A power unit fault adaptive bypass control method and system

Abstract

This invention discloses a power unit fault adaptive bypass control method and system, relating to the field of bypass control technology. It includes: acquiring real-time parameter information for each power unit to obtain a power unit information set, the power unit information set including at least one power unit information item; acquiring bypass information corresponding to the power unit, including bypass quantity information and bypass load information, to obtain a unit bypass item. Based on the combination of real-time fault level items and predicted fault level items, this invention preheats the unit bypass items, including voltage preheating and current preheating, generating unit preheating items, setting preheating adjustment level items, and matching load preheating values ​​according to the fault level to dynamically adjust the bypass state. This ensures that at the moment of actual power unit fault, the bypass is already in a semi-active state and can immediately take over the load, avoiding switching delays or impacts.

Description

Technical Field

[0001] This invention relates to the field of bypass control technology, specifically to an adaptive bypass control method and system for power unit faults. Background Technology

[0002] A power unit refers to the basic power module that makes up a large power electronic converter, such as a multilevel inverter or a modular multilevel converter (MMC). Each unit typically contains semiconductor switching devices, such as IGBTs, drive circuits, capacitors, and heat sinks, and is responsible for completing a portion of the power conversion task. It is the core component of the entire device. Multiple power units are connected in series or parallel to achieve high-voltage, high-power power conversion. When a power unit fails, it may become unable to work normally due to various reasons such as overvoltage, overcurrent, overheating, component aging, or drive failure. This may lead to equipment shutdown and affect system operation. Timely detection and handling of faults are crucial to ensuring system reliability. Bypass refers to providing an alternative path in the circuit to bypass the normal operating path of a faulty component or unit. The purpose is to remove the faulty unit from the main current path to avoid affecting the overall operation or causing greater damage. This is usually achieved by setting a bypass switch inside or outside the power unit. When a power unit fails, a bypass is needed to resolve the issue.

[0003] The DC-DC circuit with bypass function and the bypass control method disclosed in patent publication number CN118381293A obtain a first voltage value by acquiring the voltage value of the first side of the DC-DC circuit during a first preset time; determine the target set value corresponding to the second side of the DC-DC circuit; determine the voltage difference between the first voltage value and the target set value to obtain a first voltage difference; when the first voltage difference is less than the first preset voltage difference, determine the duration for which the first voltage difference in the DC-DC circuit is less than the first preset voltage difference to obtain a first duration; when the first duration is greater than the first preset duration, control the DC-DC circuit to enter the bypass working state. In this way, when the deviation between the voltage on the first side of the DC-DC circuit (i.e., the low-voltage side voltage) and the target set value (the voltage set value on the high-voltage side) is less than the first preset voltage difference, the DC-DC circuit can be controlled to enter the bypass working state, so that the voltage on the first side and the voltage on the second side of the DC-DC circuit are equal. This effectively avoids the problem of unstable operation of the DC-DC circuit due to the small deviation between the target set value and the first side voltage, and improves the stability of the DC-DC circuit.

[0004] When the above-mentioned and similar technical solutions make bypass selection, traditional fault detection methods usually rely on real-time monitoring of key parameters such as current, voltage, and temperature, and comparing them with preset thresholds. When the monitored value exceeds or falls below the threshold, the system determines that a fault has occurred and activates the corresponding protection measures. However, the bypass is only activated after the fault occurs, which is a passive remedy. Therefore, the passive response bypass decision will be further delayed, and the early warning information before the fault occurs cannot be fully utilized. Summary of the Invention

[0005] The purpose of this invention is to provide a power unit fault adaptive bypass control method and system to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a power unit fault adaptive bypass control method, comprising:

[0007] Obtain real-time parameter information for each power unit to obtain a power unit information set, which includes at least one power unit information item corresponding to a power unit.

[0008] Obtain the bypass information corresponding to the power unit, including the number of bypasses and the bypass load information, to obtain the unit bypass item;

[0009] A digital twin model of the power unit is created based on the power unit information set, resulting in the creation of model items. At the same time, the historical parameter information of the power unit is obtained, resulting in the historical information set of the power unit. The historical parameter information includes historical faults and fault parameters corresponding to the historical faults.

[0010] Based on the historical information set of the power unit, at least two fault levels are created, each fault level corresponding to fault parameters of different intensity ranges, to obtain fault level items, obtain the matching fault level of the power unit, and then obtain real-time fault level items.

[0011] Based on the matching results between the created model items and the historical information set of the power unit, the predicted fault level of the power unit is output, and the predicted fault level item is obtained. Based on the combination of the real-time fault level item and the predicted fault level item, the unit bypass item is preheated, including voltage preheating and current prediction, and the unit preheating item is obtained. Thus, bypass selection and bypass preheating based on the joint analysis of digital twin model and real-time data are realized.

[0012] Furthermore, the real-time parameter information of the power unit includes voltage parameters, current parameters, temperature parameters, and switching state parameters. The method for obtaining the power unit information set includes:

[0013] The system consists of a data acquisition layer, an edge processing layer, and a platform layer. Voltage and current are sampled based on the data acquisition layer to obtain the first sampling item. Temperature is sampled based on the data acquisition layer to obtain the second sampling item. Switching states are sampled based on the data acquisition layer to obtain the third sampling item, thus obtaining the sampled dataset.

[0014] The sampling dataset is timestamped based on the edge processing layer to obtain a data synchronization set. The platform layer creates a real-time dashboard to display and store the data synchronization set, thereby obtaining the power unit information set.

[0015] Furthermore, the method for obtaining the unit bypass term includes:

[0016] Obtain the bypass quantity information corresponding to the power unit to obtain bypass information items. Perform bypass tests based on the bypass information items. Based on the bypass test results, eliminate non-functional bypasses to obtain filtered bypass items.

[0017] By filtering bypass items, bypass attribute information is obtained, including specification information and the corresponding load limit information, and then bypass load items are obtained.

[0018] Furthermore, the method for obtaining the model item includes:

[0019] The system is structured into a network layer, a platform layer, and an application layer. The network layer enables low-latency data transmission and synchronization between physical units and virtual models. The platform layer supports the construction, training, simulation, and updating of digital twin models, providing data governance and computational support. The application layer provides users with visual monitoring, early warning, and fault diagnosis interaction.

[0020] Behavioral simulation is performed based on physical laws. Based on Ohm's law for circuits, the heat conduction equation, and the power loss formula, a mathematical model of the power unit is established, a simulation model is built, the power unit information set is input, and the simulation results of temperature change and power loss are output and compared with the power unit information set.

[0021] Historical parameter information of the power unit is obtained, an LSTM deep learning model is trained, and edge computing is used to align the data between the power unit and the virtual model. A closed loop of data acquisition, model update, and visualization feedback is established to obtain the model creation item.

[0022] Furthermore, the method for obtaining the historical information set of the power unit includes:

[0023] Set an acquisition threshold, which is a fixed period value. Based on the acquisition threshold, acquire the historical fault names of the power unit to obtain historical fault items.

[0024] Based on the historical fault items, the fault parameters corresponding to the historical fault items are obtained, and then the fault parameter items are obtained. The fault parameter items are matched with the historical fault items, and then the historical information set of the power unit is obtained.

[0025] Furthermore, the method for obtaining the fault level item includes:

[0026] Based on the extreme value data of fault parameter fluctuations in the historical information set of the power unit, the parameter fluctuation difference is obtained, and then the parameter difference item is obtained.

[0027] Based on the parameter difference term, the fault parameters are divided into intervals to obtain at least two fault parameter intervals. The fault parameter intervals are used as different fault levels to obtain the fault level term.

[0028] Furthermore, the method for obtaining the real-time fault level item includes:

[0029] The fluctuation amplitude of the power unit information set is obtained to obtain the real-time fluctuation item. The real-time fluctuation item is matched with the fault level item to obtain the matching comparison data, and then the real-time fault level item is obtained.

[0030] Furthermore, the method for obtaining the predicted fault level item includes:

[0031] Using the power unit information set as data input, the predicted parameter data is output by creating model terms to obtain output data terms. Based on the matching results between the output data terms and the historical information set of the power unit, the predicted fault level of the power unit is obtained, resulting in the predicted fault level term.

[0032] Furthermore, the method for obtaining the unit preheating term includes:

[0033] At least two preheating adjustment level items are set, each corresponding to a different load preheating value. Based on the combination result of the real-time fault level item and the predicted fault level item, the pre-adjustment level corresponding to the preheating adjustment level item is obtained, and then the matching load preheating value is obtained. The unit bypass is preheated based on the load preheating value, and then the unit preheating item is obtained.

[0034] A power unit fault adaptive bypass control system, using the aforementioned power unit fault adaptive bypass control method, includes:

[0035] Data acquisition module: Acquires real-time parameter information for each power unit, obtains a power unit information set, which includes at least one power unit information item corresponding to a power unit, acquires bypass information corresponding to the power unit, including bypass quantity information and bypass load information, and obtains unit bypass item;

[0036] Model creation module: Creates a digital twin model of the power unit based on the power unit information set, obtains the model creation item, and obtains the historical parameter information of the power unit to obtain the historical information set of the power unit. The historical parameter information includes historical faults and fault parameters corresponding to the historical faults.

[0037] Fault diagnosis module: Based on the historical information set of the power unit, at least two fault levels are created. Each fault level corresponds to fault parameters with different intensity ranges to obtain fault level items. The matching fault level of the power unit is obtained, and then the real-time fault level items are obtained.

[0038] Bypass preheating module: Based on the matching results of the created model items and the historical information set of the power unit, the predicted fault level of the power unit is output to obtain the predicted fault level item. Based on the combination of the real-time fault level item and the predicted fault level item, the bypass item of the unit is preheated, including voltage preheating and current prediction, to obtain the unit preheating item.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] This power unit fault adaptive bypass control method and system, by introducing a digital twin model and historical data analysis, solves the problem that traditional passive response strategies cannot capture the gradual fault process. Combining physical law simulation and deep learning models, it achieves dynamic simulation and prediction of the power unit's operating state. The model is based on matching real-time power unit information set and historical parameter information to output the predicted fault level. This allows the system to identify abnormal patterns before the fault actually occurs. Compared with traditional methods, this mechanism significantly improves the early warning capability. The system no longer relies solely on threshold triggering, but uses a data-driven prediction model to discover potential problems in advance, thereby initiating preventive measures before the fault worsens. This proactive prediction reduces the risk of system downtime and improves the overall fault tolerance efficiency.

[0041] Simultaneously, based on the combined results of real-time fault level items and predicted fault level items, the unit bypass items are preheated, including voltage preheating and current preheating, generating unit preheating items, setting preheating adjustment level items, and matching load preheating values ​​according to fault levels to dynamically adjust the bypass state. This ensures that at the moment of actual power unit failure, the bypass is already in a semi-activated state and can immediately take over the load, avoiding switching delays or impacts. In addition, the preheating process also includes fault detection, and bypasses that cannot be preheated are eliminated, further improving redundancy reliability. Compared with traditional threshold triggering strategies, this adaptive preheating mechanism utilizes the collaborative analysis of digital twin models and historical data, which not only reduces bypass activation time but also optimizes resource utilization and prevents overload. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the data acquisition module of the present invention;

[0043] Figure 2 This is a schematic diagram of the model creation module process of the present invention;

[0044] Figure 3 This is a schematic diagram of the fault diagnosis module of the present invention;

[0045] Figure 4 This is a schematic diagram of the bypass preheating module of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Bypass selection, as an important fault-tolerance mechanism, aims to ensure the continuous and stable operation of a system by switching to a backup path or redundant module when a part of the system fails. Traditional fault detection methods play a crucial role in bypass selection, with the goal of timely and accurate identification of faults in the system, thereby triggering corresponding bypass strategies. However, existing fault detection methods, especially those that rely on real-time monitoring of key parameters and comparison with preset thresholds, have significant limitations in bypass selection applications. Traditional fault detection methods typically rely on real-time monitoring of key parameters such as current, voltage, and temperature. These parameters reflect the operating status of the system and are considered early indicators of fault occurrence. When the monitored values ​​exceed or fall below preset thresholds, the system determines that a fault has occurred and initiates corresponding protective measures, including triggering bypass selection. However, this passive response strategy cannot fully utilize early warning information before a fault occurs. In many cases, the occurrence of a fault is not sudden but rather a gradual process. During this process, some parameters of the system may exhibit abnormalities. While current or temperature parameters fluctuate but haven't reached a set threshold, this application provides an adaptive bypass control method for power unit faults. By introducing a digital twin model and historical data analysis, it overcomes the limitation of traditional passive response strategies in capturing the gradual fault progression. Combining physical law simulation and deep learning models, it achieves dynamic simulation and prediction of the power unit's operating state. This model matches real-time power unit information with historical parameter information to output a predicted fault level. This allows the system to identify abnormal patterns before a fault actually occurs. For example, when current or temperature parameters fluctuate gradually but don't exceed a threshold, the model can predict the fault level that may occur in the next 5 minutes or 1 hour. Compared to traditional methods, this mechanism significantly improves early warning capabilities. The system no longer relies solely on threshold triggering but utilizes a data-driven predictive model to detect potential problems in advance. For example, in IGBT power module applications, the model can analyze historical fault parameters and combine them with real-time data synchronization sets to output predictive results, thereby initiating preventative measures before the fault worsens. This proactive prediction reduces system downtime risk and improves overall fault tolerance efficiency. Figures 1-4 As shown, it includes steps S100-S800.

[0048] Step S100: Obtain the real-time parameter information of each power unit to obtain a power unit information set, which includes at least one power unit information item corresponding to a power unit.

[0049] It is important to note that, such as Figure 1As shown, the real-time parameter information of the power unit includes voltage parameters, current parameters, temperature parameters, and switching status parameters. The method for obtaining the power unit information set includes: setting up a sampling layer, an edge processing layer, and a platform layer; sampling voltage and current based on the sampling layer to obtain the first sampling item; sampling temperature based on the sampling layer to obtain the second sampling item; sampling switching status based on the sampling layer to obtain the third sampling item, thereby obtaining the sampling dataset; synchronizing the sampling dataset with timestamps based on the edge processing layer to obtain the data synchronization set; and creating a real-time dashboard on the platform layer to display and store the data synchronization set, thereby obtaining the power unit information set.

[0050] Specifically, the acquisition layer includes a voltage acquisition unit, a current acquisition unit, a temperature acquisition unit, and a switch status acquisition unit. The switch status acquisition unit is used to determine whether the power unit is in an enabled state. Therefore, the acquisition layer acquires the voltage, current, temperature, and switch status parameters of the power unit and then transmits the data to the edge processing layer. The edge processing layer is located close to the device and is responsible for time alignment of the different acquired signals to ensure that the voltage, current, temperature, and status data at the same moment are matched, forming a data synchronization set. Then, the synchronized data is received by the cloud or a central server, a real-time monitoring interface is created for users to view, and the data is stored in a database, ultimately forming a structured power unit information set.

[0051] In the specific implementation process, it is now necessary to collect data from an IGBT power module. The acquisition layer is set as follows: a voltage sensor collects the DC bus voltage to obtain the first sampling item; a current transformer collects the output current to obtain the first sampling item; a thermocouple collects the module substrate temperature to obtain the second sampling item; and a driver board signal collects the IGBT switching status to obtain the third sampling item. These data are sampled multiple times per second. An edge processing layer is created through an industrial control computer next to the device to receive all the sampled data. It was found that the temperature sensor data is a few milliseconds slower than the voltage and current data. The edge layer adjusts all the data to the same time base according to the precise timestamp, such as aligning to the 0th millisecond of every second, and generates a data synchronization set. Then, the cloud platform receives the synchronized dataset and displays the module's voltage waveform, current waveform, temperature curve, and switching status in real time on the web interface. At the same time, the synchronized dataset is stored in a time series database for subsequent analysis and model training, thereby obtaining the power unit information set.

[0052] Step S200: Obtain the bypass information corresponding to the power unit, including bypass quantity information and bypass load information, to obtain the unit bypass item.

[0053] It is important to note that, such as Figure 1As shown, the method for obtaining the bypass item of the unit includes: obtaining the bypass quantity information corresponding to the power unit to obtain the bypass information item; performing bypass tests based on the bypass information item; removing non-working bypasses based on the bypass test results to obtain the filtered bypass item; obtaining bypass attribute information based on the filtered bypass item, including specification information and load limit information corresponding to the specification information, and then obtaining the bypass load item.

[0054] Specifically, first, the total number of spare bypasses designed for the target power unit in the system configuration is obtained to obtain the bypass information item. Then, these bypasses are tested periodically or as needed, such as by injecting a small current to check whether they are conducting. Bypasses that fail the test are marked and excluded from the available list to obtain the filtered bypass items, i.e., the reliable spare bypass list. For the remaining available bypasses after filtering, their technical specifications are obtained, especially their maximum load capacity, such as current and power, to obtain the bypass load item.

[0055] In the specific implementation process, there is an important server power module designed with redundant bypasses. Through system configuration query, it is found that the module has two backup bypasses. The system automatically performs self-tests on the two bypasses: a very small test current is applied to bypass 1 to check whether it can conduct normally and whether the voltage drop is within the allowable range. Assuming that bypass 1 passes the test and bypass 2 fails the test, the system removes bypass 2, resulting in the filtered bypass item, that is, only bypass 1 is available. The system reads the specifications or database record of bypass 1, the rated current is 400A, and the maximum voltage is 1200V. Based on this, it calculates or looks up the maximum sustainable load power under the current system voltage, for example, 480kW, and then obtains the bypass load item.

[0056] Step S300: Create a digital twin model of the power unit based on the power unit information set to obtain the created model item.

[0057] It is important to note that, such as Figure 2 As shown, the method for obtaining the model item creation includes: setting up a network layer, a platform layer, and an application layer; realizing low-latency data transmission and synchronization between physical units and virtual models based on the network layer; carrying out the construction, training, simulation, and updating of the digital twin model based on the platform layer, providing data governance and computational support; and providing users with visual monitoring, early warning, and fault diagnosis interaction based on the application layer; performing behavioral simulation based on physical laws; establishing a mathematical model of the power unit based on Ohm's law of circuits, the heat conduction equation, and the power loss formula; building a simulation model; inputting the power unit information set; outputting the simulation results of temperature change and power loss; and comparing them with the power unit information set; obtaining historical parameter information of the power unit; training an LSTM deep learning model; and simultaneously using edge computing to achieve data alignment between the power unit and the virtual model, establishing a closed loop of data acquisition, model update, and visualization feedback, thereby obtaining the model item creation.

[0058] Specifically, the process begins with a 3D spatial mapping of the physical entity. The model recreates the physical structure of the power unit, including core components such as IGBT modules, capacitors, heat sinks, and terminals. The appearance and layout are replicated at a 1:1 scale. SolidWorks / Blender is used for high-precision modeling, and HT for Web is used for lightweight processing to ensure smooth loading in the browser. PBR materials are used to render the realistic texture of the device, and each component is associated with its corresponding parameters. For example, IGBT modules are associated with temperature parameters, and terminals with voltage / current parameters. Components automatically highlight and change color when parameters are abnormal. Behavioral simulation is then performed based on physical laws. A mathematical model of the power unit is established based on Ohm's law, the heat conduction equation, and the power loss formula. A simulation model is built using MATLAB / Simulink, inputting the power unit information set and outputting simulation results of temperature changes and power losses, which are then compared with the power unit information set. Historical parameter information of the power unit is acquired to train an LSTM deep learning model. Edge computing is also used to align the data between the power unit and the virtual model, establishing a closed loop of data acquisition, model updates, and visualization feedback.

[0059] Step S400: Obtain historical parameter information of the power unit to obtain a historical information set of the power unit. The historical parameter information includes historical faults and fault parameters corresponding to the historical faults.

[0060] It is important to note that, such as Figure 2 As shown, the method for obtaining the historical information set of the power unit includes: setting an acquisition threshold, the acquisition threshold being a fixed period value; acquiring the historical fault names of the power unit based on the acquisition threshold to obtain historical fault items; acquiring the fault parameters corresponding to the historical fault items based on the historical fault items to obtain fault parameter items; matching the fault parameter items with the historical fault items to obtain the historical information set of the power unit.

[0061] Specifically, the acquisition threshold is set to one year. The names and types of all fault events that have occurred in the power unit within this time range are acquired, such as "overcurrent protection", "drive board failure", "heat dissipation failure", etc., to obtain historical fault items. Then, for each identified historical fault event, the key operating parameters at the time of the fault and the period before and after are searched and recorded to obtain fault parameter items, such as the current peak before the fault, the temperature value at the time of the fault, voltage drop, etc. The historical fault events are matched one by one with their corresponding detailed parameter snapshots to form a complete historical information set of the power unit.

[0062] In the specific implementation process, the historical faults of a motor drive unit in a steel rolling mill were analyzed. The acquisition threshold was set at 12 months. A search of the database logs revealed that the unit had experienced three fault events in the past 12 months: output overcurrent protection trigger, IGBT bridge arm shoot-through fault, and radiator over-temperature alarm. Key operating parameters were obtained for each event. For event 1, the current waveform 5 seconds before the fault was extracted, showing violent fluctuations with a peak value reaching 150% of the rated value, and a slight drop in DC voltage at the time of the fault. For event 2, the drive signal status at the moment of the fault was extracted, showing abnormality, extremely high short-circuit current, and previous switching frequency records. For event 3, the temperature rise curve 1 hour before the fault was extracted, showing an abnormal slope, low cooling fan speed, and average load current, thus obtaining the historical information set of the power unit.

[0063] Step S500: Create at least two fault levels based on the historical information set of the power unit, with each fault level corresponding to fault parameters of different intensity ranges, to obtain the fault level item.

[0064] It is important to note that, such as Figure 3 As shown, the method for obtaining the fault level item includes: obtaining the parameter fluctuation difference based on the extreme value data of fault parameter fluctuation in the historical information set of the power unit, and then obtaining the parameter difference item; dividing the fault parameters into intervals based on the parameter difference item, and then obtaining at least two fault parameter intervals, using the fault parameter intervals as different fault levels, and then obtaining the fault level item.

[0065] Specifically, from the historical information set of the power unit, the maximum fluctuation value of each fault parameter, such as peak current and maximum temperature, is found when or before the fault occurs. The range of these fluctuation values ​​is calculated to obtain the parameter difference item. Based on the calculated range of parameter fluctuation difference, the fluctuation amplitude of the fault parameter is divided into three intervals, each interval representing a different fault severity, namely normal, moderate, and severe.

[0066] In the specific implementation process, the extreme values ​​of fault parameter fluctuations obtained from the historical information set of the power unit were as follows: extreme temperature 80℃, extreme temperature 50℃; extreme voltage 230V, extreme voltage 200V; extreme current 1300A, extreme current 1000A. The temperature difference was obtained as 30℃, the voltage difference as 30V, and the current difference as 300A. The fluctuation range of the fault parameters was divided into three intervals, each representing a different fault severity: normal, moderate, and severe. At this time, the temperature fluctuation range was divided into 50℃-60℃, 60℃-70℃, and 70℃-80℃; the voltage fluctuation range was divided into 200V-210V, 210V-220V, and 220V-230V; and the current fluctuation range was divided into 1000A-1100A, 1100A-1200A, and 1200A-1300A.

[0067] Step S600: Obtain the matching fault level of the power unit, and then obtain the real-time fault level item.

[0068] It is important to note that, such as Figure 3 As shown, the method for obtaining the real-time fault level item includes: obtaining the fluctuation amplitude of the power unit information set to obtain the real-time fluctuation item; matching the real-time fluctuation item with the fault level item to obtain matching comparison data; and then obtaining the real-time fault level item.

[0069] Specifically, the real-time fluctuation amplitude of the power unit information set is obtained to obtain the real-time fluctuation item. The real-time fluctuation item is matched with the fault level item to obtain the real-time fault level item.

[0070] Step S700: Based on the matching results between the created model item and the historical information set of the power unit, output the predicted fault level of the power unit to obtain the predicted fault level item.

[0071] It is important to note that, such as Figure 4 As shown, the method for obtaining the predicted fault level item includes: using the power unit information set as data input, creating model items to output predicted parameter data, obtaining output data items, and obtaining the predicted fault level of the power unit based on the matching result between the output data items and the historical information set of the power unit, thus obtaining the predicted fault level item.

[0072] Specifically, the latest power unit information set is used as input data and provided to the model creation item. The digital twin model runs its algorithm based on the input historical and real-time data sequences to predict key parameter values ​​for a future period, such as the next 5 minutes or 1 hour, including predicted peak current and maximum temperature. These predicted values ​​constitute the output data item. The model output data item is compared and analyzed with the historical information set of the power unit to find the similarity between the predicted parameter patterns and historical fault patterns. Based on the above matching results, it is determined which fault parameter range the predicted future parameters are most likely to fall into. This predicted future fault severity is the predicted fault level item.

[0073] Step S800: Based on the combined results of the real-time fault level item and the predicted fault level item, the unit bypass item is preheated, including voltage preheating and current prediction, to obtain the unit preheating item.

[0074] It is important to note that, such as Figure 4 As shown, the method for obtaining the unit preheating item includes: setting at least two preheating adjustment level items, each corresponding to a different load preheating value; obtaining the pre-adjustment level corresponding to the preheating adjustment level item based on the combination result of the real-time fault level item and the predicted fault level item; obtaining the matching load preheating value; preheating the unit bypass based on the load preheating value; and thus obtaining the unit preheating item. This achieves bypass selection and bypass preheating based on the joint analysis of digital twin model and real-time data.

[0075] Specifically, two preheating adjustment levels are set, corresponding to the medium and severe fault levels respectively, namely medium and high. Each level is set with different load preheating values, including current and voltage values. The set current and voltage values ​​correspond to the voltage range and current range of the fault level, respectively. Based on the combination of the real-time fault level and the predicted fault level, the pre-adjustment level corresponding to the preheating adjustment level is obtained, and then the matching load preheating value is obtained. The unit bypass is preheated based on the load preheating value. When a unit bypass that cannot be preheated occurs, it means that there is a problem with the unit bypass. At this time, the unit bypass is removed, and the normal unit bypass is kept preheated at all times, so that the unit bypass can be activated at the moment when the power unit fails.

[0076] A power unit fault adaptive bypass control system, employing the aforementioned power unit fault adaptive bypass control method, includes: a data acquisition module: acquiring real-time parameter information for each power unit to obtain a power unit information set, the power unit information set including at least one power unit information item; acquiring bypass information corresponding to the power unit, including bypass quantity information and bypass load information, to obtain a unit bypass item; a model creation module: creating a digital twin model of the power unit based on the power unit information set, obtaining a model creation item; simultaneously acquiring historical parameter information of the power unit to obtain a power unit historical information set, the historical parameter... The data information includes historical faults and corresponding fault parameters; the fault judgment module: creates at least two fault levels based on the historical information set of the power unit, each fault level corresponding to fault parameters of different intensity ranges, obtains fault level items, obtains the matching fault level of the power unit, and then obtains the real-time fault level items; the bypass preheating module: based on the matching result of the created model items and the historical information set of the power unit, outputs the predicted fault level of the power unit, obtains the predicted fault level items, and preheats the unit bypass items based on the combination result of the real-time fault level items and the predicted fault level items, including voltage preheating and current prediction, to obtain the unit preheating items.

[0077] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. A power unit fault adaptive bypass control method, comprising: Obtain real-time parameter information for each power unit to obtain a power unit information set, which includes at least one power unit information item corresponding to a power unit. Its characteristic is that it further includes: Obtain the bypass information corresponding to the power unit, including the number of bypasses and the bypass load information, to obtain the unit bypass item; A digital twin model of the power unit is created based on the power unit information set, resulting in the creation of model items. At the same time, the historical parameter information of the power unit is obtained, resulting in the historical information set of the power unit. The historical parameter information includes historical faults and fault parameters corresponding to the historical faults. Based on the historical information set of the power unit, at least two fault levels are created, each fault level corresponding to fault parameters of different intensity ranges, to obtain fault level items, obtain the matching fault level of the power unit, and then obtain real-time fault level items. Based on the matching results between the created model items and the historical information set of the power unit, the predicted fault level of the power unit is output, and the predicted fault level item is obtained. Based on the combination of the real-time fault level item and the predicted fault level item, the unit bypass item is preheated, including voltage preheating and current prediction, and the unit preheating item is obtained. Thus, bypass selection and bypass preheating based on the joint analysis of digital twin model and real-time data are realized.

2. The power unit fault adaptive bypass control method according to claim 1, characterized in that: The real-time parameter information of the power unit includes voltage parameters, current parameters, temperature parameters, and switching status parameters. The method for obtaining the power unit information set includes: The system consists of a data acquisition layer, an edge processing layer, and a platform layer. Voltage and current are sampled based on the data acquisition layer to obtain the first sampling item. Temperature is sampled based on the data acquisition layer to obtain the second sampling item. Switching states are sampled based on the data acquisition layer to obtain the third sampling item, thus obtaining the sampled dataset. The sampling dataset is timestamped based on the edge processing layer to obtain a data synchronization set. The platform layer creates a real-time dashboard to display and store the data synchronization set, thereby obtaining the power unit information set.

3. The power unit fault adaptive bypass control method according to claim 1, characterized in that: The method for obtaining the unit bypass item includes: Obtain the bypass quantity information corresponding to the power unit to obtain bypass information items. Perform bypass tests based on the bypass information items. Based on the bypass test results, eliminate non-functional bypasses to obtain filtered bypass items. By filtering bypass items, bypass attribute information is obtained, including specification information and the corresponding load limit information, and then bypass load items are obtained.

4. The power unit fault adaptive bypass control method according to claim 1, characterized in that: The methods for obtaining the model items include: The system is structured into a network layer, a platform layer, and an application layer. The network layer enables low-latency data transmission and synchronization between physical units and virtual models. The platform layer supports the construction, training, simulation, and updating of digital twin models, providing data governance and computational support. The application layer provides users with visual monitoring, early warning, and fault diagnosis interaction. Behavioral simulation is performed based on physical laws. Based on Ohm's law for circuits, the heat conduction equation, and the power loss formula, a mathematical model of the power unit is established, a simulation model is built, the power unit information set is input, and the simulation results of temperature change and power loss are output and compared with the power unit information set. Historical parameter information of the power unit is obtained, an LSTM deep learning model is trained, and edge computing is used to align the data between the power unit and the virtual model. A closed loop of data acquisition, model update, and visualization feedback is established to obtain the model creation item.

5. The power unit fault adaptive bypass control method according to claim 1, characterized in that: The method for obtaining the historical information set of the power unit includes: Set an acquisition threshold, which is a fixed period value. Based on the acquisition threshold, acquire the historical fault names of the power unit to obtain historical fault items. Based on the historical fault items, the fault parameters corresponding to the historical fault items are obtained, and then the fault parameter items are obtained. The fault parameter items are matched with the historical fault items, and then the historical information set of the power unit is obtained.

6. The power unit fault adaptive bypass control method according to claim 1, characterized in that: The method for obtaining the fault level item includes: Based on the extreme value data of fault parameter fluctuations in the historical information set of the power unit, the parameter fluctuation difference is obtained, and then the parameter difference item is obtained. Based on the parameter difference term, the fault parameters are divided into intervals to obtain at least two fault parameter intervals. The fault parameter intervals are used as different fault levels to obtain the fault level term.

7. The power unit fault adaptive bypass control method according to claim 1, characterized in that: The method for obtaining the real-time fault level item includes: The fluctuation amplitude of the power unit information set is obtained to obtain the real-time fluctuation item. The real-time fluctuation item is matched with the fault level item to obtain the matching comparison data, and then the real-time fault level item is obtained.

8. The power unit fault adaptive bypass control method according to claim 1, characterized in that: The method for obtaining the predicted fault level item includes: Using the power unit information set as data input, the predicted parameter data is output by creating model terms to obtain output data terms. Based on the matching results between the output data terms and the historical information set of the power unit, the predicted fault level of the power unit is obtained, resulting in the predicted fault level term.

9. The power unit fault adaptive bypass control method according to claim 1, characterized in that: The method for obtaining the unit preheating term includes: At least two preheating adjustment level items are set, each corresponding to a different load preheating value. Based on the combination result of the real-time fault level item and the predicted fault level item, the pre-adjustment level corresponding to the preheating adjustment level item is obtained, and then the matching load preheating value is obtained. The unit bypass is preheated based on the load preheating value, and then the unit preheating item is obtained.

10. A power unit fault adaptive bypass control system, characterized in that: The power unit fault adaptive bypass control method according to any one of claims 1-9 includes: Data acquisition module: Acquires real-time parameter information for each power unit, obtains a power unit information set, which includes at least one power unit information item corresponding to a power unit, acquires bypass information corresponding to the power unit, including bypass quantity information and bypass load information, and obtains unit bypass item; Model creation module: Creates a digital twin model of the power unit based on the power unit information set, obtains the model creation item, and obtains the historical parameter information of the power unit to obtain the historical information set of the power unit. The historical parameter information includes historical faults and fault parameters corresponding to the historical faults. Fault diagnosis module: Based on the historical information set of the power unit, at least two fault levels are created. Each fault level corresponds to fault parameters with different intensity ranges to obtain fault level items. The matching fault level of the power unit is obtained, and then the real-time fault level items are obtained. Bypass preheating module: Based on the matching results of the created model items and the historical information set of the power unit, the predicted fault level of the power unit is output to obtain the predicted fault level item. Based on the combination of the real-time fault level item and the predicted fault level item, the bypass item of the unit is preheated, including voltage preheating and current prediction, to obtain the unit preheating item.

Images