Overheat risk identification system and method for welding machine power module

Abstract

The invention relates to the technical field of state monitoring and fault early warning of welding equipment, and discloses an overheating risk identification system and method for a welding machine power module. The overheating risk identification method is applied to the monitoring equipment, and specifically comprises the following steps: S11, obtaining a monitoring instruction of a power module, collecting various operation parameters and environment parameters of the power module at the same time to form a dynamic monitoring data set, and performing timestamp alignment on the dynamic monitoring data set by adopting a synchronous clock, the parameter time consistency collected at the same time is ensured; and S12, acquiring a coupling analysis model which completes pre-training, performing normalization and weighting processing calculation on the dynamic monitoring data set in real time, and confirming the overheating risk level of the power module in the current state. According to the method, the temperature, the electrical parameters and the heat dissipation data are synchronously fused by adopting the multi-parameter coupling analysis model, and the heat dissipation efficiency attenuation factor is introduced to carry out trend prediction, so that the beforehand early warning function is realized, and the timeliness and accuracy of risk identification are improved.

Description

A system and method for identifying overheating risks in welding machine power modules Technical Field

[0001] This invention relates to the technical field of welding equipment condition monitoring and fault early warning, and more specifically, to an overheating risk identification system and method for welding machine power modules. Background Technology

[0002] In welding equipment, overheat protection for power modules mainly relies on relatively simple monitoring methods and passive response mechanisms. The most common technical solution is to install one or more temperature sensors on the heat sink or housing of the power module to directly measure its surface temperature. The system presets a fixed temperature threshold. When the monitored temperature reaches or exceeds the threshold, an alarm signal is triggered or the main circuit power is directly cut off to prevent the module from burning out due to overheating. Some improved solutions introduce monitoring of the cooling fan speed. When the temperature rises, the fan speed is increased to enhance heat dissipation. In addition, some more advanced welding machine designs monitor the output current and use it as one of the reference factors for calculating module heat generation. Combined with a simple thermal model, the junction temperature is estimated. However, these technologies have a single monitoring dimension, usually only focusing on individual parameters such as temperature or current. The judgment logic is also relatively simple and direct, lacking the assessment of the module's operating status and the causes of overheating risks.

[0003] The existing technical solutions suffer from several key technical problems: First, single temperature threshold alarms exhibit significant lag. Thermal resistance exists between the internal chip junction temperature and the casing temperature measurement point of the power module, requiring time for temperature transfer. By the time the sensor alarms, the chip may have already endured considerable overheating stress, resulting in irreversible performance damage or lifespan reduction. Second, relying on a single or few parameters easily leads to false alarms. For example, in low-temperature environments with short-term high-current operation, the module temperature may not exceed the limit, but the instantaneous electrical stress already poses a risk. Furthermore, the slow temperature rise due to the performance degradation of the cooling fan may remain undetected for a long time before reaching the fixed threshold, accumulating risk. Third, the lack of effective risk classification often necessitates forced shutdown after an alarm to cool the internal power module, failing to maintain production continuity while ensuring safety. Therefore, to address these issues, a solution capable of accurately and proactively identifying overheating risks is urgently needed for overheating risk identification in welding machine power modules. Summary of the Invention

[0004] The purpose of this invention is to provide an overheating risk identification system and method for welding machine power modules. By employing a multi-parameter coupled analysis model to simultaneously integrate temperature, electrical parameters and heat dissipation data, and introducing a heat dissipation efficiency attenuation factor for trend prediction, the invention achieves an early warning function, significantly improving the timeliness and accuracy of risk identification, and aims to solve the problems in the prior art.

[0005] This invention is implemented as follows: a method for identifying overheating risks of a welding machine power module, applied to monitoring equipment, specifically including the following steps: S11: Obtaining monitoring instructions from the power module, simultaneously collecting various operating parameters and environmental parameters of the power module to form a dynamic monitoring dataset, and using a synchronous clock to timestamp-align the dynamic monitoring dataset to ensure time consistency of parameters collected simultaneously; S12: Obtaining a pre-trained coupled analysis model, performing real-time normalization and weighted processing calculations on the dynamic monitoring dataset to confirm the overheating risk level of the power module under the current state, and dynamically correcting the predicted temperature rise rate based on the current trend of the overheating risk level, identifying potential overheating trends in the early stages of heat dissipation performance degradation; S13: Adaptively executing corresponding emergency protection according to graded responses, with each level of response action accompanied by log recording. The log records include response time, trigger parameters, execution actions, and subsequent module status tracking, forming a closed-loop control archive; S14: After each thermal risk level event occurs, the monitoring equipment automatically extracts the complete operating data sequence, environmental data sequence, and control command sequence within a set time window before and after the event, performs correlation analysis, and outputs the analysis results. The analysis results indicate the main causes, secondary factors, and related evidence parameters, and cluster the causes based on the analysis results. For power modules with the same causes, they will be marked as key monitoring objects; S15: Based on the thermal risk level event analysis result set, a remaining life prediction model for the power module is constructed. The remaining life prediction model uses a combination of degradation trajectory fitting and stress acceleration to calculate the life of the power module after the thermal risk level event, and outputs a health management step report and equipment maintenance plan for the power module.

[0006] Furthermore, in S11, the monitoring command of the power module is acquired, and various operating parameters and environmental parameters of the power module are collected simultaneously. This includes: receiving the monitoring command of the power module, which originates from manual triggering by the operator or a preset periodic monitoring plan; after parsing the command, determining the specific identifier of the module to be monitored, the type of parameters required for monitoring, and the initial data acquisition frequency; and synchronously activating various sensors deployed on the power module body and its surrounding environment according to the command requirements to perform high-speed parallel acquisition of multi-dimensional parameters. The operating parameters include the module surface temperature, estimated internal junction temperature, real-time operating current, real-time operating voltage, and switching frequency. The environmental parameters include the temperature difference between the heat sink inlet and outlet, the cooling fan speed, the coolant flow rate, and the ambient temperature and humidity.

[0007] Furthermore, the dynamic monitoring dataset is timestamped using a synchronous clock to ensure the time consistency of parameters collected simultaneously. This includes: using the IEEE 1588 protocol with a synchronous clock source to provide a unified time reference for all data acquisition channels; through strict alignment and verification of timestamps, ensuring that temperature, current, voltage, and wind speed parameters collected by sensors from different physical locations at the same absolute time point can be accurately correlated, forming a dynamic monitoring dataset with strict time consistency; for the strong electromagnetic interference generated by the power module of the high-frequency switch, a combination of hardware filtering and software algorithm is applied to the current and voltage signals to extract effective signal features that reflect the real load state; through multiple temperature sensors distributed on the power module housing, substrate, solder joints, and heat sink, combined with the periodic calibration of the infrared thermal imager, a two-dimensional temperature field distribution map is generated and updated in real time to display the hot spot location and heat dissipation path status; high-frequency acquisition of no less than 100Hz is used in the highly dynamic arc initiation, welding, and arc termination stages to capture transient features, while low-frequency acquisition of 1-10Hz is used in the stable standby stage.

[0008] Further, in S12, the pre-trained coupling analysis model is acquired, and the dynamic monitoring dataset is normalized and weighted in real time. This includes: calling the pre-trained multi-parameter coupling analysis model from the storage unit of the monitoring device, which receives the timestamp-aligned dynamic monitoring dataset as input in real time, normalizing each parameter, and then dynamically adjusting the weight coefficients of each parameter according to the specific type of the current power module and its long-term working history. Based on this, the multi-parameter coupling analysis model constructs a comprehensive risk assessment with real-time temperature as the risk core, current and voltage as dynamic load factors, and heat dissipation parameters as adjustment factors. The function determines the current overheating risk level by continuously calculating the real-time deviation between the comprehensive risk function value and the preset safety threshold. The multi-parameter coupled analysis model introduces a heat dissipation efficiency attenuation factor. This factor dynamically corrects the predicted value of the power module temperature rise rate by analyzing the historical trends of the cooling fan speed, air duct pressure difference, and coolant flow rate. It identifies potential overheating trends in the early stages of a slight decrease in heat dissipation performance. At the same time, it estimates the junction temperature fluctuation inside the power module chip by combining the transient thermal impedance curve of the power module with the real-time calculated switching losses, in order to capture the instantaneous overheating risk caused by short-term overload.

[0009] Furthermore, a comprehensive risk function is constructed, with real-time temperature as the core risk, current and voltage as dynamic load factors, and heat dissipation parameters as adjustment factors. This comprehensive risk function R(t) = core temperature risk term + dynamic electrical stress term - heat dissipation efficiency adjustment term + historical risk accumulation term; where the core temperature risk term = Dynamic electrical stress term = Heat dissipation performance adjustment item = Historical risk accumulation item = R(t) is the comprehensive overheating risk index calculated at time t. It is a dimensionless scalar and is directly triggered by comparing it with the preset warning threshold Rwarn, alarm threshold Ralarm, and protection threshold Rprotect. α(t), β(t), γ(t), and δ(t) are dynamic adaptive weighting coefficients. Their initial values ​​are preset according to the type of power module and are adjusted online adaptively based on the operating history during operation. T c (t) represents the real-time temperature of the key point obtained directly through multi-point temperature measurement; T j-max dT represents the maximum allowable junction temperature of the power module chip. c (t) / dt represents the instantaneous rate of change of temperature at the key point, used to capture the rapid upward trend of temperature; k1 and k2 are the normalization coefficients for temperature and rate of temperature rise; I(t), V ce (t) represents the real-time operating current and saturation voltage drop of the power module; I rated V rated The rated current and rated voltage of the power module; F sw (t) represents the real-time switching frequency; k f η is the switching frequency impact factor, used to quantify the additional losses caused by high-frequency switching and reflect the level of instantaneous power loss; cooling (t) is the heat dissipation efficiency decay factor, which is calculated based on historical trend data of cooling fan speed, air duct pressure difference, and coolant flow rate. When a slow decline in heat dissipation performance is detected, η cooling The value of (t) decreases from 1, causing the overall risk to rise earlier, thus achieving early warning; S(t) and F(t) are the real-time cooling fan speed and coolant flow rate; S rated ,F rated These correspond to the rated speed and rated flow rate; ω1 and ω2 are the weighting coefficients for air-cooled and liquid-cooled parameters; D history (t) is the historical risk density function, whose value is calculated by weighting the number of overheating warning events that have occurred recently, their duration, and their severity.

[0010] Furthermore, in S13, corresponding emergency protection is adaptively executed according to the graded response, including: when the risk level is the warning threshold Rwarn (Level 1), visual and auditory prompts are issued through the human-machine interface, and a snapshot of the current operating parameters is recorded to prompt maintenance personnel to conduct preventive checks, but without interfering with the normal operation of the equipment; when the risk level rises to the alarm threshold Ralarm (Level 2), the load adjustment strategy is automatically triggered, and the output current or duty cycle is dynamically limited through the welding machine main control unit to make the power module operating point deviate from the overheat critical zone, while strengthening the output of the heat dissipation system; when the risk level reaches the protection threshold Rprotect (Level 3), a power-off command is immediately generated, the power supply to the power module is cut off in the hardware protection circuit, and the fault state is locked, waiting for manual reset.

[0011] Furthermore, in S14, the analysis results indicate the primary cause, secondary factors, and related evidence parameters, and cluster the causes of the analysis results, including: the generated structured analysis report clearly distinguishes the primary and secondary factors leading to the overheating risk, and lists the key evidence parameters supporting the judgment and their abnormal values ​​one by one, and automatically clusters and labels the analysis results according to the type of cause, the subsystems involved, and the abnormal mode of the parameters; all relevant data of this risk event, including the complete analysis results and label categories, are associated and stored with the long-term operation archive of the power module, and the cumulative workload, thermal cycle count, and the changing trend of key electrical parameters of the power module over time are continuously recorded and updated.

[0012] Furthermore, in S15, the remaining life prediction model applies a combination of degradation trajectory fitting and stress acceleration to calculate the life of the power module after a thermal risk level event. This includes: using historical overheating risk event records as key stress inputs, using cumulative workload and thermal cycle counts as cumulative damage measures, and monitoring key parameter drift as a performance degradation characterization. Through fusion analysis of the monitoring dataset, quantitative health status indicators are dynamically calculated and output. These health status indicators include the current health index, predicted remaining life, recommended inspection cycle, and recommended replacement time window. The remaining life prediction model outputs life prediction results directly integrated into the equipment health management system and linked with the enterprise's computerized maintenance management system. When the system determines that the predicted remaining life of the power module is lower than a preset safety threshold, or that the health index shows an accelerated deterioration trend in the short term, the maintenance process is automatically triggered.

[0013] Compared with existing technologies, the overheating risk identification system and method for welding machine power modules provided by this invention have the following beneficial effects: 1. By adopting a multi-parameter coupled analysis model to simultaneously integrate temperature, electrical parameters and heat dissipation data, and introducing a heat dissipation efficiency decay factor for trend prediction, a pre-warning function is realized, significantly improving the timeliness and accuracy of risk identification. At the same time, the established three-level gradient response mechanism and closed-loop log archive overcome the shortcomings of traditional extreme measures such as power outages, and can maintain production continuity to the maximum extent while ensuring safety, forming a traceable risk control closed loop; 2. By constructing a risk knowledge graph-based cause clustering label and hybrid life prediction model, not only can the root cause of overheating be accurately located, but the remaining lifespan of the power module can also be probabilistically predicted, providing a scientific basis for preventive replacement. In addition, the introduction of cloud-edge collaborative visualized digital twin services enables the risk identification algorithm to be continuously iterated and optimized, and realizes remote, intuitive and interactive management of equipment status, thereby upgrading the single risk warning function into an active health management system covering the entire life cycle of the module, significantly improving the overall reliability, availability and maintainability of the equipment.

[0014] An overheating risk identification system for a welding machine power module, used to execute the aforementioned overheating risk identification method, the system includes: a data acquisition module for synchronously acquiring multi-dimensional operating parameters and environmental parameters of the power module to form a timestamp-aligned dynamic monitoring dataset; a risk assessment module for calling a pre-trained coupled analysis model to normalize and weight the dynamic monitoring dataset, assess and output the overheating risk level in real time; an early warning control module for executing graded response control and generating logs based on the overheating risk level; a risk tracing module for extracting multi-dimensional data sequences after a risk event for correlation analysis and outputting a structured report with causal clustering labels; and a lifespan prediction module for constructing a lifespan prediction model based on historical risk events and performance data, generating a health management report and linking it to a maintenance plan.

[0015] Specifically, the risk assessment module includes: a model calculation unit, used to run the coupled analysis model and perform data normalization and risk function calculation; and an adaptive adjustment unit, used to dynamically adjust the weight coefficients and heat dissipation efficiency attenuation factors in the model according to the power module type and working history, so as to achieve adaptive optimization of risk assessment parameters. Attached Figure Description

[0016] Figure 1 is a flowchart illustrating the overheating risk identification method for a welding machine power module proposed in this invention; Figure 2 is a flowchart illustrating the process of using a synchronous clock to timestamp-align the dynamic monitoring dataset in the overheating risk identification method for a welding machine power module proposed in this invention; Figure 3 is a structural schematic diagram of the overheating risk identification system for a welding machine power module proposed in this invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0019] In the accompanying drawings of this embodiment, the same or similar reference numerals correspond to the same or similar components. In the description of this invention, it should be understood that if terms such as "upper," "lower," "left," and "right" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting this invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0020] Referring to Figures 1-2, a method for identifying overheating risks of a welding machine power module, applied to monitoring equipment, specifically includes the following steps: S11: Obtaining monitoring commands from the power module, simultaneously collecting various operating parameters and environmental parameters of the power module to form a dynamic monitoring dataset, and aligning the timestamps of the dynamic monitoring dataset using a synchronous clock to ensure the time consistency of parameters collected simultaneously. This provides a high-quality data foundation with strict time synchronization for subsequent multi-parameter coupled analysis, avoiding analysis errors caused by data asynchrony; wherein, obtaining monitoring commands from the power module and simultaneously collecting various operating parameters and environmental parameters of the power module... This includes: receiving monitoring commands from the power module, which originate from manual triggering by the operator or a preset periodic monitoring plan; after parsing the commands, determining the specific identifier of the module to be monitored, the types of parameters required for monitoring, and the initial data acquisition frequency; synchronously activating various sensors deployed on the power module itself and its surrounding environment according to the command requirements, and performing high-speed parallel acquisition of multi-dimensional parameters, including module surface temperature, estimated internal junction temperature, real-time operating current, real-time operating voltage, and switching frequency; and environmental parameters including the temperature difference between the radiator inlet and outlet, cooling fan speed, coolant flow rate, and ambient temperature and humidity; S 12: The pre-trained coupled analysis model is acquired, and real-time normalization and weighting of the dynamic monitoring dataset are performed to calculate and confirm the overheating risk level of the power module under the current state. Based on the current trend of the overheating risk level, the predicted temperature rise rate is dynamically corrected. Potential overheating trends are identified in the early stages of declining heat dissipation performance. By fusing multi-dimensional electrical, thermal, and heat dissipation data and performing forward-looking trend correction, a paradigm shift from single-threshold alarms to multi-parameter proactive risk assessment is achieved, significantly improving the timeliness and accuracy of early warnings. Specifically, the pre-trained coupled analysis model is acquired, and real-time normalization and weighting of the dynamic monitoring dataset are performed. The weighting calculation includes: calling a pre-trained multi-parameter coupling analysis model from the storage unit of the monitoring equipment, which is used to receive the dynamic monitoring dataset aligned with timestamps in real time as input, normalizing each parameter, and then dynamically adjusting the weight coefficients of each parameter according to the specific type of the current power module and its long-term working history; on this basis, the multi-parameter coupling analysis model constructs a comprehensive risk function with real-time temperature as the risk core, current and voltage as dynamic load factors, and heat dissipation parameters as adjustment factors. By continuously calculating the real-time deviation of the comprehensive risk function value from the preset safety threshold, the current overheating risk level is determined.The multi-parameter coupled analysis model incorporates a heat dissipation efficiency attenuation factor. This factor dynamically corrects the predicted rate of temperature rise of the power module by analyzing the historical trends of cooling fan speed, airflow pressure difference, and coolant flow rate. It identifies potential overheating trends in the early stages of a slight decrease in heat dissipation performance. Simultaneously, by combining the transient thermal impedance curve of the power module with real-time calculated switching losses, it estimates the junction temperature fluctuations within the power module's chip to capture the risk of instantaneous overheating caused by short-term overload. A comprehensive risk function is constructed, with real-time temperature as the core risk, current and voltage as dynamic load factors, and heat dissipation parameters as adjustment factors. This function includes: Comprehensive Risk Function R(t) = Core Temperature Risk Term + Dynamic Electrical Stress Term - Heat Dissipation Efficiency Adjustment Term + Historical Risk Accumulation Term; where, the core temperature risk term = ; Dynamic electrical stress term = Heat dissipation performance adjustment item = Historical risk accumulation item = R(t) is the comprehensive overheating risk index calculated at time t. It is a dimensionless scalar and is directly triggered by comparing it with the preset warning threshold Rwarn, alarm threshold Ralarm, and protection threshold Rprotect. α(t), β(t), γ(t), and δ(t) are dynamic adaptive weighting coefficients. Their initial values ​​are preset according to the type of power module and are adjusted online adaptively based on the operating history during operation. T c (t) represents the real-time temperature of the key point obtained directly through multi-point temperature measurement; T j-max dT represents the maximum allowable junction temperature of the power module chip. c (t) / dt represents the instantaneous rate of change of temperature at the key point, used to capture the rapid upward trend of temperature; k1 and k2 are the normalization coefficients for temperature and rate of temperature rise; I(t), V ce (t) represents the real-time operating current and saturation voltage drop of the power module; I rated V rated The rated current and rated voltage of the power module; F sw (t) represents the real-time switching frequency; k f η is the switching frequency impact factor, used to quantify the additional losses caused by high-frequency switching and reflect the level of instantaneous power loss; cooling (t) is the heat dissipation efficiency decay factor, which is calculated based on historical trend data of cooling fan speed, air duct pressure difference, and coolant flow rate. When a slow decline in heat dissipation performance is detected, η cooling The value of (t) decreases from 1, causing the overall risk to rise earlier, thus achieving early warning; S(t) and F(t) are the real-time cooling fan speed and coolant flow rate; S rated ,F ratedThese correspond to the rated speed and rated flow rate; ω1 and ω2 are the weighting coefficients for air-cooled and liquid-cooled parameters; D history (t) is the historical risk density function, whose value is calculated by weighting the number, duration, and severity of recent overheating warning events; S13: Based on the graded response, corresponding emergency protection is adaptively executed, and each level of response action is accompanied by log recording. The log recording includes response time, trigger parameters, execution actions, and subsequent module status tracking, forming a closed-loop control archive. This gradient response mechanism achieves seamless connection from warning, adjustment to protection, maximizing the continuity of production operations while ensuring module safety, and forming a complete action archive that is traceable and auditable; S14: After each thermal risk level event occurs, the monitoring equipment automatically extracts the complete operating data sequence, environmental data sequence, and control command sequence within the set time window before and after the event, performs correlation analysis, and outputs the analysis results. The analysis results indicate the main causes, secondary factors, and related evidence parameters, and clusters the causes based on the analysis results. For the same causes in power modules, they will be marked as key monitoring. The controlled object's in-depth source tracing analysis function goes beyond simple alarm recording, accurately locating the root cause of the fault and accumulating knowledge, providing direct data support for optimizing maintenance strategies and making targeted improvements; S15: Based on the set of thermal risk level event analysis results, a remaining life prediction model for the power module is constructed. The remaining life prediction model uses a combination of degradation trajectory fitting and stress acceleration to calculate the life of the power module after a thermal risk level event, outputting a health management step report and equipment maintenance plan for the power module, linking discrete risk events with long-term performance degradation, realizing a leap from short-term risk warning to long-term life prediction, providing a scientific basis for predictive maintenance and spare parts management, and helping to achieve full life cycle health management of assets. By adopting a multi-parameter coupled analysis model to simultaneously integrate temperature, electrical parameters and heat dissipation data, and introducing a heat dissipation efficiency decay factor for trend prediction, a pre-warning function is realized, significantly improving the timeliness and accuracy of risk identification.

[0021] In S11 of this embodiment, the dynamic monitoring dataset is timestamped using a synchronous clock to ensure the time consistency of parameters collected at the same time. This includes: using the IEEE 1588 protocol with a synchronous clock source to provide a unified time reference for all data acquisition channels; through strict alignment and verification of timestamps, ensuring that temperature, current, voltage, and wind speed parameters collected by sensors from different physical locations at the same absolute time point can be accurately correlated, forming a dynamic monitoring dataset with strict time consistency; for the strong electromagnetic interference generated by the power module of the high-frequency switch, a combination of hardware filtering and software algorithm is applied to the current and voltage signals to extract effective signal features that reflect the real load state; through multiple temperature sensors distributed on the power module housing, substrate, solder joints, and heat sink, combined with the periodic calibration of the infrared thermal imager, a two-dimensional temperature field distribution map is generated and updated in real time to display the hot spot location and heat dissipation path status; high-frequency acquisition of no less than 100Hz is used to capture transient features during the highly dynamic arc initiation, welding, and arc termination stages, while low-frequency acquisition of 1-10Hz is used during the stable standby stage.

[0022] In S13 of this embodiment, the corresponding emergency protection is adaptively executed according to the graded response, including: when the risk level is the warning threshold Rwarn of the first-level warning, visual and auditory prompts are issued through the human-machine interface, and a snapshot of the current operating parameters is recorded to prompt the operation and maintenance personnel to perform preventive checks, but without interfering with the normal operation of the equipment; when the risk level rises to the alarm threshold Ralarm of the second-level alarm, the load adjustment strategy is automatically triggered, and the output current or duty cycle is dynamically limited through the welding machine main control unit to make the power module operating point deviate from the overheat critical zone, while strengthening the output of the heat dissipation system; when the risk level reaches the protection threshold Rprotect of the third-level emergency, a power-off command is immediately generated, the power supply to the power module is cut off in the hardware protection circuit, and the fault state is locked, waiting for manual reset.

[0023] In S14 of this embodiment, the analysis results indicate the primary cause, secondary factors, and related evidence parameters, and cluster the causes of the analysis results. This includes: the generated structured analysis report clearly distinguishes the primary and secondary factors leading to the overheating risk, and lists the key evidence parameters supporting the judgment and their abnormal values ​​one by one. It also automatically clusters and labels the analysis results according to the type of cause, the subsystems involved, and the abnormal mode of the parameters. All relevant data of this risk event, including the complete analysis results and label categories, are associated and stored with the long-term operation archive of the power module, and the cumulative workload, thermal cycle count, and the changing trend of key electrical parameters such as saturation voltage drop of the power module over time are continuously recorded and updated.

[0024] In S15 of this embodiment, the remaining life prediction model applies a combination of degradation trajectory fitting and stress acceleration to calculate the life of the power module after a thermal risk level event. This includes: using historical overheating risk event records as key stress inputs, using cumulative workload and thermal cycle count as cumulative damage measures, and monitoring key parameter drift as a performance degradation characterization. By fusing and analyzing the monitoring dataset, the model dynamically calculates and outputs quantitative health status indicators, including the current health index, predicted remaining life, recommended inspection cycle, and recommended replacement time window. The remaining life prediction model outputs life prediction results directly integrated into the equipment health management system and linked with the enterprise's computerized maintenance management system. When the system determines that the predicted remaining life of the power module is lower than a preset safety threshold, or that the health index shows an accelerated deterioration trend in the short term, the maintenance process is automatically triggered.

[0025] This technical solution overcomes the shortcomings of traditional extreme measures such as power outages by establishing a three-level gradient response mechanism and closed-loop log archives. It can maintain production continuity to the maximum extent while ensuring safety, and forms a traceable risk control closed loop.

[0026] Referring to Figure 3, an overheating risk identification system for a welding machine power module is used to execute the overheating risk identification method described above. The system includes: a data acquisition module for synchronously acquiring multi-dimensional operating parameters and environmental parameters of the power module to form a timestamp-aligned dynamic monitoring dataset; a risk assessment module for calling a pre-trained coupled analysis model to normalize and weight the dynamic monitoring dataset, assess and output the overheating risk level in real time; an early warning control module for executing graded response control and generating logs based on the overheating risk level; a risk tracing module for extracting multi-dimensional data sequences after a risk event for correlation analysis and outputting a structured report with causal clustering labels; and a lifespan prediction module for constructing a lifespan prediction model based on historical risk events and performance data, generating a health management report and linking it to a maintenance plan. By constructing a causal clustering label and a hybrid lifespan prediction model based on a risk knowledge graph, the system can not only accurately locate the root cause of overheating but also probabilistically predict the remaining lifespan of the power module, providing a scientific basis for preventative replacement decisions.

[0027] In this embodiment, the risk assessment module includes: a model calculation unit for running a coupled analysis model and performing data normalization and risk function calculation; and an adaptive adjustment unit for dynamically adjusting the weight coefficients and heat dissipation efficiency attenuation factor in the model according to the power module type and working history, thereby achieving adaptive optimization of risk assessment parameters. By using a multi-parameter coupled analysis model to simultaneously integrate temperature, electrical parameters and heat dissipation data, and introducing a heat dissipation efficiency attenuation factor for trend prediction, a pre-warning function is achieved, significantly improving the timeliness and accuracy of risk identification.

[0028] This technical solution introduces a cloud-edge collaborative visualized digital twin service, enabling continuous iteration and optimization of risk identification algorithms and achieving remote, intuitive, and interactive management of device status. This transforms a single risk warning function into a proactive health management system covering the entire lifecycle of the module, significantly improving the overall reliability, availability, and maintainability of the equipment.

[0029] In this embodiment, the entire operation process can be automated by computer control, and sensors are set up in each operation stage to provide signal feedback and ensure that the steps are performed in sequence. These are all conventional knowledge of current automation control, and will not be elaborated on in this embodiment.

[0030] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for identifying overheating risks in a welding machine power module, characterized in that, The method, applied to monitoring equipment, specifically includes the following steps: S11: Obtain monitoring instructions from the power module, simultaneously collect various operating parameters and environmental parameters of the power module to form a dynamic monitoring dataset, and use a synchronous clock to timestamp-align the dynamic monitoring dataset to ensure the time consistency of parameters collected simultaneously; S12: Obtain the coupled analysis model with the pre-trained model, perform real-time normalization and weighted processing calculations on the dynamic monitoring dataset, confirm the overheating risk level of the power module under the current state, and dynamically correct the predicted temperature rise rate based on the current trend of the overheating risk level, identifying potential overheating trends in the early stages of heat dissipation performance degradation; S13: Adaptively execute corresponding emergency protection according to the graded response, and each level of response action is accompanied by log recording, the log recording including response time, trigger parameters, etc. S14: After each thermal risk level event occurs, the monitoring equipment automatically extracts the complete operating data sequence, environmental data sequence, and control command sequence within a set time window before and after the event, performs correlation analysis, and outputs the analysis results. The analysis results indicate the main causes, secondary factors, and related evidence parameters, and cluster the causes based on the analysis results. For power modules with the same causes, they will be marked as key monitoring objects. S15: Based on the thermal risk level event analysis result set, a remaining life prediction model for the power module is constructed. The remaining life prediction model uses a combination of degradation trajectory fitting and stress acceleration to calculate the life of the power module after the thermal risk level event, and outputs a health management step report and equipment maintenance plan for the power module.

2. The overheating risk identification method for a welding machine power module as described in claim 1, characterized in that, In S11, the monitoring command of the power module is acquired, and various operating parameters and environmental parameters of the power module are collected simultaneously. This includes: receiving the monitoring command of the power module, which originates from manual triggering by the operator or a preset periodic monitoring plan; after parsing the command, determining the specific identifier of the module to be monitored, the type of parameters required for monitoring, and the initial data acquisition frequency; and synchronously activating various sensors deployed on the power module body and its surrounding environment according to the command requirements to perform high-speed parallel acquisition of multi-dimensional parameters. The operating parameters include the module surface temperature, estimated internal junction temperature, real-time operating current, real-time operating voltage, and switching frequency. The environmental parameters include the temperature difference between the heat sink inlet and outlet, the cooling fan speed, the coolant flow rate, and the ambient temperature and humidity.

3. The overheating risk identification method for a welding machine power module as described in claim 2, characterized in that, Furthermore, the dynamic monitoring dataset is timestamped using a synchronous clock to ensure the time consistency of parameters collected simultaneously. This includes: using the IEEE 1588 protocol with a synchronous clock source to provide a unified time reference for all data acquisition channels; through strict alignment and verification of timestamps, ensuring that temperature, current, voltage, and wind speed parameters collected by sensors from different physical locations at the same absolute time point can be accurately correlated, forming a dynamic monitoring dataset with strict time consistency; for the strong electromagnetic interference generated by the power module of the high-frequency switch, a combination of hardware filtering and software algorithm is applied to the current and voltage signals to extract effective signal features reflecting the real load state; through multiple temperature sensors distributed on the power module housing, substrate, solder joints, and heat sink, combined with the periodic calibration of the infrared thermal imager, a two-dimensional temperature field distribution map is generated and updated in real time to display the hot spot location and heat dissipation path status; high-frequency acquisition of no less than 100Hz is used in the highly dynamic arc initiation, welding, and arc termination stages to capture transient features, while low-frequency acquisition of 1-10Hz is used in the stable standby stage.

4. The overheating risk identification method for a welding machine power module as described in claim 3, characterized in that, In S12, the pre-trained coupling analysis model is acquired, and the dynamic monitoring dataset is normalized and weighted in real time. This includes: calling the pre-trained multi-parameter coupling analysis model from the storage unit of the monitoring device, which receives the timestamp-aligned dynamic monitoring dataset as input in real time, normalizing each parameter, and then dynamically adjusting the weight coefficients of each parameter according to the specific type of the current power module and its long-term working history; based on this, the multi-parameter coupling analysis model constructs a comprehensive risk function with real-time temperature as the risk core, current and voltage as dynamic load factors, and heat dissipation parameters as adjustment factors. The system continuously calculates the real-time deviation between the comprehensive risk function value and the preset safety threshold to determine the current overheating risk level. The multi-parameter coupled analysis model introduces a heat dissipation efficiency attenuation factor. This factor dynamically corrects the predicted value of the power module temperature rise rate by analyzing the historical trends of the cooling fan speed, air duct pressure difference, and coolant flow rate. It identifies potential overheating trends in the early stages of a slight decrease in heat dissipation performance. At the same time, it combines the transient thermal impedance curve of the power module with the real-time calculated switching losses to estimate the junction temperature fluctuation inside the power module chip, thereby capturing the instantaneous overheating risk caused by short-term overload.

5. The overheating risk identification method for a welding machine power module as described in claim 4, characterized in that, A comprehensive risk function is constructed, with real-time temperature as the core risk, current and voltage as dynamic load factors, and heat dissipation parameters as adjustment factors. This comprehensive risk function R(t) = core temperature risk term + dynamic electrical stress term - heat dissipation efficiency adjustment term + historical risk accumulation term; where, the core temperature risk term = Dynamic electrical stress term = Heat dissipation performance adjustment item = Historical risk accumulation item = R(t) is the comprehensive overheating risk index calculated at time t. It is a dimensionless scalar and is directly triggered by comparing it with the preset warning threshold Rwarn, alarm threshold Ralarm, and protection threshold Rprotect. α(t), β(t), γ(t), and δ(t) are dynamic adaptive weighting coefficients. Their initial values ​​are preset according to the type of power module and are adjusted online adaptively based on the operating history during operation. T c (t) represents the real-time temperature of the key point obtained directly through multi-point temperature measurement; T j-max dT represents the maximum allowable junction temperature of the power module chip. c (t) / dt represents the instantaneous rate of change of temperature at the key point, used to capture the rapid upward trend of temperature; k1 and k2 are the normalization coefficients for temperature and rate of temperature rise; I(t), V ce (t) represents the real-time operating current and saturation voltage drop of the power module; I rated V rated The rated current and rated voltage of the power module; F sw (t) represents the real-time switching frequency; k f η is the switching frequency impact factor, used to quantify the additional losses caused by high-frequency switching and reflect the level of instantaneous power loss; cooling (t) is the heat dissipation efficiency decay factor, which is calculated based on historical trend data of cooling fan speed, air duct pressure difference, and coolant flow rate. When a slow decline in heat dissipation performance is detected, η cooling The value of (t) decreases from 1, causing the overall risk to rise earlier, thus achieving early warning; S(t) and F(t) are the real-time cooling fan speed and coolant flow rate; S rated ,F rated These correspond to the rated speed and rated flow rate; ω1 and ω2 are the weighting coefficients for air-cooled and liquid-cooled parameters; D history (t) is the historical risk density function, whose value is calculated by weighting the number of overheating warning events that have occurred recently, their duration, and their severity.

6. The overheating risk identification method for a welding machine power module as described in claim 5, characterized in that, In S13, corresponding emergency protection is adaptively executed according to the graded response, including: when the risk level is the warning threshold Rwarn (Level 1), visual and auditory prompts are issued through the human-machine interface, and a snapshot of the current operating parameters is recorded to prompt maintenance personnel to perform preventive checks, but without interfering with the normal operation of the equipment; when the risk level rises to the alarm threshold Ralarm (Level 2), the load adjustment strategy is automatically triggered, and the output current or duty cycle is dynamically limited through the welding machine main control unit to make the power module operating point deviate from the overheat critical zone, while strengthening the output of the heat dissipation system; when the risk level reaches the protection threshold Rprotect (Level 3), a power-off command is immediately generated, the power supply to the power module is cut off in the hardware protection circuit, and the fault state is locked, waiting for manual reset.

7. The overheating risk identification method for a welding machine power module as described in claim 6, characterized in that, In S14, the analysis results indicate the primary cause, secondary factors, and related evidence parameters, and cluster the causes of the analysis results. This includes: the generated structured analysis report clearly distinguishes between the primary and secondary factors leading to the overheating risk, and lists the key evidence parameters supporting the judgment and their abnormal values ​​one by one. It also automatically clusters and labels the analysis results based on the type of cause, the subsystems involved, and the abnormal parameter patterns. All relevant data of this risk event, including the complete analysis results and label categories, are associated and stored with the long-term operation records of the power module, and the cumulative workload, thermal cycle count, and the changing trends of key electrical parameters of the power module over time are continuously recorded and updated.

8. The overheating risk identification method for a welding machine power module as described in claim 7, characterized in that, In S15, the remaining life prediction model applies a combination of degradation trajectory fitting and stress acceleration to calculate the life of the power module after a thermal risk level event. This includes: using historical overheating risk event records as key stress inputs, using cumulative workload and thermal cycle counts as cumulative damage measures, and monitoring key parameter drift as a performance degradation characterization. By fusing and analyzing the monitoring dataset, the model dynamically calculates and outputs quantitative health status indicators, including the current health index, predicted remaining life, recommended inspection cycle, and recommended replacement time window. The remaining life prediction model's output life prediction results are directly integrated into the equipment health management system and linked with the enterprise's computerized maintenance management system. When the system determines that the predicted remaining life of the power module is lower than a preset safety threshold, or that the health index shows an accelerated deterioration trend in the short term, the maintenance process is automatically triggered.

9. An overheating risk identification system for a welding machine power module, characterized in that, The system, used to execute the overheating risk identification method according to any one of claims 1-8, comprises: a data acquisition module for synchronously acquiring multi-dimensional operating parameters and environmental parameters of the power module to form a timestamp-aligned dynamic monitoring dataset; a risk assessment module for calling a pre-trained coupled analysis model to normalize and weight the dynamic monitoring dataset, and to assess and output the overheating risk level in real time; an early warning control module for executing graded response control and generating logs according to the overheating risk level; a risk tracing module for extracting multi-dimensional data sequences for correlation analysis after a risk event and outputting a structured report with causal clustering labels; and a lifespan prediction module for constructing a lifespan prediction model based on historical risk events and performance data, generating a health management report, and linking it to a maintenance plan.

10. The overheating risk identification system for a welding machine power module as described in claim 9, characterized in that, The risk assessment module includes: a model calculation unit, used to run the coupled analysis model and perform data normalization and risk function calculation; and an adaptive adjustment unit, used to dynamically adjust the weight coefficients and heat dissipation efficiency attenuation factor in the model according to the power module type and working history, so as to achieve adaptive optimization of risk assessment parameters.

Images