Comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing

Abstract

The present invention discloses a comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing, which relates to the technical field of dry-type transformers. The multimodal sensing subsystem pre-deploys multimodal sensing equipment in the dry-type transformer to collect the operating status of the dry-type transformer to generate a set of eigenvalue vectors. The local structure analysis subsystem constructs a frequency-energy two-dimensional response spectrum to determine the triggering of the mechanical response mechanism and identify the loose hot zone. The disturbance analysis subsystem identifies the jump points where sudden changes occur based on the loose hot zone, and according to the positions of the jump points, screens out the jump points where the saturation zone is driven by structural disturbance and shifted, and obtains the drift area. The dissipative coupling subsystem combines the drift area with the loose hot zone to identify the coupled excitation area, and combines the eigenvalue vector set to evaluate the degree of influence of the energy consumption in the coupled excitation area on the total energy consumption in the dry-type transformer, thereby realizing the execution of the dynamic structural adjustment mechanism.

Description

Technical Field

[0001] The present invention relates to the technical field of dry-type transformers, and in particular to a comprehensive monitoring system for energy consumption of dry-type transformers based on multi-modal sensing. Background Art

[0002] As a key component of power distribution, dry-type transformers are widely used in urban distribution networks, data centers, rail transit, and other scenarios requiring high safety and stability. During operation, they are subject to the interplay of multiple physical factors, including structural vibration, electromagnetic disturbances, and thermal coupling. Loose internal structures or disrupted magnetic flux paths can easily lead to localized energy consumption anomalies, resulting in reduced overall efficiency and the accumulation of potential faults.

[0003] In the actual operation of dry-type transformers, core loosening is often caused by mechanical vibration or aging of the fixing clamps. It is common in urban power distribution rooms, photovoltaic grid-connected equipment, or industrial automation control systems. This structural degradation does not immediately affect the overall power supply function in the early stages, but it will trigger a series of nonlinear cross-anomalies:

[0004] On the one hand, the instability of the core will lead to the enhancement of local vibration energy release, resulting in structural vibration coupling energy consumption;

[0005] On the other hand, this type of disturbance will cause the originally symmetrically closed magnetic flux channels to reorganize and form new high-density magnetic concentration areas, which is the phenomenon of magnetic saturation zone migration;

[0006] This type of problem is long-term latent and multi-dimensionally disturbing, manifesting as structural displacement after the iron core is loaded and vibrated, the originally uniform and closed magnetic flux distribution is reorganized, and the magnetic density in the local area increases. Existing monitoring methods mostly rely on current, voltage and hot spot temperature parameters, which are difficult to accurately reflect the variation of magnetic flux path and local iron loss changes, and ignore the coupling identification of energy consumption anomalies caused by structural micro-changes. Summary of the Invention

[0007] In view of the above-mentioned problems existing in the prior art, the present application provides a comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing.

[0008] The embodiment of the present disclosure provides a comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing, including:

[0009] The multimodal sensing subsystem deploys multimodal sensing equipment in the dry-type transformer in advance to collect the operating status of the dry-type transformer and generate a set of eigenvalue vectors;

[0010] The local structural analysis subsystem constructs a two-dimensional frequency-energy response map to determine the triggering of the mechanical response mechanism and identify loosening hot spots;

[0011] The disturbance analysis subsystem identifies the jump points where sudden changes occur based on the loose hot zone. Based on the locations of the jump points, it selects the jump points where the saturation zone is shifted by the structural disturbance and obtains the drift area.

[0012] The dissipative coupling subsystem combines the drift area and the loose hot zone to identify the coupled excitation area. Combined with the eigenvalue vector set, it can evaluate the impact of the energy consumption in the coupled excitation area on the total energy consumption in the dry-type transformer and realize the execution of the dynamic structural adjustment mechanism.

[0013] Optionally, the multimodal sensing subsystem includes a deployment unit and a collection unit;

[0014] A deployment unit is used to deploy multimodal sensing equipment inside the dry-type transformer, including electromagnetic field density probes, three-dimensional vibration sensor arrays, and thermal imaging probes. A unified clock reference module is integrated into each modal sensing device to achieve timing alignment for data acquisition.

[0015] The acquisition unit is used to use multimodal sensing equipment to collect vibration acceleration signals, magnetic flux density, structural vibration energy density and magnetic flux disturbance energy density at various positions in real time, and generate a eigenvalue vector set after data preprocessing.

[0016] Optionally, the local structure analysis subsystem includes a map generation unit, a mechanism triggering unit, and an identification unit;

[0017] A spectrum generation unit collects vibration acceleration signals in real time using a three-dimensional vibration sensor array to form a time-domain vibration data stream, performs signal preprocessing on the time-domain vibration data stream, performs time-segment window processing on the preprocessed time-domain vibration data stream, applies a window function to each time segment for weighting, and performs a fast Fourier transform on the time-domain vibration data stream of each windowed time segment to obtain spectrum results of multiple time periods. The spectrum results of the multiple time periods are spliced ​​on the time axis to generate a frequency-energy two-dimensional response spectrum;

[0018] The mechanism trigger unit extracts the modal resonance peak under each frequency band condition from the frequency-energy two-dimensional response spectrum, compares it with the initial baseline modal frequency, calculates the modal shift ratio, and sets a continuous judgment window. If the modal shift ratio exceeds the preset shift threshold within the continuous judgment window, it indicates that the modal shift is abnormal, and the mechanical response mechanism is triggered.

[0019] The identification unit receives the mechanical response mechanism and uses the weighted sum of the vibration energy concentration and the modal shift ratio to construct the structural looseness index under different time segment conditions. It also stabilizes the judgment through time domain averaging to obtain the stabilization result, and compares the stabilization result with the preset looseness threshold. If the stabilization result exceeds the looseness threshold, the corresponding position will be regarded as a loose hot zone, and a spatial thermal map will be generated based on the loose hot zone.

[0020] Optionally, the disturbance analysis subsystem includes a zone unit, a jump unit, a disturbance unit, and a trajectory unit;

[0021] The zone unit is centered on the loose hot zone and a circle of grids is expanded outward as the boundary zone. The boundary zone and the loose hot zone are combined to generate the space-time field coverage area.

[0022] The jump unit constructs a local magnetic flux density space-time field within the space-time field coverage area and uses the slope mutation discrimination method to identify the instantaneous jump position. The specific contents are as follows:

[0023] According to the local magnetic flux density space-time field, a time series is extracted at each location to calculate the first-order time derivative of the magnetic flux density;

[0024] Extract the first-order time derivative of the maximum magnetic flux density from the time series. If the first-order time derivative of the maximum magnetic flux density exceeds the mean of the first-order time derivative of the maximum magnetic flux density at the corresponding position in the historical period, it is determined that a sudden jump occurs at the corresponding position and marked as a jump point.

[0025] According to the jump point, record the jump point information, including time, position and magnetic flux density.

[0026] Optionally, the disturbance unit presets a critical threshold value of magnetic flux density saturation. If the magnetic flux density at the jump point exceeds the critical threshold value of magnetic flux density saturation and is within the loosening hot zone, it is determined that the saturation zone is driven and offset by the structural disturbance; if the magnetic flux density at the jump point exceeds the critical threshold value of magnetic flux density saturation but is not within the loosening hot zone, it is determined that the generation of the saturation zone is a non-mechanical driving source.

[0027] The trajectory unit determines the migration path of the saturation area in space over time according to the jump point where the saturation area is driven by structural disturbance, and obtains the drift trajectory vector.

[0028] Optionally, the dissipative coupling subsystem includes a coupling analysis unit, a quantitative analysis unit, and a comparison and optimization unit;

[0029] The coupling analysis unit determines the drift region based on the drift trajectory vector and superimposes the drift region with the spatial thermal map to obtain the overlap. The overlap is used to reflect whether there is a high degree of consistency between the structural vibration excitation and the magnetic flux anomaly. Specifically: ,in, is the overlap, is the drift area, It is the part with loose hot zone in the spatial heat map. is the area of ​​spatial overlap, is the region of spatial union; if the overlap exceeds the preset overlap threshold, the region of spatial overlap is marked as a coupled excitation region.

[0030] Optionally, the quantitative analysis unit extracts features from the eigenvalue vector set to obtain the structural vibration energy density and magnetic flux disturbance energy density at different positions in the dry-type transformer. Combined with the coupling excitation zone, the energy consumption behavior caused by structural vibration excitation in the coupling excitation zone is analyzed. The structural vibration energy density and magnetic flux disturbance energy density in the coupling excitation zone are summed using a weighted summation algorithm to evaluate the coupling energy consumption per unit time in the coupling excitation zone of the current dry-type transformer. Specifically, ,in, is the coupling energy consumption per unit time in the coupling excitation region, is the coordinate point of the coupling excitation region, is the coupled excitation region, The structural vibration at the coordinate point The vibration energy density per unit time, is the magnetic flux disturbance at the coordinate point The magnetic energy dissipation density, and All are weight values;

[0031] Optionally, a comparison and optimization unit compares the coupling energy consumption per unit time of the coupling excitation zone with the total energy consumption per unit time of the dry-type transformer to calculate the energy consumption per unit time ratio. The energy consumption per unit time ratio is used to reflect the degree of influence of the energy consumption in the coupling excitation zone on the total energy consumption in the dry-type transformer. If the energy consumption per unit time ratio exceeds a predetermined ratio threshold, it is determined that the coupling excitation zone at the corresponding time point enters an abnormal dissipation critical state, and the coupling excitation zone is marked as a structural loosening-driven magnetic flux drift abnormal state, which is transmitted to the background operation platform to execute a structural adjustment mechanism for the coupling excitation zone. The structural adjustment mechanism includes suggestions for reinforcement of the core support points and local re-application of structural bonding materials.

[0032] Beneficial effects of the present invention:

[0033] (1) The local structural analysis subsystem generates a two-dimensional frequency-energy response spectrum, and combined with modal response offset detection, it can effectively identify the structural excitation characteristic area (i.e., loose hot zone) caused by core loosening, further improving the ability to identify the latent degradation behavior of the early mechanical structure. The perturbation analysis subsystem further identifies the magnetic flux density jump point based on the loose hot zone, and comprehensively determines whether the magnetic flux is in a saturated state and whether it is driven by structural perturbations, thereby accurately extracting the drift trajectory of the saturation zone and obtaining the drift area, realizing the causal relationship modeling between magnetic flux behavior and structural variation. Through the dissipative coupling subsystem, the structural vibration excitation area and the magnetic flux drift area are spatially overlapped and analyzed to construct a coupled excitation area. The energy ratio between the coupling energy consumption per unit time and the total energy consumption of the system in this area is evaluated in combination with the eigenvalue vector set. If the energy consumption ratio is too high, it can be determined that the area has entered an abnormal dissipative critical state, and a dynamic structural adjustment mechanism recommendation is triggered. Compared with traditional single-modal monitoring methods, this system can realize joint monitoring and linkage feedback judgment in three dimensions: structure, magnetic flux, and energy consumption. It has higher hidden danger identification sensitivity, stronger path tracing capability, and higher accuracy in locating abnormal energy consumption sources, thereby improving the energy efficiency management and reliability assurance level during the operation of dry-type transformers.

[0034] (2) The disturbance analysis subsystem realizes the causal coupling diagnosis capability between the magnetic flux behavior and structural state of the dry-type transformer by constructing a magnetic flux disturbance analysis mechanism centered on structural anomalies. By using the zone unit, a grid is extended outward based on the identified loose hot zone to form a space-time field coverage area, which effectively covers the magnetic flux distribution range that may be affected by structural disturbances and provides a precise spatial boundary for magnetic flux behavior analysis. The jump unit constructs a magnetic flux density space-time field based on the local magnetic flux density signal in the space-time field coverage area, extracts the time series of each position and calculates the first-order derivative. The slope mutation discrimination method is used to identify the maximum derivative point and determine it as the instantaneous magnetic flux jump position. This method can identify early magnetic flux dynamic abnormality signals. Combined with whether the jump point is in the loose hot zone, it can distinguish and judge the magnetic flux saturation behavior driven by structural disturbance from the magnetic flux disturbance caused by non-structural sources, thereby effectively judging whether the current magnetic flux saturation phenomenon is induced by the loose core structure, improving the accuracy of fault attribution. This mechanism can not only avoid false alarms but also enhance the analysis consistency of the structure-flux physical chain. The trajectory unit further extracts the spatial coordinates of the saturation region that change with time based on the time sequence of the jump points, and forms a drift trajectory vector for tracking the dynamic migration path of the saturation region.

[0035] (3) The present invention determines the degree of coupling between structural excitation and magnetic flux disturbance in physical space through quantitative calculation of overlap. This analysis process enhances the physical logic support for fault mechanism identification and can effectively screen out coupling excitation areas with highly consistent structure-magnetic flux behavior, providing a clear regional basis for subsequent energy consumption analysis. The quantitative analysis unit further extracts the structural vibration energy density and magnetic flux disturbance energy density of each coordinate point in the coupling excitation area from the eigenvalue vector set, and constructs a unit time coupling energy consumption index through weighted summation. This enables the system to evaluate the specific contribution of local coupling dissipation inside the dry-type transformer to the overall energy consumption based on multi-modal physical response, and can quantitatively determine the composite energy consumption process caused by the interaction between structural fatigue and magnetic flux saturation in a certain area. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application.

[0037] Figure 1 This is a block diagram of the comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing of the present invention. DETAILED DESCRIPTION

[0038] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0039] In photovoltaic inverter grid-connected scenarios, dry-type transformers often experience frequent current fluctuations, causing the core's magnetic flux density to briefly exceed the saturation threshold. This can induce abnormal eddy currents and increase energy consumption. Although surface temperature rise is minimal, this can still cause localized core fatigue or insulation degradation over long periods of time. Existing systems often rely on a single current or temperature signal and lack the ability to detect and identify this hidden energy consumption mechanism, saturation zone drift. Therefore, this paper proposes a comprehensive dry-type transformer energy consumption monitoring system based on multimodal sensing.

[0040] Example 1

[0041] like Figure 1 As shown, the embodiment of the present invention proposes a comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing, including:

[0042] The multimodal sensing subsystem deploys multimodal sensing equipment in the dry-type transformer in advance to collect the operating status of the dry-type transformer and generate a set of eigenvalue vectors;

[0043] The local structural analysis subsystem constructs a two-dimensional frequency-energy response map to determine the triggering of the mechanical response mechanism and identify loosening hot spots;

[0044] The disturbance analysis subsystem identifies the jump points where sudden changes occur based on the loose hot zone. Based on the locations of the jump points, it selects the jump points where the saturation zone is shifted by the structural disturbance and obtains the drift area.

[0045] The dissipative coupling subsystem combines the drift area and the loose hot zone to identify the coupled excitation area. Combined with the eigenvalue vector set, it can evaluate the impact of the energy consumption in the coupled excitation area on the total energy consumption in the dry-type transformer and realize the execution of the dynamic structural adjustment mechanism.

[0046] In this embodiment, by deploying multi-source heterogeneous sensing devices within dry-type transformers and combining structural and magnetic flux modal information, this system achieves the joint perception and analysis of microstructural changes, magnetic flux disturbances, and localized energy consumption anomalies during operation. Compared to traditional single-parameter monitoring methods, this system can proactively identify latent abnormal operating conditions, such as latent structural loosening, saturation zone shifts, and energy consumption drift, even in non-fault conditions. By analyzing the contribution of energy consumption, it determines whether these areas have a substantial impact on system efficiency, thereby triggering structural adjustment mechanisms such as support point reinforcement or localized re-application of adhesive, achieving the systematic optimization goals of early diagnosis, precise intervention, and delayed degradation.

[0047] For example, a dry-type transformer in a data center's power supply system operated under high load fluctuations. While traditional monitoring methods could provide temperature rise alarms, during a nighttime server cluster switchover, the system, using a three-dimensional vibration sensor array and magnetic flux density probes, detected a slight shift in the modal frequency of a certain region of the core, followed by a sudden hourly jump in the flux density slope. The disturbance analysis subsystem further identified that the flux value at the transition point in this region had reached critical saturation and was accompanied by spatial drift. The dissipative coupling subsystem superimposed this drift trajectory with the heat map and confirmed that this point was a coupling excitation region, with the coupling energy consumption per unit time increasing to 34.2% of the total system power fluctuation. The system immediately flagged this as an abnormal loose drive flux drift condition and provided structural adjustment recommendations to the operations and maintenance platform. While the equipment did not enter an alarm state during this process, the system proactively identified the abnormal energy consumption trend, avoiding potential insulation degradation and energy waste, further improving equipment stability and lifecycle safety margins.

[0048] Among them, when the short-term magnetic flux density exceeds the saturation point of the iron core, abnormal eddy currents will be generated, but there will be no obvious temperature rise on the surface. Therefore, the traditional method of using temperature rise alarm may lead to the phenomenon of delayed detection of abnormalities.

[0049] By building a multimodal dry-type transformer monitoring system that integrates perception, reasoning, and feedback, it is possible to dynamically identify the problem of magnetic flux saturation area drift, visualize the hidden losses of "no temperature rise on the surface" but "increased internal energy consumption", realize real-time dynamic risk assessment and adaptive optimization feedback, and intervene in advance to adjust control measures to prevent local overheating and insulation performance degradation.

[0050] Example 2

[0051] Please refer to Figure 1 ,Specifically: the multimodal perception subsystem includes a deployment unit and a collection unit;

[0052] A deployment unit is used to deploy multimodal sensing equipment inside the dry-type transformer, including electromagnetic field density probes, three-dimensional vibration sensor arrays, and thermal imaging probes. A unified clock reference module is integrated into each modal sensing device to achieve timing alignment for data acquisition.

[0053] This clock reference module is used to provide high-precision time synchronization signals for sensing devices such as electromagnetic field density probes and three-dimensional vibration sensor arrays. Through the unified clock reference provided by this module, each sensor can trigger data collection at a unified sampling time and generate data frames with global timestamps, ensuring that multimodal sensing information is comparable and fusible in the same time domain, laying a data foundation for subsequent modal collaboration analysis and energy consumption behavior modeling.

[0054] Electromagnetic field density probes will be installed at the intersection of multiple magnetic flux conduction paths outside the dry-type transformer core structure to collect local instantaneous magnetic flux density values ​​and their spatial gradient distribution characteristics. A three-dimensional vibration sensor array will deploy high-sensitivity acceleration sensors along vibration-sensitive areas such as the connection between the core frame and the fixings to collect multi-axial structural vibration response signals and extract indicators such as modal frequency, amplitude response, and energy distribution.

[0055] The acquisition unit is used to use multimodal sensing equipment to collect vibration acceleration signals, magnetic flux density, structural vibration energy density and magnetic flux disturbance energy density at various positions in real time, and generate a eigenvalue vector set after data preprocessing.

[0056] Data preprocessing includes noise removal, missing value filling, and data smoothing operations. The missing value filling methods include mean filling, median filling, interpolation filling, and regression filling.

[0057] The local structure analysis subsystem includes a graph generation unit, a mechanism triggering unit, and an identification unit;

[0058] A spectrum generation unit collects vibration acceleration signals in real time using a three-dimensional vibration sensor array to form a time-domain vibration data stream, performs signal preprocessing on the time-domain vibration data stream, performs time-segment window processing on the preprocessed time-domain vibration data stream, applies a window function (such as a Hamming window) to weight each time segment, and performs a fast Fourier transform on the time-domain vibration data stream of each windowed time segment (i.e., time-domain vibration data within a fixed-length time segment) to obtain multi-time period spectrum results. The spectrum results of the multiple time periods are spliced ​​on the time axis to generate a frequency-energy two-dimensional response spectrum;

[0059] Time segment window processing is to divide the pre-processed time domain vibration data stream into multiple fixed-length time segments (for example, 1 second), and apply a window function (such as a Hamming window) to each segment for weighting to reduce spectrum leakage. Fast Fourier transform is then performed on each windowed time segment signal (i.e., time domain vibration data within a fixed-length time segment) to obtain spectrum results for multiple time segments. The spectrum results of multiple time segments are then spliced ​​on the time axis to generate a three-dimensional frequency-time-energy array. Based on the three-dimensional frequency-time-energy array, a two-dimensional frequency-energy response spectrum is generated.

[0060] Signal preprocessing: Perform the following processing operations on the original vibration signal to remove noise and redundant information and improve the accuracy of spectrum calculation: remove the DC component; use sliding average or median filtering to remove low-frequency noise; and use a bandpass filter to limit the frequency band (such as 5–500 Hz) to match the core resonance frequency band.

[0061] The mechanism trigger unit extracts the modal resonance peak under each frequency band condition from the frequency-energy two-dimensional response spectrum, compares it with the initial baseline modal frequency, calculates the modal shift ratio, and sets a continuous judgment window. If the modal shift ratio exceeds the preset shift threshold within the continuous judgment window, it indicates that the modal shift is abnormal, and the mechanical response mechanism is triggered.

[0062] Frequency band refers to a series of frequency components obtained after FFT transformation within a time segment;

[0063] Extracting the modal resonance peak under each frequency band condition means extracting the frequency point with the maximum energy in the corresponding time period, which is the modal resonance peak of the time period;

[0064] The initial baseline modal frequency is the resonant modal frequency of the system corresponding to the corresponding sensor point or measurement area when there is no structural looseness. The steps to obtain it are:

[0065] 1. Select a time window when the equipment is in a structurally intact state after being put into operation or after maintenance;

[0066] 2. Enable multimodal sensing systems (especially 3D vibration sensors);

[0067] 3. Collect natural operating vibration signals for a period of time (without obvious impact or disturbance);

[0068] 4. Perform FFT analysis on the collected time-domain vibration data to extract the resonance peak frequency;

[0069] 5. Store the frequency peak sequence as a modal baseline library to obtain the baseline modal frequency;

[0070] The modal shift ratio is used to measure whether the current mode has a significant frequency drift compared to the initial mode. It is obtained using the following formula: ,in, is the modal shift ratio, is the modal resonance peak, is the initial baseline modal frequency;

[0071] The identification unit receives the mechanical response mechanism and uses the weighted sum of the vibration energy concentration and the modal shift ratio to construct the structural looseness index under different time segment conditions. It also stabilizes the judgment through time domain averaging to obtain the stabilization result, and compares the stabilization result with the preset looseness threshold. If the stabilization result exceeds the looseness threshold, the corresponding position will be regarded as a loose hot zone, and a spatial thermal map will be generated based on the loose hot zone.

[0072] Vibration energy concentration is the ratio of main frequency energy to full frequency band energy, which reflects the degree of energy focus in a certain frequency band. The specific method of obtaining it is: ,in, is the vibration energy concentration, f is the frequency number, It is the energy corresponding to the main peak frequency, indicating the frequency point with the largest energy on the frequency axis in the fast Fourier transform The corresponding magnitude square, is the sum of the energy of all frequencies in the current time segment, is the vibration energy density on the spectral component corresponding to the frequency f, if A significant increase, such as from 10% to 40%, indicates that the system vibration energy is being abnormally focused, which is common in the resonant coupling of loose parts.

[0073] The structural looseness index is a time-series quantitative index with spatial attributes. It reflects whether the corresponding location shows modal frequency shift and abnormal vibration energy accumulation at a certain spatial position and time period. The specific method of obtaining it is: structural looseness index = vibration energy concentration multiplied by the corresponding weight value + modal shift ratio multiplied by the corresponding weight value;

[0074] In actual operation, dry-type transformers often face disturbances caused by the frequent start-up and shutdown of external equipment. These disturbances are transmitted to the iron core through the infrastructure, causing its tiny structure to loosen, resulting in loose hot spots.

[0075] The spatial heat map will mark the location of each sensor in the three-dimensional vibration sensor array as a loose hot zone;

[0076] The time domain averaging method is used for stabilization judgment, that is, to calculate the average value of the structural looseness index under different frequency band conditions; the stabilization result is the average value of the structural looseness index at the corresponding position within the monitoring period.

[0077] The main purpose of the local structural analysis subsystem is to identify whether the core has modal frequency drift (i.e., modal response change) due to slight mechanical looseness, determine whether this drift causes vibration energy redistribution, and then construct a structural looseness index.

[0078] In this embodiment, the comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing proposed in the present invention has a good modular structure and timing coupling mechanism, and can collect, dynamically analyze, respond to and identify abnormal structural and magnetic flux behaviors under multiple physical modes from the source, and realize rapid judgment of abnormal energy consumption status and output of structural adjustment suggestions.

[0079] First, the multimodal sensing subsystem deploys electromagnetic field density probes, a three-dimensional vibration sensor array, a structural acoustics array, and thermal imaging probes to build a sensor network covering multiple physical dimensions. All sensors integrate a unified clock reference module, providing a high-precision synchronized clock signal for different modal acquisitions. This ensures consistent global timestamps for all modal data, ensuring temporal consistency and fusion of cross-modal information in subsequent analysis.

[0080] For example, when a part of the core becomes loose due to mechanical fatigue, that area will exhibit characteristics such as increased vibration, magnetic flux disturbance, and localized heating. Multimodal synchronous acquisition allows these weak signals to be observed simultaneously, providing a basis for subsequent abnormal behavior determination. Secondly, the local structure analysis subsystem uses a spectrum generation unit to construct a two-dimensional frequency-energy response spectrum from the three-dimensional vibration acceleration signal. Its core logic is as follows:

[0081] 1) Perform pre-processing operations such as DC removal and filtering on the original vibration signal;

[0082] 2) Segmenting by fixed length (e.g., 1 second) and windowing (Hamming window) to reduce spectral leakage;

[0083] 3) Perform a fast Fourier transform (FFT) on each data segment, construct a three-dimensional frequency-time-energy array, and project it to generate a two-dimensional spectrum. This spectrum serves as the key basis for the system to identify mechanical modal changes and can intuitively display the evolution process of the vibration modal resonance peak.

[0084] The mechanism trigger unit then calculates the modal offset ratio based on the modal frequency changes extracted from the spectrum. Continuously exceeding the offset threshold can trigger the mechanical response mechanism. The essence of this process is to convert frequency domain behavior into a structural response evidence chain, realizing the intelligent conversion from "data" to "events."

[0085] Furthermore, based on the trigger mechanism, the identification unit integrates the modal shift ratio and vibration energy concentration, constructing a structural looseness index through a weighted combination. It then uses a time-domain averaging strategy across multiple frequency bands and time windows to stabilize the output results, forming a structural looseness stabilization determination result. For example, if a sensor point consistently exhibits high energy concentration and modal shift across multiple frequency bands, its corresponding location will be identified as a loosening hotspot and annotated in the spatial heat map. The spatial heat map can reflect the state spatial distribution of each sensor in the multimodal sensor array, achieving a cross-dimensional projection from time-frequency characteristics to spatial abnormal hot spots, forming an intuitive visual representation of the structural state.

[0086] In summary, through the above-mentioned continuous and closed-loop processing flow of deployment acquisition, frequency domain analysis, modal response identification and stabilization evaluation, the present invention can effectively solve the problems existing in traditional dry-type transformer monitoring, such as single mode, time misalignment, and delayed early abnormality identification. The system can identify and warn in advance under various non-ideal operating conditions, and is particularly suitable for identifying energy consumption anomalies caused by minor loosening, resonance excitation, and magnetic flux drift, effectively extending equipment life and improving operational reliability and energy efficiency.

[0087] Example 3

[0088] Please refer to Figure 1 ,Specifically: the disturbance analysis subsystem includes the zone unit, jump unit, disturbance unit and trajectory unit;

[0089] The zone unit is centered on the loose hot zone and a circle of grids is expanded outward as the boundary zone. The boundary zone and the loose hot zone are combined to generate the space-time field coverage area.

[0090] The jump unit constructs a local magnetic flux density space-time field within the space-time field coverage area and uses the slope mutation discrimination method to identify the instantaneous jump position. Its purpose is to find the position and time when the magnetic flux density changes suddenly with time (i.e., the slope rises or falls sharply);

[0091] The specific contents are:

[0092] According to the local magnetic flux density space-time field, a time series is extracted at each location to calculate the first-order time derivative (slope) of the magnetic flux density;

[0093] Extract the first-order time derivative of the maximum magnetic flux density from the time series. If the first-order time derivative of the maximum magnetic flux density exceeds the mean of the first-order time derivative of the maximum magnetic flux density at the corresponding position in the historical period, it is determined that a sudden jump occurs at the corresponding position and marked as a jump point.

[0094] According to the jump point, record the jump point information, including time, position and magnetic flux density.

[0095] The disturbance unit has a preset magnetic flux density saturation critical threshold. If the magnetic flux density at the jump point exceeds the magnetic flux density saturation critical threshold and is within the loosening hot zone, it is determined that the saturation zone is offset by the structural disturbance. If the magnetic flux density at the jump point exceeds the magnetic flux density saturation critical threshold but is not within the loosening hot zone, it is determined that the saturation zone is generated by a non-mechanical driving source.

[0096] The displacement of the saturation zone driven by structural disturbance reflects that the originally uniform or relatively stable magnetic flux distribution inside the transformer core is caused by loose structure or slight deformation at a certain place, which leads to the change of geometry or magnetic resistance of the magnetic flux path, thereby causing the spatial migration of the position of the local magnetic flux density concentration point (i.e., the saturation zone).

[0097] When the structure is healthy, the position of the saturation zone is relatively stable. However, when the structure is disturbed, such as uneven core clamping force, loose fasteners, etc., the magnetic circuit will be nonlinearly redistributed, causing the saturation point originally located in a certain area to move to another place.

[0098] On the contrary, if the magnetic flux density at the trip point does not exceed the magnetic flux density saturation critical threshold, it means that the current trip point does not belong to the saturation region;

[0099] The trajectory unit determines the migration path of the saturation area in space over time according to the jump point where the saturation area is driven by structural disturbance, and obtains the drift trajectory vector.

[0100] The drift trajectory vector records the movement trajectory coverage area of ​​the saturation area in space over time, and the position coordinates of the saturation area at each moment;

[0101] The saturation region is where the magnetic flux density exceeds the critical threshold of magnetic flux density saturation;

[0102] In the dry-type transformer core, when the core material enters the magnetic saturation zone, the saturation zone will produce an increase in nonlinear eddy current dissipation, the magnetic flux distribution will be broken, and it will turn into uneven local focusing. The local eddy current will be significantly enhanced, resulting in hot spots and local energy consumption. Under long-term operation, it will cause serious problems such as thermal fatigue, insulation breakdown, and performance degradation.

[0103] The local magnetic flux density space-time field is a three-dimensional function relationship, which represents the magnetic flux density distribution evolving over time in a certain structural area (i.e., spatial position) (x, y). By placing magnetic flux density probes at various positions in the loose area of ​​the structure (transformer), the magnetic flux density values ​​of all positions are recorded at each sampling moment, and then reconstructed into a continuous magnetic flux density space-time field function through interpolation (space-time convolution or space-time interpolation algorithm can be used), finally forming the local magnetic flux density space-time field;

[0104] In this embodiment, the disturbance analysis subsystem proposed in the present invention realizes the spatiotemporal analysis modeling and structural coupling tracing judgment of the magnetic flux disturbance behavior in the dry-type transformer through the integrated zone unit, jump unit, disturbance unit and trajectory unit. The system can establish a high-sensitivity and precise tracking magnetic flux drift identification mechanism starting from the loose hot zone in a combination of spatial stratification and time evolution, effectively enhancing the perception and interpretation capabilities of local nonlinear energy consumption anomalies.

[0105] In terms of specific logic, the zone unit is centered around the identified loose hotspot, extending a circle of grids outward as the boundary zone. By combining the boundary zone with the loose hotspot, a spatiotemporal field coverage zone is constructed, defining the spatial boundaries for subsequent disturbance analysis. This not only focuses on key areas, but also reduces data redundancy and improves computational efficiency. For example, when the 3D vibration sensor array detects a clear structural loose hotspot at the sensor position in the upper right corner of the dry-type transformer core, the system automatically forms a boundary zone within the surrounding grids, centered around this area. These are then combined to form a coverage zone for magnetic flux disturbance analysis.

[0106] In the coverage area, the jump unit constructs a magnetic flux density time series curve based on each coordinate point, and identifies the points where the local magnetic flux changes dramatically by calculating its first-order time derivative (i.e., magnetic flux slope). It further extracts the maximum slope value from each time series and compares it with the average maximum value of the position in the historical period. If it exceeds the significance level, it can be determined that a sudden jump has occurred at that position, which is marked as a jump point, and its time, position, and magnetic flux density are recorded.

[0107] For example, at a certain sensor node (x3, y4) in the coverage area, the magnetic flux slope reaches 1.7 T / s at the current moment, while the historical maximum average slope is only 0.9 T / s. Based on this, the system automatically marks this point as a jump point, indicating that an abnormal magnetic flux disturbance has occurred at this location in the current cycle.

[0108] The perturbation unit compares the magnetic flux density at each jump point with a preset flux saturation threshold. If the value exceeds the threshold and is spatially located within the loosening hotspot, the saturation behavior is determined to be driven by structural perturbations. If the location is not within the loosening hotspot, the saturation source is considered to be non-mechanically driven. Conversely, if the magnetic flux density at the jump point does not reach the saturation threshold, it is not considered a saturation zone. For example, if the flux threshold is set to 1.5T and the magnetic flux value at the jump point is 1.8T and is located within the structural hotspot, it indicates that the formation of the magnetic flux saturation zone is caused by structural perturbations, and it can be further determined to be a structurally driven magnetic flux drift.

[0109] During continuous system operation, the trajectory unit tracks the spatial location and occurrence time of all identified structurally driven trip points and constructs a drift trajectory vector that evolves over time. This vector records the coordinate migration path of the magnetic flux saturation zone at each moment in a sequence, forming a continuous motion trajectory, which in turn maps the dynamic evolution of the saturation zone. For example, the system collects the distribution of trip points for 10 consecutive minutes to form a trajectory sequence (x3, y4) → (x4, y5) → (x5, y6). The system automatically constructs the corresponding drift trajectory vector, indicating that the magnetic flux saturation zone has shifted from the center of the transformer to the corners, warning of possible structural support instability.

[0110] In summary, the disturbance analysis subsystem further realizes the closed-loop identification of the causal chain between structural behavior and magnetic flux disturbances, can accurately locate complex magnetic flux disturbance phenomena to the source of structural loosening, and supports the reconstruction and visualization tracking of the evolution trajectory of the magnetic flux saturation zone; enhances the system's real-time recognition capability of hidden abnormal states, provides a decision-making basis for subsequent structural adjustments and heat loss compensation, and avoids the evolution of structural cumulative faults into sudden damage.

[0111] Example 4

[0112] Please refer to Figure 1 ,Specifically: the dissipative coupling subsystem includes a coupling analysis unit, a ,quantitative analysis unit, and a comparison and optimization unit;

[0113] The coupling analysis unit determines the drift area based on the drift trajectory vector and overlays the drift area with the spatial thermal map to obtain the overlap. The overlap is used to reflect whether there is a high degree of consistency between the structural vibration excitation and the magnetic flux anomaly, which is a possible source of energy consumption surge. Specifically: ,in, is the overlap, is the drift area, It is the part with loose hot zone in the spatial heat map. is the area of ​​spatial overlap, is the region of spatial union; if the overlap exceeds the preset overlap threshold, the region of spatial overlap is marked as a coupled excitation region.

[0114] The drift region is the area covered by the spatiotemporal trajectory formed by multiple jump points in time evolution; the drift trajectory vector is a set of spatial jump points with a time order, usually expressed as a vector sequence, and the drift region is the area swept by this set of vectors in space;

[0115] The quantitative analysis unit extracts features from the eigenvalue vector set to obtain the structural vibration energy density and magnetic flux disturbance energy density at different positions in the dry-type transformer. Combined with the coupling excitation zone, the energy consumption behavior caused by structural vibration excitation in the coupling excitation zone is analyzed. The structural vibration energy density and magnetic flux disturbance energy density in the coupling excitation zone are summed using a weighted summation algorithm to evaluate the coupling energy consumption per unit time in the current dry-type transformer coupling excitation zone. Specifically, ,in, is the coupling energy consumption per unit time in the coupling excitation region, is the coordinate point of the coupling excitation region, is the coupled excitation region, The structural vibration at the coordinate point The vibration energy density per unit time, is the magnetic flux disturbance at the coordinate point The magnetic energy dissipation density, and Both are weight values, used to adjust the contribution of the two to the coupling energy consumption;

[0116] Structural vibration energy density indicates the trend of local structural vibration energy consumption. Its monitoring and calculation methods are as follows:

[0117] By using a three-dimensional micro-accelerometer array installed on the surface of the transformer core, the vibration signal a(t) is collected in real time. The position of each sensor point is analyzed, and the structural vibration energy density can be estimated using the unit mass vibration energy formula, which is: ,in, is the equivalent mass (which can be approximated as unit mass or obtained through finite element model), is the velocity vector obtained by integrating the acceleration signal; , 、 、 is the acceleration signal on different axes;

[0118] Can be simplified to: ,in, and is the edge time point in each time period;

[0119] The flux disturbance energy density represents the heat loss caused by local flux disturbance. It is monitored and calculated as follows:

[0120] Use a magnetic flux density probe or Hall effect sensor array, deployed outside the core or in the core opening area, to collect the magnetic flux density and its rate of change in real time. The local eddy current loss caused by the magnetic flux disturbance can be approximated by the following expression to estimate the magnetic flux disturbance energy density: , is the rate of change of magnetic flux density, is the material conductivity (transformer core material parameter), D is the core thickness, It is a proportional symbol in mathematics; it is an expression that approximately estimates the eddy current energy consumption density per unit volume, reflecting that the more drastic the change in magnetic flux density per unit time, the stronger the conductivity of the material, and the thicker the conductive path, the greater the local electromagnetic dissipation (eddy current heat) generated.

[0121] Specifically, the coupling energy consumption per unit time in the coupling excitation zone is used to quantitatively evaluate the comprehensive energy consumption behavior caused by structural disturbances and magnetic flux anomalies in the coupling excitation zone, and to determine their impact on the system's operating efficiency. This is more practical than looking at magnetic flux or vibration alone, and is an energy causal determination method that integrates multi-modal physical fields.

[0122] In this embodiment, the dissipative coupling subsystem accurately identifies coupling excitation regions where magnetic flux anomalies are induced by structural disturbances by establishing a spatial overlap between drift regions and structural loosening hotspots, further enabling a quantitative assessment of coupling energy consumption. Specifically, the coupling analysis unit analyzes the spatial coverage area (i.e., the drift region) formed by the drift trajectory vectors and overlays it with the spatial heat map generated earlier by the vibration identification unit. This calculation calculates the spatial overlap between the two regions and further determines the likelihood of coupling excitation. For example, if structural vibration caused by core loosening in a dry-type transformer region is identified as a loosening hotspot in the spatial heat map, and the magnetic flux behavior analysis in this region identifies multiple time-evolving transition points, the drift trajectory formed by these transition points eventually overlaps the hotspot. At this point, the overlap between the two regions increases further, indicating the presence of a structural-flux coupling driving mechanism in this region, posing a potential risk of high energy consumption.

[0123] The system automatically marks the area as a coupled excitation zone. Based on this, the quantitative analysis unit extracts the multimodal feature vector set within the coupled excitation zone, obtaining the structural vibration energy density and magnetic flux disturbance energy density at each coordinate point. The system then uses weight parameters to adjust their contribution to total energy consumption. Finally, the coupled energy consumption per unit time is calculated through weighted summation. This indicator is more physically consistent and has greater significance for the root cause of faults than traditional methods that only monitor magnetic flux anomalies or vibration intensity. The logic of this step is to construct a coupled space using vibration and magnetic flux information, extract and attribute the energy consumption sources in the coupled space, and finally output a quantifiable coupled energy consumption indicator as a basis for judgment.

[0124] Compared to existing methods, this system can quickly determine whether a surge in energy consumption is caused by coupling effects during actual operation, avoiding misjudgments. It can also quantitatively assess the impact of a specific area on the overall transformer energy consumption, assisting in maintenance decisions such as structural reinforcement or operating condition adjustments. This method thus provides the ability to trace the linkage between structural disturbances and magnetic flux offsets, enhancing the causal clarity of fault identification. It also provides a dual quantitative basis for spatial location and energy consumption intensity, improving the scientific nature of maintenance priority assessments.

[0125] Example 5

[0126] Please refer to Figure 1 Specifically: the comparison and optimization unit compares the coupling energy consumption per unit time of the coupling excitation zone with the total energy consumption per unit time of the dry-type transformer to calculate the energy consumption per unit time ratio. The energy consumption per unit time ratio is used to reflect the degree of influence of the energy consumption in the coupling excitation zone on the total energy consumption in the dry-type transformer. If the energy consumption per unit time ratio exceeds the preset ratio threshold, it means that the coupling area has a significant impact on the system fluctuation, and the coupling excitation zone at the corresponding time point is judged to enter the abnormal dissipation critical state, and the coupling excitation zone is marked as a structural loose-driven magnetic flux drift abnormal state, which is transmitted to the background operation platform, and the structural adjustment mechanism is executed on the coupling excitation zone. The structural adjustment mechanism includes suggestions for reinforcement of the core support points and local re-application of structural bonding materials.

[0127] The total energy consumption per unit time of the dry-type transformer is obtained by installing a power sensor at the input end of the transformer (such as the low-voltage distribution side);

[0128] In this embodiment, the comparison and optimization unit, based on a quantitative comparison mechanism between the energy consumption per unit time of the coupling excitation zone and the total energy consumption per unit time of the dry-type transformer as a whole, enables real-time identification and dynamic control triggering of the coupling region's impact on system energy consumption anomalies. By establishing an energy consumption per unit time ratio indicator, this unit enables the system to clearly determine whether the combined energy consumption generated by structural vibration and magnetic flux disturbance within the coupling excitation zone has reached a critical threshold that substantially impacts the transformer system's operating efficiency. When the coupling energy consumption per unit time ratio exceeds a preset threshold, the system automatically determines the corresponding coupling excitation zone as being in a critical state of abnormal dissipation and further marks it as a state of abnormal magnetic flux drift driven by structural looseness. This information is then transmitted to the backend operating platform via a command interface, driving the precise activation of the structural adjustment mechanism.

[0129] Based on the identified abnormal position, the structural adjustment mechanism can automatically recommend solutions for core support point reinforcement or local repair of structural bonding materials at the corresponding location, thus forming a complete closed-loop path from perception → analysis → identification → response without relying on manual maintenance. This process is highly timely and intelligent, effectively reducing the long-term energy efficiency degradation caused by the failure to respond to structural micro-loosening in a timely manner, and improving the overall lifecycle management level of the equipment. Therefore, this comparison and optimization unit not only improves the dry-type transformer system's ability to analyze complex coupled abnormal energy consumption, but also implements closed-loop control logic for predictive maintenance of structural faults. It has significant beneficial effects such as strong intelligent identification, fast response mechanism, and high effectiveness of structural regulation. It is suitable for the safe operation and maintenance of dry-type transformers under high operating loads or critical application scenarios.

[0130] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. The comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing is characterized by: include, The multimodal sensing subsystem deploys multimodal sensing equipment in the dry-type transformer in advance to collect the operating status of the dry-type transformer and generate a set of eigenvalue vectors; The local structural analysis subsystem constructs a two-dimensional frequency-energy response map to determine the triggering of the mechanical response mechanism and identify loosening hot spots; The disturbance analysis subsystem identifies the jump points where sudden changes occur based on the loose hot zone. Based on the locations of the jump points, it selects the jump points where the saturation zone is shifted by the structural disturbance and obtains the drift area. The disturbance analysis subsystem includes zone unit, jump unit, disturbance unit and trajectory unit; The zone unit is centered on the loose hot zone and a circle of grids is expanded outward as the boundary zone. The boundary zone and the loose hot zone are combined to generate the space-time field coverage area. The jump unit constructs a local magnetic flux density space-time field within the space-time field coverage area and uses the slope mutation discrimination method to identify the instantaneous jump position. The specific contents are as follows: According to the local magnetic flux density space-time field, a time series is extracted at each location to calculate the first-order time derivative of the magnetic flux density; Extract the first-order time derivative of the maximum magnetic flux density from the time series. If the first-order time derivative of the maximum magnetic flux density exceeds the mean of the first-order time derivative of the maximum magnetic flux density at the corresponding position in the historical period, it is determined that a sudden jump occurs at the corresponding position and marked as a jump point. According to the jump point, record the jump point information, including time, position and magnetic flux density; The disturbance unit has a preset magnetic flux density saturation critical threshold. If the magnetic flux density at the jump point exceeds the magnetic flux density saturation critical threshold and is within the loosening hot zone, it is determined that the saturation zone is offset by the structural disturbance. If the magnetic flux density at the jump point exceeds the magnetic flux density saturation critical threshold but is not within the loosening hot zone, it is determined that the saturation zone is generated by a non-mechanical driving source. The trajectory unit determines the migration path of the saturation area in space over time according to the jump point where the saturation area is driven by structural disturbance, and obtains the drift trajectory vector; The dissipative coupling subsystem combines the drift area and the loose hot zone to identify the coupled excitation area. Combined with the eigenvalue vector set, it can evaluate the impact of the energy consumption in the coupled excitation area on the total energy consumption in the dry-type transformer and realize the execution of the dynamic structural adjustment mechanism.

2. The comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing according to claim 1 is characterized in that: The multimodal sensing subsystem includes a deployment unit and a collection unit; A deployment unit is used to deploy multimodal sensing equipment inside the dry-type transformer, including electromagnetic field density probes, three-dimensional vibration sensor arrays, and thermal imaging probes. A unified clock reference module is integrated into each modal sensing device to achieve timing alignment for data acquisition. The acquisition unit is used to use multimodal sensing equipment to collect vibration acceleration signals, magnetic flux density, structural vibration energy density and magnetic flux disturbance energy density at various positions in real time, and generate a eigenvalue vector set after data preprocessing.

3. The comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing according to claim 2 is characterized in that: The local structure analysis subsystem includes a graph generation unit, a mechanism triggering unit, and an identification unit; A spectrum generation unit collects vibration acceleration signals in real time using a three-dimensional vibration sensor array to form a time-domain vibration data stream, performs signal preprocessing on the time-domain vibration data stream, performs time-segment window processing on the preprocessed time-domain vibration data stream, applies a window function to each time segment for weighting, and performs a fast Fourier transform on the time-domain vibration data stream of each windowed time segment to obtain spectrum results of multiple time periods. The spectrum results of the multiple time periods are spliced ​​on the time axis to generate a frequency-energy two-dimensional response spectrum; The mechanism trigger unit extracts the modal resonance peak under each frequency band condition from the frequency-energy two-dimensional response spectrum, compares it with the initial baseline modal frequency, calculates the modal shift ratio, and sets a continuous judgment window. If the modal shift ratio exceeds the preset shift threshold within the continuous judgment window, it indicates that the modal shift is abnormal, and the mechanical response mechanism is triggered. The identification unit receives the mechanical response mechanism and uses the weighted sum of the vibration energy concentration and the modal shift ratio to construct the structural looseness index under different time segment conditions. It also stabilizes the judgment through time domain averaging to obtain the stabilization result, and compares the stabilization result with the preset looseness threshold. If the stabilization result exceeds the looseness threshold, the corresponding position will be regarded as a loose hot zone, and a spatial thermal map will be generated based on the loose hot zone.

4. The comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing according to claim 3 is characterized in that: The dissipative coupling subsystem includes a coupling analysis unit, a quantitative analysis unit, and a comparison and optimization unit; The coupling analysis unit determines the drift region based on the drift trajectory vector and superimposes the drift region with the spatial thermal map to obtain the overlap. The overlap is used to reflect whether there is a high degree of consistency between the structural vibration excitation and the magnetic flux anomaly. Specifically: ,in, is the overlap, is the drift area, It is the part with loose hot zone in the spatial heat map. is the area of ​​spatial overlap, is the region of spatial union; if the overlap exceeds the preset overlap threshold, the region of spatial overlap is marked as a coupled excitation region.

5. The comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing according to claim 4 is characterized in that: The quantitative analysis unit extracts features from the eigenvalue vector set to obtain the structural vibration energy density and magnetic flux disturbance energy density at different positions in the dry-type transformer. Combined with the coupling excitation zone, the energy consumption behavior caused by structural vibration excitation in the coupling excitation zone is analyzed. The structural vibration energy density and magnetic flux disturbance energy density in the coupling excitation zone are summed using a weighted summation algorithm to evaluate the coupling energy consumption per unit time in the current dry-type transformer coupling excitation zone. Specifically, ,in, is the coupling energy consumption per unit time in the coupling excitation region, is the coordinate point of the coupling excitation region, is the coupled excitation region, The structural vibration at the coordinate point The vibration energy density per unit time, is the magnetic flux disturbance at the coordinate point The magnetic energy dissipation density, and All are weight values.

6. The comprehensive monitoring system for energy consumption of dry-type transformers based on multimodal sensing according to claim 5 is characterized in that: The comparison and optimization unit compares the coupling energy consumption per unit time of the coupling excitation zone with the total energy consumption per unit time of the dry-type transformer to calculate the energy consumption per unit time ratio. The energy consumption per unit time ratio is used to reflect the degree of influence of the energy consumption in the coupling excitation zone on the total energy consumption in the dry-type transformer. If the energy consumption per unit time ratio exceeds the preset ratio threshold, it is determined that the coupling excitation zone at the corresponding time point enters the abnormal dissipation critical state, and the coupling excitation zone is marked as a structural loosening-driven magnetic flux drift abnormal state, which is transmitted to the background operation platform to execute the structural adjustment mechanism for the coupling excitation zone. The structural adjustment mechanism includes reinforcement of the core support points and local re-application of structural bonding materials.

Images