Fault prediction and health management method for box-type substation

Abstract

This invention relates to the field of fault prediction and health management technology, and in particular provides a method for fault prediction and health management of prefabricated substations. The method includes: generating an initial energy distribution map by aggregating data through an electrical signal acquisition network; removing background noise from the initial energy distribution map to obtain a purified energy distribution map; aligning and superimposing the thermal, vibrational, and electromagnetic distribution submaps according to their time series to identify abnormal correlation patterns in spatial location and temporal phase among the thermal, vibrational, and electromagnetic distribution submaps, forming an abnormal energy coupling feature map; and comparing the abnormal energy coupling feature map point-by-point with the energy coupling tolerance curve marked in the original design drawings of the prefabricated substation to calculate the deviation between the actual coupling values ​​of thermal, vibrational, and electromagnetic energies at each spatial point and the design tolerance values. This invention improves the accuracy and timeliness of fault early warning.

Description

Technical Field

[0001] This invention relates to the field of fault prediction and health management technology, and in particular to a method for fault prediction and health management of prefabricated substations. Background Technology

[0002] As a critical terminal distribution device in the power system, the operational reliability of prefabricated substations directly impacts power supply quality and user safety. Due to their compact internal space and high equipment integration, they are subjected to long-term load variations, environmental corrosion, and electromagnetic interference, resulting in a relatively high failure rate. Traditional operation and maintenance models mainly rely on single methods such as periodic inspections, infrared thermography, and partial discharge detection, which are insufficient for early and accurate fault warnings and health status assessments. With the development of smart grids and condition-based maintenance technologies, higher demands are placed on fault prediction and health management of prefabricated substations, urgently requiring technical methods that can integrate multiple physical quantities to achieve early anomaly identification and remaining life prediction.

[0003] Currently, fault monitoring and health management technologies for prefabricated substations mainly include the following categories:

[0004] Monitoring technologies based on a single physical quantity, such as monitoring temperature distribution with an infrared thermal imager, monitoring mechanical vibration with a vibration sensor, or monitoring partial discharge with an ultra-high frequency sensor, are usually designed for specific fault types and can effectively identify obvious anomalies that have already formed, but they are difficult to capture early, weak, complex fault signs.

[0005] Multi-sensor data fusion technology involves deploying various types of sensors inside the substation to collect signals such as temperature, vibration, and electromagnetic fields. These signals are then fused through a data layer or feature layer for comprehensive analysis. Existing methods often employ simple threshold alarms or statistical models, such as support vector machines and neural networks, for anomaly detection, lacking in-depth modeling of the coupling mechanisms of multiple physical fields.

[0006] Model-based prediction techniques use thermodynamic, structural dynamics, or electromagnetic field simulation models, combined with real-time data, to extrapolate the state of equipment. However, they require high model accuracy and computational resources, and are difficult to adapt to the complex and ever-changing operating conditions and environmental disturbances in actual operation.

[0007] Trend analysis techniques based on historical data extract equipment performance degradation trends by analyzing time series data, such as moving averages, exponential smoothing, and regression analysis. Existing methods often focus on extrapolating trends from single parameters, failing to effectively integrate the influence of multi-physics interactions on the degradation process.

[0008] Disadvantages of existing technology:

[0009] 1. The monitoring dimension is singular and it is difficult to cover the complex fault mechanism. Existing technologies usually monitor physical quantities such as heat, vibration and electromagnetic independently, and lack synchronous analysis and identification of the coupling relationship between multiple physical fields. However, the early faults of box-type substations often manifest as the coordinated abnormality of multiple physical quantities. Single-dimensional monitoring is prone to missing key signs, resulting in delayed early warning.

[0010] 2. The data fusion layer is shallow and the feature extraction is insufficient. Existing multi-sensor fusion methods mostly stay at the level of data superposition or simple feature splicing, and fail to conduct deep signal separation and correlation analysis based on the physical propagation characteristics of different energy forms, such as velocity and attenuation law. This results in high feature expression redundancy and incomplete extraction of coupling information, which affects the accuracy of subsequent state assessment.

[0011] 3. Anomaly identification relies on thresholds and lacks dynamic adaptability. Existing methods mostly use fixed or statistical thresholds for anomaly judgment, which are difficult to adapt to the dynamic changes of equipment under different loads, environments, and aging stages. Threshold settings that are too sensitive are prone to false alarms, while those that are too lenient may miss early anomalies, lacking a fine-grained characterization of the continuous evolution of equipment status.

[0012] 4. The prediction model is too simplified and does not take into account the multi-field coupling effect. Existing trend prediction methods are mostly based on linear or simple nonlinear extrapolation of a single parameter, and do not fully consider the comprehensive impact of the coupling of multiple physical fields such as heat, mechanics, and electromagnetic on the equipment degradation rate. As a result, the remaining life prediction results often deviate from the actual failure process and are difficult to support accurate preventive maintenance decisions.

[0013] 5. Lack of coordinated analysis of spatial distribution and temporal phase: Existing technologies for processing monitoring data mostly focus on independent analysis of the time or spatial dimensions, failing to simultaneously mine spatiotemporal correlation features. Early signs of faults often manifest as abnormal coupling patterns between specific spatial locations and specific temporal phases; spatiotemporal separation analysis methods are prone to losing crucial diagnostic information.

[0014] 6. The system is highly complex and its engineering applicability is limited. Although some methods based on precision simulation or complex machine learning have theoretical advantages, they have high requirements for data quality, computing resources and professional debugging, making it difficult to apply them stably on a large scale in scenarios such as prefabricated substations where the field environment is complex and the operation and maintenance resources are limited.

[0015] In summary, existing technologies have significant shortcomings in areas such as deep fusion of multiphysics fields, extraction of spatiotemporal correlation features, dynamic threshold adaptation, and modeling of coupling mechanisms, which hinder further improvements in fault prediction and health management of prefabricated substations. This invention aims to systematically address these problems through methods such as multilayer energy conversion films, multi-field coupling feature mapping, and dynamic trajectory prediction. Summary of the Invention

[0016] To achieve the above objectives, the present invention adopts the following technical solution:

[0017] One aspect of the present invention provides a method for fault prediction and health management of prefabricated substations, comprising the following steps:

[0018] A multi-layer energy conversion film is pre-laid on the inner surface of the prefabricated substation shell to convert the thermal radiation, mechanical vibration and electromagnetic fluctuations generated during the operation of the prefabricated substation into corresponding electrical signals in real time; after being collected by the electrical signal acquisition network, an initial energy distribution map is generated; the initial energy distribution map is then processed to remove background noise to obtain a purified energy distribution map;

[0019] Based on the different transmission speeds and attenuation characteristics of thermal, vibrational, and electromagnetic energies within a prefabricated substation, the purification energy distribution map is decomposed into independent thermal, vibrational, and electromagnetic distribution sub-maps. These sub-maps are then aligned according to time series and superimposed for comparison to identify abnormal correlation patterns in spatial location and temporal phase among them, forming an abnormal energy coupling characteristic map.

[0020] The abnormal energy coupling characteristic map is compared point by point with the energy coupling tolerance curve marked in the original design drawings of the prefabricated substation. The deviation between the actual coupling value of thermal, vibration and electromagnetic energy at each spatial point and the design tolerance value is calculated. Based on the magnitude of the deviation and its distribution density in the overall structure of the prefabricated substation, the current health level of the prefabricated substation is determined through a preset health status judgment. Based on the historical trend of the deviation, the time range in which key components may reach the failure threshold is calculated, forming a fault type prediction list and the remaining available time.

[0021] This invention utilizes a multi-layer energy conversion film to achieve real-time electrical signal conversion of thermal radiation, mechanical vibration, and electromagnetic fluctuations. After data acquisition and noise processing, a purified energy distribution map is generated. Based on the differences in energy transfer rates and attenuation characteristics, the map is decomposed into three independent sub-maps: thermal, vibrational, and electromagnetic. Through time series alignment and overlay comparison, abnormal correlation patterns in spatial location and temporal phase are identified, forming an abnormal energy coupling feature map. By comparing the abnormal energy coupling feature map point by point with the design tolerance curve, the deviation of the actual coupling value from the design tolerance is calculated. Combined with distribution density and historical trends, the current health level is assessed, and the failure time range of key components is predicted. This embodiment achieves multi-physical quantity fusion monitoring of the operating status of prefabricated substations. Through early identification and quantitative analysis of abnormal energy coupling, the accuracy and timeliness of fault early warning are improved, providing data support for preventive maintenance. Attached Figure Description

[0022] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0023] Figure 1 This is a flowchart of the method for fault prediction and health management of prefabricated substations provided in Embodiment 1 of the present invention;

[0024] Figure 2 This is a schematic diagram of the fault prediction and health management method for prefabricated substations provided in Embodiment 1 of the present invention.

[0025] Figure 3 This is a process diagram of generating the initial energy distribution map provided in Embodiment 2 of the present invention;

[0026] Figure 4 This is a process diagram of forming an anomalous energy coupling feature map provided in Embodiment 4 of the present invention;

[0027] Figure 5 This is a flowchart illustrating the process of forming a fault type prediction list and remaining available time, as provided in Embodiment 7 of the present invention.

[0028] Figure 6 A block diagram of the electronic device provided by the present invention;

[0029] Figure 7 A block diagram of a computer-readable storage medium provided for this invention. Detailed Implementation

[0030] The technical solutions of the present invention will now be described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0031] Hereinafter, the terms "first," "second," etc., are used for descriptive convenience only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0032] In this invention, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a fixed mechanical connection, a detachable mechanical connection, or an integral part; or, "connection" can be a direct connection or an indirect connection through an intermediate medium. Furthermore, unless otherwise explicitly specified and limited, the term "coupling" should be interpreted broadly. For example, "coupling" can be a direct electrical connection, such as physical contact and electrical conduction between two components; it can also be understood as an electrical connection between different components in a circuit structure through physical lines capable of transmitting electrical signals, such as copper foil or wires on a printed circuit board (PCB), to transmit electrical signals; or, "coupling" can be an indirect electrical connection between two components through an intermediate medium; or, "coupling" can be an electrical connection between two components in a non-contact manner, such as an electrical connection between two components using capacitive coupling to transmit electrical signals.

[0033] In this embodiment of the invention, directional terms such as "up," "down," "left," and "right" may be defined relative to the orientation of the components shown in the accompanying drawings. It should be understood that these directional terms can be relative concepts, used for relative description and clarification, and can change accordingly depending on the orientation of the components in the accompanying drawings.

[0034] Example 1: As Figure 1 As shown in the figure, this invention provides a method for fault prediction and health management of prefabricated substations, comprising the following steps:

[0035] Step S100: A multi-layer energy conversion film is pre-laid on the inner surface of the box-type substation shell to convert the thermal radiation, mechanical vibration and electromagnetic fluctuations generated during the operation of the box-type substation into corresponding electrical signals in real time; after being collected by the electrical signal acquisition network, an initial energy distribution map is generated; the initial energy distribution map is subjected to background noise stripping to obtain a purified energy distribution map;

[0036] Step S200: Based on the different transmission speeds and attenuation characteristics of thermal, vibrational, and electromagnetic energy within the prefabricated substation, the purification energy distribution map is decomposed into independent thermal, vibrational, and electromagnetic distribution sub-maps. The thermal, vibrational, and electromagnetic distribution sub-maps are aligned according to time series and superimposed for comparison to identify abnormal correlation patterns in spatial location and temporal phase among the thermal, vibrational, and electromagnetic distribution sub-maps, forming an abnormal energy coupling feature map.

[0037] Step S300: Compare the abnormal energy coupling characteristic map with the energy coupling tolerance curve marked in the original design drawings of the prefabricated substation point by point, and calculate the deviation between the actual coupling value of thermal, vibration and electromagnetic energy at each spatial point and the design tolerance value; based on the magnitude of the deviation and its distribution density in the overall structure of the prefabricated substation, determine the current health level of the prefabricated substation through the preset health status judgment, and based on the historical trend of the deviation, calculate the time range in which key components may reach the failure threshold, and form a fault type prediction list and remaining available time.

[0038] Among them, a prefabricated substation refers to a complete set of power distribution equipment, such as high-voltage switchgear, transformers, and low-voltage distribution devices, which are installed in a closed, movable box structure according to a certain wiring scheme. As the terminal node that directly distributes electrical energy to end users, its internal space is compact, the electromagnetic environment is complex, and it is subject to load changes and environmental erosion for a long time, making it a critical facility prone to failure. It is both the carrier for laying energy conversion films and the core object of fault prediction and health management. Multilayer energy conversion films refer to composite functional layers attached to the inner wall of the substation, which are composed of a thermoelectric material layer sensitive to thermal radiation, a piezoelectric material layer sensitive to mechanical vibration, and a magnetostrictive material layer sensitive to electromagnetic fluctuations. Through physical contact, it can capture the three sources of fault energy inside the prefabricated substation in situ: abnormal heating, vibration caused by mechanical loosening or deformation, and electromagnetic fluctuations caused by insulation deterioration. It converts macroscopic physical quantities into processable electrical signals, thus constructing the physical basis for fault feature extraction. Thermal radiation, mechanical vibration, and electromagnetic fluctuations refer to three inherent physical phenomena accompanying the operation of prefabricated substations, and are also key indicators reflecting the health status of the equipment. Thermal radiation mainly originates from conductor losses and ferromagnetic losses; mechanical vibration is mainly caused by magnetostriction of the iron core, electrodynamic stress, and loosening of fasteners; electromagnetic fluctuations are closely related to voltage and current changes, partial discharge, and ferroresonance. Synchronous monitoring aims to cover early signs of three typical faults: overheating, loosening, and discharge. The initial energy distribution map refers to a three-dimensional spatiotemporal data model generated by collecting data from multiple energy conversion films and aggregating it through a signal network. Using the internal spatial coordinates of the prefabricated substation as the planar dimension, thermal, vibration, and electromagnetic energy intensities as the numerical dimension, and time as the sequence dimension, it records the original comprehensive energy field distribution of the substation at a specific moment. It is the fundamental data source for all analysis and processing, containing information on the actual equipment status and background interference. Transmission speed and attenuation characteristics refer to the differentiated physical laws exhibited by heat, vibration, and electromagnetic waves as they propagate through specific media in a prefabricated substation, such as air, insulating oil, and solid structural components. For example, electromagnetic waves propagate at near the speed of light, vibration waves propagate more slowly in solids, and heat conduction is even slower, with attenuation proportional to the square of the distance. These inherent differences serve as the technical basis for separating the mixed comprehensive energy field into independent physical fields in signal processing. The thermal distribution sub-map, vibrational distribution sub-map, and electromagnetic distribution sub-map refer to three independent spatiotemporal distribution models obtained after separating the purified comprehensive energy spectrum based on the differences in transmission speed and attenuation characteristics of heat, vibration, and electromagnetic energy. The thermal distribution sub-map reflects the temperature rise gradient of various parts of the equipment; the vibrational distribution sub-map reflects the dynamic stress distribution of the mechanical structure; and the electromagnetic distribution sub-map reflects the distortion and disturbance distribution of the electric or magnetic field. These three sub-maps are the basis for identifying anomalies in single physical fields and anomalies in multi-field coupling.Anomaly correlation patterns in spatial location and temporal phase: These refer to the non-design-related correspondences between different physical fields at specific locations and time periods, identified through overlay analysis of the thermal, vibrational, and electromagnetic state distribution sub-maps after aligning time series. For example, a fastener or spatial location may simultaneously experience a peak vibration and a sudden temperature rise during peak load or temporal phase, with a high correlation between their trends. Correlation patterns reveal complex fault mechanisms that cannot be detected by single physical field monitoring and are the core basis for forming anomaly energy coupling characteristic maps.

[0039] The thermoelectric material layer refers to a thin-film functional layer made by sintering semiconductor thermoelectric compounds. Its structure consists of a thermoelectric stack array formed by alternating p-type and n-type thermoelectric units on a flexible substrate and connected in series. The working principle is based on the Seebeck effect. When there is a temperature gradient along the thickness direction in the layer, the charge carriers in the thermoelectric units diffuse from the hot end to the cold end, forming a potential difference in the material. This results in a DC voltage signal that is proportional to the temperature gradient being output on the electrodes at both ends of the layer, enabling in-situ sensing of the intensity and distribution of thermal radiation inside the box-type substation. The piezoelectric material layer refers to a flexible piezoelectric film made by casting and polarization treatment of lead zirconate titanate piezoelectric ceramic powder and polymer composite. Its structure includes a continuous phase formed by oriented piezoelectric fibers or piezoelectric particles embedded in the polymer matrix. The working principle is the positive piezoelectric effect. When the layer is subjected to dynamic stress generated by mechanical vibration, the internal cell of the piezoelectric phase deforms, causing the positive and negative charge centers to separate. The charge or voltage signal related to the strain and strain rate is induced on the electrodes on the upper and lower surfaces of the layer, and the mechanical vibration waveform is converted into the corresponding electrical signal output in real time. Magnetostrictive material layers refer to amorphous or nanocrystalline thin strips made from iron-gallium alloys or rare-earth iron-based alloys through vacuum melting and rapid solidification. Their structure exhibits continuous magnetic thin film layers with magnetic anisotropy. The working principle is based on the coupling mechanism of the magnetostrictive and inverse magnetostrictive effects. When the layer is placed in an alternating magnetic field generated by electromagnetic fluctuations, the magnetic domains inside the material rotate and shift, causing periodic changes in the material's dimensions. Simultaneously, this alternating stress induces changes in the magnetic flux density within the layer through the inverse effect. This induces an electromotive force related to the magnetic field strength and rate of change on the induction coils arranged around the layer or on the electrodes at both ends of the layer, enabling real-time capture of electromagnetic fluctuation characteristics. The three layers are stacked in the following order: the thermoelectric conversion layer is tightly attached to the shell surface, the vibration-electric conversion layer is in the middle, and the magnetoelectric conversion layer is on the innermost side. The layers are isolated by an insulating adhesive layer to prevent electrical signal crosstalk, collectively forming a composite sensing structure capable of synchronously responding to three physical energies.

[0040] In the above embodiments, this embodiment achieves real-time electrical signal conversion of thermal radiation, mechanical vibration, and electromagnetic fluctuations through multi-layer energy conversion films. After acquisition and noise processing, a purified energy distribution map is generated. Based on the differences in energy transfer speed and attenuation characteristics, the map is decomposed into three independent sub-maps: thermal, vibrational, and electromagnetic. Through time series alignment and superposition comparison, abnormal correlation patterns in spatial location and temporal phase are identified, forming an abnormal energy coupling feature map. By comparing the abnormal energy coupling feature map with the design tolerance curve point by point, the deviation of the actual coupling value from the design tolerance is calculated. Combined with the distribution density and historical trend, the current health level is assessed and the failure time range of key components is predicted. This embodiment realizes multi-physical quantity fusion monitoring of the operating status of prefabricated substations. Through early identification and quantitative analysis of abnormal energy coupling, the accuracy and timeliness of fault early warning are improved, providing data support for preventive maintenance.

[0041] Example 2: As Figure 3 As shown, based on Example 1, the process of generating the initial energy distribution map in step S100 of this embodiment of the invention specifically includes the following steps:

[0042] Step S101: The electrical signals collected by the acquisition network are categorized and reorganized according to the physical properties of the composite functional layer in the multilayer energy conversion film from which each electrical signal originates and its fixed coding identifier in the acquisition channel, forming thermal response signal group, vibration response signal group and electromagnetic response signal group respectively.

[0043] Step S102: Assign each discrete electrical signal value in the thermal response signal group to the internal spatial coordinates of the shell of the source composite functional layer, and fit the continuous distribution surface of the thermal radiation intensity on the shell surface at the sampling time of the discrete electrical signal through surface reconstruction; at the same time, construct the continuous distribution surface of vibration intensity and the continuous distribution surface of electromagnetic wave intensity using the vibration response signal group and the electromagnetic response signal group respectively.

[0044] Step S103: Overlay the continuous distribution surfaces of thermal radiation, vibration and electromagnetic fluctuation generated at the same discrete electrical signal sampling time according to the same spatial coordinate reference to form an instantaneous comprehensive energy distribution snapshot at the sampling time; record and store the instantaneous comprehensive energy distribution snapshots of all consecutive sampling times in chronological order to construct an initial energy distribution map with spatial location as the planar dimension, energy type as the hierarchical dimension, and time series as the depth dimension.

[0045] In the above embodiments, this embodiment classifies and reorganizes electrical signals based on their physical properties and coded identifiers to form thermal response, vibration response, and electromagnetic response signal groups; it achieves the classification and extraction of multiple types of energy signals; it combines thermal response signal values ​​with the spatial coordinates of the source location, and fits a continuous distribution surface of thermal radiation intensity on the shell surface through surface reconstruction; simultaneously, it constructs continuous distribution surfaces of vibration intensity and electromagnetic wave intensity based on the vibration and electromagnetic signal groups respectively; it transforms discrete signals into a continuous spatial distribution model, improving the accuracy and continuity of the energy field description. The three intensity distribution surfaces at the same moment are superimposed according to a unified spatial coordinate reference to form an instantaneous comprehensive energy distribution snapshot, which is recorded and stored in a time series, constructing an initial energy distribution map containing spatial location, energy type, and time dimension; it realizes the fusion expression of multi-physics energy in the spatiotemporal domain, providing a structured and traceable data foundation for energy state analysis and optimization.

[0046] Example 3: Based on Example 2, the process of fitting the continuous distribution surface of the thermal radiation intensity on the shell surface at the sampling time of the discrete electrical signal through surface reconstruction in step S102 of this embodiment of the invention specifically includes the following steps:

[0047] Step S1021: The value of each discrete electrical signal in the thermal response signal group is corrected to the thermal radiation intensity value at the corresponding spatial coordinate point according to the calibration conversion curve of the thermoelectric material in its composite functional layer, and bound to the internal spatial coordinate of the shell at the spatial coordinate point to form a set of heat dissipation intensity data tuples with coordinate identifiers.

[0048] Step S1022: Based on the thermal conductivity coefficient of the shell material and the shell geometry, calculate the thermal influence weight distribution of each heat dissipation intensity data element on any position on the shell surface, and map the heat intensity value of each heat dissipation intensity data element to the entire shell surface according to the corresponding weight distribution to generate multiple single-source thermal radiation distribution fields.

[0049] Step S1023: Superimpose the thermal radiation distribution fields contributed by all single sources under the same spatial reference, and perform an equilibrium correction based on energy conservation on the energy accumulation caused by multi-source contributions in the superimposed field, and finally obtain the continuous distribution surface of thermal radiation intensity on the shell surface at the sampling time.

[0050] In step S1021, each discrete electrical signal value in the thermal response signal group is corrected to a thermal radiation intensity value. The calculation formula is as follows:

[0051]

[0052] in, Indicates the first The electrical signal value output by the thermoelectric material layer corresponding to each sampling point is expressed in volts. The calibration conversion factor of the thermoelectric material layer is expressed in watts per square meter per volt, and is determined by a joint calibration experiment on the Seebeck coefficient and thermal radiation response of the material before it leaves the factory. To compensate for the intercept, the unit is watts per square meter, used to eliminate zero-point errors caused by ambient temperature deviations; the formula accurately maps the original electrical signal to the thermal radiation intensity value at the sampling point, ensuring the physical accuracy of the spatial distribution reconstruction;

[0053] In step S1022, the weight distribution of the thermal influence of each heat dissipation intensity data tuple on any location on the shell surface is calculated, and the calculation formula for the thermal radiation distribution field contributed by a single source is as follows:

[0054]

[0055] In the formula, This represents the spatial coordinate vector of any point on the surface of the shell. For the first The coordinates of each sampling point; represents the thermal conductivity of the shell material, measured in watts per meter Kelvin; The characteristic attenuation length, in meters, is determined by the shell geometry and is obtained through finite element thermal simulation fitting. The Euclidean distance between the two points is given by the formula, which is based on the Green's function of the heat conduction point source and introduces an exponential decay term to describe the physical law of heat diffusion on the surface of a finite-sized shell. Its function is to distribute the heat radiation intensity of the discrete point source to the entire surface according to the actual heat conduction path, thereby generating a single source contribution field that conforms to the thermodynamic principle.

[0056] In step S1023, the contribution fields of all single sources are superimposed and energy conservation corrections are applied to obtain the calculation formula for the continuous distribution surface:

[0057]

[0058]

[0059]

[0060]

[0061] in, This represents the total number of sampling points; For the first The area of ​​each sampling point, represented by a micro-element, is in square meters and is determined by the sensor layout density. Represents the entire surface of the shell; The area is a infinitesimal element; the preliminary total field is first obtained by superposition. Then calculate the total heat power of all sampling points. With the initial total field integral power Finally, the ratio of the two is used as a global correction factor and multiplied onto the initial field to obtain the final continuous distribution surface. The purpose of this set of formulas is to ensure that the reconstructed thermal radiation distribution not only conforms to the physical diffusion law in terms of spatial morphology, but also satisfies the law of conservation in terms of overall energy, avoiding the false increase or decrease of energy caused by the superposition of multiple sources, thereby obtaining an accurate and reliable continuous thermal radiation surface.

[0062] In the above embodiments, this embodiment achieves high-precision reconstruction from finite discrete points to a continuous surface. The entire process integrates sensor calibration data, material thermal property parameters, and geometric topology information, and introduces physical constraints into the mathematical interpolation, so that the generated continuous distribution surface not only conforms to the statistical characteristics of the measurement data but also satisfies the physical laws of heat conduction; it significantly improves the equivalence of the spatial resolution of the thermal radiation field, and can more realistically reflect the thermal state distribution of the shell surface, providing a reliable full-field data foundation for thermal analysis, fault diagnosis, or performance evaluation.

[0063] Example 4: Figure 4 As shown, based on Example 1, the process of forming an abnormal energy coupling feature map in step S200 of this embodiment of the invention specifically includes the following steps:

[0064] Step S201: For the thermal distribution sub-map, vibrational distribution sub-map, and electromagnetic distribution sub-map that have been aligned with the time series, extract the spatial distribution gradient and curvature change of energy intensity in each sub-map, locate the spatial point set where the energy intensity shows abrupt changes or extreme values ​​in the local region, and label each point with its corresponding time phase label, thereby generating the thermal singular feature point set, vibrational singular feature point set, and electromagnetic singular feature point set respectively;

[0065] Step S202: Match the positions of the three singular feature point sets in a unified three-dimensional spatial coordinate system. Using the spatial coordinates of each singular feature point as a reference, search for feature points in the other two singular feature point sets whose spatial coordinate difference is less than the preset neighborhood radius. Combine all singular feature points that meet the spatial proximity condition to form multiple spatially overlapping feature groups composed of thermal, vibrational, and electromagnetic singular points. Record the spatial coordinates of the overlapping position and the original time phase label corresponding to each point in each spatially overlapping feature group.

[0066] Step S203: For each spatially overlapping feature group, extract the waveform curves of the three energy intensities changing with time at the location from the time series of the thermal distribution submap, vibrational distribution submap, and electromagnetic distribution submap; by comparing the fluctuation shape of the waveform curves and the order of peak appearance, select feature groups with highly similar waveform shapes and a fixed leading or lagging relationship in time phase; mark the selected feature groups as anomalous energy coupling regions, and integrate the spatial range, participating energy type, and waveform phase relationship of all anomalous regions into an anomalous energy coupling feature map.

[0067] In the above embodiments, this embodiment locates local energy abrupt changes or extreme points by extracting the spatial gradient and curvature changes of energy intensity from the thermal, vibrational, and electromagnetic state distribution sub-maps, and labels them with time phase tags. This achieves data reduction from continuous distribution fields to key feature points while preserving the spatial anomaly locations and their dynamic temporal information, providing an accurate spatiotemporal reference point set for multiphysics correlation analysis. Secondly, the three singular feature point sets are spatially matched in a unified three-dimensional coordinate system. By searching for spatially neighboring points using a preset neighborhood radius, a spatially overlapping feature group composed of thermal, vibrational, and electromagnetic singularities is formed. Spatial correlation rules for cross-physics features are established, effectively identifying the co-occurrence regions of different energy forms in space, avoiding mismatches caused by sensor placement errors or coordinate system deviations, and improving the spatial positioning accuracy of multiphysics coupling events. Next, for each spatially overlapping feature group, waveform curves showing the time-varying effects of the three energy intensities at the corresponding locations are extracted. By comparing the similarity of waveform morphology and the temporal phase relationship of peak occurrence, feature groups with fixed leading or lagging relationships are selected. A correlation discrimination in the time dimension is introduced, which can distinguish between random spatial co-occurrence events and real anomalous events with causal or coupling relationships. The joint criterion of waveform morphology and phase relationship enhances the robustness of feature selection and reduces misjudgments caused by noise or irrelevant fluctuations. Finally, the selected feature groups are marked as anomalous energy coupling regions, and their spatial range, energy type, and waveform phase relationship are integrated to form a structured anomalous energy coupling feature map. This map integrates the spatial distribution of anomalous events, the types of physical fields involved, and the temporal characteristics of dynamic interactions, realizing a standardized representation of multi-physics field anomalous coupling modes. It provides interpretable and traceable multi-dimensional feature evidence for system status monitoring, fault early warning, and mechanism analysis, supports anomaly classification and root cause tracing based on coupling modes, and improves the precision of health management of complex systems.

[0068] Example 5: Based on Example 4, the process of selecting feature groups with highly similar waveforms and fixed leading or lagging time phases in step S203 of this embodiment of the invention specifically includes the following steps:

[0069] Step S2031: For each spatially overlapping feature group, the waveform curves of the three energy intensities (thermal, vibrational, and electromagnetic) that are extracted over time are symbolically encoded to represent their fluctuation patterns. Based on the rising and falling trends and inflection point distribution of the waveform curves in each sampling period, the symbolic encoding converts the continuous waveform curve shape into a symbol sequence composed of several rising segments, falling segments, and stable segments arranged in chronological order, generating a unique morphological feature identifier for each energy waveform.

[0070] Step S2032: Compare the morphological feature identifiers of the three energy waveforms within the same spatial overlap feature group, and select spatial overlap feature groups with completely identical symbol sequences; for the three waveform curves within the spatial overlap feature group, take the time of the energy peak as the marker point, measure and record the time offset of the peak time of the thermal waveform relative to the peak time of the vibrating waveform, and the time offset of the peak time of the electromagnetic waveform relative to the peak time of the thermal waveform, to form the phase offset vector of the spatial overlap feature group;

[0071] Step S2033: Based on the sign and absolute value of each offset in the phase offset vector, identify spatially overlapping feature groups whose offsets maintain a constant sign throughout the entire time series and whose absolute value fluctuation range is less than a preset stable threshold; determine the spatially overlapping feature groups as target feature groups that meet the requirements of highly similar waveform morphology and a fixed leading or lagging relationship in time phase, and separate them from all spatially overlapping feature groups as effective inputs for integration into an abnormal energy coupling feature map.

[0072] In the above embodiments, this embodiment achieves accurate identification of target feature groups with stable coupling modes from a large number of spatially overlapping feature groups; it integrates symbolic matching of waveform morphology and quantitative stability analysis of time phase, taking into account the characteristics of coupling events in both morphological similarity and temporal regularity, thereby screening out anomalous regions that truly have multi-physics field synergistic change laws; the screening results provide a rigorously verified data foundation for constructing a highly reliable anomalous energy coupling feature map, which helps to further analyze the coupling mechanism and achieve accurate state assessment and fault diagnosis.

[0073] Example 6: Based on Example 5, the process of converting the continuous waveform curve shape into a symbol sequence composed of several rising segments, falling segments, and stationary segments arranged in chronological order in step S2031 of this embodiment of the invention specifically includes the following steps:

[0074] Step S20311: For each spatially overlapping feature group, the waveform curves of the three energy intensities (thermal, vibrational, and electromagnetic) as a function of time are extracted and sampled in segments at fixed time intervals. The continuous waveform curves are discretized into a set of sample segments arranged in chronological order. For each sample segment, the energy intensity difference between the start and end points of the sample segment is calculated. Based on the sign and absolute value of the energy intensity difference, the sample segment is initially labeled as an upward trend segment, a downward trend segment, or a stationary trend segment.

[0075] Step S20312: Arrange the upward trend segments, downward trend segments, and stable trend segments obtained after initial labeling in chronological order to form an initial trend segment sequence; merge adjacent segments of the same type in the initial trend segment sequence, merging segments with the same trend label that appear consecutively into a single trend segment; at the same time, remove isolated trend segments in the sequence whose duration is shorter than the preset duration to obtain a simplified trend segment sequence.

[0076] Step S20313: Assign a fixed symbol to each trend segment in the simplified trend segment sequence, where the upward trend segment corresponds to the first symbol, the downward trend segment corresponds to the second symbol, and the stationary trend segment corresponds to the third symbol; connect all the symbol identifiers of the trend segments according to their chronological order to form the morphological feature identifier of the waveform curve.

[0077] In the above embodiments, this embodiment transforms continuous waveform curves into compact symbol sequences, which significantly reduces data dimensionality and complexity while preserving core morphological topological features; it enhances the robustness and comparability of waveform feature expression, laying a reliable foundation for rapid and accurate matching of waveform morphologies across physical fields.

[0078] Example 7: As Figure 5 As shown, based on Example 1, the process of forming a fault type prediction list and remaining available time in step S300 of this embodiment of the invention specifically includes the following steps:

[0079] Step S301: Arrange the deviations between the actual coupling values ​​of thermal, vibrational and electromagnetic energies at each spatial point and the design tolerance values ​​according to time sequence to construct a historical evolution sequence of the deviations at each spatial point; superimpose and merge the historical evolution sequences of the deviations at all spatial points within the coverage area of ​​the same key component of the prefabricated substation to form a comprehensive deviation evolution trajectory of the key component.

[0080] Step S302: Divide the evolution trajectory of the comprehensive deviation of each key component into stages. Based on the slope change and fluctuation amplitude of the trajectory, divide it into a stable fluctuation stage, a linear growth stage and an accelerated growth stage in sequence. Extract the starting point of the accelerated growth stage as the warning start time and use the local growth rate of the trajectory in the accelerated growth stage as the degradation rate characterization value of the key component.

[0081] Step S303: Based on the degradation rate characterization value of each key component and the failure threshold line on the energy coupling tolerance curve marked on the original design drawings of the key component, calculate the time length required to extend along the degradation rate from the current sampling time to intersect with the failure threshold line; mark the time length as the remaining available time; at the same time, according to the correspondence between the spatial location of the starting point of the accelerated growth stage and the surrounding energy coupling anomaly mode, match the corresponding fault type from the preset fault type lookup table, and combine the remaining available time with the fault type to form a fault type prediction list.

[0082] In the above embodiments, this embodiment constructs a historical evolution sequence of spatial point deviation amplitudes and superimposes it into a comprehensive trajectory of key components to achieve multi-physical quantity fusion monitoring of equipment status. Integrating discrete spatial point data along the time dimension forms a continuous state evolution path, overcoming the randomness interference of single-point monitoring and improving the overall integrity and stability of state representation. Based on the stage division and feature extraction of trajectory morphology, a quantitative identification mechanism for equipment degradation processes is established. By analyzing the trajectory slope and fluctuation characteristics, the critical transition point from steady-state operation to accelerated degradation can be accurately located, providing an objective basis for early warning triggering. The calculation of degradation rate characterization values ​​transforms abstract state changes into quantifiable dynamic parameters. Combining design tolerance curves and real-time degradation rates, time-dimensional prediction of fault development is achieved. By calculating the theoretical time for the current state to extend to the failure threshold, static threshold comparison is transformed into dynamic lifespan estimation. Simultaneously, utilizing the mapping relationship between spatial location and abnormal patterns, early classification and identification of fault types are achieved. The final output prediction list correlates fault types with remaining available time, forming decision support data that combines qualitative and quantitative information. The list can provide clear time window guidance for maintenance strategy formulation, and at the same time guide the preparation of targeted maintenance plans through fault type prediction.

[0083] In summary, this embodiment achieves closed-loop prediction capability from real-time monitoring to early warning by connecting multi-source data fusion, state trajectory analysis, dynamic threshold comparison and pattern matching, thereby improving the systematicness and timeliness of fault prediction in prefabricated substations.

[0084] Example 8: Based on Example 7, the process of calculating the time required for the degradation rate to extend from the current sampling time to intersect with the failure threshold line in step S303 of this embodiment of the invention specifically includes the following steps:

[0085] Step S3031: Take the local growth rate of the accelerated growth stage in the evolution trajectory of the comprehensive deviation amplitude of each key component as the extension slope, and take the comprehensive deviation amplitude value at the current sampling time as the extension starting point to construct a forward-extending degradation trend ray in the comprehensive deviation amplitude-time two-dimensional coordinate system; the direction of the degradation trend ray is determined by the extension direction of the trajectory within the accelerated growth stage, and its extension path strictly follows the local direction of the trajectory within the accelerated growth stage.

[0086] Step S3032: Extract the failure threshold line corresponding to the spatial location of the key component from the energy coupling tolerance curve of the original design drawings; the failure threshold line is represented as a constant amplitude straight line parallel to the time axis in the comprehensive deviation amplitude-time two-dimensional coordinate system, and the amplitude value of the constant amplitude straight line is the upper limit of the allowable deviation amplitude marked on the original design drawings of the key component.

[0087] Step S3033: Proceed the constructed degradation trend ray step by step along the time axis. Calculate the comprehensive deviation amplitude value on the ray at each step point and compare the comprehensive deviation amplitude value with the constant amplitude value on the extracted failure threshold line point by point. When the comprehensive deviation amplitude value on the degradation trend ray is equal to or exceeds the constant amplitude value on the failure threshold line for the first time, record the time coordinate corresponding to the step point. Determine the time span between the time coordinate and the time coordinate at the current sampling time as the time length required to extend along the degradation rate from the current sampling time to intersect with the failure threshold line.

[0088] In the above embodiments, this embodiment achieves deterministic derivation from historical degradation behavior to future failure time by connecting trend modeling, threshold anchoring, numerical solution and time quantization. It combines the dynamic evolution of equipment status with static design standards to form a life prediction technology path based on physical mechanisms and mathematical extrapolation, which improves the objectivity and interpretability of remaining life assessment.

[0089] Example 9: Based on Example 8, the process of calculating the comprehensive deviation amplitude value on the ray at each propulsion step point in step S3033 of this embodiment of the invention specifically includes the following steps:

[0090] Step S30331: Based on the extension direction of the degradation trend ray, starting from the ray's starting point, using the fluctuation period of the trajectory within the accelerated growth phase as the step length, determine the time coordinates of each advancement step point sequentially along the positive time axis; the difference between the time coordinate of each advancement step point and the starting point time coordinate is an integer multiple of the advancement step length.

[0091] Step S30332: For each advance step point with a determined time coordinate, substitute the time coordinate of the advance step point into the linear extension relationship represented by the degradation trend ray; the linear extension relationship is uniquely determined by the ray starting point coordinate and the local direction of the trajectory within the acceleration growth stage. After substitution, the comprehensive deviation amplitude value on the ray corresponding to the advance step point is obtained.

[0092] Step S30333: Compare the comprehensive deviation amplitude value calculated on the ray at each advancement step point with the actual fluctuation envelope of the trajectory within the time interval during the accelerated growth phase; if the comprehensive deviation amplitude value falls within the range of the actual fluctuation envelope, retain the comprehensive deviation amplitude value as the effective comprehensive deviation amplitude value of the current advancement step point; if the effective comprehensive deviation amplitude value exceeds the actual fluctuation envelope, use the upper or lower boundary value of the actual fluctuation envelope as the corrected comprehensive deviation amplitude value, thus completing the determination of the comprehensive deviation amplitude value on the ray at the advancement step point.

[0093] In the above embodiments, this embodiment uses the fluctuation cycle of the accelerated growth phase as the step length to determine the advancement point, thereby aligning the time axis discretization with the actual fluctuation characteristics of the equipment. This ensures that the time resolution of the prediction calculation matches the inherent rhythm of the equipment's state changes, avoiding omissions of key state inflection points or computational redundancy due to improper step length selection. By substituting the time coordinates of the advancement point into a linear extension relationship determined based on the local trajectory of historical data, the theoretical amplitude calculation for future state points is achieved. Following the established degradation trend model, the mathematical continuity of the amplitude value at each prediction point is guaranteed, ensuring logical consistency in the trend extrapolation over time. By introducing the actual fluctuation envelope to compare and correct the theoretical calculation values, a physical rationality verification mechanism for trend extrapolation results was established. The pure mathematical extrapolation results were constrained to the statistical range of the recent actual fluctuations of the equipment, preventing the predicted values ​​from deviating from the actual possible fluctuation range due to the idealized assumptions of the linear model. By correcting the predicted values ​​that exceed the range based on the envelope boundary, the fusion of model prediction and historical experience data was achieved. When the theoretical extrapolation value exceeds the historical fluctuation limit, the envelope boundary is used as the correction value, which not only preserves the trend direction but also limits the prediction amplitude to the observed behavior pattern of the equipment, thereby enhancing the engineering credibility of the prediction results.

[0094] In summary, this embodiment constructs a prediction value generation method that balances the rigor of mathematical model extrapolation with the rationality of actual physical behavior through steps such as period-aligned discretization, model-based continuous calculation, envelope constraint verification, and boundary correction. While maintaining the trend extension logic, empirical constraints are applied through historical data envelope bands, making the prediction results both dynamic based on theoretical extrapolation and robust based on observation data, thereby improving the reliability and practicality of the state prediction stage in remaining lifetime assessment.

[0095] Example 10: Based on Example 9, the linear extension relationship in step S30332 of this embodiment is uniquely determined by the coordinates of the ray starting point and the local trajectory within the accelerated growth stage. Specifically, it includes the following steps:

[0096] Step S303321: Bind the current comprehensive deviation amplitude value to its corresponding time coordinate to form a coordinate pair of the ray starting point in the two-dimensional coordinate system; the time coordinate in the coordinate pair is taken from the latest sampling time after the extracted warning start time, and the comprehensive deviation amplitude coordinate is taken from the value on the comprehensive deviation amplitude evolution trajectory corresponding to the latest sampling time.

[0097] Step S303322: From the comprehensive deviation amplitude evolution trajectory during the accelerated growth phase, extract a trajectory segment with the warning start time as the starting point and the current sampling time as the ending point; perform directional feature extraction on the trajectory segment to obtain the overall extension direction angle of the trajectory segment in the time-amplitude plane. The overall extension direction angle is uniquely represented by the angle between the directed line segment pointing from the segment start point to the segment end point and the positive direction of the time axis.

[0098] Step S303323: Using the coordinates of the ray's starting point as a fixed base point and the obtained overall extension direction angle as the fixed direction of the ray, construct an infinite straight line in the time-amplitude plane that starts from the base point and extends outward along the overall extension direction angle; the infinite straight line represents the linear extension relationship represented by the deterioration trend ray, and each point on it satisfies the geometric constraint of linear extension along the fixed direction angle with the base point as the center.

[0099] In the above embodiments, this embodiment establishes a deterministic linear extrapolation model based on the macroscopic characteristics of recent historical trajectories through steps such as real-time anchor point setting, recent trend direction extraction, linear model construction, and geometric constraint solidification. Starting from the latest state and guided by the recent net change direction, a trend ray with a clear mathematical expression is generated, providing a stable and repeatable prediction baseline for the calculation of the remaining duration, and ensuring the rigor and consistency of the life assessment process at the model level.

[0100] Example 11: Based on Example 10, the process of obtaining the overall extension direction angle of the trajectory segment in the time-amplitude plane in step S303322 of this embodiment of the invention specifically includes the following steps:

[0101] Step S3033221: For the intercepted trajectory segment with the warning start time as the starting point and the current sampling time as the ending point, identify and remove outliers caused by instantaneous interference based on the amplitude difference between adjacent data points within the trajectory segment; at the same time, use a sliding window average to convert the trajectory segment into a central trend line that reflects the overall trend of change;

[0102] Step S3033222: On the central trend line, extract the first data point corresponding to the warning start time as the direction reference starting point, and extract the last data point corresponding to the current sampling time as the direction reference ending point; using the direction reference starting point as the base point, construct a directed line segment pointing to the direction reference ending point;

[0103] Step S3033223: Take the coordinates of the direction reference starting point on the time axis as the rotation center, and make the directed line segment virtually rotate around the rotation center in the time-amplitude plane until the directed line segment completely coincides with the positive direction of the time axis; record the plane angle through which the directed line segment rotates during the virtual rotation, and use the plane angle as the overall extension direction angle of the trajectory segment.

[0104] In the above embodiments, the three steps of this embodiment sequentially complete trajectory purification and trend extraction, feature line segment construction, and direction angle quantification. The overall process realizes the transformation from the original trajectory fragment to stable and measurable directional features; it can eliminate local interference and accurately capture the macroscopic extension direction of the trajectory within the observation period, providing consistent and reliable parameter basis for subsequent analysis or decision-making based on directional features.

[0105] Example 12: Based on Example 11, the process of converting a trajectory segment into a central trend line reflecting the overall trend of change using sliding window averaging in step S3033221 of this embodiment of the invention specifically includes the following steps:

[0106] Step S30332211: Determine the time span of the sliding window based on the fluctuation cycle of the trajectory during the accelerated growth phase; the time span is three times the length of the fluctuation cycle, and the window covers at least three complete fluctuation cycles.

[0107] Step S30332212: Starting from the initial data point of the trajectory segment, the sliding window is gradually moved along the positive time axis with a single sampling interval as the step size; at each position of the sliding window, the comprehensive deviation amplitude value of all data points within the coverage area of ​​the sliding window is extracted, and after arranging them in time order, the amplitude value of the data point at the middle position is taken as the representative value of the sliding window to form the trend point corresponding to the position of the sliding window;

[0108] Step S30332213: Arrange all trend points corresponding to the sliding window positions in chronological order according to the time sequence of their window center time; and connect adjacent trend points using an interpolation method based on local linearity preservation to form a continuous and smooth central trend line that reflects the overall trend of the trajectory segment.

[0109] In the above embodiments, this embodiment determines the sliding window time span based on the fluctuation cycle. Its technical effect is to ensure that the window length can cover the typical fluctuation patterns of the trajectory, ensuring that the trend extraction process has sufficient inclusiveness for periodic changes, avoiding oversensitivity to local fluctuations due to an excessively short window, or obscuring important trend features due to an excessively long window. Step S30332212 uses the median value of the comprehensive deviation amplitude sequence of all data points within the sliding window as the representative value. This can resist the influence of extreme values ​​or residual outliers within the window, and is more robust than the mean or endpoint values, making the trend points better reflect the concentrated trend of amplitude within the window range. Step S30332213 arranges the trend points of each window in chronological order and connects them into a continuous curve using an interpolation method based on local linearity preservation. While preserving the overall trend represented by the trend points, it ensures a smooth and natural transition of the central trend line between adjacent trend points, avoiding step-like or discontinuous jumps, thus forming a trend characterization line that reflects both the macroscopic evolution of the trajectory and maintains local coherence.

[0110] In summary, this embodiment achieves an effective transformation from a noisy and fluctuating original trajectory to a continuous and smooth central trend line by using adaptive window settings, robust median representative point selection, and interpolation connections that preserve local linearity. It significantly suppresses high-frequency fluctuations and random interference while preserving the main change pattern of the trajectory during the observation period, providing a clear and stable trend benchmark for directional feature extraction.

[0111] Example 13: Based on Example 12, the process of connecting adjacent trend points using an interpolation method based on local linearity preservation in step S30332213 of this embodiment of the invention specifically includes the following steps:

[0112] Step S303322131: From the trend point sequence arranged in chronological order, extract each pair of adjacent trend points in sequence, and denot them as the preceding trend point and the subsequent trend point respectively; using the preceding trend point as the base point and the subsequent trend point as the target point, obtain the local change feature vector between the two points in the time-amplitude plane, which is composed of the time difference and amplitude difference between the two points.

[0113] Step S303322132: Based on the obtained local change feature vector, construct a virtual sliding connecting rod within the time interval between the preceding trend point and the subsequent trend point; the starting point of the sliding connecting rod is fixed at the preceding trend point, and its direction is consistent with the direction of the local change feature vector. The sliding connecting rod extends forward along the time axis at a constant rate until its endpoint coincides with the subsequent trend point.

[0114] Step S303322133: Record the endpoint position of the sliding connecting rod at each moment during the extension process to form a series of continuous position points; connect the position points in chronological order to generate interpolation line segments between the preceding trend points and the subsequent trend points; repeat the operation for all adjacent trend point pairs, and connect all the generated interpolation line segments end to end to form a continuous and smooth central trend line, which maintains the direction of change of the original trajectory segment within a local range.

[0115] In the above embodiments, this embodiment constructs connecting line segments with consistent direction and natural transition between adjacent trend points by preserving local linearity. The overall technical effect is to generate a central trend line that accurately reflects the changing direction of the trajectory segment in each local time period, while maintaining global continuity and smoothness. This provides a trend foundation with good geometric consistency and sufficient noise suppression for calculating the overall extension direction angle.

[0116] Figure 6 A block diagram of an exemplary electronic device suitable for implementing embodiments of the present invention is shown.

[0117] Electronic devices may include a central processing unit / microprocessor / main control chip; and a storage medium coupled to the central processing unit / microprocessor / main control chip, wherein computer-executable instructions are stored for performing the steps of various methods of embodiments of the present invention when executed by a processor.

[0118] The central processing unit / microprocessor / main control chip may include, but is not limited to, one or more processors or microprocessors.

[0119] Storage media may include, but are not limited to, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, and computer storage media (such as hard disks, floppy disks, solid-state drives, removable disks, CD-ROMs, DVD-ROMs, Blu-ray discs, etc.).

[0120] In addition, the electronic device may include (but is not limited to) a data bus, an input / output bus / external bus / device bus, a display, and input / output devices (e.g., keyboard, mouse, speaker, etc.).

[0121] The central processing unit / microprocessor / main control chip can communicate with external devices via wired or wireless networks (not shown) through input / output buses / external buses / device buses.

[0122] The storage medium may also store at least one computer-executable instruction for performing the steps of various functions and / or methods in the embodiments described herein when the central processing unit / microprocessor / main control chip is running.

[0123] In one embodiment, the at least one computer-executable instruction may also be compiled into or comprise a software product, wherein one or more computer-executable instructions are executed by a processor to perform the steps of the various functions and / or methods in the embodiments described herein.

[0124] Figure 7 A schematic diagram of a computer-readable storage medium according to an embodiment of the present invention is shown.

[0125] like Figure 7 As shown, instructions, such as computer-readable instructions, are stored on a non-transitory computer-readable storage medium. When the computer-readable instructions are executed by a processor, the various methods described above can be performed. The non-transitory computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-transitory non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. For example, the non-transitory computer-readable storage medium can be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions stored on the non-transitory computer-readable storage medium, the various methods described above can be performed.

[0126] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0127] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0128] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0129] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the methods of the various embodiments of this invention through a computer device (which may be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0130] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for fault prediction and health management of prefabricated substations, characterized in that, Includes the following steps: The system receives an abnormal energy coupling feature spectrum containing anomalous correlation patterns in spatial location and temporal phase between thermal, vibrational, and electromagnetic distribution sub-maps. It then compares this abnormal energy coupling feature spectrum point-by-point with the energy coupling tolerance curve marked on the original design drawings of the prefabricated substation. The deviation between the actual coupling values ​​of thermal, vibrational, and electromagnetic energies at each spatial point and the design tolerance values ​​is calculated. Based on the magnitude of the deviation and its distribution density in the overall structure of the prefabricated substation, the system determines the current health level of the prefabricated substation through a preset health status assessment. Furthermore, based on the historical trend of the deviation, it estimates the time range within which key components may reach their failure threshold, generating a fault type prediction list and remaining available time.

2. The method for fault prediction and health management of prefabricated substations as described in claim 1, characterized in that, The process of creating a list of predicted fault types and remaining availability includes the following steps: The deviations between the actual coupling values ​​of thermal, vibrational, and electromagnetic energies at each spatial point and the design tolerance values ​​are arranged in chronological order to construct a historical evolution sequence of the deviations at each spatial point. The historical evolution sequences of the deviations at all spatial points within the coverage area of ​​the same key component of the prefabricated substation are superimposed and fused to form a comprehensive deviation evolution trajectory of the key component. The evolution trajectory of the comprehensive deviation of each key component is divided into stages based on the slope change and fluctuation amplitude of the trajectory. The stage is divided into a stable fluctuation stage, a linear growth stage, and an accelerated growth stage. The starting point of the accelerated growth stage is extracted as the warning start time, and the local growth rate of the trajectory within the accelerated growth stage is used as the characterization value of the deterioration rate of the key component. Based on the degradation rate characterization value of each key component, and combined with the failure threshold line on the energy coupling tolerance curve marked on the original design drawings of the key components, the time length required to extend along the degradation rate from the current sampling time to intersect with the failure threshold line is calculated; the time length is marked as the remaining available time; at the same time, based on the correspondence between the spatial location of the starting point of the accelerated growth stage and the surrounding energy coupling anomaly mode, the corresponding failure type is matched from the preset failure type lookup table, and the remaining available time and failure type are combined to form a failure type prediction list.

3. The method for fault prediction and health management of prefabricated substations as described in claim 2, characterized in that, The process of calculating the time required for the degradation rate to extend from the current sampling moment to intersect with the failure threshold line includes the following steps: Using the local growth rate of the accelerated growth phase in the evolution trajectory of the comprehensive deviation amplitude of each key component as the extension slope, and taking the comprehensive deviation amplitude value at the current sampling time as the extension starting point, a deterioration trend ray extending forward is constructed in the comprehensive deviation amplitude-time two-dimensional coordinate system. Extract the failure threshold line corresponding to the spatial location of the key components from the energy coupling tolerance curve of the original design drawings; The failure threshold line is represented as a constant amplitude straight line parallel to the time axis in the comprehensive deviation amplitude-time two-dimensional coordinate system. The amplitude value of the constant amplitude line is the upper limit of the allowable deviation amplitude marked on the original design drawings of the key components. The constructed degradation trend ray is progressively advanced along the time axis. At each advancement step, the comprehensive deviation amplitude value on the ray is calculated, and the comprehensive deviation amplitude value is compared point by point with the constant amplitude value on the extracted failure threshold line. When the comprehensive deviation amplitude value on the degradation trend ray is equal to or exceeds the constant amplitude value on the failure threshold line for the first time, the time coordinate corresponding to the advancement step is recorded. The time span between the time coordinate and the time coordinate at the current sampling moment is determined as the time length required to extend along the degradation rate from the current sampling moment to intersect with the failure threshold line.

4. The method for fault prediction and health management of prefabricated substations as described in claim 3, characterized in that, The process of calculating the overall deviation magnitude on the ray at each advance step point includes the following steps: Based on the extension direction of the deterioration trend ray, starting from the ray's origin, the time coordinates of each advancement step point are determined sequentially along the positive time axis, using the fluctuation period of the trajectory within the accelerated growth phase as the step length. For each advance step point with a defined time coordinate, substitute the time coordinate of the advance step point into the linear extension relationship represented by the degradation trend ray; The linear extension relationship is uniquely determined by the coordinates of the ray's starting point and the local trajectory within the accelerated growth phase. Substituting these values ​​yields the comprehensive deviation amplitude value on the ray corresponding to the advancement step point. The comprehensive deviation amplitude value on the ray calculated at each advancement step point is compared with the actual fluctuation envelope of the trajectory within the time interval during the accelerated growth phase. If the comprehensive deviation amplitude value falls within the actual fluctuation envelope range, then the comprehensive deviation amplitude value is retained as the effective comprehensive deviation amplitude value of the current advancement step point; If the effective comprehensive deviation amplitude value exceeds the actual fluctuation envelope, the upper or lower boundary value of the actual fluctuation envelope is used as the corrected comprehensive deviation amplitude value to complete the determination of the comprehensive deviation amplitude value on the ray at the advance step point.

5. The method for fault prediction and health management of prefabricated substations as described in claim 4, characterized in that, The linear extension relationship is uniquely determined by the coordinates of the ray's starting point and the local trajectory within the accelerated growth phase, and includes the following steps: The comprehensive deviation amplitude value at the current moment is bound to its corresponding time coordinate to form a coordinate pair of the ray starting point in the two-dimensional coordinate system; the time coordinate in the coordinate pair is taken from the latest sampling moment after the extracted warning start time, and the comprehensive deviation amplitude coordinate is taken from the value on the comprehensive deviation amplitude evolution trajectory corresponding to the latest sampling moment. From the evolution trajectory of the comprehensive deviation amplitude during the accelerated growth phase, a trajectory segment is extracted with the warning start time as the starting time and the current sampling time as the ending time; directional feature extraction is performed on the trajectory segment to obtain the overall extension direction angle of the trajectory segment in the time-amplitude plane. The overall extension direction angle is uniquely represented by the angle between the directed line segment from the segment start time to the segment end time and the positive direction of the time axis. Using the coordinates of the ray's starting point as a fixed base point and the obtained overall extension direction angle as the fixed direction of the ray, an infinite straight line is constructed in the time-amplitude plane, starting from the base point and extending outward along the overall extension direction angle. The infinite straight line represents the linear extension relationship represented by the deterioration trend ray, and each point on it satisfies the geometric constraint of linear extension along the fixed direction angle with the base point as the center.

6. The method for fault prediction and health management of prefabricated substations as described in claim 5, characterized in that, The process of obtaining the overall extension direction angle of a trajectory segment in the time-amplitude plane includes the following steps: For the captured trajectory segment starting from the warning start time and ending at the current sampling time, outliers caused by instantaneous interference are identified and removed based on the amplitude difference between adjacent data points within the trajectory segment; at the same time, a sliding window average is used to convert the trajectory segment into a central trend line that reflects the overall trend of change. On the central trend line, the first data point corresponding to the start time of the warning is extracted as the starting point of the direction reference, and the last data point corresponding to the current sampling time is extracted as the ending point of the direction reference. Using the starting point of the direction reference as the base point, construct a directed line segment pointing towards the ending point of the direction reference; Using the coordinates of the starting point on the time axis as the center of rotation, the directed line segment is virtually rotated around the center of rotation in the time-amplitude plane until the directed line segment completely coincides with the positive direction of the time axis; the plane angle rotated by the directed line segment during the virtual rotation is recorded, and the plane angle is used as the overall extension direction angle of the trajectory segment.

7. The method for fault prediction and health management of prefabricated substations as described in claim 6, characterized in that, The process of converting a trajectory segment into a central trend line reflecting the overall trend of change using a sliding window averaging includes the following steps: Based on the fluctuation cycle of the trajectory during the accelerated growth phase, the time span of the sliding window is determined; the time span is three times the length of the fluctuation cycle, and the window covers at least three complete fluctuation cycles. Starting from the initial data point of the trajectory segment, the sliding window is gradually moved along the positive time axis with a single sampling interval as the step size. At each position of the sliding window, the comprehensive deviation amplitude value of all data points within the coverage area of ​​the sliding window is extracted. After arranging them in chronological order, the amplitude value of the data point at the middle position is taken as the representative value of the sliding window to form the trend point corresponding to the position of the sliding window. Arrange all trend points corresponding to the sliding window positions in chronological order according to the time of their window center; and connect adjacent trend points using an interpolation method based on local linearity preservation to form a continuous and smooth central trend line that reflects the overall trend of the trajectory segment.

8. The method for fault prediction and health management of prefabricated substations as described in claim 7, characterized in that, The process of connecting adjacent trend points using interpolation based on local linearity preservation includes the following steps: From the trend point sequence arranged in chronological order, each pair of adjacent trend points is extracted sequentially and denoted as the preceding trend point and the subsequent trend point, respectively. Using the preceding trend point as the base point and the subsequent trend point as the target point, the local change feature vector between the two points in the time-amplitude plane is obtained, which is composed of the time difference and amplitude difference between the two points. Based on the obtained local change feature vector, a virtual sliding connecting rod is constructed in the time interval between the preceding trend point and the subsequent trend point. The starting point of the sliding connecting rod is fixed at the preceding trend point, and its direction is consistent with the direction of the local change feature vector. The sliding connecting rod extends forward along the time axis at a constant rate until its endpoint coincides with the subsequent trend point. Record the endpoint position of the sliding connecting rod at each moment during the extension process to form a series of continuous position points; Connect the position points in chronological order to generate interpolation segments between previous and subsequent trend points; repeat the operation for all adjacent trend point pairs, and connect all the generated interpolation segments end to end to form a continuous and smooth central trend line. The central trend line maintains the direction of change of the original trajectory segment within a local range.

9. The method for fault prediction and health management of prefabricated substations as described in claim 1, characterized in that, It also includes pre-laying multiple layers of energy conversion films on the inner surface of the prefabricated substation shell to convert the thermal radiation, mechanical vibration and electromagnetic fluctuations generated during the operation of the prefabricated substation into corresponding electrical signals in real time; after being collected by the electrical signal acquisition network, an initial energy distribution map is generated; the initial energy distribution map is then processed to remove background noise to obtain a purified energy distribution map.

10. The method for fault prediction and health management of prefabricated substations as described in claim 1, characterized in that, It also includes decomposing the purification energy distribution map into independent thermal, vibrational, and electromagnetic distribution sub-maps based on the different transmission speeds and attenuation characteristics of thermal, vibrational, and electromagnetic energy within the prefabricated substation; aligning the thermal, vibrational, and electromagnetic distribution sub-maps according to time series and superimposing and comparing them to identify abnormal correlation patterns in spatial location and temporal phase among the thermal, vibrational, and electromagnetic distribution sub-maps, forming an abnormal energy coupling feature map.

Images