Method and system for monitoring electrical equipment

Abstract

This application discloses a monitoring method and system for electrical equipment. The method includes: acquiring the thermal radiation distribution characteristics and insulation degradation acoustic characteristics of electrical equipment in a locomotive; acquiring the locomotive's operating status parameters; determining the target weights for the thermal radiation distribution characteristics and the insulation degradation acoustic characteristics based on the operating status parameters; and generating a fused early warning decision result based on the thermal radiation distribution characteristics, the insulation degradation acoustic characteristics, and their respective target weights. This application effectively eliminates normal dynamic fluctuations (such as load fluctuations and electromagnetic shocks) under complex and variable locomotive dynamic conditions (e.g., load fluctuations and electromagnetic shocks), significantly improving early warning accuracy and thus enhancing the system's active protection capabilities.

Description

Technical Field

[0001] This application belongs to the field of locomotive monitoring technology, and in particular relates to a monitoring method and system for electrical equipment. Background Technology

[0002] With the rapid evolution of global rail transit technology, high-power AC drive electric locomotives and high-speed trains have become the core carriers of modern logistics and rail passenger transport. As the core architecture of the locomotive power system, the operational stability of key electrical equipment such as main transformers, traction converters, and high-voltage electrical cabinets directly determines driving safety, operational efficiency, and maintenance costs. Because these onboard electrical devices operate in extremely harsh environments with high voltage, high current, strong electromagnetic interference, and severe mechanical vibration, their insulation systems and electrical connections are highly susceptible to performance degradation under the coupling of multiple physical fields. This degradation typically manifests as weak partial discharge signals or abnormal localized thermal radiation. If these signals are not detected and accurately warned of in a timely manner, they can easily lead to insulation breakdown, fires, or even serious traffic accidents.

[0003] In related technologies, online monitoring of vehicle electrical equipment mainly employs static multimodal fusion technology. This involves using infrared visual sensors to capture the thermal radiation distribution characteristics of the electrical equipment surface, monitoring for localized overheating caused by increased contact resistance or overload, and using ultrasonic sensors to collect acoustic characteristics in the 20kHz-150kHz high-frequency band, monitoring for partial discharge pulses generated by internal insulation degradation. A classic weighted summation method is used, for example, assigning a fixed weight of 0.6 to the thermal radiation distribution characteristics and a fixed weight of 0.4 to the acoustic characteristics, attempting to reduce the false alarm rate of a single sensor through complementary information from different dimensions. However, under the complex and variable dynamic conditions of locomotives (such as load fluctuations and electromagnetic shocks), the above scheme cannot achieve accurate decoupling between fault characteristics and normal dynamic fluctuations, leading to inaccurate warnings and frequent false alarms. Summary of the Invention

[0004] The purpose of this application is to provide a monitoring method and system for electrical equipment to solve the problems of inaccurate early warning and frequent false alarms in related technologies.

[0005] To achieve the above objectives, the embodiments of this application adopt the following technical solutions: In a first aspect, embodiments of this application provide a method for monitoring electrical equipment, comprising: acquiring thermal radiation distribution characteristics and acoustic insulation degradation characteristics of electrical equipment in a locomotive; acquiring operating status parameters of the locomotive; determining target weights for the thermal radiation distribution characteristics and the acoustic insulation degradation characteristics respectively based on the operating status parameters; and generating a fusion early warning decision result based on the thermal radiation distribution characteristics, the acoustic insulation degradation characteristics, and their respective target weights.

[0006] Secondly, embodiments of this application provide a monitoring system for electrical equipment, comprising: a multimodal sensing module for acquiring thermal radiation distribution characteristics and insulation degradation acoustic characteristics of electrical equipment in a locomotive; a working condition sensing module for acquiring operating status parameters of the locomotive; a dynamic fusion calculation module for determining target weights for the thermal radiation distribution characteristics and the insulation degradation acoustic characteristics respectively based on the operating status parameters; and generating a fusion early warning decision result based on the thermal radiation distribution characteristics, the insulation degradation acoustic characteristics, and their respective target weights.

[0007] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects: Based on the real-time operating status parameters of the locomotive, the embodiments of this application dynamically determine the target weights of different modal characteristics. At the algorithm level, a deep dynamic decoupling of "locomotive physical dynamic conditions" and "algorithm weights" is achieved, that is, the fault characteristics and normal dynamic fluctuations are accurately decoupled. Under the varied and complex locomotive dynamic conditions (such as load fluctuations and electromagnetic shocks), normal dynamic fluctuations (such as load fluctuations and electromagnetic shocks) are effectively eliminated, the accuracy of early warning is greatly improved, and thus the active protection capability of the system is enhanced. Attached Figure Description

[0008] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a method for monitoring electrical equipment, provided as an embodiment of this application; Figure 2 A schematic diagram illustrating an application scenario of a monitoring method for electrical equipment provided in one embodiment of this application; Figure 3 A flowchart illustrating a method for monitoring electrical equipment, provided as another embodiment of this application; Figure 4 A schematic diagram of the structure of a monitoring system for electrical equipment is provided as an embodiment of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0010] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein. Furthermore, "and / or" in this application indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship. It should be noted that all data involved in this application was obtained with the user's authorization.

[0011] In related technologies, static multimodal fusion technology is used for online monitoring of electrical equipment. However, a deeper analysis of the actual operating characteristics of locomotives reveals that directly transplanting the monitoring logic derived from the static power grid into the highly dynamic onboard environment presents profound and irreconcilable technical contradictions. 1) Fluctuations in physical load lead to serious false positives (i.e., operating noise interference). When locomotives are performing actual transport tasks, their operating state exhibits strong dynamic non-stationary characteristics. Frequent switching between traction, braking, coasting, and phase-splitting conditions causes drastic and random fluctuations in the load current of electrical equipment. For example, when a locomotive performs high-power traction tasks such as heavy-load climbing or high-speed starting, the continuous high current in the main circuit generates significant Joule heating. The temperature rise caused by this normal ohmic loss often appears highly similar to overheating caused by early electrical connection loosening in infrared thermal radiation images. Static multimodal fusion technology in related technologies is based on preset fixed weights and lacks deep perception of the locomotive's underlying real-time physical dynamic parameters (e.g., it does not recognize traction current), thus failing to effectively separate normal load fluctuations from actual potential faults at the principle level. This coupling of "operating condition noise" makes the system prone to misjudging normal operating temperature rise as a serious thermal fault, leading to frequent false alarms.

[0012] 2) Complex electromagnetic environments lead to confusion in acoustic signal recognition. In complex electromagnetic environments, when a locomotive passes through a phase-splitting zone or experiences a momentary arcing impact due to pantograph disconnection, it radiates extremely strong high-frequency electromagnetic pulses and acoustic interference into the onboard environment. These transient impact signals often exhibit significant overlap in time-frequency domain characteristics with the actual partial discharge signals accompanying insulation degradation within electrical equipment. Static multimodal fusion techniques, when dealing with such highly uncertain "conflicting evidence," suffer from a rigid and inflexible confidence (i.e., weight) allocation mechanism. This prevents real-time adjustments to the contribution of each modal evidence (i.e., feature) based on the instantaneous operating conditions, thus failing to reduce the weight of interfering modal evidence in real time. Consequently, the fused decision result is highly susceptible to being misled by transient physical impacts, severely reducing the robustness of the early warning system.

[0013] In summary, the static multimodal fusion technology in related technologies cannot accurately decouple fault characteristics from normal dynamic fluctuations under complex and variable locomotive dynamic conditions (such as load fluctuations and electromagnetic shocks), leading to inaccurate warnings and frequent false alarms. Therefore, this application proposes a monitoring method and system for electrical equipment that can effectively eliminate normal dynamic fluctuations (such as load fluctuations and electromagnetic shocks) under complex and variable locomotive dynamic conditions (such as load fluctuations and electromagnetic shocks), significantly improving warning accuracy and thus enhancing the system's active protection capabilities.

[0014] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0015] Figure 1 This is a flowchart illustrating a method for monitoring electrical equipment, provided as an embodiment of this application. Figure 1 As shown, the monitoring method for electrical equipment according to an embodiment of this application may specifically include the following steps: S101, acquire the thermal radiation distribution characteristics and insulation degradation acoustic characteristics of the electrical equipment in the locomotive.

[0016] In this embodiment of the application, the entity executing the electrical equipment monitoring method is an electrical equipment monitoring system, which can be installed in an electronic device. This electronic device can be a terminal device or a server. The terminal device can be a mobile phone, tablet computer, desktop computer, laptop, vehicle-mounted device, etc.; the server can be a standalone server or a server cluster composed of multiple servers. In the field of locomotive monitoring technology, this electrical equipment monitoring system can be installed within the electrical equipment monitoring system.

[0017] Through the multimodal sensing module 101 (such as...) Figure 2(As shown) This acquires the thermal radiation distribution characteristics of key connection points and insulation surfaces of electrical equipment in the locomotive, and the acoustic characteristics of insulation degradation caused by ultrasonic partial discharge signals resulting from insulation degradation within the electrical equipment. Specifically, this can be achieved through the infrared vision sensor array 102 in the multimodal sensing module (such as... Figure 2 (As shown) Real-time acquisition of infrared thermal radiation image data streams from electrical equipment, and acquisition of the corresponding thermal radiation distribution characteristics. The ultrasonic sensor (i.e., ultra-high frequency acoustic sensor) array 103 in the multimodal sensing module can also real-time acquisition of insulation degradation acoustic signal (i.e., ultrasonic partial discharge acoustic signal) data streams from electrical equipment, and acquisition of the corresponding insulation degradation acoustic characteristics. The infrared thermal radiation image data stream and the ultrasonic signal data stream can be aligned at the hardware level via FPGA.

[0018] Furthermore, spatial domain median filtering can be applied to the sensing data acquired by the multimodal sensing module to eliminate image shift blur and mechanical vibration noise. Specifically, for infrared thermal radiation images, spatial domain filtering algorithms can be used to eliminate random noise and image shift blur caused by severe locomotive vibration; for acoustic signals of insulation degradation, a bandpass filter set between 20kHz and 150kHz can be used to remove low-frequency interference noise generated by mechanical vibrations such as wheel-rail friction and fan operation.

[0019] The infrared vision sensor array employs a long-wavelength uncooled focal plane array detector. The core photosensitive element typically has a pixel resolution of 640×480 or higher, operating in a wavelength range of 8 micrometers (μm) to 14 μm. This wavelength effectively penetrates some of the smoke and dust in the locomotive's operating environment and exhibits extremely high capture efficiency for thermal radiation energy generated by electrical equipment. Its thermal sensitivity is better than 50 milliklvin (mK), ensuring the generation of high-contrast thermal radiation distribution images even under minute temperature differences. The detector is externally equipped with a germanium single-crystal infrared optical lens, coated with an anti-reflection film and a diamond-like carbon film to resist physical wear and chemical corrosion during operation. The infrared vision sensor array transmits the raw thermal radiation image data stream to the backend in real time via a Gigabit Ethernet (GigE Vision) interface. The specific installation location of the infrared vision sensor array is optimized, allowing it to be fixed to the top of the electrical cabinet or the transformer observation window using a bracket with a vibration-damping structure, ensuring that the optical axis center covers the high-voltage bushing, oil drain line, and busbar connection.

[0020] The ultrasonic sensor array can be composed of multiple highly sensitive piezoelectric ceramic sensors. These sensors utilize high-performance lead barium titanate material and are arranged in a symmetrical, ring-shaped configuration. They employ the time difference of arrival (TDOA) principle to spatially locate partial discharge sources, with a response frequency covering 20 kHz to 150 kHz. This allows for precise coverage of ultrasonic partial discharge signals caused by insulating dielectric polarization or breakdown in high-voltage cable joints and transformer bushings inside electric locomotives. Each piezoelectric ceramic sensor is equipped with a low-noise preamplifier circuit, whose gain can be dynamically adjusted within the range of 20 dB to 60 dB according to the background noise level, ensuring that weak acoustic evidence of partial discharge can be clearly extracted.

[0021] S102, Obtain the operating status parameters of the locomotive.

[0022] In this embodiment of the application, the working condition sensing module 201 (such as...) can be used. Figure 2 (As shown) to obtain the locomotive's operating status parameters (i.e., dynamic operating condition parameters). The operating condition perception module, as a key node for obtaining the locomotive's underlying operating logic, can access the Multifunction Vehicle Bus (MVB) 202 in the locomotive communication network through redundant interfaces at the physical layer. Figure 2 As shown, following the IEC61375 standard protocol, it has the ability to analyze process data and message data in real time, extracting the locomotive's operating status parameters in real time. The operating condition perception module integrates a dedicated MVB control chip. Through hardware filtering and address mapping of bus messages, the data sampling period can be set within the range of 10ms to 50ms, ensuring synchronization with the timing of the locomotive's electric traction control system.

[0023] Among them, the operating status parameters constitute a multi-dimensional operating condition vector, which may include, but is not limited to, the real-time traction current reflecting the load intensity of the electrical system. I t The pantograph grid voltage reflects the stability of energy input. U p The real-time speed of the locomotive is related to the convective heat transfer conditions on the surface of the associated equipment. v The main circuit breaker status and brake cylinder pressure value are used to assist in determining whether the locomotive is in traction, coasting, or braking state.

[0024] In addition, to ensure strict synchronization between operating status parameters and sensing data, the operating condition sensing module can use cyclic redundancy check and bidirectional time synchronization mechanism to align the timestamps of operating status parameters with the timestamps of sensing data, ensuring that each frame of sensor data can match the precise current, voltage and speed values ​​at the moment of its generation.

[0025] It should be noted that, in addition to being extracted from the MVB train bus, operating status parameters can also be obtained by directly measuring real-time current and voltage by adding an independent Hall current / voltage sensor to the electrical circuit, thus replacing the bus data input.

[0026] S103, based on the operating state parameters, determine the target weights of the thermal radiation distribution characteristics and the acoustic characteristics of insulation degradation, respectively.

[0027] In this embodiment of the application, the dynamic fusion computing module 301 (such as...) Figure 2 (As shown) Based on the operating status parameters obtained in step S102, the target weights of the thermal radiation distribution characteristics and the acoustic characteristics of insulation degradation are determined respectively. The dynamic fusion computing module establishes high-speed, low-latency data interaction links with the multimodal sensing module and the operating condition sensing module, respectively, typically using PCIe or dual-path gigabit Ethernet for physical connection. The dynamic fusion computing module can use an industrial-grade multi-core processor, including a high-performance ARM core for task scheduling and logic control, and a built-in hardware acceleration unit (NPU / GPU) for performing deep learning preprocessing and complex matrix operations.

[0028] Furthermore, such as Figure 3 As shown, step S103, "determining the target weights of the thermal radiation distribution characteristics and the acoustic characteristics of insulation degradation respectively based on the operating state parameters," may specifically include the following steps: S301, based on the operating state parameters, calculate the first compensation operator for characterizing load temperature rise deviation and the second compensation operator for characterizing transient voltage impact.

[0029] In this embodiment, the dynamic fusion calculation module can construct and update the dynamic weight compensation operator in real time based on the operating status parameters obtained in step S102. Specifically, the dynamic weight compensation operator may include a first compensation operator α1 for characterizing load temperature rise deviation and a second compensation operator α2 for characterizing transient voltage surges. The first compensation operator α1 is used to compensate for temperature rise deviation caused by load fluctuations. The second compensation operator α2 is used to suppress electromagnetic pulse interference.

[0030] The first compensation operator α1 can be generated through the following steps: when the real-time traction current is continuously exceeded by a preset traction current threshold and the duration is greater than a preset duration threshold, the expected temperature rise of the electrical equipment is calculated; the first compensation operator is generated according to the expected temperature rise, and the first compensation operator is configured to nonlinearly reduce the preset weight of the thermal radiation distribution characteristics when the deviation between the actual temperature rise and the expected temperature rise is within a preset deviation threshold.

[0031] Specifically, the dynamic fusion computing module incorporates a Joule thermophysical model of electrical equipment. When the operating condition sensing module detects that the locomotive is in a high-current traction condition, i.e., the real-time traction current... I t When the current exceeds a preset traction current threshold (e.g., 1.2 times the rated current) for an extended period, and the duration exceeds a preset duration threshold, the dynamic fusion calculation module calculates the expected temperature rise of the electrical equipment through integral calculation. The calculation formula is as follows:

[0032] in, R The equivalent contact resistance of electrical equipment varies with temperature. h To determine the locomotive's real-time speed v Dynamically corrected convective heat transfer coefficient, S For heat dissipation area, T ( t To monitor temperature in real time, T amb For ambient temperature, C For the specific heat capacity of the material, m For the quality of electrical equipment.

[0033] The above calculation formula fully considers the Joule heating effect, convective heat transfer, and the heat capacity characteristics of the material itself.

[0034] Based on the calculated expected temperature rise A first compensation operator α1 is generated. The first compensation operator is configured to: when the actual temperature rise and the expected temperature rise are... When the deviation is within a preset deviation threshold, the preset weight of the thermal radiation distribution feature (i.e., the basic probability assignment of the thermal radiation distribution feature in the fusion decision) is non-linearly reduced. , H (This indicates the "thermal failure" assumption). When the actual temperature rise and the expected temperature rise... When highly matched, α1 tends to a smaller value, thereby nonlinearly suppressing the weight of infrared thermal radiation features in the fusion decision and preventing normal temperature rise caused by high load from being falsely reported as thermal fault.

[0035] It should be noted that in the engineering implementation of automotive microprocessors, the above-mentioned calculations for the expected temperature rise are not included. The continuous integral formula can be equivalently replaced by a discrete-time accumulation algorithm based on a fixed sampling period; alternatively, a multi-dimensional look-up table (LUT) containing traction current, locomotive speed, and ambient temperature can be generated in advance through offline simulation, and the expected temperature rise can be directly output through table lookup and interpolation algorithms at runtime. The LUT table is a commonly used data structure in software engineering. By pre-calculating the results of complex formulas and storing them in a table, the results can be directly retrieved by simply inputting the condition index at runtime, greatly reducing the CPU's computational burden. It is the best engineering alternative to the integral formula in the embodiments of this application.

[0036] The second compensation operator α2 can be generated through the following steps: when the locomotive is detected to be in a transient voltage impact condition caused by phase separation or pantograph offline, the grid voltage change rate of the pantograph is calculated; the second compensation operator is generated based on the grid voltage change rate, and the second compensation operator is configured to increase the preset weight of the acoustic characteristics of insulation degradation when the grid voltage change rate exceeds a preset change rate threshold.

[0037] Specifically, when the operating condition sensing module detects that the locomotive is in a transient voltage surge condition caused by phase separation or pantograph disconnection, the dynamic fusion calculation module calculates the pantograph's grid voltage change rate. Based on the calculated grid voltage change rate A second compensation operator α2 is generated. The second compensation operator α2 is configured to determine that there is extremely strong broadband electromagnetic interference in the current acoustic environment when the rate of change of the grid voltage exceeds a preset rate of change threshold (i.e., the external electromagnetic noise is enhanced). These interference signals may overlap with partial discharge signals in the frequency domain. At this time, the second compensation operator α2 intervenes to increase the preset weight of the acoustic characteristics of insulation degradation, that is, to increase the discrimination weight of the ultrasonic sensor array for specific pulse phase characteristics. By increasing the conflict suppression coefficient in the acoustic evidence chain, the transient arc interference of non-partial discharge nature is stripped away, ensuring the determinism of the fusion early warning decision result.

[0038] S302, the preset weights of the thermal radiation distribution features are corrected according to the first compensation operator to obtain the target weights of the thermal radiation distribution features.

[0039] S303, the preset weights of the acoustic features of insulation degradation are corrected according to the second compensation operator to obtain the target weights of the acoustic features of insulation degradation.

[0040] Furthermore, the monitoring method for electrical equipment in this application embodiment may also include the following steps: acquiring the optical window contamination status of an infrared visual sensor used to collect the thermal radiation distribution characteristics; when the optical window contamination status exceeds a preset limit, triggering an adaptive protection and cleaning module to perform physical cleaning action on the surface of the optical window of the infrared visual sensor.

[0041] Specifically, the operating environment of vehicle-mounted sensors is more demanding and variable compared to ground-based base stations. When a vehicle travels at high speed through tunnels, deserts, or encounters severe weather such as rain, snow, or fog, the optical windows of the infrared vision sensor arrays installed on the exterior of the vehicle or in specific compartments are highly susceptible to damage from dust, oil, and frost. This physical optical shielding and signal attenuation leads to a sharp drop in the signal-to-noise ratio of thermal radiation distribution characteristics, resulting in severe feature submersion. In static fusion technologies, the algorithm logic relies entirely on preset assumptions of input completeness, failing to detect the sensor's own "sub-health" state or to adaptively reduce weights or compensate for real-time performance degradation. Once a core sensing mode fails due to environmental factors, it often triggers the logical collapse of the entire fusion and early warning chain.

[0042] Therefore, the monitoring system for electrical equipment in this application embodiment may further include an adaptive protection and cleaning module 401 (such as...). Figure 2 As shown), this adaptive protection and cleaning module is an electromechanical-hydraulic-pneumatic integrated subsystem, installed outside the optical window of the infrared vision sensor array. Figure 2 As shown, the adaptive protection and cleaning module 401 may specifically include: a pneumatic cleaning air curtain 402 connected to the locomotive's main air duct and equipped with a slit nozzle, an electrically heated defrosting wire 404 embedded in the edge of the optical window 403, and an ambient temperature and humidity sensor 405 installed inside the infrared vision sensor protective cover.

[0043] The dynamic fusion computing module also has a window health status assessment logic, which assesses the optical window contamination status of the infrared vision sensor in real time. When the optical window contamination status exceeds the preset limit and the ambient temperature and humidity sensor indicates a high humidity or low temperature frosting environment, the pneumatic cleaning air curtain in the adaptive protection and cleaning module is automatically triggered to open, and the optical window surface is washed with high frequency using compressed air of 0.4MPa to 0.6MPa.

[0044] The state of contamination of the optical window can be assessed through the following steps: extracting the spatial correlation and edge intensity gradient of the infrared thermal image acquired by the infrared vision sensor. According to the gradient G The optical window's contamination status is assessed in real time. If the contamination status exceeds a preset limit, i.e., the edge intensity gradient... G Below the preset resolution threshold.

[0045] The above edge intensity gradient G The calculation employed an optimized Sobel operator. At a specific code implementation level, the system divides the thermal radiation image into multiple sub-regions and calculates the horizontal (…) for each sub-region. x ) and vertical ( y The gradient amplitude in the direction of the noise level is measured. If the average gradient value of the central monitoring area is lower than the preset background noise benchmark value, the window is considered dirty. The opening time of the pneumatic cleaning curtain is set to a cyclic spray mode, i.e., spraying for 3 seconds, pausing for 2 seconds, for a total of 5 cycles. This method utilizes the impact force of compressed air more effectively than continuous spraying, while also reducing the pressure on the locomotive's main air source.

[0046] In a specific application scenario, when a locomotive enters a damp tunnel or encounters rain or snow, condensation or dirt may form on the surface of the optical window, leading to decreased contrast and blurred edges in the infrared thermal radiation image. If the assessed edge intensity gradient G is less than 60% of the normal threshold, and the ambient temperature and humidity sensor indicates that the current ambient humidity is higher than 85% or the temperature is lower than 3 degrees Celsius, the pneumatic cleaning curtain is automatically triggered to open. At this time, using the locomotive's own compressed air at 0.4MPa to 0.6MPa, a high-speed scouring air film is formed on the surface of the optical window through specially designed slit nozzles.

[0047] It should be noted that the pneumatic cleaning curtain can be replaced by "miniature high-frequency ultrasonic vibration glass piezoelectric ceramic" to shake off dirt, or by the traditional "mechanical miniature wiper spraying water" for cleaning.

[0048] In addition, the monitoring method for electrical equipment in this application embodiment may further include the following steps: calculating a third compensation operator to characterize the attenuation of the window signal based on the state of contamination of the optical window; correspondingly, the above step S202 "correcting the preset weight of the thermal radiation distribution feature according to the first compensation operator to obtain the target weight of the thermal radiation distribution feature" may specifically include the following steps: correcting the preset weight of the thermal radiation distribution feature according to the first compensation operator and the third compensation operator to obtain the target weight of the thermal radiation distribution feature.

[0049] Specifically, the dynamic fusion calculation module calculates a third compensation operator α3 to characterize the signal attenuation of the optical window based on the state of contamination of the optical window. The third compensation operator α3 is configured to forcibly apply a preset weight to the thermal radiation distribution characteristics during the physical self-cleaning process of the optical window and during the transition period before the optical window's transmittance recovers. The weights are lowered to a preset lower limit (e.g., below 0.2), and the control of the early warning decision logic is smoothly transferred to the acoustic sensing mode, which is unaffected by optical contamination. After self-cleaning is completed and the image edge intensity gradient G recovers to the normal monitoring threshold (i.e., image sharpness is restored), the weight penalty is revoked through a smooth transition algorithm, restoring normal fusion. The third compensation operator α3 is used for anti-optical attenuation and cleaning compensation.

[0050] For example, in the application scenarios mentioned above, such as when a locomotive enters a damp tunnel or encounters rain or snow, a third compensation operator α3 is generated within 3-5 seconds of jet cleaning. This reduces the weight of the distorted infrared data to below 0.2, allowing the ultrasonic channel to take over 100% of the early warning decision-making authority. After cleaning, the image gradient returns to normal, and the infrared data weight is smoothly adjusted back, achieving extremely high fault tolerance and self-healing capabilities for the early warning system.

[0051] S104, Based on the thermal radiation distribution characteristics, the acoustic characteristics of insulation degradation, and their respective target weights, a fusion early warning decision result is generated.

[0052] In this embodiment, the dynamic fusion calculation module can generate a fusion early warning decision result based on the thermal radiation distribution characteristics and the acoustic characteristics of insulation degradation obtained in step S101, and the target weights of the thermal radiation distribution characteristics and the acoustic characteristics of insulation degradation determined in step S103. The fusion early warning decision result may specifically include a target fault proposition and its corresponding confidence level.

[0053] As a feasible implementation method, the thermal radiation distribution characteristics, the acoustic characteristics of insulation degradation, and their respective target weights can be input into the improved DS evidence theory fusion model for decision synthesis to obtain the fusion early warning decision result output by the improved DS evidence theory fusion model.

[0054] Specifically, the multimodal fusion model used in this application is an improved Dempster-Shafer Theory (DS) fusion model based on conflict allocation optimization. This improved DS fusion model searches for the optimal solution that minimizes the total uncertainty of the system through iterative computation. First, it extracts the relative temperature difference and the temperature evolution gradient over time at the electrical equipment connection points from infrared thermal radiation images using a lightweight deep convolutional neural network. Simultaneously, it extracts probability density function features characterizing the discharge energy distribution—discharge amplitude, phase distribution, and discharge repetition frequency—from ultrasonic signals using high-frequency envelope analysis and phase resolution algorithms (PRPD / PRPS). During the fusion calculation stage, a dynamic weight compensation operator is introduced to correct the weights of each modal evidence (i.e., the basic confidence function (Mass function)). The correction logic is defined as follows: The algorithm calculates the basic probability assignments (i.e., confidence levels) for different fault propositions. It also employs nonlinear weighted smoothing to remove false conflict evidence caused by load fluctuations or electromagnetic shocks. For local conflicts arising from interference during the synthesis process, it abandons traditional global normalization and adopts a conflict allocation strategy based on evidence similarity. It calculates the correlation coefficient matrix between various evidence sources and allocates the conflict probability according to the corrected confidence ratio to propositions with support higher than a preset value, rather than directly classifying them into uncertain terms. This effectively eliminates false conflict evidence caused by drastic changes in locomotive operating conditions (such as temperature rise due to sudden current surges and acoustic signal mismatch), thus obtaining accurate fusion basic probability assignments (i.e., confidence levels). During this process, if the dynamic weight compensation operator indicates that the current period is one of abnormal fluctuations, the improved DS evidence theory algorithm automatically increases its tolerance for "uncertainty" propositions until a warning result with sufficient confidence is obtained.

[0055] DS evidence theory is an uncertainty reasoning theory that can integrate evidence from multiple independent information sources (such as infrared and acoustic) (Mass function confidence assignment), but it is prone to failure when dealing with highly conflicting evidence. This application improves upon it by introducing a dynamic weight compensation operator.

[0056] PRPD (Phase Resolved Partial Discharge): Projecting the occurrence time of the partial discharge signal onto the corresponding phase of the AC power frequency voltage (e.g., 50Hz), the resulting three-dimensional spectrum (amplitude-phase-number) is the core method for identifying the type of partial discharge.

[0057] To improve the computational efficiency of the dynamic fusion computing module, this application embodiment achieves a lightweight modification of the DS evidence theory algorithm, reducing the entire fusion decision cycle to within 200ms. Since locomotive electrical faults (especially thermal runaway) typically possess a certain thermal inertia, this response speed is sufficient to meet real-time protection requirements. In terms of hardware implementation, the system's mean time between failures (MTBF) has been tested to exceed 15,000 hours and complies with the EN50155 standard for railway electronic equipment, demonstrating strong resistance to vibration, shock, and wide-temperature operation.

[0058] It should be noted that, in addition to using the improved DS evidence theory, the multimodal fusion algorithm can also use dynamic Bayesian networks (DBN), fuzzy logic inference matrices, or deep neural network Transformer models based on spatiotemporal attention mechanisms. It is only necessary to decouple and bind the input layer of its weight matrix with the locomotive operating state parameters.

[0059] Furthermore, the monitoring method for electrical equipment in this application embodiment also includes the following steps: determining the corresponding early warning threshold level based on the reliability level in the fusion early warning decision result; and controlling the locomotive's actuator to execute the corresponding intervention action command based on the early warning threshold level.

[0060] Specifically, the tiered early warning execution module 501 (such as...) Figure 2 (As shown) via the locomotive microcomputer control interface 502 (e.g.) Figure 2 As shown, the hierarchical early warning execution module achieves closed-loop linkage with the locomotive central control unit. Based on the confidence level in the fusion early warning decision results, the module matches the corresponding early warning threshold level and outputs the corresponding intervention action command to the locomotive's actuator through the locomotive microcomputer control interface according to the early warning threshold level. The actuator executes the intervention action command, realizing closed-loop control from perception and decision to execution.

[0061] The hierarchical early warning execution module transforms complex fusion early warning decision results into intuitive hierarchical early warning logic. This logic can be set as follows: Level 1 Early Warning: Corresponding to abnormal thermal trends. When it is determined that electrical equipment exhibits a slight thermal degradation trend but has not yet reached a dangerous level, the hierarchical early warning execution module displays an early warning message on the onboard terminal and simultaneously generates a maintenance work order containing fault coordinates and characteristic data to the ground maintenance platform via the 4G / 5G wireless network; Level 2 Early Warning: Corresponding to determined moderate fault characteristics. When it is determined that insulation degradation has entered an irreversible stage, the hierarchical early warning execution module sends a notification to the locomotive control logic (i.e., the locomotive central control). The unit sends a load reduction command to forcibly limit the traction power of the tested branch to below 60% of the rated power, so as to delay the insulation degradation process by reducing the heat load; Level 3 warning: corresponding to critical failure state, such as severe partial discharge or extremely high-speed thermal runaway state, the graded warning execution module outputs a hardware logic level signal, which triggers the locomotive main circuit breaker to perform an emergency trip action through a hard-wired circuit to achieve active safety protection. For example, the graded warning execution module directly drives the hardware relay to output a high-level trip signal to the locomotive main circuit breaker protection circuit to achieve millisecond-level rapid power cut-off and ensure the safety of locomotive and personnel.

[0062] When sending commands to the locomotive central control unit (TCMS), the graded early warning execution module can employ a dual-channel verification mechanism. The first channel transmits digital messages via the MVB bus, containing fault codes and detailed parameters; the second channel outputs a level signal via hardwiring as a backup for the trip command. This redundancy design ensures that the system's active safety protection functions can still be reliably executed when the communication network is busy or malfunctions.

[0063] In specific implementation cases, when the locomotive is under long-distance heavy-load traction conditions, the traction current... It The current remains above 1500A. At this point, the temperature of the converter busbar detected by the infrared visual sensor rises to 85 degrees Celsius. If the traditional warning threshold (e.g., 75 degrees Celsius) is applied, a high-temperature warning will be triggered. However, in the scheme of this embodiment, the dynamic fusion calculation module calculates that the steady-state temperature rise after 30 minutes of continuous application of 1500A current should be 85±3 degrees Celsius. Therefore, the calculated first compensation operator α1 lowers the "fault" confidence level of the infrared mode to an extremely low level. Since the ultrasonic sensor array does not detect a partial discharge signal higher than the background noise at this time, the fusion warning decision result ultimately remains in the "normal operating condition" state, thereby avoiding unnecessary emergency shutdown.

[0064] Conversely, if, under a light load condition with a traction current of only 500A, the infrared visual sensor detects the same high temperature of 85 degrees Celsius, and the ultrasonic sensor array monitors pulse discharge signals exceeding 50 times per second, the dynamic fusion calculation module will determine that the actual temperature rise far exceeds the physical model's prediction. In this case, the first compensation operator α1 will significantly increase the weight of the infrared evidence. Combined with the high-confidence support of acoustic evidence, the DS synthesis result will quickly converge to the "serious fault" proposition. The graded early warning execution module will immediately send an emergency trip signal to the TCMS via the MVB bus, limiting the impact of the fault to its nascent stage.

[0065] To verify the technical superiority of the monitoring method for electrical equipment in the embodiments of this application, the following demonstration is conducted through specific embodiments, comparative examples, and comparative analysis of experimental data.

[0066] In this embodiment, the electrical equipment monitoring system is installed inside the traction converter compartment of an electric locomotive, focusing on monitoring the cable joints on the output side of the converter. During a heavy-load operation test spanning 200 kilometers, the locomotive ran on a continuous steep gradient, and the traction current... It consistently maintains a high current of 1600A, far exceeding the rated current of 1200A.

[0067] As a comparative example, a traditional fixed-threshold early warning system was used, with an infrared early warning threshold set at 80 degrees Celsius. During the test, when the cable joint temperature rose to 86 degrees Celsius with increasing load, convective heat transfer was limited by the low train speed and could not dissipate heat in time. The comparative system immediately triggered a level two alarm, leading to unplanned load reduction on the locomotive.

[0068] The monitoring system using this embodiment of the application, with its operating condition sensing module, captured the 1600A current and low vehicle speed conditions in real time. The dynamic fusion calculation module calculated, using the Joule thermophysical model, that under these conditions, due to... R Value of temperature rise effect and small hThe theoretically expected temperature rise of the cable joint should be between 84 and 89 degrees Celsius. Therefore, the first compensation operator α1 calculated by the system significantly reduces the fault confidence of the infrared thermal radiation image. At this time, the amplitude of the partial discharge signal monitored by the ultrasonic sensor array remains at the background noise level (<20dB), and no abnormal discharge pulses are detected. After fusion with the improved DS evidence theory, the probability of the final output "normal" proposition is 0.92. Therefore, the monitoring system of this embodiment successfully identified the temperature rise as a normal load temperature rise and did not issue a false alarm.

[0069] Subsequently, when the locomotive entered a light-load coasting state (current only 200A), the system detected the same temperature rise of 86 degrees Celsius. At this time, since the physical model calculated an expected temperature rise of only 45 degrees Celsius, the actual temperature rise deviated significantly from the expected temperature rise. The first compensation operator immediately increased the weight of the infrared evidence. At the same time, the ultrasonic sensor array detected obvious ultrasonic pulses, and the PRPD spectrum showed typical internal insulation discharge characteristics. The two pieces of evidence were fused with high weights, quickly generating a level-three warning command. The main circuit breaker tripped within 120ms. Subsequent disassembly and inspection revealed severe insulation erosion inside the cable joint.

[0070] Table 1 below provides a detailed comparison of the performance data of the electrical equipment monitoring system of this application and the traditional monitoring system under different operating conditions: Table 1. Performance Comparison Test Data of the Early Warning System of the Present Invention and Traditional Systems

[0071] The data comparison in Table 1 clearly shows that in Scenario 1, the traditional system, unable to identify the physical logic of load and temperature rise, generated typical false alarms under heavy load; while in Scenario 2, under excessive phase electromagnetic shock, the traditional system could not distinguish between transient arc interference and actual partial discharge. The embodiments of this application, by introducing dynamic weight compensation operators α1 and α3, exhibit extremely high robustness in both of these key scenarios. Particularly in Scenario 3, the unique adaptive protection and cleaning module of this application, combined with the third compensation operator α3, solves the engineering problem of optical monitoring's susceptibility to failure in extreme and harsh environments.

[0072] It should be noted that the monitoring method for electrical equipment in this application is not only applicable to electric locomotives and EMUs, but can also be extended to the field of electrical safety assurance for urban rail transit and maglev trains, and has extremely high engineering application value and significant social benefits.

[0073] In terms of structural design, the embodiments of this application achieve accurate monitoring of the status of electrical equipment around the clock and under all operating conditions through the collaborative work of multiple modules; in terms of algorithm implementation, the decision confidence problem under strong interference environment is solved through dynamic weight compensation and improved evidence theory; in terms of engineering application, the long-term reliable operation of the early warning system is achieved through adaptive protection and hierarchical closed-loop control.

[0074] In summary, the electrical equipment monitoring method of this application dynamically determines the target weights of different modal characteristics based on the real-time operating status parameters of the locomotive. This achieves deep dynamic decoupling between the "locomotive physical dynamics conditions" and the "algorithm weights" at the algorithm level, thus achieving precise decoupling between fault characteristics and normal dynamic fluctuations. Under complex and variable locomotive dynamics conditions (such as load fluctuations and electromagnetic shocks), it effectively eliminates normal dynamic fluctuations (such as load fluctuations and electromagnetic shocks), significantly improving early warning accuracy and thereby enhancing the system's active protection capabilities. The optical window self-cleaning feedback mechanism effectively eliminates interference from severe weather, further improving early warning accuracy. This represents a leap from passive monitoring to active safety intervention, directly participating in locomotive power control and protection logic through a tiered early warning strategy, significantly reducing the probability of electrical equipment accidents.

[0075] Figure 4 This is a schematic diagram of the structure of a monitoring system for electrical equipment provided in one embodiment of this application. Figure 4 As shown, the electrical equipment monitoring system 400 of this application embodiment may specifically include: a multimodal sensing module 101, a working condition sensing module 201, and a dynamic fusion calculation module 301. Wherein: The multimodal sensing module 101 is used to acquire the thermal radiation distribution characteristics and insulation degradation acoustic characteristics of electrical equipment in the locomotive; The operating condition sensing module 201 is used to acquire the operating status parameters of the locomotive; The dynamic fusion calculation module 301 is used to determine the target weights of the thermal radiation distribution feature and the acoustic feature of insulation degradation respectively based on the operating status parameters; and to generate a fusion early warning decision result based on the thermal radiation distribution feature, the acoustic feature of insulation degradation and their respective target weights.

[0076] In this embodiment of the application, the specific process by which each module in the electrical equipment monitoring system implements its function can be found in the relevant descriptions in the above embodiments of the electrical equipment monitoring method, and will not be repeated here.

[0077] In summary, the electrical equipment monitoring system of this application dynamically determines the target weights of different modal characteristics based on the real-time operating status parameters of the locomotive. This achieves deep dynamic decoupling between the "locomotive physical dynamics conditions" and the "algorithm weights" at the algorithm level, thus achieving precise decoupling between fault characteristics and normal dynamic fluctuations. Under complex and variable locomotive dynamics conditions (such as load fluctuations and electromagnetic shocks), it effectively eliminates normal dynamic fluctuations (such as load fluctuations and electromagnetic shocks), significantly improving early warning accuracy and thereby enhancing the system's active protection capabilities. The optical window self-cleaning feedback mechanism effectively eliminates interference from severe weather, further improving early warning accuracy. This represents a leap from passive monitoring to active safety intervention, directly participating in locomotive power control and protection logic through a tiered early warning strategy, significantly reducing the probability of electrical equipment accidents.

[0078] This application also provides an electronic device. For example... Figure 5 As shown, the electronic device 500 can vary considerably due to differences in configuration or performance. It may include one or more processors 501 and memory 502, with memory 502 storing one or more programs or instructions. Memory 502 may be temporary or persistent storage. The application program stored in memory 502 may include one or more modules (not shown), each module including a series of computer-executable instructions for the electronic device 500. Furthermore, processor 501 may be configured to communicate with memory 502 and execute the series of computer-executable instructions stored in memory 502 on the electronic device 500. The electronic device 500 may also include one or more power supplies 503, one or more wired or wireless network interfaces 504, one or more input / output interfaces 505, and one or more keyboards 506.

[0079] Specifically, in the embodiments of this application, the electronic device includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor. When the program or instructions are executed by the processor, they implement the steps of any of the above-described embodiments of the monitoring method for electrical devices.

[0080] The electronic device in this application dynamically determines the target weights of different modal characteristics based on the real-time operating status parameters of the locomotive. This achieves deep dynamic decoupling between the "locomotive physical dynamics" and the "algorithm weights" at the algorithm level, thus precisely decoupling fault characteristics from normal dynamic fluctuations. Under complex and variable locomotive dynamics (e.g., load fluctuations, electromagnetic shocks), it effectively eliminates normal dynamic fluctuations (e.g., load fluctuations, electromagnetic shocks), significantly improving early warning accuracy and enhancing the system's active protection capabilities. Through an optical window self-cleaning feedback mechanism, it effectively eliminates interference from severe weather, further improving early warning accuracy. This represents a leap from passive monitoring to active safety intervention, directly participating in locomotive power control and protection logic through a tiered early warning strategy, significantly reducing the probability of electrical equipment accidents.

[0081] This application also proposes a readable storage medium storing one or more computer programs or instructions, which, when executed by a processor in an electronic device, enable the processor in the electronic device to perform the steps of any of the above-described embodiments of the monitoring method for electrical devices.

[0082] The readable storage medium of this application embodiment dynamically determines the target weights of different modal characteristics based on the real-time operating status parameters of the locomotive. This achieves deep dynamic decoupling between the "locomotive physical dynamics condition" and the "algorithm weights" at the algorithm level, thus achieving precise decoupling between fault characteristics and normal dynamic fluctuations. Under complex and variable locomotive dynamics conditions (such as load fluctuations and electromagnetic shocks), it effectively eliminates normal dynamic fluctuations (such as load fluctuations and electromagnetic shocks), significantly improving early warning accuracy and thereby enhancing the system's active protection capabilities. Through an optical window self-cleaning feedback mechanism, it effectively eliminates interference from severe weather, further improving early warning accuracy. This represents a leap from passive monitoring to active safety intervention, directly participating in locomotive power control and protection logic through a tiered early warning strategy, significantly reducing the probability of electrical equipment accidents.

[0083] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0084] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0085] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0086] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0087] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0088] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0089] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0090] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0091] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0092] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0093] This application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0094] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0095] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for monitoring electrical equipment, characterized in that, include: To obtain the thermal radiation distribution characteristics and acoustic characteristics of insulation degradation of electrical equipment in locomotives; Obtain the locomotive's operating status parameters; Based on the operating status parameters, the target weights for the thermal radiation distribution characteristics and the acoustic characteristics of insulation degradation are determined respectively. Based on the thermal radiation distribution characteristics, the acoustic characteristics of insulation degradation, and their respective target weights, a fusion early warning decision result is generated.

2. The method according to claim 1, characterized in that, The step of determining the target weights for the thermal radiation distribution characteristics and the acoustic characteristics of insulation degradation, respectively, based on the operating state parameters, includes: Based on the operating state parameters, a first compensation operator for characterizing load temperature rise deviation and a second compensation operator for characterizing transient voltage impact are calculated respectively. The target weight of the thermal radiation distribution feature is obtained by correcting the preset weight of the thermal radiation distribution feature according to the first compensation operator. The preset weights of the acoustic features of insulation degradation are corrected according to the second compensation operator to obtain the target weights of the acoustic features of insulation degradation.

3. The method according to claim 2, characterized in that, Calculating the first compensation operator based on the operating status parameters includes: When the real-time traction current is continuously exceeded by a preset traction current threshold and the duration is greater than a preset duration threshold, the expected temperature rise of the electrical equipment is calculated. The first compensation operator is generated based on the expected temperature rise. The first compensation operator is configured to nonlinearly reduce the preset weight of the thermal radiation distribution characteristics when the deviation between the actual temperature rise and the expected temperature rise is within a preset deviation threshold.

4. The method according to claim 2, characterized in that, The calculation of the second compensation operator based on the operating status parameters includes: When the locomotive is detected to be in a transient voltage surge condition caused by phase separation or pantograph disconnection, the grid voltage change rate of the pantograph is calculated. The second compensation operator is generated based on the grid voltage change rate. The second compensation operator is configured to increase the preset weight of the acoustic characteristics of insulation degradation when the grid voltage change rate exceeds a preset change rate threshold.

5. The method according to claim 2, characterized in that, Also includes: Obtain the dirt status of the optical window of the infrared vision sensor used to collect the thermal radiation distribution characteristics; When the optical window becomes contaminated beyond a preset limit, the adaptive protection and cleaning module is triggered to perform a physical cleaning action on the surface of the optical window of the infrared vision sensor.

6. The method according to claim 5, characterized in that, Also includes: Based on the state of contamination of the optical window, a third compensation operator is calculated to characterize the attenuation of the window signal; The step of correcting the preset weights of the thermal radiation distribution features according to the first compensation operator to obtain the target weights of the thermal radiation distribution features includes: The target weight of the thermal radiation distribution feature is obtained by correcting the preset weight of the thermal radiation distribution feature based on the first compensation operator and the third compensation operator.

7. The method according to claim 5, characterized in that, The step of obtaining the optical window contamination status of the infrared visual sensor used to collect the thermal radiation distribution characteristics includes: Extract the spatial correlation and edge intensity gradient of the infrared thermal image acquired by the infrared vision sensor; The optical window's contamination status is assessed in real time based on the gradient.

8. The method according to claim 1, characterized in that, The step of generating a fused early warning decision result based on the thermal radiation distribution characteristics, the acoustic characteristics of insulation degradation, and their respective target weights includes: The thermal radiation distribution characteristics, the acoustic characteristics of insulation degradation, and their respective target weights are input into the improved DS evidence theory fusion model for decision synthesis, and the fusion early warning decision result output by the improved DS evidence theory fusion model is obtained.

9. The method according to claim 1, characterized in that, The fusion-based early warning decision-making results include the target fault proposition and the corresponding reliability level; The method further includes: Based on the reliability level in the fusion early warning decision results, the corresponding early warning threshold level is determined; Based on the warning threshold level, the locomotive's actuators are controlled to execute corresponding intervention commands.

10. A monitoring system for electrical equipment, characterized in that, include: A multimodal sensing module is used to acquire the thermal radiation distribution characteristics and insulation degradation acoustic characteristics of electrical equipment in the locomotive; The operating condition sensing module is used to acquire the operating status parameters of the locomotive; The dynamic fusion calculation module is used to determine the target weights of the thermal radiation distribution feature and the acoustic feature of insulation degradation respectively based on the operating status parameters; and to generate a fusion early warning decision result based on the thermal radiation distribution feature, the acoustic feature of insulation degradation and their respective target weights.

Images