature of the m-th historical infrared thermal image of the winding region in the i-th time window.

3. The method for power equipment maintenance and analysis based on the Internet of Things according to claim 2, characterized in that: Step S2 includes: Step S2-1: For the historical temperature sequence of the casing region in the i-th time window Historical temperature sequence of winding region Perform power spectral density estimation to obtain the sequence Power spectral density estimation ,sequence Power spectral density estimation and the historical temperature sequence of the casing area Historical temperature sequence of winding region Cross-power spectral density estimation ; Step S2-2: Based on the power spectral density estimate obtained in step S2-1, the amplitude squared coherence method is used to calculate the historical temperature sequence of the casing region in the i-th time window. Historical temperature sequence of the winding region in the i-th time window The coherence coefficient is calculated using the following formula: ; Step S2-3: Calculate the coherence coefficients at several frequency points in the frequency domain. The average value is used as the single historical coherence coefficient for the i-th time window. ; Step S2-4: Traverse the set of historical infrared thermal image sequences For all n historical infrared thermal image sequences, repeat step S2-1 to obtain a set of historical coherence coefficients consisting of n individual historical coherence coefficients. ,in This represents the single historical coherence coefficient for the first time window. This represents the single historical coherence coefficient for the second time window. This represents the single historical coherence coefficient for the i-th time window. Represents the single historical coherence coefficient for the nth time window; Step S2-5: For the set of historical coherence coefficients Perform statistical analysis and calculate its mean. and standard deviation and with the mean Centered on the target, and with a width of k times the standard deviation, the dynamic threshold range for coherence anomalies is determined as follows: , where k is a preset constant.

4. The method for power equipment maintenance and analysis based on the Internet of Things according to claim 3, characterized in that: Step S3 includes: Step S3-1: Acquire real-time infrared thermal images of the transformer within the current time window, where the time window length is t; thus forming a real-time infrared thermal image sequence. The real-time infrared thermal imaging sequence It contains m real-time infrared thermal images arranged in chronological order; Step S3-2: Traverse the real-time infrared thermal image sequence The real-time infrared thermal image in the image is used to perform the following operations: Step S3-2-1: Calculate the average temperature of all pixels within the sleeve region to obtain the temperature value of the sleeve region; then, use the infrared thermal image sequence of the current time window... The casing area temperature values ​​at all times are arranged in chronological order to form the real-time temperature sequence of the casing area for the current time window, denoted as . ,in This indicates the temperature of the first real-time infrared thermal image of the casing area within the current time window. This indicates the temperature of the second real-time infrared thermal image of the casing area within the current time window. This indicates the temperature of the m-th real-time infrared thermal image of the casing area in the current time window; Step S3-2-2: Calculate the average temperature of all pixels within the winding region to obtain the temperature value of the winding region; then, use the infrared thermal image sequence of the current time window... The winding region temperature values ​​at all times are arranged in chronological order to form the real-time temperature sequence of the winding region for the current time window, denoted as . ,in This indicates the temperature of the first real-time infrared thermal image of the winding region within the current time window. This indicates the temperature of the second real-time infrared thermal image of the winding region within the current time window. This indicates the temperature of the m-th winding region in the current time window using real-time infrared thermal imaging. Step S3-3: Calculate the real-time temperature sequence of the casing region for the current time window using the same amplitude squared coherence method as in step S2-1. Real-time temperature sequence of the winding region in the current time window The coherence coefficient between them; calculate the average value of the coherence coefficients at several frequency points in the frequency domain, and take the average value as the real-time coherence coefficient, denoted as . .

5. The method for power equipment maintenance and analysis based on the Internet of Things according to claim 4, characterized in that: Step S4 includes: Step S4-1: Calculate the real-time coherence coefficient obtained in step S3. The coherence anomaly dynamic threshold range determined in step S2 Compare; Step S4-2: When the real-time coherence coefficient Exceeding the dynamic threshold range of coherence anomalies When the transformer is found to be in an abnormal state, an early warning is issued.

6. A power equipment maintenance and analysis system based on the Internet of Things (IoT), used to execute the power equipment maintenance and analysis method based on the IoT as described in any one of claims 1-5, characterized in that: The system includes: Historical data acquisition and processing module, historical coherence analysis module, real-time monitoring and analysis module, and early warning judgment and output module; The historical data acquisition and processing module is used to acquire historical infrared thermograms of the transformer within several consecutive time windows under normal operating conditions, forming several historical infrared thermogram sequences; to obtain historical infrared thermograms of the bushing area and the winding area from the historical infrared thermograms, forming a region pair; and to extract historical temperature data of the bushing area and the winding area from the historical infrared thermogram sequences, forming historical temperature sequences of the bushing area and historical temperature sequences of the winding area. The historical coherence analysis module is used to calculate the coherence coefficient between the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region within several time windows, forming a set of historical coherence coefficients; statistical analysis is performed on the set of historical coherence coefficients to determine the normal fluctuation range, which serves as the dynamic threshold interval for coherence anomalies of the region pair. The real-time monitoring and analysis module is used to acquire real-time infrared thermal images of the transformer within a continuous time window, forming a real-time infrared thermal image sequence; extract the real-time temperature sequences of the bushing area and winding area from the real-time infrared thermal image sequence, and calculate their coherence coefficient; The early warning judgment and output module is used to compare the real-time coherence coefficient with the coherence anomaly threshold. When the real-time coherence coefficient exceeds the dynamic threshold range of coherence anomaly, an early warning is issued.

7. The method for power equipment maintenance and analysis based on the Internet of Things according to claim 6, characterized in that: The historical data acquisition and processing module includes: The sequence generation unit is used to acquire historical infrared thermal images of the transformer under normal operating conditions for n consecutive time windows, where the length of the time window is t; thereby forming a set of n historical infrared thermal image sequences. A region pair definition unit is used to obtain the historical infrared thermal image of the bushing region and the historical infrared thermal image of the winding region from the historical infrared thermal image, and define them as a region pair. The temperature sequence generation unit is used to traverse the historical infrared thermal image sequence of each time window in the historical infrared thermal image sequence set, calculate the average temperature of all pixels in the sleeve region and the winding region, and form the historical temperature sequence of the sleeve region and the historical temperature sequence of the winding region, respectively.

8. The method for power equipment maintenance and analysis based on the Internet of Things according to claim 7, characterized in that: The historical coherence analysis module includes: The power spectrum estimation unit is used to estimate the power spectral density of the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region for each time window, and obtain the corresponding power spectral density estimate and cross-power spectral density estimate. The coherence calculation unit, based on power spectral density estimation, uses the amplitude squared coherence method to calculate the coherence coefficient between the historical temperature sequence of the bushing region and the historical temperature sequence of the winding region. The coefficient averaging unit is used to calculate the average value of the coherence coefficients at several frequency points in the frequency domain, and this average value is used as the single historical coherence coefficient for the corresponding time window. The coefficient set generation unit is used to traverse all historical infrared thermal image sequences and generate a historical coherence coefficient set consisting of multiple single historical coherence coefficients. The threshold determination unit is used to perform statistical analysis on the historical coherence coefficient set, calculate its mean and standard deviation, and determine the dynamic threshold range of coherence anomalies.

9. The method for power equipment maintenance and analysis based on the Internet of Things according to claim 8, characterized in that: The real-time monitoring and analysis module includes: The real-time sequence generation unit is used to acquire real-time infrared thermal images of the transformer within the current time window and form a real-time infrared thermal image sequence. The real-time temperature extraction unit is used to traverse the real-time infrared thermal image sequence, calculate the average temperature of all pixels in the sleeve area and the winding area, and form the real-time temperature sequence of the sleeve area and the real-time temperature sequence of the winding area respectively. The real-time coherence calculation unit uses the same amplitude squared coherence method as the historical coherence analysis to calculate the coherence coefficient between the real-time temperature sequence of the bushing region and the real-time temperature sequence of the winding region, and calculates the average value of the coherence coefficient at several frequency points in the frequency domain, and uses this average value as the real-time coherence coefficient.

10. The method for power equipment maintenance and analysis based on the Internet of Things according to claim 9, characterized in that: The early warning judgment and output module includes: The comparison and judgment unit is used to compare the real-time coherence coefficient with the dynamic threshold range of coherence anomalies. The early warning information generation unit determines that the transformer is in an abnormal state and issues an early warning when the real-time coherence coefficient exceeds the dynamic threshold range of coherence anomaly.

Images