Urban rail train air conditioner double-redundancy variable frequency power supply control method and system

Abstract

The application provides a kind of urban rail train air conditioner double redundancy variable frequency power supply control method and system, it is related to emergency protection circuit device technical field, method includes: with the time sequence data of vehicle outside environment as dynamic boundary condition, multiple carriages are cooperated to control optimization solution, generate K refrigeration distribution time sequence baseline;In the execution of differentiated refrigeration process, time sequence state tracking is carried out, K indoor environment dynamic parameter flow and K multimode power state data are obtained;Power health trend analysis is carried out, and K power redundancy reserve prediction value is output;Demand deviation identification is carried out, and K refrigeration power dynamic compensation is output;Through distributed consensus negotiation algorithm, the double-dimensional resilience control of K double redundancy variable frequency power supply is executed.The application solves the technical problem that the urban rail train air conditioner power supply control of prior art is generally using load distribution based on preset rules, cannot respond to the dynamic change factors of carriage in real time, affects the overall energy efficiency of air conditioning system.

Description

Technical Field

[0001] This invention relates to the field of emergency protection circuit device technology, specifically to a dual-redundant variable frequency power supply control method and system for urban rail train air conditioning. Background Technology

[0002] The air conditioning system of urban rail trains plays a crucial role in regulating the temperature of the carriages and providing a comfortable riding environment. The air conditioning system not only needs to cope with complex changes in the external environment, but also needs to make real-time adjustments based on dynamic factors such as passenger density and temperature inside the train. Therefore, how to improve the stability, efficiency and adaptability of the air conditioning system has become a technical problem that urgently needs to be solved.

[0003] Traditional air conditioning systems often rely on a single power source, and their operation is affected by factors such as power supply performance fluctuations, environmental changes, and changes in vehicle load. However, with the advancement of power electronics technology, the use of dual-redundant inverter power supplies has become an effective solution. Dual redundancy systems improve the reliability of power supply and reduce the impact of single-point failures on system stability.

[0004] Existing urban rail train air conditioning systems typically employ load distribution based on preset rules, or timed control schemes in certain situations. These methods cannot respond in real time to dynamic changes in factors such as temperature and passenger density within the carriages. Due to the lack of dynamic load adjustment capabilities, the load distribution between different carriages may be uneven, leading to overload of air conditioning in some carriages while energy is wasted in other carriages, thus affecting the overall efficiency of the system. Summary of the Invention

[0005] This application provides a dual-redundant variable frequency power supply control method and system for urban rail train air conditioning, aiming to solve the technical problem that the existing urban rail train air conditioning power supply control usually adopts load distribution based on preset rules, which cannot respond to dynamic changes in the carriage in real time and affects the overall energy efficiency of the air conditioning system.

[0006] The first aspect of this application discloses a method for controlling dual-redundant inverter power supplies for air conditioning in urban rail trains. The method includes: using time-series data of the external environment as dynamic boundary conditions, performing multi-car collaborative control optimization based on K passenger density time-series baselines and K car temperature time-series baselines of K connected cars in the urban rail train to generate K cooling distribution time-series baselines; and K edge dual-redundant control units, during the process of controlling the K dual-redundant inverter power supplies to perform differentiated cooling in the K connected cars using the K cooling distribution time-series baselines, controlling the K connected cars and K... The timing state tracking of the dual-redundant inverter power supply yields K in-vehicle environmental dynamic parameter streams and K multi-modal power supply state data. Based on the K multi-modal power supply state data, power health trend analysis is performed, outputting K predicted power redundancy margin values. Demand deviation identification is performed based on the K in-vehicle environmental dynamic parameter streams, outputting K dynamic compensation amounts for cooling power. The K edge dual-redundant control units execute two-dimensional resilience control of the K dual-redundant inverter power supplies through a distributed consensus negotiation algorithm based on the K predicted power redundancy margin values ​​and the K dynamic compensation amounts for cooling power.

[0007] The second aspect of this application discloses a dual-redundant inverter power supply control system for urban rail transit trains. This system is used in the aforementioned dual-redundant inverter power supply control method for urban rail transit trains. The system includes: a control optimization solution module, used to perform multi-car collaborative control optimization solution based on K passenger density time-series baselines and K car temperature time-series baselines of K connected cars in the urban rail transit train, using external environmental time-series data as dynamic boundary conditions, to generate K cooling distribution time-series baselines; and a time-series state tracking module, used to track the K edge dual-redundant control units during the process of controlling the K dual-redundant inverter power supplies to perform differentiated cooling of the K connected cars using the K cooling distribution time-series baselines. The system tracks the timing status of the connected carriage and K dual-redundant inverter power supplies to obtain K in-vehicle environmental dynamic parameter streams and K multi-modal power supply status data. A power health trend analysis module performs power health trend analysis based on the K multi-modal power supply status data and outputs K predicted power redundancy margin values. A demand deviation identification module identifies demand deviations based on the K in-vehicle environmental dynamic parameter streams and outputs K dynamic cooling power compensation values. A dual-dimensional resilience control module enables the K edge dual-redundant control units to execute dual-dimensional resilience control of the K dual-redundant inverter power supplies using a distributed consensus negotiation algorithm based on the K predicted power redundancy margin values ​​and the K dynamic cooling power compensation values.

[0008] One or more technical solutions provided in this application have at least the following beneficial effects:

[0009] Based on time-series data of the external environment, passenger density, and cabin temperature, a multi-cabin collaborative control optimization solution is used to generate a cooling distribution time-series baseline for each cabin. This allows for the rational allocation of cooling resources according to the different needs of different cabins, avoiding over-cooling or unbalanced loads, thereby improving the energy efficiency and comfort of the air conditioning system. K edge dual-redundant control units control K dual-redundant inverter power supplies to perform differentiated cooling for each cabin and track the timing status. The dual redundancy ensures reliable system operation even in the event of a power failure, preventing air conditioning system failure due to a single power supply failure. The timing status tracking ensures accurate adjustments to the system under environmental changes in each cabin, ensuring a comfortable cabin environment. Finally, by analyzing the power supply health status data from K multimodal power supplies, the system outputs predicted power redundancy margins. Power health trend analysis can monitor the power supply health status in real time and predict redundancy margins. Early identification of power failures or performance degradation helps optimize power configuration, avoid power overload, and improve system reliability and predictability. Based on the dynamic parameter flow of the in-vehicle environment, it identifies demand deviations and outputs dynamic compensation for cooling power to cope with changes in environmental demand. Through this intelligent dynamic compensation, the air conditioning system can maintain a comfortable temperature in the cabin under different passenger loads and external environmental conditions. K edge dual-redundant control units execute dual-dimensional resilience control through a distributed consensus negotiation algorithm based on predicted power redundancy margins and dynamic compensation for cooling power. This ensures cross-cabin load coordination. The distributed consensus algorithm guarantees coordination and consistency among multiple control units, avoiding the impact of errors or lags of a single control unit on the operation of the entire system. Dual-dimensional resilience control not only enhances the system's fault recovery capability but also adjusts the control strategy in real time according to actual operating conditions, ensuring the stable operation of the air conditioning system in each cabin.

[0010] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0011] Figure 1 This is a schematic flowchart of a dual-redundant inverter power supply control method for urban rail train air conditioning, provided in an embodiment of this application.

[0012] Figure 2 This is a schematic diagram of a dual-redundant inverter power supply control system for urban rail train air conditioning, provided as an embodiment of this application.

[0013] Figure labeling: Control optimization solution module 10, timing state tracking module 20, power health trend analysis module 30, demand deviation identification module 40, two-dimensional resilience control module 50. Detailed Implementation

[0014] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0015] Example 1, as Figure 1 As shown in the figure, this application provides a method for controlling a dual-redundant inverter power supply for an urban rail train air conditioning system. The method includes:

[0016] A100: Using the time series data of the external environment as dynamic boundary conditions, multi-car collaborative control optimization is performed based on the K passenger density time series baselines and K car temperature time series baselines of K connected cars in the urban rail train, generating K cooling distribution time series baselines.

[0017] Vehicle external environment time series data refers to data on changes in the external environment over time, such as temperature and humidity. This data is used to provide external constraints and dynamic conditions for the entire control process. For example, when the outside temperature changes, the cooling demand of the air conditioning system will also change, and the time series data of the external environment becomes the dynamic boundary condition.

[0018] The K passenger density time-series baselines for K connected carriages refer to the baseline data of passenger density (number of passengers per square meter) in each carriage at different times. These data are used to assess the load situation in the carriages, because more passengers will lead to more heat generation, thereby increasing the demand for air conditioning. The K carriage temperature time-series baselines refer to the baseline data of temperature changes in each carriage over time. These data represent the temperature distribution of different carriages at different times and are the basic data for controlling the air conditioning temperature.

[0019] Multi-carriage collaborative control optimization refers to using the aforementioned carriage data and optimization algorithms to collaboratively control the air conditioning systems of each carriage, ensuring a balance of cooling demand across different carriages. Based on passenger density, carriage temperature, and external environmental data, the cooling demand of each carriage is calculated, and corresponding cooling allocation time-series baselines are generated. These time-series baselines indicate the cooling capacity required by each carriage at different time periods. Using these baselines, the cooling ratio of each carriage can be dynamically adjusted in future operations to ensure that the temperature in each carriage remains within a comfortable range.

[0020] A200: During the process of using the K cooling distribution timing baseline to control the K dual-redundant frequency converters to perform differentiated cooling of the K connected carriages, the K edge dual-redundant control units perform timing state tracking of the K connected carriages and the K dual-redundant frequency converters to obtain K in-vehicle environment dynamic parameter streams and K multi-modal power supply state data.

[0021] K edge dual-redundant control units refer to control devices installed in each carriage, responsible for executing the control tasks of the air conditioning system. Each control unit has two redundant modules, so even if one module fails, the other module can continue to operate. The K edge dual-redundant control units control the corresponding K dual-redundant inverter power supplies based on the generated K cooling distribution timing baselines. The inverter power supplies dynamically adjust the power of the air conditioning compressor according to the cooling demand, thereby adjusting the temperature of each carriage. The cooling demand of each carriage will vary due to differences in passenger density and carriage temperature. This differentiated cooling process means that each carriage independently adjusts its cooling power according to its specific cooling needs, rather than simply implementing uniform control across the entire vehicle.

[0022] Time-series status tracking refers to real-time monitoring of the operating status of the air conditioning system and power supply in each compartment. During this process, the dynamic parameters of the in-vehicle environment include variables such as temperature, humidity, and passenger density, while the multi-modal power status data includes information such as power output, voltage, and frequency. The flow and changes of these data will affect the adjustment method of the cooling process.

[0023] A300: Based on the K multi-mode power state data, perform power health trend analysis and output K predicted power redundancy margin values.

[0024] By analyzing K multimodal power supply state data, the health status of each power module is assessed. This health analysis can employ machine learning algorithms, regression analysis, or degradation models to predict power supply performance trends. The goal of health trend analysis is to identify signs of power supply performance degradation, predict potential future failures or performance declines, and ensure stable system operation. During the analysis, the current performance of the power supply is compared with a preset health reference baseline to calculate performance degradation. If the power supply health deteriorates, the corresponding cooling capacity will decrease. Health trend analysis outputs K predicted power redundancy margin values. These predicted values ​​reflect how much redundant power each power supply will have remaining under possible future load changes, which helps assess whether power supply load adjustments or configuration optimization are needed.

[0025] A400: Based on the K in-vehicle environment dynamic parameter streams, demand deviation is identified, and K dynamic compensation values ​​for cooling power are output.

[0026] Based on K dynamic parameters of the in-vehicle environment, demand deviation is identified, that is, the deviation between the current cooling demand of the carriage and the preset baseline is calculated. For example, the deviation between the actual temperature and the set temperature in the carriage, or the deviation between the actual passenger density and the preset benchmark value, are all considered as part of the demand deviation. These deviations mean changes in the current cooling demand and need to be compensated in the control strategy.

[0027] Based on the identified demand deviations, the system calculates the dynamic compensation amount of cooling power required for each carriage. This compensation amount represents the dynamically adjusted cooling power needed to restore or maintain a comfortable temperature in the carriage. For example, when the passenger density in a carriage increases, the heat load inside the carriage also increases, thus requiring more cooling power; when the temperature in the carriage is too high, the cooling power also needs to be increased accordingly. Through this calculation of dynamic compensation, the air conditioning system can adjust its output in real time to ensure a stable temperature in the carriage.

[0028] A500: The K edge dual-redundant control units execute dual-dimensional resilience control of the K dual-redundant frequency converters through a distributed consensus negotiation algorithm based on the K predicted power redundancy margins and the K dynamic compensation values ​​for cooling power.

[0029] K predicted power redundancy margins and K dynamic cooling power compensation values ​​are used as input data for K edge dual-redundant control units. These control units dynamically adjust the power load and cooling power based on this information to ensure system stability and efficiency. The goal of dual-dimensional resilience control is to enhance system stability and reliability, ensuring normal operation even under changes in power supply or environment. Through dual-dimensional resilience control, control units not only handle the temperature demands of local carriages but also consider power redundancy margins and fault recovery capabilities, ensuring stable system operation even in the face of external environmental changes or power failures. For example, when the cooling demand of a carriage increases or power redundancy decreases, the load of other carriages is dynamically adjusted to ensure the overall system's healthy and efficient operation. The distributed consensus negotiation algorithm enables communication and cooperation among control units, allowing each edge dual-redundant control unit to reach a consistent control objective. In this process, each control unit independently calculates and proposes its own control commands, which are then shared and negotiated through the network to ultimately achieve a globally optimized set of control commands, thereby realizing coordinated load scheduling between carriages.

[0030] Furthermore, the K edge dual-redundant control units execute two-dimensional resilience control of the K dual-redundant frequency converters based on the K predicted power redundancy margins and K dynamic cooling power compensation values ​​through a distributed consensus negotiation algorithm. The method also includes:

[0031] A510: The K edge dual-redundant control units encapsulate K carriage identifiers, K power redundancy margin prediction values, and K cooling power dynamic compensation values ​​into K local state vectors; A520: The K edge dual-redundant control units broadcast the K local state vectors through the vehicle network so that the K edge dual-redundant control units receive the global state vector; A530: Based on the global state vector, the K edge dual-redundant control units perform distributed collaborative computing in parallel to achieve consistent state allocation and output K multi-objective control instruction sets; A540: After verifying the safety interlock by broadcasting the K multi-objective control instruction sets through the vehicle network, the K edge dual-redundant control units atomically execute the cross-carriage load collaborative scheduling of the K dual-redundant frequency converters.

[0032] The carriage identifier is a unique identifier for each carriage, used to distinguish different carriages; the power redundancy margin prediction value represents the remaining available power of each power module, predicting the future load capacity of the power supply; the dynamic cooling power compensation amount represents the amount of cooling power that needs to be dynamically adjusted, reflecting the deviation between the temperature and passenger density inside the carriage. K edge dual-redundant control units encapsulate the above data as K local state vectors.

[0033] Each edge dual-redundant control unit broadcasts its local state vector via an in-vehicle network, such as a CAN bus-based network or other in-vehicle communication protocols. This step ensures that all edge dual-redundant control units can receive information from other edge dual-redundant control units, enabling global coordination and collaborative computing. Control data from all compartments is integrated to form a global state vector, serving as the basis for global coordinated computing.

[0034] Upon receiving the global state vector, each edge dual-redundant control unit performs parallel computation based on this information. Distributed collaborative computation means that each edge dual-redundant control unit performs computation independently, but they share information to reach consensus. In this way, globally optimal control can be achieved, rather than considering each car individually. The goal of parallel computation is to achieve consistent state allocation, that is, to ensure that the cooling load of all cars is reasonably allocated and that the redundancy of each power supply is effectively utilized. Through this collaborative computation, multiple edge dual-redundant control units can balance the load, optimize power usage, and achieve collaborative control between cars. Each edge dual-redundant control unit outputs a multi-objective control command set based on the computation results. These commands include not only cooling control signals for each car but also load scheduling commands across cars to adjust the power distribution between cars.

[0035] K multi-objective control command sets are broadcast to all K edge dual-redundant control units via the vehicle network. First, a safety interlock verification is performed to ensure that the execution of all commands does not lead to system conflicts or dangers. This verification includes monitoring the operating status of each control unit to ensure that load coordination and scheduling in each compartment does not cause instability to the system. For example, if a power module is close to full load, an interlock mechanism is used to prevent overload or adjust the load of other power modules. Atomized execution means that each control unit must fully execute an operation without interruption, ensuring the consistency and stability of the entire system. Cross-compartment load coordination and scheduling means that the cooling and power loads of each compartment are coordinated, ensuring reasonable load distribution across all compartments and preventing overload or power failure in any single compartment.

[0036] Furthermore, the K edge dual-redundant control units, based on the global state vector, perform distributed collaborative computation in parallel to achieve consistent state allocation and output K multi-objective control instruction sets. The method also includes:

[0037] S1: The K edge dual-redundant control units generate K initial control instruction sets based on the global state vector; S2: The K initial control instruction sets are broadcast through the vehicle network so that the K edge dual-redundant control units can receive the global initial instruction sets; S3: The K edge dual-redundant control units perform local collaborative optimization calculations in parallel based on the global initial instruction sets and the global state vector, and output K updated control instruction sets; S4: After calculating the K difference norms of the K initial control instruction sets and the K updated control instruction sets for convergence evaluation, the K edge dual-redundant control units broadcast the K updated control instruction sets through the vehicle network so that the K edge dual-redundant control units can receive the global updated instruction sets; Iterate steps S3 to S4 until the local convergence evaluation feedback of the K edge dual-redundant control units all meet the preset consistency threshold, and output the K multi-objective control instruction sets.

[0038] Based on the global state vector, each edge dual-redundant control unit generates a corresponding initial control instruction set. These initial instruction sets include two types of instructions: local load allocation instructions, which indicate how to allocate the cooling load within each car; and cross-car power coordination instructions, which indicate how to coordinate power allocation between cars to ensure that the load of each car does not exceed its power supply capacity.

[0039] Through the vehicle network, K edge dual-redundant control units broadcast the generated K initial control command sets. Each control unit receives the initial control command sets from other control units through the broadcast mechanism, which serve as the global initial command set and the starting point for global collaborative computing.

[0040] After receiving the global initial instruction set and global state vector, each edge dual-redundant control unit performs local collaborative optimization calculations based on this information. At this point, each control unit adjusts its control commands according to the specific conditions of the carriage, such as the current temperature deviation, passenger density, and power redundancy, to optimize the cooling load and power distribution. These calculations are parallel, meaning that each control unit optimizes independently, but they use the same global information. This allows the optimization process to be performed simultaneously across multiple control units, thereby improving computational efficiency. After each control unit performs local optimization, it outputs an updated control instruction set. These updated instruction sets are adjusted compared to the initial instruction set to more accurately balance load and power and make the coordination between carriages more efficient.

[0041] After outputting the updated control command set, a convergence evaluation is performed. This evaluation calculates the difference norm between the initial and updated control command sets. The difference norm measures the gap between the two sets of commands and is evaluated by calculating their vector difference or other similar mathematical metrics. Based on the calculated difference norm, it is determined whether the preset convergence condition has been met. If the difference norm is less than a preset threshold, it indicates that the optimal control state is approaching, and optimization can stop. If the difference is still large, optimization calculations continue. After the convergence evaluation is completed, all edge dual-redundant control units broadcast the updated control command set via the vehicle network. Each control unit receives the global updated command set from other control units, ensuring that the air conditioning and power load distribution in each compartment is adjusted synchronously.

[0042] Between steps S3 (performing local collaborative optimization calculations and outputting updated control instruction sets) and S4 (calculating the difference norm for convergence evaluation), an iterative process is entered. In each iteration, a new updated control instruction set is calculated based on the current control instruction set and the global state vector. Then, the difference between the updated and initial instruction sets is evaluated. If the difference is large, it indicates that the instruction set needs further adjustment. The preset consistency threshold is a preset value representing the convergence criterion that must be achieved. When the difference norm of all control units is below this threshold, it indicates that convergence has been achieved. At this point, K multi-objective control instruction sets are output. These instruction sets contain load allocation for each car and power coordination instructions across cars, enabling globally optimal cooling load allocation and power management.

[0043] Furthermore, the multi-objective control instruction set includes local load allocation instructions and cross-carriage power coordination instructions.

[0044] Local load allocation instructions refer to how each control unit distributes the cooling load to the carriages it is responsible for. Each carriage determines its required cooling capacity based on factors such as its internal temperature and passenger density. These instructions specify how much cooling resources should be allocated to each carriage. Cross-carriage power coordination instructions refer to how to coordinate power between different carriages. For example, when the cooling demand of a carriage increases, it will lead to an increase in the power load of that carriage. At this time, it is necessary to dispatch redundant power from other carriages to maintain the stable operation of the entire vehicle system. These instructions guide how to balance the power distribution between carriages.

[0045] Furthermore, the K edge dual-redundant control units perform local cooperative optimization calculations in parallel based on the global initial instruction set and the global state vector, outputting K updated control instruction sets. The method also includes:

[0046] S31: The second edge dual-redundant control unit extracts control instructions from neighboring units based on the global initial instruction set to obtain the first initial control instruction set of the first edge dual-redundant control unit and the third initial control instruction set of the third edge dual-redundant control unit; S32: With the objectives of minimizing the cooling power deviation of the carriage, maximizing the redundancy utilization rate, and minimizing the difference between neighboring instructions, the local conflict resolution of the second initial control instruction set, the first initial control instruction set, and the third initial control instruction set is performed, and the second updated control instruction set is output.

[0047] The neighboring units of the second edge dual-redundant control unit are the first edge dual-redundant control unit and the third edge dual-redundant control unit. The second edge dual-redundant control unit extracts the control instructions of the neighboring units from the global initial instruction set, that is, it extracts the first initial control instruction set and the third initial control instruction set. These instructions of the neighboring units will provide the basis for subsequent conflict resolution.

[0048] Minimizing the cooling power deviation of this car means adjusting the cooling load to minimize the difference between the actual cooling power of the car and the required cooling power; maximizing redundancy utilization means ensuring that redundant power is fully utilized and avoiding waste by rationally scheduling the power load; minimizing neighbor command differences means minimizing command differences with neighboring cars to ensure the efficiency and consistency of cross-car power coordination.

[0049] Because each carriage has different cooling needs and power loads, commands between different carriages may conflict. For example, the load allocation of one carriage may not match that of its neighboring carriages. Through local conflict resolution, the control unit adjusts the commands according to the aforementioned objectives to resolve these conflicts and ensure optimal load allocation and power scheduling for each carriage. After conflict resolution is completed, the second edge dual-redundant control unit outputs a second updated control command set. This command set includes adjusted local load allocation and cross-carriage power coordination commands. These updated command sets reflect the optimal solution after conflict resolution.

[0050] Furthermore, using the time-series data of the external environment as dynamic boundary conditions, and based on the K passenger density time-series baselines and K car temperature time-series baselines of the K connected cars in the urban rail train, multi-car cooperative control optimization is performed to generate K cooling distribution time-series baselines. The method also includes:

[0051] A110: After spatiotemporally aligning the time-series data of the external environment, the first passenger density time-series baseline, and the first carriage temperature time-series baseline, fine-grained segmentation using a sliding time window is used to obtain multiple overlapping time-series scene segments; A120: Similarity matching is performed on the multiple time-series scene segments in the control case library to retrieve multiple historical control segments; A130: The multiple historical control segments are smoothed to generate the first preliminary cooling allocation curve for the first connected carriage; A140: After constructing the K preliminary cooling allocation curves for the K connected carriages by analogy, conflict resolution of the K preliminary cooling allocation curves is performed according to the thermal coupling relationship of the K connected carriages, and the K cooling allocation time-series baselines are output.

[0052] The time-series data of the external environment, the first passenger density time-series baseline, and the first carriage temperature time-series baseline are spatiotemporally aligned to ensure they can be compared and used on the same time base. This is because there is a time difference between the external environment and the in-vehicle environment, and interpolation and alignment methods are needed to align data from different sources with the time axis to ensure consistency. The sliding time window technique divides the time-series data into multiple overlapping time periods, each called a time-series scene segment. Fine-grained segmentation refers to these very small time windows, which can capture short-term trends in data changes. Each scene segment corresponds to changes in the external environment, passenger density, and carriage temperature within a short period.

[0053] The control case library is a database storing historical control data. Each historical control segment contains past air conditioning control strategies under specific environmental and load conditions. These historical cases are cooling allocation schemes generated based on different external environmental conditions and internal passenger compartment temperatures, passenger densities, and other factors. Temporal scene segment similarity matching compares multiple generated temporal scene segments with historical control segments in the control case library. Using similarity metrics such as Euclidean distance and cosine similarity, it finds the historical control segment most similar to the current temporal scene segment in terms of external environmental conditions, passenger density, and passenger compartment temperature changes. Through similarity matching, multiple historical control segments are retrieved from the control case library. These segments provide a reference for the current temporal scene segment, helping to determine the appropriate cooling allocation strategy under similar environmental and load conditions.

[0054] The purpose of smoothing is to eliminate fluctuations in historical control segments, making the control curve smoother and avoiding unstable control decisions caused by short-term data fluctuations. Smoothing methods include moving averages, weighted averages, or other smoothing algorithms to reduce irregular changes in the data. The first preliminary cooling allocation curve represents the cooling demand and allocation in historical control segments. This curve reflects how the cooling load is allocated under specific environmental and cabin conditions, providing a basis for subsequent optimization and adjustment.

[0055] By analogy with the process of obtaining the first preliminary cooling allocation curve, corresponding preliminary cooling allocation curves are constructed for the other carriages. The thermal coupling relationship among the K connected carriages refers to the mutual influence of heat between different carriages; for example, a temperature rise in one carriage will affect the temperature distribution of adjacent carriages. Conflicts may exist among the generated K preliminary cooling allocation curves, such as the cooling demand of some carriages being inconsistent with that of adjacent carriages, leading to uneven load distribution. Through conflict resolution, the load distribution of each carriage is adjusted to ensure the overall balance of the system. Conflict resolution can be adjusted based on the thermal coupling relationship between carriages, optimizing the cooling power scheduling among carriages. After conflict elimination, K cooling allocation timing baselines are output; these baselines represent the final cooling allocation scheme for each carriage.

[0056] Furthermore, based on the K in-vehicle environmental dynamic parameter streams, demand deviation is identified, and K dynamic compensation values ​​for cooling power are output. The method also includes:

[0057] A410: Decompose the dynamic parameter flow of the first in-vehicle environment to obtain the real-time temperature flow and the real-time passenger density flow of the first compartment; A420: Based on the time series information of the dynamic parameter flow of the first in-vehicle environment, map and segment the first reference passenger density sequence and the first reference temperature sequence from the first passenger density time series baseline and the first compartment temperature time series baseline; A430: Compare the real-time temperature flow and the first reference temperature sequence of the first compartment to calculate the temperature deviation sequence of the first compartment; A440: Compare the real-time passenger density flow and the first reference passenger density sequence of the first compartment to calculate the passenger density deviation sequence; A450: Input the temperature deviation sequence and the passenger density deviation sequence of the first compartment into the prefitted heat load regression function to calculate the first heat load deviation sequence; A460: Calculate and output the first dynamic compensation amount of cooling power based on the first heat load deviation sequence.

[0058] The first in-vehicle environment dynamic parameter stream refers to a series of data on the changes in the in-vehicle environment over time. The first in-vehicle environment dynamic parameter stream is decomposed into more detailed parts, specifically into the first carriage real-time temperature stream and the first carriage real-time passenger density stream. The first carriage real-time temperature stream represents the temperature changes over time at different locations in the carriage; the first carriage real-time passenger density stream represents the number of passengers (or passenger density) over time.

[0059] The first in-vehicle environmental dynamic parameter stream contains time-series information, indicating the recording time points of the compartment temperature and passenger density. Based on this time-series information, corresponding first reference passenger density sequences and first reference temperature sequences are mapped and segmented from the first passenger density time-series baseline and the first compartment temperature time-series baseline. The first reference passenger density sequence represents the ideal passenger density change trend, and the first reference temperature sequence represents the ideal compartment temperature change trend, reflecting the temperature level that the compartment should maintain based on factors such as the external environment and passenger density. This mapping and segmentation process ensures that the real-time dynamic data inside the vehicle can be interfaced with the reference sequences in the baseline data, providing a basis for subsequent control and optimization.

[0060] By comparing the real-time temperature flow of the first carriage with the first reference temperature sequence, the carriage temperature deviation at each moment is calculated. The carriage temperature deviation refers to the difference between the real-time carriage temperature and the ideal reference temperature, and is used to adjust the cooling capacity of the air conditioning system. The first carriage temperature deviation sequence is deviation data that changes over time, used to determine whether the current carriage temperature has reached the predetermined target and how much adjustment is needed.

[0061] By comparing the real-time passenger density flow of the first carriage with the first reference passenger density sequence, the passenger density deviation at each moment is calculated. The passenger density deviation refers to the difference between the actual passenger density and the ideal passenger density of the carriage at each moment. The first passenger density deviation sequence is the deviation data that changes over time and is used to determine whether the carriage is overloaded, i.e., the passenger density is too high, so that the system can automatically adjust the cooling output to cope with the possible increase in heat load.

[0062] The heat load regression function is a mathematical model derived from historical data or theoretical models through regression analysis. This function describes the impact of cabin temperature and passenger density on the heat load of the air conditioning system. The pre-fitted heat load regression function has been trained and adjusted based on previous data, and can convert the input temperature deviation and passenger density deviation into heat load deviation. Taking the first cabin temperature deviation sequence and the first passenger density deviation sequence as input, the heat load regression function calculates the first heat load deviation sequence based on the current temperature deviation and passenger density deviation, reflecting the degree of deviation of the cabin's cooling demand from the preset value due to changes in cabin temperature and passenger density.

[0063] Based on the first heat load deviation sequence, the required cooling power compensation is calculated. If the heat load deviation is positive, it indicates that the temperature inside the carriage is too high or the passenger density is too high, requiring an increase in cooling power; conversely, if the deviation is negative, the cooling power needs to be reduced. The output of the first dynamic compensation amount of cooling power is to adjust the cooling output of the air conditioning system so that the actual temperature and passenger density inside the carriage are as close as possible to the target value.

[0064] Furthermore, based on the K multi-mode power state data, a power health trend analysis is performed to output K predicted power redundancy margin values. The method also includes:

[0065] A310: Construct a first multimodal characteristic time-series curve using first multimodal power state data; A320: Compare the first multimodal characteristic time-series curve with the first multimodal health reference baseline to fit the first multimodal performance degradation curve; A330: Perform cross-modal fusion of the first multimodal performance degradation curve based on preset degradation correlation weight allocation to output a first power health trend curve; A340: Use the first cooling allocation time-series baseline as the load demand, combine it with the first power health trend curve to perform probabilistic simulation, and output a first power redundancy margin prediction value.

[0066] Power state data is arranged in chronological order and plotted as a curve to describe the trend of power performance changes over time. By combining power data from different modes, a first multimodal characteristic time series curve is constructed. This curve can show the changes in power state at different points in time, as well as the power's operating efficiency and health status.

[0067] The actual collected first multimodal characteristic time-series curve is compared with the preset first multimodal health reference baseline to identify deviations in power supply performance and analyze the gap between the current operating state and the normal operating state. Through the comparison process, the trend of power supply performance degradation is identified, and based on the deviation between the actual data and the health baseline, a first multimodal performance degradation curve is fitted. This curve reflects the trend of power supply performance gradually degrading over time.

[0068] Attenuation-related weight allocation refers to assigning corresponding weights to each mode based on the degradation status of different power modules. Each mode has different importance in the health assessment, thus requiring weight allocation. Weight allocation is based on historical data, failure modes, or power module characteristics. For example, voltage changes have a greater impact on the power supply's health, therefore they are given higher weights during fusion. Cross-modal fusion refers to combining the attenuation curves of different modes to form a comprehensive health trend. After cross-modal fusion, a first power supply health trend curve is output, which comprehensively reflects the overall health status of the power supply and potential future performance changes.

[0069] Probabilistic simulations are performed by combining the first cooling allocation timing baseline and the first power supply health trend curve. The simulation process integrates the degradation of power supply performance and changes in cooling demand to predict the possible future power redundancy margin of the power supply. The probabilistic simulation covers the impact of power supply performance changes on cooling demand satisfaction under different conditions, simulating the system performance in various scenarios. Based on the simulation results, the predicted value of the first power redundancy margin is output, that is, the predicted remaining redundant power of the power supply at a certain point in the future.

[0070] Furthermore, the method also includes:

[0071] A150: Retrieve the line operation sequence position from the train dispatching platform according to the operation number of the urban rail train; A160: Perform fine-grained matching and segmentation of the external environment time sequence data from the environmental data monitoring platform based on the line operation sequence position; A170: Fit the K passenger density time sequence baselines and K carriage temperature time sequence baselines using the line operation time window of the line operation sequence position as the rolling time reference window.

[0072] Each urban rail train has a unique operation number during scheduling and operation, which identifies its operational status. This operation number is associated with the train's operating time, route, and carriage configuration, and is used to query operational information related to the train. The train dispatching platform is a central management system responsible for tracking the real-time location, running sequence, and route information of urban rail trains. Based on the train's operation number, the system retrieves the train's route running sequence location from the train dispatching platform, i.e., the train's current and future location and time.

[0073] The environmental data monitoring platform is a system used to monitor external environmental parameters in real time. It provides time-series data of the external environment, recording external environmental conditions based on geographical location and time changes. Fine-grained matching segmentation refers to accurately matching the time-series data of the external environment obtained from the environmental data monitoring platform according to the train's operational sequence position. This subdivides the environmental data into multiple small time windows, ensuring that each time segment precisely corresponds to the actual time-series position of the train. Through this segmentation, it is ensured that the environmental changes faced by the train during operation are synchronized with its actual operating path, providing real-time external environmental information for subsequent decisions such as those made by the air conditioning system.

[0074] The location data of the train's route provides its path and time. Based on this location data, a rolling time reference window is determined, which represents the train's operating range within a specific time period. The rolling time reference window indicates the train's operating range during a certain time period and is used to synchronize the train's operating status with environmental changes. Through the rolling time reference window, K passenger density time-series baselines and K carriage temperature time-series baselines are fitted to ensure that these baseline data reflect the actual needs of the train at different locations and times. The passenger density time-series baseline represents the change in the number of passengers or passenger density in the carriages during different time periods; the carriage temperature time-series baseline represents the ideal temperature change trend inside the carriages.

[0075] Example 2, based on the same inventive concept as the dual-redundant inverter power supply control method for urban rail train air conditioning in the aforementioned examples, such as... Figure 2 As shown in the figure, this application embodiment provides a dual-redundant inverter power supply control system for urban rail train air conditioning, the system comprising:

[0076] The control optimization solution module 10 is used to perform multi-car collaborative control optimization based on the external environmental time-series data as dynamic boundary conditions, and according to the K passenger density time-series baselines and K car temperature time-series baselines of the K connected cars in the urban rail train, to generate K cooling distribution time-series baselines. The time-series state tracking module 20 is used to track the time-series states of the K connected cars and the K dual-redundant frequency converters during the process of using the K cooling distribution time-series baselines to control the K dual-redundant frequency converters to perform differentiated cooling in the K connected cars, thereby obtaining K in-car environmental dynamic parameter flows and... K multimodal power supply status data; a power health trend analysis module 30, used to perform power health trend analysis based on the K multimodal power supply status data, and output K predicted values ​​of power redundancy margin; a demand deviation identification module 40, used to identify demand deviations based on the K in-vehicle environment dynamic parameter streams, and output K dynamic compensation values ​​of cooling power; a dual-dimensional resilience control module 50, used by the K edge dual-redundant control units to execute dual-dimensional resilience control of the K dual-redundant frequency converters based on the K predicted values ​​of power redundancy margin and the K dynamic compensation values ​​of cooling power through a distributed consensus negotiation algorithm.

[0077] Furthermore, the two-dimensional toughness control module 50 is used to perform the following operation steps:

[0078] The K edge dual-redundant control units encapsulate K carriage identifiers, K predicted power redundancy margins, and K dynamic cooling power compensation values ​​into K local state vectors. The K edge dual-redundant control units broadcast the K local state vectors through the vehicle network so that the K edge dual-redundant control units can receive a global state vector. Based on the global state vector, the K edge dual-redundant control units perform distributed collaborative computation in parallel to achieve consistent state allocation and output K multi-objective control instruction sets. After verifying the safety interlock by broadcasting the K multi-objective control instruction sets through the vehicle network, the K edge dual-redundant control units atomically execute the cross-carriage load collaborative scheduling of the K dual-redundant frequency converters.

[0079] Furthermore, the two-dimensional toughness control module 50 is used to perform the following operation steps:

[0080] S1: The K edge dual-redundant control units generate K initial control instruction sets based on the global state vector; S2: The K initial control instruction sets are broadcast through the vehicle network so that the K edge dual-redundant control units can receive the global initial instruction sets; S3: The K edge dual-redundant control units perform local collaborative optimization calculations in parallel based on the global initial instruction sets and the global state vector, and output K updated control instruction sets; S4: After calculating the K difference norms of the K initial control instruction sets and the K updated control instruction sets for convergence evaluation, the K edge dual-redundant control units broadcast the K updated control instruction sets through the vehicle network so that the K edge dual-redundant control units can receive the global updated instruction sets; Iterate steps S3 to S4 until the local convergence evaluation feedback of the K edge dual-redundant control units all meet the preset consistency threshold, and output the K multi-objective control instruction sets.

[0081] Furthermore, the multi-objective control instruction set includes local load allocation instructions and cross-carriage power coordination instructions.

[0082] Furthermore, the two-dimensional toughness control module 50 is used to perform the following operation steps:

[0083] S31: The second edge dual-redundant control unit extracts control instructions from neighboring units based on the global initial instruction set to obtain the first initial control instruction set of the first edge dual-redundant control unit and the third initial control instruction set of the third edge dual-redundant control unit; S32: With the objectives of minimizing the cooling power deviation of the carriage, maximizing the redundancy utilization rate, and minimizing the difference between neighboring instructions, the local conflict resolution of the second initial control instruction set, the first initial control instruction set, and the third initial control instruction set is performed, and the second updated control instruction set is output.

[0084] Furthermore, the control optimization solution module 10 is used to perform the following operation steps:

[0085] After spatiotemporally aligning the time-series data of the external environment, the first passenger density time-series baseline, and the first carriage temperature time-series baseline, a sliding time window fine-grained segmentation is used to obtain multiple overlapping time-series scene segments. Similarity matching is performed on these multiple time-series scene segments in a control case library to retrieve multiple historical control segments. These historical control segments are smoothed to generate a first preliminary cooling allocation curve for the first connected carriage. By analogy, K preliminary cooling allocation curves for the K connected carriages are constructed, and conflict resolution is performed on the K preliminary cooling allocation curves based on the thermal coupling relationship of the K connected carriages, outputting the K cooling allocation time-series baselines.

[0086] Furthermore, the demand deviation identification module 40 is used to perform the following operation steps:

[0087] The dynamic parameter flow of the first in-vehicle environment is decomposed to obtain the real-time temperature flow and the real-time passenger density flow of the first compartment. Based on the time series information of the dynamic parameter flow of the first in-vehicle environment, the first reference passenger density sequence and the first reference temperature sequence are mapped and segmented from the first passenger density time series baseline and the first compartment temperature time series baseline. The real-time temperature flow and the first reference temperature sequence of the first compartment are compared to calculate the temperature deviation sequence of the first compartment. The real-time passenger density flow and the first reference passenger density sequence of the first compartment are compared to calculate the passenger density deviation sequence of the first compartment. The temperature deviation sequence and the passenger density deviation sequence of the first compartment are input into a pre-fitted heat load regression function to calculate the heat load deviation sequence. The first cooling power dynamic compensation amount is calculated and output based on the first heat load deviation sequence.

[0088] Furthermore, the power health trend analysis module 30 is used to perform the following operation steps:

[0089] A first multimodal characteristic time-series curve is constructed using first multimodal power state data; the first multimodal characteristic time-series curve is compared with a first multimodal health reference baseline to fit a first multimodal performance degradation curve; cross-modal fusion of the first multimodal performance degradation curve is performed based on a preset degradation correlation weight allocation to output a first power health trend curve; the first cooling allocation time-series baseline is used as the load demand, and probabilistic simulation is performed in combination with the first power health trend curve to output a first power redundancy margin prediction value.

[0090] Furthermore, the control optimization solution module 10 is used to perform the following operation steps:

[0091] Based on the operation number of the urban rail train, the line operation sequence position is retrieved from the train dispatching platform; based on the line operation sequence position, fine-grained matching and segmentation of the external environment time sequence data is performed from the environmental data monitoring platform; using the line operation time window of the line operation sequence position as the rolling time reference window, the K passenger density time sequence baselines and K carriage temperature time sequence baselines are fitted.

[0092] Through the foregoing detailed description of a dual-redundant inverter power supply control method for urban rail train air conditioning, those skilled in the art can clearly understand the dual-redundant inverter power supply control system for urban rail train air conditioning in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to in the method section.

[0093] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for controlling a dual-redundant inverter power supply for an air conditioning system in urban rail trains, characterized in that, The method includes: Using the time series data of the external environment as dynamic boundary conditions, the multi-car cooperative control optimization solution is performed based on the K passenger density time series baselines and K car temperature time series baselines of K connected cars in the urban rail train, generating K cooling distribution time series baselines. After spatiotemporally aligning the time-series data of the external environment, the first passenger density time-series baseline, and the first carriage temperature time-series baseline, multiple overlapping time-series scene segments are obtained by fine-grained segmentation using a sliding time window. The control case library is used to perform similarity matching on the multiple time-series scene segments in order to retrieve multiple historical control segments; Smooth the multiple historical control segments to generate the first preliminary cooling distribution curve for the first connected carriage; After constructing the K preliminary cooling distribution curves for the K connected carriages by analogy, the conflict resolution of the K preliminary cooling distribution curves is performed according to the thermal coupling relationship of the K connected carriages, and the K cooling distribution timing baselines are output. During the process of using the K cooling distribution timing baselines to control the K dual-redundant frequency converters to perform differentiated cooling for the K connected carriages, the K edge dual-redundant control units perform timing state tracking of the K connected carriages and the K dual-redundant frequency converters to obtain K in-vehicle environment dynamic parameter streams and K multi-modal power supply state data. Based on the K multimodal power state data, a power health trend analysis is performed, and K predicted power redundancy margin values ​​are output. The first multimodal characteristic time-series curve is constructed using the first multimodal power state data; The first multimodal health reference baseline is used to compare the first multimodal characteristic time-series curve, and the first multimodal performance degradation curve is fitted. Based on the preset attenuation correlation weight allocation, the first multimodal performance attenuation curve is fused across modes to output the first power supply health trend curve. Using the first cooling distribution timing baseline as the load demand, and combining it with the first power health trend curve, a probabilistic simulation is performed to output the predicted value of the first power redundancy margin. Based on the K in-vehicle environment dynamic parameter streams, demand deviation is identified, and K dynamic compensation values ​​for cooling power are output. The K edge dual-redundant control units execute the dual-dimensional resilience control of the K dual-redundant frequency converters through a distributed consensus negotiation algorithm based on the K predicted power redundancy margins and the K dynamic compensation values ​​for cooling power. The K edge dual-redundant control units encapsulate K carriage identifiers, the K power redundancy margin prediction values, and the K cooling power dynamic compensation values ​​as K local state vectors; The K edge dual-redundant control units broadcast the K local state vectors through the vehicle network so that the K edge dual-redundant control units can receive the global state vector; The K edge dual-redundant control units perform distributed collaborative computation in parallel based on the global state vector to achieve consistent state allocation and output K multi-objective control instruction sets; After verifying the safety interlock by broadcasting the K multi-target control instruction sets through the vehicle network, the cross-carriage load collaborative scheduling of the K dual-redundant frequency converters is atomically executed by the K edge dual-redundant control units.

2. The method for controlling a dual-redundant inverter power supply for an urban rail train air conditioning system as described in claim 1, characterized in that, The K edge dual-redundant control units perform distributed collaborative computation in parallel based on the global state vector to achieve consistent state allocation and output K multi-objective control instruction sets. The method further includes: S1: The K edge dual-redundant control units generate K initial control instruction sets based on the global state vector; S2: Broadcast the K initial control command sets via the vehicle network so that the K edge dual-redundant control units receive the global initial command sets; S3: The K edge dual-redundant control units perform local collaborative optimization calculations in parallel based on the global initial instruction set and the global state vector, and output K update control instruction sets; S4: After calculating the K difference norms of the K initial control command sets and the K updated control command sets for convergence evaluation, the K edge dual-redundant control units broadcast the K updated control command sets through the vehicle network so that the K edge dual-redundant control units receive the global update command set; Iterate through steps S3 to S4 until the local convergence evaluation feedback of the K edge dual-redundant control units all meet the preset consistency threshold, and then output the K multi-objective control instruction sets.

3. The method for controlling a dual-redundant inverter power supply for an urban rail train air conditioning system as described in claim 2, characterized in that, The multi-objective control instruction set includes local load allocation instructions and cross-carriage power coordination instructions.

4. The method for controlling a dual-redundant inverter power supply for an urban rail train air conditioning system as described in claim 3, characterized in that, The K edge dual-redundant control units perform local collaborative optimization calculations in parallel based on the global initial instruction set and the global state vector, outputting K updated control instruction sets. The method further includes: The second edge dual-redundant control unit extracts neighbor unit control instructions based on the global initial instruction set to obtain the first initial control instruction set of the first edge dual-redundant control unit and the third initial control instruction set of the third edge dual-redundant control unit. With the goals of minimizing the cooling power deviation of the carriage, maximizing the redundancy utilization rate, and minimizing the differences between neighboring instructions, local conflict resolution is performed on the second initial control instruction set, the first initial control instruction set, and the third initial control instruction set, and the second updated control instruction set is output.

5. A method for controlling a dual-redundant inverter power supply for an urban rail train air conditioning system as described in claim 1, characterized in that, Based on the K in-vehicle environmental dynamic parameter streams, demand deviation is identified, and K dynamic compensation values ​​for cooling power are output. The method further includes: Decompose the dynamic parameter flow of the first car interior environment to obtain the real-time temperature flow and the real-time passenger density flow of the first car compartment; Based on the time-series information of the first in-vehicle environment dynamic parameter stream, the first reference passenger density sequence and the first reference temperature sequence are mapped and segmented from the first passenger density time-series baseline and the first compartment temperature time-series baseline. Compare the real-time temperature flow of the first carriage with the first reference temperature sequence to calculate the temperature deviation sequence of the first carriage. Compare the real-time passenger density flow of the first carriage with the first reference passenger density sequence to calculate the first passenger density deviation sequence; Input the first carriage temperature deviation sequence and the first passenger density deviation sequence into the prefitted heat load regression function to calculate the first heat load deviation sequence; The first dynamic compensation amount of cooling power is calculated and output based on the first heat load deviation sequence.

6. The method for controlling a dual-redundant inverter power supply for an urban rail train air conditioning system as described in claim 1, characterized in that, The method further includes: Based on the operation number of the urban rail train, retrieve the line operation sequence position from the train dispatching platform; Based on the time sequence location of the route during operation, fine-grained matching and segmentation of the time sequence data of the external environment are performed from the environmental data monitoring platform; Using the line operation time window at the line operation sequence position as the rolling time reference window, the K passenger density time series baselines and K carriage temperature time series baselines are fitted.

7. A dual-redundant inverter power supply control system for urban rail train air conditioning, characterized in that, For implementing the dual-redundant inverter power supply control method for urban rail train air conditioning according to any one of claims 1-6, the system comprises: The control optimization solution module is used to perform multi-car collaborative control optimization solution based on the time series data of the external environment as dynamic boundary conditions, and the time series baselines of passenger density and temperature of K connected carriages in the urban rail train, to generate K cooling distribution time series baselines. The timing state tracking module is used to track the timing state of the K connected carriages and the K dual-redundant frequency converters during the process of using the K cooling distribution timing baselines to control the K dual-redundant frequency converters to perform differentiated cooling of the K connected carriages, thereby obtaining K in-vehicle environmental dynamic parameter streams and K multi-modal power supply status data. The power health trend analysis module is used to perform power health trend analysis based on the K multi-mode power status data and output K power redundancy margin prediction values. The demand deviation identification module is used to identify demand deviations based on the K in-vehicle environment dynamic parameter streams and output K dynamic compensation values ​​for cooling power. The dual-dimensional resilience control module is used by the K edge dual-redundant control units to execute dual-dimensional resilience control of the K dual-redundant frequency converters based on the K predicted power redundancy margins and K dynamic compensation values ​​for cooling power through a distributed consensus negotiation algorithm.

Images