A multi-parameter joint fault analysis system for transformers

Abstract

The present application relates to the technical field of power equipment monitoring and fault analysis, and particularly relates to a multi-parameter joint fault analysis system for transformers. The multi-parameter joint fault analysis system for transformers comprises the following modules: a data acquisition module, a baseline compensation module, a feature extraction module and a fault discrimination module. The data acquisition module is used for continuously acquiring online DGA monitoring data and top-layer oil temperature time-series monitoring data of the transformer. The baseline compensation module is used for performing working condition interference compensation according to the online DGA monitoring data and the top-layer oil temperature time-series data. The present application separates fault gas production residual error through two-stage baseline difference correction, and combines a three-layer verification system of time-series offset, hysteresis area and carbon oxide increment to realize fault analysis of the transformer, thereby improving the discrimination accuracy of transformer core thermal fault diagnosis.

Description

Technical Field

[0001] This invention relates to the field of power equipment monitoring and fault analysis technology, and in particular to a multi-parameter joint fault analysis system for transformers. Background Technology

[0002] Dissolved gas analysis (DGA) is the mainstream method for transformer fault diagnosis. Both multi-point grounding contact thermal faults in the core and thermal aging of winding insulation cause a continuous increase in CO and CO2 levels in the oil, and the gas composition characteristics of these two types of faults highly overlap. Traditional diagnostic methods relying solely on DGA components cannot distinguish faults originating from the same gas source, easily leading to misjudgments and missed detections, resulting in poor diagnostic reliability. Conventional monitoring schemes that synchronously acquire oil temperature do not compensate for operating condition interference. Gas solubility shifts caused by oil temperature changes and baseline gas production from normal insulation thermal aging can mask true abnormal signals, making it difficult to extract effective fault identification criteria. Furthermore, existing technologies lack long-cycle time-series analysis logic, cannot quantify the response differences between gas production and oil temperature, and cannot distinguish fault causes through asymmetric loss characteristics during temperature rise and fall processes. They can only rely on instantaneous gas concentration to judge equipment status, resulting in low diagnostic confidence. Summary of the Invention

[0003] Therefore, it is necessary to provide a multi-parameter joint fault analysis system for transformers to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, a multi-parameter joint fault analysis system for transformers is provided, the system comprising the following modules:

[0005] The data acquisition module is used to continuously collect online DGA monitoring data and top oil temperature time-series monitoring data of the transformer;

[0006] The baseline compensation module is used to perform operating condition interference compensation based on online DGA monitoring data and top oil temperature time series data, calculate the dissolved gas concentration component corresponding to the temperature, calculate the basic gas production component under normal aging of the equipment, and generate a normalized abnormal gas production residual sequence.

[0007] The feature extraction module is used to select continuous time-series data throughout the year, perform time-series operations on the abnormal gas production residual sequence and top oil temperature time-series data, and output phase shift features; construct a scatter plot using top oil temperature and gas production residual and solve for the hysteresis region area parameter;

[0008] The fault identification module is used to determine faults based on the incremental carbon oxide composition, phase shift characteristics, and hysteresis region area parameters. When the incremental carbon oxide composition is the main growth component, the timing calculation outputs a non-zero phase shift. The closed hysteresis region formed by the scatter plot is identified as the core contact overheating fault region, and the power outage detection and maintenance period corresponding to the core thermal expansion state is output based on the phase shift characteristics.

[0009] The beneficial effects of the present invention are as follows:

[0010] I. This invention establishes a two-level baseline difference correction process. It replaces the solubility migration difference value at each time-series node based on the oil dissolved gas comparison data table. Then, it separates the normal aging gas production component based on the insulation paper baseline gas production value, independently outputting the abnormal gas production residual sequence corresponding to core bonding heating. This layered difference calculation achieves layered separation of two types of steady-state gas production signals—insulating oil temperature expansion and contraction, and insulation paper natural aging—from the core overheating fault gas production signal. It physically separates the two types of homogeneous gas production components. The same set of carbon monoxide and carbon dioxide concentration data can be decomposed into background steady-state components and fault abnormal components, eliminating the problem of component characteristic confusion caused by the superposition of two types of non-fault gas production.

[0011] II. This invention constructs a complete set of time-series feature extraction and judgment systems, including annual time-series offset traversal calculation, hysteresis region area quantification calculation, and three-layer parameter linkage verification. The phase offset duration of gas production relative to oil temperature is obtained through equal-step global offset traversal. The hysteresis closed contour area is obtained by accumulating isothermal wide-zone units, and three-layer synchronous verification is completed by combining the annual carbon oxide increment. This complete set of quantitative time-series features can fully capture the hysteresis response relationship between gas production and oil temperature during the core's annual expansion and contraction cycle, as well as the asymmetric change law of gas production during heating and cooling stages. It relies on multi-dimensional quantitative parameters to complete state differentiation, synchronously locking the maintenance time-series interval with the strongest defect signal coupling, and achieving accurate classification of two types of operating conditions: normal thermal decomposition of insulating paper and multi-point contact overheating of the core. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of a multi-parameter joint fault analysis system for transformers.

[0013] Figure 2 This is a schematic diagram illustrating the synchronous acquisition of transformer top-level oil temperature and dissolved gas in the oil according to one embodiment.

[0014] Figure 3 A flowchart illustrating the calculation of the concentration compensation of five characteristic gases in the oil at the top oil temperature in one embodiment;

[0015] Figure 4 This is a diagram illustrating the interface of the transformer multi-parameter joint fault analysis system of the present invention;

[0016] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0018] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0019] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0020] To achieve the above objectives, please refer to Figures 1 to 4 A multi-parameter joint fault analysis system for transformers, the system 10 includes the following modules: data acquisition module 101, baseline compensation module 102, feature extraction module 103 and fault discrimination module 104;

[0021] Preferably, the data acquisition module is used to continuously acquire online DGA monitoring data and top oil temperature time-series monitoring data of the transformer;

[0022] Optionally, the data acquisition module includes the following for continuously acquiring online DGA monitoring data and top oil temperature time-series monitoring data of the transformer:

[0023] Read the online DGA monitoring data generated by the oil intake pipeline of the transformer tank. The online DGA monitoring data includes independent concentration records of dissolved carbon monoxide, carbon dioxide, methane, ethylene and acetylene in the insulating oil.

[0024] Read the time-series monitoring data of the top oil temperature generated by the heat conduction of the metal shell of the transformer tank cover. The reading action is performed synchronously according to the adaptively adjustable periodic acquisition interval.

[0025] The oil tank's oil intake pipeline is connected to the circulating insulating oil inside the transformer tank. The insulating oil continuously wets the outer layer of the iron core covered with insulating paper, and receives the carbon oxides decomposed by the insulating paper when heated, which are then retained inside the oil to form a corresponding concentration record.

[0026] The obtained online DGA monitoring data and top oil temperature time-series monitoring data are uniformly bound with a unique time sequence identifier. After binding, the data is transmitted to the baseline compensation module to perform operating condition interference compensation processing.

[0027] In one embodiment, the data acquisition module is activated to perform synchronous data acquisition. The acquisition interval is adaptively set between 30 and 120 minutes, and data acquisition is triggered synchronously according to the set time nodes. The existing oil sampling pipeline in the transformer tank continuously draws circulating insulating oil from within the tank, generating real-time online DGA monitoring data. This data synchronously records the independent concentration values ​​of carbon monoxide, carbon dioxide, methane, ethylene, and acetylene in the insulating oil. A temperature acquisition element is attached and fixed to the outer surface of the transformer tank cover metal shell, with the element's contact area completely in contact with the metal shell. The metal shell transfers heat from the transformer's internal oil to the acquisition element, which continuously records temperature values ​​at different times, generating top-layer oil temperature time-series monitoring data hourly. The circulating insulating oil inside the transformer continuously wets the outer layer of insulating paper covering the core. The carbon oxides produced by the thermal decomposition of the insulating paper directly dissolve into the insulating oil fluid, and changes in gas concentration are synchronously recorded in the corresponding monitoring data fields in real time.

[0028] After a single cycle of synchronous acquisition is completed, a unique time series identifier is configured for the online DGA monitoring data and top oil temperature time series monitoring data generated in the current period. The time series identifier adopts a combination format of eight-digit date code plus six-digit time code. Each set of synchronous data from the same source is bound and encapsulated to generate a standardized set of time series datasets. The set of time series datasets is directly pushed to the baseline compensation module through a preset internal transmission link.

[0029] In another embodiment, after receiving the complete time-series dataset, the baseline compensation module retrieves a standard dissolved gas reference data table for the mineral insulating oil of the transformer. The data table covers a temperature range from 0°C to 100°C, with each 1°C gradient corresponding to a set of specific gas solubility parameters. The top-level oil temperature values ​​of each time-series node within the complete time-series dataset are extracted, and the solubility migration is obtained by matching the gas solubility parameters at the corresponding temperatures in the data table. For the concentration values ​​of the five types of gases in the online DGA monitoring data, a time-series difference calculation is performed between each value and the solubility migration at the corresponding time point. The calculated values ​​replace the original collected concentration values.

[0030] See Figure 2The main component shown is an oil-immersed transformer. A temperature sensing element is attached to the transformer enclosure cover to collect top-layer oil temperature data. An oil sampling pipeline connects to the lower part of the transformer, linking to an online DGA monitoring component to collect dissolved gas data from the oil sample. Both types of monitoring data are fed into a synchronous acquisition unit, which is set to an adaptive acquisition interval of 30 to 120 minutes. All acquired data enters a time-series identifier binding unit, where each data point is assigned a YYYYMMDD timestamp plus an HHMMSS timestamp. Data with time-series identifiers is then pushed outwards and flows into the baseline compensation processing unit.

[0031] Preferably, the baseline compensation module is used to perform operating condition interference compensation based on online DGA monitoring data and top oil temperature time series data, calculate the dissolved gas concentration component corresponding to the temperature, calculate the basic gas production component under normal aging of the equipment, and generate a normalized abnormal gas production residual sequence.

[0032] Optionally, the dissolved gas concentration component corresponding to the calculated temperature in the baseline compensation module is specifically as follows:

[0033] Retrieve the top layer oil temperature time-series monitoring data generated by heat conduction from the metal shell of the transformer tank cover, and retrieve the preset oil-dissolved gas comparison data table matching mineral insulating oil;

[0034] According to the quantitative conversion logic of Henry's Law, the amount of gas solubility migration caused by oil temperature change is calculated at each time node. The difference between the online DGA monitoring data generated by the oil tank pipeline and the amount of gas solubility migration is calculated line by line.

[0035] The result of the difference calculation is used to replace the original online DGA monitoring data to correct the dissolved gas concentration component caused by the temperature expansion and contraction of the insulating oil.

[0036] In one embodiment, the baseline compensation module retrieves the top-layer oil temperature time-series monitoring data generated by the thermal conduction of the transformer tank cover metal shell, and simultaneously retrieves a pre-stored oil-gas dissolution comparison data table for general mineral insulating oil compatible with transformer equipment. This data table has a preset temperature coverage range of 0℃ to 100℃, with a fixed gradient of 1℃ to complete the level division. Each temperature level independently corresponds to the standard solubility parameters of five dissolved gases: carbon monoxide, carbon dioxide, methane, ethylene, and acetylene. The internal parameters of the data table are consistent with the type of insulating oil added to the transformer on-site. According to the time sequence arrangement of the top-layer oil temperature time-series monitoring data, the real-time oil temperature value is extracted for each time-series node, and the solubility parameters of the five gases at the corresponding temperature level are retrieved by matching the oil-gas dissolution comparison data table. The amount of gas solubility migration corresponding to the oil temperature fluctuation at each time-series node is calculated.

[0037] For example, a practical demonstration of difference calculation was carried out by selecting a conventional timing node for stable transformer operation. At the timing node corresponding to an oil temperature of 50℃, the solubility migration of various gases at that temperature was obtained by referring to a table. The solubility migration of carbon dioxide was 10μL / L, the solubility migration of carbon monoxide was 6μL / L, and the solubility migration of methane was 3μL / L. The online DGA monitoring data corresponding to the same time point was retrieved. The original concentration of carbon dioxide in this data set was 52 μL / L, the original concentration of carbon monoxide was 27 μL / L, and the original concentration of methane was 15 μL / L. The item-by-item difference calculation was performed sequentially: 52 μL / L minus 10 μL / L yielded a corrected concentration of carbon dioxide of 42 μL / L, 27 μL / L minus 6 μL / L yielded a corrected concentration of carbon monoxide of 21 μL / L, and 15 μL / L minus 3 μL / L yielded a corrected concentration of methane of 12 μL / L. Ethylene and acetylene were calculated using the same item-by-item difference calculation method, combined with their respective corresponding solubility migration and original monitoring concentrations.

[0038] In another embodiment, after each time-series node completes the sub-item difference calculation for the five gas categories, the calculated corrected concentration values ​​are used to replace the original recorded data of the same time series and gas category within the online DGA monitoring data in situ. The data replacement operation proceeds line by line according to the sequence of time-series nodes, with each time-series node's data being calculated and replaced independently.

[0039] See Figure 3 The process begins by retrieving a standard dissolved gas reference table. The table parameters cover the temperature range from 0 to 100°C, with a set of corresponding parameters set in 1°C increments. The reference table data is matched with the collected top-layer oil temperature values, and the matching operation retrieves the solubility migration of various gases at the corresponding temperatures. For five characteristic gases—CO, CO2, CH4, C2H4, and C2H2—item-by-item difference calculations are performed, generating compensation data. This compensation data is output in chart form, and the charts contain data used for analyzing gas production trends and determining equipment malfunctions.

[0040] Optionally, the baseline compensation module calculates the basic gas production component during normal aging of the equipment as follows:

[0041] Extract time-series monitoring data of top oil temperature generated by heat conduction from the metal casing of the transformer tank cover throughout the entire year;

[0042] The baseline gas production value of the natural decomposition of the outer layer of the iron core insulation paper is quantitatively calculated based on the Arrhenius equation. The outer layer of the iron core insulation paper completely covers the surface of the iron core metal laminations, and the baseline gas production value corresponds to the corresponding value of the gas produced by the stable decomposition of the insulation paper under the condition that the iron core laminations are not attached and are heated.

[0043] The difference between the online DGA monitoring data after solubility correction and the baseline gas production value is calculated at each time node to output the basic gas production quantity corresponding to the normal aging of the equipment corresponding to the core bonding heating.

[0044] In one embodiment, time-series monitoring data of top oil temperature generated by heat conduction from the metal shell of the transformer tank cover throughout the entire natural year are extracted, and oil temperature record data corresponding to continuous time-series nodes throughout the year are collected. All monitoring data are generated based on the correspondence between the heat conduction of the metal shell of the tank cover and the oil temperature. The time-series data spans the transformer's annual operating conditions and fully matches the complete cyclical change process of the core metal laminations' heating and cooling. The outer surface of the core metal laminations is completely covered with an insulating paper structure, and the insulating paper is kept in a fully covered and adhered state. The time-series monitoring data of top oil temperature collected throughout the year serves as the basic parameter for the quantitative calculation of the thermal decomposition of the insulating paper.

[0045] It should be noted that the preset calibration parameters for gas production during normal aging of transformer core insulation paper are retrieved. These parameters correspond to the standard gas production values ​​under conditions where the core laminations do not have localized adhesion heating or abnormal overheating. The calibration parameters are adapted to the normal operating oil temperature range of transformers from 25℃ to 90℃. Different temperature nodes correspond to specific stable thermal decomposition benchmark gas production values ​​for insulation paper. The annual top-layer oil temperature time-series monitoring data are matched node by node. For each oil temperature time-series node, the benchmark gas production value of insulation paper at the corresponding temperature level is retrieved, and the data is collected to form a benchmark gas production time-series dataset for the entire year.

[0046] For example, a practical exercise of gas-specific differential calculations was conducted at any time point throughout the year. Taking the 55℃ oil temperature time point as an example, the baseline carbon dioxide production value for normal aging of insulating paper at this time point is 8 μL / L, and the baseline carbon monoxide production value is 4 μL / L. The online DGA monitoring data for this time point, after solubility correction, yields a carbon dioxide concentration of 42 μL / L and a carbon monoxide concentration of 21 μL / L. Performing gas-specific differential calculations, subtracting 8 μL / L from 42 μL / L yields a carbon dioxide-specific production component of 34 μL / L, and subtracting 4 μL / L from 21 μL / L yields a carbon monoxide-specific production component of 17 μL / L. Methane, ethylene, and acetylene gases were calculated using the same logic, combining the corresponding time point baseline production values ​​with the corrected online DGA monitoring data to complete the item-specific differential calculation operation.

[0047] Of particular importance is that the continuous time-series data in the feature extraction module specifically includes:

[0048] Extract all online DGA monitoring data and top oil temperature time-series monitoring data bound to time-series identifiers within the complete natural year. Only select data segments whose time-series identifier time axis completely spans the complete cycle of heating expansion and cooling contraction of the iron core metal laminations, excluding local data that only cover the local deformation stage of a single season.

[0049] The data segments whose timing markers were broken due to the blockage of the insulating oil delivery path and the interruption of heat conduction of the box cover were isolated. Only the complete set of data with uninterrupted timing marker connection and synchronous and stable oil dissolution and shell thermal conduction status were retained.

[0050] The selected complete set of data is directed to the timing processing unit as the sole input carrier for computation.

[0051] In one embodiment, all online DGA monitoring data and top-layer oil temperature time-series monitoring data bound with unique time-series identifiers are extracted across the entire calendar year, and a complete set of monitoring data generated by continuous time-series nodes throughout the year is collected. Using the physical deformation cycle of the core metal laminations as the screening criterion, the annual time-series data coverage interval is divided. Continuous data segments that completely encompass the entire process of heating and expansion and cooling and contraction of the core metal laminations are selected, while local time-series data segments containing only a single heating or cooling deformation stage are removed. Time-series monitoring data corresponding to one complete expansion-contraction cycle are retained.

[0052] The annual time-series data after preliminary screening were used to verify the integrity of the time series. For the periods of interruption of oil delivery caused by the start-up and shutdown of the insulating oil delivery path, pipeline maintenance, and abnormal heat exchange structure, as well as the periods of interruption of heat conduction caused by contact misalignment of the temperature measurement structure of the cover shell and external interference, the data segments with broken time series markers, data record gaps and misalignments caused by these two types of conditions were located. All abnormal data segments were isolated and archived, and the isolated data segments were not included in the subsequent time series calculation process. The complete set of online DGA monitoring data and top oil temperature time series monitoring data with continuous connection of time series markers, continuous and stable dissolved gas collection status in oil, and continuous and consistent thermal conductivity measurement status of the cover shell were retained.

[0053] In another embodiment, annual complete deformation cycle data screening parameters are set, with a continuous acquisition duration of no less than 180 days for a single valid data segment, fully covering the transformer's summer high-temperature expansion and contraction conditions and winter low-temperature contraction conditions, matching the operating cycle corresponding to the annual maximum deformation difference of the core metal laminations. The retained complete sets of data undergo time-series alignment verification, verifying that the time sequence identifier codes and acquisition time nodes of all online DGA monitoring data and top-level oil temperature time-series monitoring data completely overlap, and are then aggregated to form a standardized annual continuous time-series dataset.

[0054] After periodic screening, anomaly isolation, and time series verification, the complete set of annual time series data is directed to the time series operation unit inside the feature extraction module. It serves as the sole input carrier for time series correlation operations and feature parameter solving. All data entering the operation unit maintains the original time series arrangement order and data field structure.

[0055] Preferably, the feature extraction module is used to select continuous time-series data throughout the year, perform time-series operations on the abnormal gas production residual sequence and top oil temperature time-series data, and output phase shift features; construct a scatter plot using top oil temperature and gas production residual and solve for the hysteresis region area parameter;

[0056] Optionally, the feature extraction module performs time-series operations on the abnormal gas production residual sequence and the top oil temperature time-series data as follows:

[0057] The abnormal gas generation residual sequence corresponding to the core bonding heating and the top oil temperature time-series monitoring data generated by the heat conduction of the transformer box cover metal shell are synchronously imported into the time-series computing unit.

[0058] The timing calculation unit is set to cover the data offset traversal interval of the annual expansion and contraction process of the iron core. The data offset matching position is switched sequentially with an equal time step, and the correlation value between the abnormal gas production residual sequence and the top oil temperature timing monitoring data is calculated position by position.

[0059] The offset duration corresponding to the maximum value of the associated values ​​is selected as the phase offset feature.

[0060] In one embodiment, the timing operation unit is configured with a data offset traversal interval adapted to the annual expansion and contraction cycle of the transformer core. The overall traversal interval is set from 0 hours to 360 hours, covering the complete deformation lag time range of the core throughout the year during heating and expansion and cooling and contraction. The traversal process uses a uniform fixed time step size to perform bit-by-bit switching and matching, with a single step size uniformly set to 6 hours. The data offset matching position is changed sequentially according to the timing progression direction. Taking the standardized traversal operation as an example, in the 0-hour offset state, the abnormal gas generation residual sequence and the top oil temperature timing monitoring data are kept completely aligned in timing nodes; in the 6-hour offset state, the top oil temperature timing monitoring data is shifted backward by 6 hours to match the abnormal gas generation residual sequence of the preceding period; in the 12-hour offset state, the top oil temperature timing monitoring data is shifted backward by 12 hours. Different timing offset combinations are completed sequentially. Throughout the process, only the relative timing arrangement positions of the two types of data are changed, while the original data values, total data volume, and timing node numbers remain unchanged.

[0061] In another embodiment, after each set of offset matching positions is locked, the correlation numerical calculation between the abnormal gas production residual sequence and the top oil temperature time series monitoring data is carried out simultaneously. Corresponding correlation values ​​are generated for each offset point, and all correlation values ​​corresponding to each 6-hour step size within the interval from 0 hours to 360 hours are completely recorded, forming multiple sets of time series offset correlation parameter sets. The global traversal calculation process continuously executes a fixed process of step offset, matching alignment, and correlation calculation until the standardization calculation of all offset intervals is completed. After the global traversal calculation is completed, a numerical comparison and screening operation is carried out on the correlation values ​​corresponding to all offset durations. The maximum value in the correlation value set is extracted, and the time series offset duration corresponding to the maximum value is locked. This offset duration is fixed as the phase offset feature corresponding to the core deformation.

[0062] Optionally, the scatter plot construction in the feature extraction module is as follows:

[0063] Traverse each set of original gas production residuals and top oil temperature time-series monitoring data that are synchronously bound to time-series identifiers within a complete natural year, assign the top oil temperature time-series monitoring data to the horizontal axis coordinate, and assign the original gas production residuals to the vertical axis coordinate to generate independent coordinate points;

[0064] Based on the two deformation processes of the iron core metal laminations—thermal expansion and cooling contraction—corresponding classification marks are added to the coordinate points.

[0065] All marked coordinate points are partitioned and stored, and the partitioned coordinate points are connected to draw the boundary of the hysteresis region.

[0066] In one embodiment, each set of original gas production residual data and top-layer oil temperature time-series monitoring data with synchronized binding and unified time-series identifiers within a complete natural year is traversed sequentially. For each set of time-series paired data, a coordinate point generation operation is performed. The horizontal axis of the Cartesian coordinate system is uniformly assigned the specific temperature value of the current period's top-layer oil temperature time-series monitoring data, and the vertical axis is uniformly assigned the specific value of the original gas production residual data of the same time-series node. Each set of time-series paired data uniquely corresponds to a discrete coordinate point. Taking an annual monitoring time-series node as an example, when the oil temperature value of the time-series node steadily increases to 52℃, the corresponding original gas production residual value is 31μL / L, generating an independent coordinate point with horizontal coordinate 52 and vertical coordinate 31; when the oil temperature value of the time-series node drops back to 41℃, the corresponding original gas production residual value is 22μL / L, generating an independent coordinate point with horizontal coordinate 41 and vertical coordinate 22. All time-series nodes throughout the year complete the generation of all discrete coordinate points according to this fixed assignment rule.

[0067] Specifically, the top-layer oil temperature trend of consecutive time-series nodes is used as a fixed criterion to classify and label all discrete coordinate points. The top-layer oil temperature values ​​of two adjacent time-series nodes are collected. If the oil temperature value of the subsequent time-series node is higher than that of the preceding time-series node, it is determined that the period is in the process of thermal expansion of the core metal laminations, and all coordinate points generated by time-series nodes within this period are uniformly recorded as thermal expansion type. If the oil temperature value of the subsequent time-series node is lower than that of the preceding time-series node, it is determined that the period is in the process of cooling and contraction of the core metal laminations, and all coordinate points generated by time-series nodes within this period are uniformly recorded as cooling and contraction type. All point markings are determined solely based on the oil temperature's temporal rise and fall trend, without any other judgment conditions.

[0068] In one embodiment, all discrete coordinate points are partitioned and categorized for storage according to two labeling categories: thermal expansion and cooling contraction. Two independent data storage intervals are established, and the corresponding labeled coordinate points are completely aggregated into their respective intervals. Coordinate points of the same category are stored centrally, while coordinate points of different categories are independently separated. All horizontal axis oil temperature parameters, vertical axis residual parameters, and classification label parameters for all coordinate points are synchronously retained without parameter deletion or numerical correction. A connection drawing operation is performed on the discrete coordinate points in the two independent storage intervals. Following a chronological order, all coordinate points in the thermal expansion interval are sequentially connected, and simultaneously, all coordinate points in the cooling contraction interval are sequentially connected according to the same chronological order.

[0069] Optionally, the drawing of the hysteresis region boundary in the feature extraction module is specifically as follows:

[0070] According to the time sequence identifier, all coordinate points with added core expansion marks are connected in series from first to last, forming a continuous trajectory of the change in gas production value during the process of the core metal laminations opening due to heating;

[0071] According to the time sequence, all coordinate points with added core shrinkage marks are connected in series from first to last, forming a continuous trajectory of the change in gas production value during the cooling and bonding process of the core metal laminations;

[0072] Connect the coordinates of the first two continuous trajectories and the coordinates of the last two continuous trajectories to generate a closed contour figure, and solve for the area parameter of the closed contour figure.

[0073] In one embodiment, all discrete coordinate points marked with core expansion are retrieved from the partitioned storage. These points are then sorted sequentially from earliest to latest according to the annual time sequence identifier bound to each coordinate point. The time sequence identifier uses a fixed format of eight-bit date code plus six-bit time code. The points are rearranged based on the value of the code. For the sorted coordinate points marked with core expansion, a linear connection and concatenation operation is performed point by point, generating a unique straight line segment between adjacent coordinate points. All points are then connected sequentially to form a complete and continuous trajectory. This trajectory corresponds to the synchronous change trajectory of the top layer oil temperature time sequence monitoring data and the original gas production residual data throughout the entire process of the core metal laminations being heated and opened. The trajectory retains the horizontal and vertical axis numerical parameters of all original coordinate points.

[0074] All discrete coordinate points marked with core shrinkage in the partitioned storage are retrieved and arranged sequentially according to the time sequence identifier of each point, following the same ordering rules. A linear series connection operation is then performed on each of the sorted coordinate points marked with core shrinkage, generating straight line segments between adjacent points. After all points are connected, a complete and independent continuous trajectory is formed. This trajectory corresponds to the synchronous change trajectory of the top layer oil temperature time sequence monitoring data and the original gas generation residual data during the entire process of core metal lamination cooling and shrinkage. The arrangement order of the trajectory points perfectly matches the actual operating sequence of the transformer.

[0075] In another embodiment, the endpoint coordinate parameters of the thermal expansion trajectory and the cooling contraction trajectory are locked. The first and last time-series coordinate points of the thermal expansion trajectory are extracted, and the first and last time-series coordinate points of the cooling contraction trajectory are extracted simultaneously. The first and last time-series coordinate points of the two trajectories are connected by a straight line, and the last time-series coordinate points of the two trajectories are also connected by a straight line. The four trajectory segments are connected to each other to form a complete closed contour figure. All boundary nodes of the closed contour figure originate from the original discrete coordinate points.

[0076] It should be noted that the planar polygon shoelace algorithm is used to calculate the area of ​​the closed contour. All boundary node coordinates are entered sequentially according to the time-series traversal order. The horizontal axis is uniformly substituted with the top oil temperature value, and the vertical axis is uniformly substituted with the original gas production residual value. The product of the horizontal axis value of adjacent coordinate points and the vertical axis value of the next node is accumulated sequentially. Then, the product of the vertical axis value of adjacent coordinate points and the horizontal axis value of the next node is accumulated again. The absolute value of the difference between the two sets of accumulated results is calculated, and half of the absolute value of the difference is taken as the final area value of the closed contour. Taking two adjacent sets of boundary points as an example, when coordinates (45, 26) and (58, 33) are involved in the calculation, the corresponding product terms are included in the overall accumulation calculation system. All boundary points are cyclically collected and the difference is calculated, outputting a unique and fixed hysteresis region area parameter.

[0077] Optionally, the solution for the hysteresis region area parameter in the feature extraction module is as follows:

[0078] Using the time-series monitoring data range of top-layer oil temperature on the horizontal axis as the dividing benchmark, the closed contour graphic is uniformly divided into multiple sets of equal-width partition units.

[0079] Read the maximum and minimum values ​​of the original gas production residual corresponding to the horizontal axis interval of each group unit;

[0080] The difference between the maximum and minimum values ​​of the original gas production residual is used as the longitudinal span of the partition unit.

[0081] The planar values ​​of each group of zoning units are calculated and accumulated. The accumulated values ​​are then transmitted synchronously to the fault diagnosis module as the area parameter of the hysteresis region.

[0082] In one embodiment, a closed contour graphic consisting of oil temperature and gas production residual is locked. Based on the complete numerical range of the top-layer oil temperature time-series monitoring data corresponding to the horizontal axis of the graphic, an equal-width partitioning operation is performed. First, the extreme temperature parameters of the entire horizontal axis of the closed contour graphic are read to determine the minimum and maximum temperatures. A fixed partition width parameter of 2°C is set, and the closed contour graphic is continuously divided according to a fixed temperature span, resulting in several groups of independent partition units with equal horizontal axis widths.

[0083] Specifically, data reading operations are performed on each of the divided equal-width partition units one by one, retrieving the coordinate point data of all closed contour boundaries within the horizontal temperature range of each partition unit. The original gas production residual values ​​corresponding to all coordinate points within a single partition unit are extracted in batches. After aggregating all residual values ​​within a single range, the maximum and minimum original gas production residual values ​​corresponding to each group of partition units are filtered and output. Extreme value parameter locking is performed for each unit, and the temperature range parameters, maximum residual value parameters, and minimum residual value parameters for each group of partition units are retained. A vertical span calculation is performed on each group of partition units by subtracting the minimum original gas production residual value from the maximum value of the original gas production residual value corresponding to the single partition unit. The resulting value is used as the vertical span parameter of that partition unit. Combining the fixed 2℃ horizontal temperature span of the partition unit with the corresponding vertical span parameter, the planar area value of the single group of partition units is calculated. The area value of a single partition unit is obtained by multiplying the horizontal temperature span by the vertical residual span. Traverse all equal-width partitions, perform planar area calculations for all units sequentially, and aggregate the area data of all independent partitions.

[0084] In another embodiment, the planar area values ​​of all partition units are continuously accumulated, and the area values ​​of all partition units are summarized and merged in sequence. The total value obtained after summarization is defined as the hysteresis region area parameter. The horizontal axis temperature partition parameters, vertical residual span parameters, single unit area parameters, and total area limit parameters corresponding to this calculation are fully preserved. The parameter data maintains a correspondence with the top oil temperature time-series monitoring data and the original gas production residual data throughout the process. Finally, the integrated hysteresis region area parameter is transmitted to the fault diagnosis module.

[0085] Preferably, the fault identification module is used to determine faults based on the incremental carbon oxide components, phase shift characteristics, and hysteresis region area parameters; when the incremental carbon oxide components are the main growth components, the timing calculation outputs a non-zero phase shift; the closed hysteresis region formed by the scatter plot is determined as the core contact overheating fault region, and the power outage detection and maintenance period corresponding to the core thermal expansion state is output based on the phase shift characteristics.

[0086] Optionally, the fault determination module includes fault determination based on the incremental carbon oxide composition, phase shift characteristics, and hysteresis region area parameters, including:

[0087] Extract online DGA monitoring data from the initial time series node and the final time series node of the complete natural year, and read the corresponding recorded values ​​of carbon monoxide, carbon dioxide, and hydrocarbon gases within the online DGA monitoring data respectively;

[0088] The difference between the initial value and the final value of each gas in the year is calculated item by item, and the differences obtained from all gases are sorted by numerical value comparison.

[0089] Select the gas type corresponding to the first position in the sorting, verify whether the gas type belongs to carbon monoxide or carbon dioxide, and generate comprehensive verification conditions for core contact overheating faults.

[0090] In one embodiment, the start and end boundaries of the complete natural year time series dataset are locked, and the standardized and corrected online DGA monitoring data corresponding to the initial and end time series nodes of the year are extracted. Both types of time series node data contain five gas concentration parameters of carbon monoxide, carbon dioxide, methane, ethylene, and acetylene with a unified recording dimension. All data time series identifiers match the complete natural year collection cycle, and the data field format and collection measurement standards are kept uniform, serving as the basic parameter set for annual gas increment accounting.

[0091] For example, the gas concentration values ​​at the beginning and end of the year are retrieved separately, and the annual difference calculation for each gas type is performed item by item. Incremental calculations are completed using fixed rules, where the annual increment for a single gas type equals the gas concentration value at the end of the year minus the gas concentration value at the beginning of the year. Standardized calculations are performed using fixed instance parameters: the initial year node has a carbon monoxide concentration of 12 μL / L, a carbon dioxide concentration of 45 μL / L, and a total hydrocarbon concentration of 18 μL / L; the final year node has a carbon monoxide concentration of 28 μL / L, a carbon dioxide concentration of 72 μL / L, and a total hydrocarbon concentration of 22 μL / L. The calculated increments are 16 μL / L for carbon monoxide, 27 μL / L for carbon dioxide, and 4 μL / L for hydrocarbons. The corresponding increments for ethylene and acetylene are calculated simultaneously.

[0092] In another embodiment, the annual incremental values ​​calculated for all five gas categories are aggregated into a unified numerical set. All incremental values ​​within the set are then sorted in descending order, and a hierarchical arrangement is generated based on the absolute value of the incremental values, resulting in a fixed sorting sequence. The gas category corresponding to the first position in the sorting sequence is locked; this category represents the gas type with the largest concentration increase within the complete natural year. A category classification verification operation is performed on the gas category corresponding to the first position to verify whether it matches the categories of carbon monoxide and carbon dioxide. If the first gas category is carbon monoxide or carbon dioxide, a first-level comprehensive verification condition corresponding to the core contact overheating fault is generated; if the first gas category is any one of the hydrocarbon gases—methane, ethylene, or acetylene—this type of verification condition is not triggered.

[0093] Optionally, the fault determination module, based on the incremental carbon oxide composition, phase shift characteristics, and hysteresis region area parameters, may also include:

[0094] Simultaneously retrieve three sets of calculation output values: carbon oxide type determination result, phase shift characteristics, and hysteresis region area parameter;

[0095] The three-layer verification is completed sequentially. The first layer verifies whether the gas with the largest difference value is carbon monoxide or carbon dioxide. The second layer verifies that the phase shift characteristic corresponding to the duration is not zero. The third layer verifies that the value corresponding to the hysteresis region area parameter is greater than zero.

[0096] If all three layers of testing pass, it is determined that there is a multi-point contact overheating fault in the core metal laminations; if any layer of testing fails, it is determined that the insulating paper only undergoes normal thermal decomposition.

[0097] In one embodiment, the fault discrimination module synchronously collects three types of standardized parameter data output by the preposed feature calculation process, and uniformly incorporates them into the fault judgment and verification system. The collected data includes the determination result of carbon oxide type, the numerical value of the phase offset characteristic duration of iron core deformation, and the numerical value of the hysteresis region area parameter of oil temperature-gas production residual error. The determination result of carbon oxide type is generated by sorting and calculating the five types of gas increment differences of annual online DGA monitoring data, the numerical value of the phase offset characteristic duration is generated by traversing calculation with a fixed step of 6 hours in the interval from 0 hour to 360 hours of the time sequence calculation unit, and the numerical value of the hysteresis region area parameter is generated by cumulative accounting of the closed contour figure equal temperature partition unit.

[0098] In another embodiment, the fault discrimination module executes a three-level fixed parameter verification process, and each level of verification completes parameter comparison independently in sequence. The first level of verification checks the gas category with the largest annual increment value, locks the gas data ranked first in the increment sequence, and confirms that the gas category belongs to the category of carbon monoxide or carbon dioxide. The second level of verification checks the specific duration value of the phase offset feature, determines that the duration value is strictly not zero, and locks the existence of a time sequence lag offset feature between oil temperature change and gas production residual change. The third level of verification checks the specific numerical value of the hysteresis region area parameter, determines that the area parameter value is greater than zero, and locks the closed hysteresis change rule between the top-layer oil temperature and the original gas production residual error.

[0099] The three-level verification adopts an all-pass judgment mechanism, and completes the standardized judgment operation with a specific parameter verification example. When the gas with the maximum annual increment is carbon monoxide, the phase offset duration is 36 hours, and the hysteresis region area is 152 square units, all three levels of verification are up to the standard; if any one of the situations occurs that the gas with the maximum increment is hydrocarbon gas, the phase offset duration is 0 hour, and the hysteresis region area is 0 square unit, the corresponding single-level verification fails. When all parameter comparisons of the three-level verification are qualified, the judgment result of the multi-point contact overheating fault of the iron core metal lamination is generated; when any level of parameter comparison in the three-level verification is unqualified, the judgment result that only the normal thermal decomposition occurs in the insulating paper wrapped on the outer layer of the iron core is generated.

[0100] It is particularly important that the specific operation of the maintenance period output calculation in the fault discrimination module is as follows:

[0101] Obtain the association data table corresponding to the phase offset feature and the lamination gap, wherein the association data table internally stores the iron core lamination interlayer gap interval data corresponding to different phase offset values one by one;

[0102] Traverse all the lamination gap data in the association data table that matches the current fault phase offset feature, and screen the continuous time sequence interval where the interlayer gap value of the lamination is in the interval extreme state;

[0103] This continuous time interval is fixed as the detection period with the highest coupling degree of the core overheating defect signal. Based on this locked time period, standardized operation and maintenance timing instructions are generated, and the power outage maintenance execution period adapted to the special investigation of transformer core contact overheating defects is output.

[0104] In one embodiment, the fault diagnosis module retrieves a pre-calibrated core lamination parameter association table. This table contains a one-to-one correspondence between phase offset characteristic duration and core lamination interlayer gap. The table covers the phase offset value range from 0 hours to 360 hours, corresponding to a core lamination interlayer gap range from 0.1 mm to 3.0 mm. All parameters are generated based on the mechanical deformation calibration of the transformer core laminations. The table is internally divided into gradient levels, with each 12-hour phase offset value serving as a level, and each level independently corresponding to a set of lamination interlayer gap value ranges.

[0105] In another embodiment, the fixed value of the current device phase offset feature output by the fault determination process is retrieved. Using this value as a matching benchmark, the phase offset feature and lamination gap association data table is traversed across the entire domain to locate the lamination gap interval parameter that perfectly matches the current phase offset feature value. Simultaneously, all consecutive time-series nodes matching the gap interval state in the complete annual time-series chain are retrieved, and all corresponding time-series nodes are collected to form multiple sets of time-series interval data. The extreme value parameter of the lamination gap corresponding to each time-series interval is read one by one, and the maximum and minimum gap values ​​within each time-series interval are distinguished to lock the consecutive time-series intervals in the interval extreme value state.

[0106] The interval screening operation was completed using a standardized parameter matching method. If the current phase offset characteristic value is 48 hours, it corresponds to the 2.1mm to 2.3mm interlayer gap interval within the matching data table. After traversing the time series nodes throughout the year, a time series interval in which the interlayer gap value is stably maintained at the maximum value of 2.3mm was obtained. This time series interval was marked as the corresponding period of coupling between core mechanical deformation and overheating defect signals. All screening operations were strictly completed according to the data table range parameters and time series node record parameters, retaining only continuous time series segments where the gap value is in an extreme steady state.

[0107] The selected and locked continuous time intervals of extreme states are designated as the detection period for matching the coupling degree of transformer core overheating defect signals. Based on these locked continuous time intervals, the corresponding time start and end codes and duration parameters are extracted to generate standardized operation and maintenance timing instructions. These instructions are bound to the corresponding time interval parameters and the extreme value parameters of the core lamination gap, outputting a power outage maintenance execution period adapted to the special investigation of transformer core contact overheating defects. The output data fully utilizes the previous phase offset characteristic parameters, lamination gap correlation parameters, and time node coding system.

[0108] See Figure 4This is the interface for a multi-parameter joint fault analysis system for transformers. The top displays the transformer equipment number and corresponding substation location, and is divided into functional zones such as overview, DGA trend, and phase analysis. The page sections display top-level oil temperature, real-time concentrations of five types of gases, and phase offset values, along with annual gas concentration trend curves and compensated gas production residual curves. The left side of the interface lists multiple real-time alarm records, while the right side displays parameters such as core operating status, maintenance time window, hysteresis area, phase offset duration, and main gas components. The bottom of the page provides an entry point for generating a complete maintenance plan.

[0109] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, 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 the equivalents of the application are intended to be included within the invention.

[0110] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A multi-parameter joint fault analysis system for transformers, characterized in that, Includes the following modules: The data acquisition module is used to continuously collect online DGA monitoring data and top oil temperature time-series monitoring data of the transformer. The baseline compensation module is used to perform operating condition interference compensation based on online DGA monitoring data and top oil temperature time series data, calculate the dissolved gas concentration component corresponding to the temperature, calculate the basic gas production component under normal aging of the equipment, and generate a normalized abnormal gas production residual sequence. The feature extraction module is used to select continuous time-series data throughout the year, perform time-series operations on the abnormal gas production residual sequence and top oil temperature time-series data, and output phase shift features; A scatter plot was constructed using the top oil temperature and gas production residual, and the area parameter of the hysteresis region was solved. The fault identification module is used to determine faults based on the incremental carbon oxide composition, phase shift characteristics, and hysteresis region area parameters. When the incremental carbon oxide composition is the main growth component, the timing calculation outputs a non-zero phase shift. The closed hysteresis region formed by the scatter plot is identified as the core contact overheating fault region, and the power outage detection and maintenance period corresponding to the core thermal expansion state is output based on the phase shift characteristics.

2. The multi-parameter joint fault analysis system for transformers according to claim 1, characterized in that, The data acquisition module continuously collects online DGA monitoring data and top oil temperature time-series monitoring data for the transformer, including: Read the online DGA monitoring data generated by the oil intake pipeline of the transformer tank. The online DGA monitoring data includes independent concentration records of dissolved carbon monoxide, carbon dioxide, methane, ethylene and acetylene in the insulating oil. Read the time-series monitoring data of the top oil temperature generated by the heat conduction of the metal shell of the transformer tank cover. The reading action is performed synchronously according to the adaptively adjustable periodic acquisition interval. The oil tank's oil intake pipeline is connected to the circulating insulating oil inside the transformer tank. The insulating oil continuously wets the outer layer of the iron core covered with insulating paper, and receives the carbon oxides decomposed by the insulating paper when heated, which are then retained inside the oil to form a corresponding concentration record. The obtained online DGA monitoring data and top oil temperature time-series monitoring data are uniformly bound with a unique time sequence identifier. After binding, the data is transmitted to the baseline compensation module to perform operating condition interference compensation processing.

3. The multi-parameter joint fault analysis system for transformers according to claim 1, characterized in that, The dissolved gas concentration component corresponding to the calculated temperature in the baseline compensation module is as follows: Retrieve the top layer oil temperature time-series monitoring data generated by heat conduction from the metal shell of the transformer tank cover, and retrieve the preset oil-dissolved gas comparison data table matching mineral insulating oil; According to the quantitative conversion logic of Henry's Law, the amount of gas solubility migration caused by oil temperature change is calculated at each time node. The difference between the online DGA monitoring data generated by the oil tank pipeline and the amount of gas solubility migration is calculated line by line. The result of the difference calculation is used to replace the original online DGA monitoring data to correct the dissolved gas concentration component caused by the temperature expansion and contraction of the insulating oil.

4. The multi-parameter joint fault analysis system for transformers according to claim 3, characterized in that, The baseline compensation module calculates the basic gas production component during normal aging of the equipment as follows: Extract time-series monitoring data of top oil temperature generated by heat conduction from the metal casing of the transformer tank cover throughout the entire year; The baseline gas production value of the natural decomposition of the outer layer of the iron core insulation paper is quantitatively calculated based on the Arrhenius equation. The outer layer of the iron core insulation paper completely covers the surface of the iron core metal laminations, and the baseline gas production value corresponds to the corresponding value of the gas produced by the stable decomposition of the insulation paper under the condition that the iron core laminations are not attached and are heated. The difference between the online DGA monitoring data after solubility correction and the baseline gas production value is calculated at each time node to output the basic gas production quantity corresponding to the normal aging of the equipment corresponding to the core bonding heating.

5. The multi-parameter joint fault analysis system for transformers according to claim 1, characterized in that, The feature extraction module performs time-series operations on the abnormal gas production residual sequence and the top oil temperature time-series data as follows: The abnormal gas generation residual sequence corresponding to the core bonding heating and the top oil temperature time-series monitoring data generated by the heat conduction of the transformer box cover metal shell are synchronously imported into the time-series computing unit. The timing calculation unit is set to cover the data offset traversal interval of the annual expansion and contraction process of the iron core. The data offset matching position is switched sequentially with an equal time step, and the correlation value between the abnormal gas production residual sequence and the top oil temperature timing monitoring data is calculated position by position. The offset duration corresponding to the maximum value of the associated values ​​is selected as the phase offset feature.

6. The multi-parameter joint fault analysis system for transformers according to claim 5, characterized in that, The scatter plot construction in the feature extraction module is as follows: Traverse each set of original gas production residuals and top oil temperature time-series monitoring data that are synchronously bound to time-series identifiers within a complete natural year, assign the top oil temperature time-series monitoring data to the horizontal axis coordinate, and assign the original gas production residuals to the vertical axis coordinate to generate independent coordinate points; Based on the two deformation processes of the iron core metal laminations—thermal expansion and cooling contraction—corresponding classification marks are added to the coordinate points. All marked coordinate points are partitioned and stored, and the partitioned coordinate points are connected to draw the boundary of the hysteresis region.

7. The multi-parameter joint fault analysis system for transformers according to claim 6, characterized in that, The drawing of hysteresis region boundaries in the feature extraction module is specifically as follows: According to the time sequence identifier, all coordinate points with added core expansion marks are connected in series from first to last, forming a continuous trajectory of the change in gas production value during the process of the core metal laminations opening due to heating; According to the time sequence, all coordinate points with added core shrinkage marks are connected in series from first to last, forming a continuous trajectory of the change in gas production value during the cooling and bonding process of the core metal laminations; Connect the coordinates of the first two continuous trajectories and the coordinates of the last two continuous trajectories to generate a closed contour figure, and solve for the area parameter of the closed contour figure.

8. The multi-parameter joint fault analysis system for transformers according to claim 7, characterized in that, The solution for the hysteresis region area parameter in the feature extraction module is as follows: Using the time-series monitoring data range of top-layer oil temperature on the horizontal axis as the dividing benchmark, the closed contour graphic is uniformly divided into multiple sets of equal-width partition units. Read the maximum and minimum values ​​of the original gas production residual corresponding to the horizontal axis interval of each group unit; The difference between the maximum and minimum values ​​of the original gas production residual is used as the longitudinal span of the partition unit. The planar values ​​of each group of zoning units are calculated and accumulated. The accumulated values ​​are then transmitted synchronously to the fault diagnosis module as the area parameter of the hysteresis region.

9. The multi-parameter joint fault analysis system for transformers according to claim 1, characterized in that, The fault diagnosis module determines faults based on the incremental carbon oxide composition, phase shift characteristics, and hysteresis region area parameters, including: Extract online DGA monitoring data from the initial time series node and the final time series node of the complete natural year, and read the corresponding recorded values ​​of carbon monoxide, carbon dioxide, and hydrocarbon gases within the online DGA monitoring data respectively; The difference between the initial value and the final value of each gas in the year is calculated item by item, and the differences obtained from all gases are sorted by numerical value comparison. Select the gas type corresponding to the first position in the sorting, verify whether the gas type belongs to carbon monoxide or carbon dioxide, and generate comprehensive verification conditions for core contact overheating faults.

10. The multi-parameter joint fault analysis system for transformers according to claim 9, characterized in that, The fault diagnosis module, which determines faults based on incremental carbon oxide composition, phase shift characteristics, and hysteresis region area parameters, also includes: Simultaneously retrieve three sets of calculation output values: carbon oxide type determination result, phase shift characteristics, and hysteresis region area parameter; The three-layer verification is completed sequentially. The first layer verifies whether the gas with the largest difference value is carbon monoxide or carbon dioxide. The second layer verifies that the phase shift characteristic corresponding to the duration is not zero. The third layer verifies that the value corresponding to the hysteresis region area parameter is greater than zero. If all three layers of testing pass, it is determined that there is a multi-point contact overheating fault in the core metal laminations; if any layer of testing fails, it is determined that the insulating paper only undergoes normal thermal decomposition.

Images