Combination mutual standby direct current power supply system

Abstract

The invention discloses a combined mutual standby DC power supply system, and relates to the technical field of power supply management, and the system comprises the steps: building a bus dynamic feature vector; performing time domain-frequency domain joint reconstruction on the dynamic feature vector of the bus, extracting an inertia coefficient, an impedance change slope and a power phase drift amount corresponding to the bus, and establishing a bus state mapping model; executing reverse mapping of the bus dynamic response by using the bus state mapping model, and generating a virtual state vector of each direct current power supply unit; according to the time sequence evolution difference of the virtual state vector, constructing a relative deviation matrix between the direct current power supply units, and identifying an abnormal power supply behavior; and carrying out dynamic power supply switching management by utilizing an identification result. The technical problems that in the prior art, the dynamic response of the direct current bus cannot be recognized in real time, and abnormal detection and switching management of the power supply unit are lagged are solved, and the technical effects of improving the power supply reliability of the system and achieving self-adaptive mutual backup and dynamic switching among multiple units are achieved.

Description

Technical Field

[0001] This invention relates to the field of power management technology, and more specifically to a combined and mutually redundant DC power supply system. Background Technology

[0002] In traditional DC power supply systems, bus voltage and current signals are mostly monitored as steady-state parameters, which makes it difficult to reflect dynamic changes under load disturbances in a timely manner, resulting in the inaccurate identification of bus dynamic response characteristics. At the same time, the status judgment between power supply units is usually based on fixed logic or periodic detection methods, lacking rapid perception and analysis of transient anomalies, resulting in lag in anomaly detection. This leads to slow response and insufficient stability during power supply switching, making it difficult to achieve efficient coordination and mutual backup among multiple power supply units. Summary of the Invention

[0003] This application provides a combined backup DC power supply system to address the technical problems in the prior art, such as the inability to identify the dynamic response of the DC bus in real time and the lag in the detection and switching management of power supply unit anomalies.

[0004] In view of the above problems, this application provides a combined backup DC power supply system.

[0005] This application provides a combined and mutually redundant DC power supply system, the system comprising: The monitoring module is used to collect voltage perturbations, instantaneous current fluctuations, and ripple spectrum characteristics of the DC bus in real time, and establish a dynamic feature vector of the bus, which characterizes the dynamic response of the bus under load disturbance. The dynamic feature reconstruction module is used to perform time-domain and frequency-domain joint reconstruction of the dynamic feature vector of the bus, extract the corresponding inertia coefficient, impedance change slope, and power phase drift of the bus, and establish a bus state mapping model based on the extraction results. The inverse mapping module is used to perform inverse mapping of the dynamic response of the bus using the bus state mapping model, and generate virtual state vectors for each DC power supply unit. The virtual state vectors include unity power gain response coefficient, instantaneous current reflection characteristics, and output stability margin. The anomaly identification module is used to construct a relative deviation matrix between DC power supply units based on the temporal evolution differences of the virtual state vectors, and identify abnormal power supply behavior. The management module is used to perform dynamic power supply switching management based on the identification results.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application acquires real-time voltage perturbations, instantaneous current fluctuations, and ripple spectrum characteristics of the DC bus to establish a dynamic feature vector for the bus. This dynamic feature vector characterizes the dynamic response of the bus under load disturbances. The dynamic feature vector is reconstructed in both the time and frequency domains to extract the corresponding inertia coefficient, impedance change slope, and power phase drift. A bus state mapping model is established based on the extraction results. The bus state mapping model is used to perform inverse mapping of the bus dynamic response, generating virtual state vectors for each DC power supply unit. These virtual state vectors include a unity power gain response coefficient, instantaneous current reflection characteristics, and output stability margin. Based on the temporal evolution differences of the virtual state vectors, a relative deviation matrix between DC power supply units is constructed to identify abnormal power supply behavior. The identification results are used for dynamic power supply switching management. This invention addresses the technical problems of the inability to identify the dynamic response of DC buses in real time and the lag in the detection and switching management of power supply unit anomalies in the prior art. By establishing a dynamic feature vector of the bus, constructing a bus state mapping model, and realizing the linkage management of virtual state inversion and anomaly identification of power supply units, it achieves the technical effects of improving the reliability of system power supply and realizing adaptive mutual backup and dynamic switching among multiple units. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 A schematic diagram of a combined and backup DC power supply system provided in this application embodiment; Figure 2 This is a schematic diagram illustrating the process of constructing a relative deviation matrix in a combined backup DC power supply system provided in an embodiment of this application.

[0009] Figure labeling: Monitoring module 11, Dynamic feature reconstruction module 12, Inverse mapping module 13, Anomaly identification module 14, Management module 15. Detailed Implementation

[0010] This application provides a combined backup DC power supply system to address the technical problems in the prior art, such as the inability to identify the dynamic response of the DC bus in real time and the lag in the detection and switching management of power supply unit anomalies. By establishing a dynamic feature vector of the bus, constructing a bus state mapping model, and realizing the linkage management of virtual state inversion and anomaly identification of power supply units, the system achieves the technical effects of improving power supply reliability and realizing adaptive backup and dynamic switching among multiple units.

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0012] It should be noted that any variation of the terms "comprising" and "having" is intended to cover non-exclusive inclusion, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such processes, systems, products, or devices.

[0013] Examples, such as Figure 1 As shown, this application provides a combined backup DC power supply system, the system comprising: Monitoring module 11 is used to collect voltage disturbances, instantaneous current fluctuations and ripple spectrum characteristics of the DC bus in real time, and establish a dynamic feature vector of the bus, which characterizes the dynamic response of the bus under load disturbance.

[0014] In this embodiment, the monitoring module 11 uses a synchronous sampling circuit to collect the bus voltage and current signals in real time by setting high-precision voltage sensors and Hall current sensors at both ends of the DC bus. In this process, the voltage sensor first detects the transient changes of the bus under load disturbance conditions to obtain voltage perturbations; then, the Hall current sensor detects the dynamic changes of the bus current to obtain instantaneous current fluctuations; finally, the collected voltage and current data are sent to a spectrum analysis circuit, where a Fast Fourier Transform (FFT) is used to extract the ripple spectrum characteristics of the bus in the frequency domain, reflecting the energy distribution characteristics and ripple features of the bus at different frequencies.

[0015] After signal acquisition and analysis, monitoring module 11 performs amplitude normalization and time alignment processing on voltage perturbations, instantaneous current fluctuations, and ripple spectrum characteristics to ensure data consistency. Finally, the three types of signals are comprehensively processed to form a bus dynamic feature vector characterizing the dynamic response characteristics of the bus. This bus dynamic feature vector is used to reflect the real-time dynamic changes of the bus under load disturbances.

[0016] The dynamic feature reconstruction module 12 is used to perform time-domain-frequency domain joint reconstruction on the dynamic feature vector of the bus, extract the corresponding inertia coefficient, impedance change slope and power phase drift of the bus, and establish a bus state mapping model based on the extraction results.

[0017] In this embodiment, when the dynamic feature reconstruction module 12 performs time-domain and frequency-domain joint reconstruction of the bus dynamic feature vector, it first performs time-domain reconstruction processing. During this process, by extracting the voltage and current change data of the bus before and after the load disturbance, and expanding them on the time axis, a voltage response curve is obtained. Then, the dynamic feature reconstruction module 12 performs fitting analysis on the rise time, overshoot amplitude, and settling time of this response curve. The rise time reflects the time required for the bus to recover its main amplitude from the start of the disturbance; the overshoot amplitude represents the maximum deviation of the voltage from the steady-state value; and the settling time represents the time it takes for the voltage to reach steady state again after the disturbance. Through comprehensive fitting of these three characteristic parameters, the parameter characterizing the inertial characteristics of the bus, namely the inertia coefficient, is calculated.

[0018] Subsequently, the dynamic feature reconstruction module 12 reconstructs the dynamic feature vector of the bus in the frequency domain. During this process, spectral analysis is performed on the voltage and current data to extract the voltage and current amplitudes at each frequency, and their ratio is calculated to obtain the equivalent impedance of the bus. Impedance variation data within the main operating frequency range are selected, and a straight line is fitted with frequency as the abscissa and impedance variation as the ordinate to obtain the average trend of impedance variation. The slope of this trend is the impedance variation slope, used to reflect the impedance response characteristics and energy transmission stability of the bus under frequency variations.

[0019] Next, the dynamic feature reconstruction module 12 compares the voltage phase and current phase obtained from the frequency domain analysis. By calculating the phase difference between the two before and after the disturbance, the power phase drift is obtained. The power phase drift represents the phase shift of the bus due to load changes during power transmission, reflecting the dynamic changes in power flow.

[0020] Finally, the dynamic feature reconstruction module 12 correlates and models the three types of feature quantities, namely inertia coefficient, impedance change slope and power phase drift, and establishes a bus state mapping model based on their change law.

[0021] The reverse mapping module 13 is used to perform reverse mapping of the bus dynamic response using the bus state mapping model to generate virtual state vectors for each DC power supply unit. The virtual state vectors include unity power gain response coefficient, instantaneous current reflection characteristics, and output stability margin.

[0022] In this embodiment, when the inverse mapping module 13 performs the inverse mapping of the bus dynamic response using the bus state mapping model, it first constructs a nonlinear reversible subset of operators to describe the correspondence between the bus input and the power supply unit output based on the mutual information relationship between the bus state mapping model and the bus dynamic feature vector. Then, the bus dynamic feature vector is input into the nonlinear reversible subset of operators, and through multi-operator collaborative iteration, the response results of each DC power supply unit under the current bus dynamic conditions are calculated, generating a virtual state candidate set containing multiple candidate states. Finally, with constraints of global power conservation of the bus, node current consistency, and transient energy balance, the virtual state candidate set is optimized and screened using multiple objectives to generate the virtual state vector of each DC power supply unit. This virtual state vector contains three types of characteristic parameters: unity power gain response coefficient, instantaneous current reflection characteristics, and output stability margin, used to characterize the actual response capability and operational stability of each power supply unit under bus dynamic disturbances.

[0023] Furthermore, in the system provided in the application embodiment, the reverse mapping module 13, which uses the bus state mapping model to perform reverse mapping of the bus dynamic response and generate virtual state vectors for each DC power supply unit, further includes: Based on the mutual information relationship between the bus state mapping model and the bus dynamic feature vector, a nonlinear reversible subset of the bus dynamic response is constructed. This nonlinear reversible subset is built by performing polynomial perturbation mapping training in the three-dimensional parameter space of inertia coefficient, impedance change rate, and power phase drift. The bus dynamic feature vector is input into the nonlinear reversible subset, and a virtual state candidate set for each DC power supply unit is generated through multi-operator collaborative iteration. The virtual state candidate set is then optimized and screened using multi-objective conditions based on the constraints of global power conservation of the bus, node current consistency, and transient energy balance, and the virtual state vector is output.

[0024] In this embodiment, the correlation between the inertia coefficient, impedance change rate, and power phase drift in the bus dynamic response process is first determined based on the mutual information relationship between the bus dynamic characteristic vector and the bus state mapping model. During this process, by analyzing bus operation data under different operating conditions, the mutual influence patterns among the three parameters are extracted, and a three-dimensional parameter space reflecting the coupling relationship between parameters is established. Within this three-dimensional parameter space, small-amplitude perturbations are applied to the inertia coefficient, impedance change rate, and power phase drift, respectively, and the changes in bus voltage, current, and power after the perturbation are collected. The relationship between parameter changes and response changes is fitted in the form of a polynomial function. Through continuity and reversibility verification of the fitted curves, a polynomial mapping relationship capable of conversion between forward and reverse directions is formed. Based on this, a nonlinear reversible subset of the bus dynamic response is constructed to achieve the reverse derivation of the bus dynamic characteristic vector to the operating characteristics of each DC power supply unit.

[0025] After constructing the nonlinear reversible operator subset, the bus dynamic characteristic vector is sequentially input into the operator subset, and the response results under different parameter combinations are obtained through multiple iterations. In each iteration, the actual values ​​of the inertia coefficient, impedance change rate, and power phase drift are substituted, and the output results are compared with the reference data of the bus state mapping model until the calculation results converge, generating a set of virtual state candidates representing the operating state of each DC power supply unit. Each virtual state candidate contains the possible power distribution characteristics, current response characteristics, and output fluctuations of the DC power supply unit under bus dynamic conditions, constituting the virtual state candidate set.

[0026] Next, physical constraint analysis is performed on the generated virtual state candidate set, verifying it based on the bus global power conservation, node current consistency, and transient energy balance conditions. Power conservation is used to confirm that the bus input power and the output power of each power supply unit remain balanced before and after the disturbance; node current consistency is used to verify the coordination of bus current distribution; and transient energy balance is used to evaluate the stability of energy exchange between the bus and power supply units. After multi-constraint screening, candidates that do not meet the conditions are eliminated, and only the operating states that meet the dynamic balance are retained.

[0027] Among the candidates that meet the constraints, the unity power gain response coefficient, instantaneous current reflection characteristics, and output stability margin are extracted. The unity power gain response coefficient is obtained by comparing the change in input power before and after the bus power disturbance with the change in output power of the power supply unit, and calculating the ratio between the two, reflecting the power supply unit's response sensitivity to power disturbances. The instantaneous current reflection characteristics are obtained by comparing the direction of change in bus voltage with the direction of change in power supply unit current, and calculating the amplitude ratio of the portion of the current opposite to the voltage direction, used to describe the transient energy feedback characteristics. The output stability margin is obtained by measuring the ratio of the maximum overshoot amplitude of the output voltage after the disturbance to the steady-state recovery time, used to evaluate the dynamic stability level of the power supply unit during the disturbance process. Finally, the inverse mapping module 13 integrates the three parameters to form a virtual state vector.

[0028] The anomaly identification module 14 is used to construct a relative deviation matrix between DC power supply units based on the temporal evolution differences of the virtual state vector, and to identify abnormal power supply behavior.

[0029] In this embodiment, when the anomaly identification module 14 constructs the relative deviation matrix between DC power supply units based on the temporal evolution differences of the virtual state vectors, it first aligns the virtual state vectors of each DC power supply unit temporally and calculates the difference sequence of power phase drift and impedance change slope. Then, it performs multi-scale wavelet decomposition and spectral entropy analysis on the difference sequence to obtain a dynamic spectral entropy matrix reflecting the dynamic characteristics differences between different power supply units. Next, it performs an orthogonal decoupling transformation on the dynamic spectral entropy matrix to construct the relative deviation matrix between DC power supply units, and uses the spectral radius or characteristic entropy of the relative deviation matrix as an analysis index to generate a first anomaly factor. By comparing the first anomaly factor with a set threshold, the anomaly identification module 14 identifies and locates abnormal power supply behavior occurring in the bus system.

[0030] Furthermore, such as Figure 2 As shown, in the system provided in the application embodiment, the anomaly identification module 14, which constructs a relative deviation matrix between DC power supply units based on the temporal evolution differences of the virtual state vector, further includes: After aligning the virtual state vectors of each DC power supply unit in time, the difference sequence of power phase drift and impedance change slope is calculated. Multi-scale wavelet decomposition and spectral entropy analysis are performed on the difference sequence to generate a dynamic spectral difference entropy matrix between each DC power supply unit pair. After orthogonal decoupling transformation of the dynamic spectral difference entropy matrix, a relative deviation matrix is ​​constructed. Threshold identification is performed on the spectral radius or characteristic entropy of the relative deviation matrix to construct a first anomaly factor. The first anomaly factor is used to identify abnormal power supply behavior.

[0031] In this embodiment, the virtual state vectors of each DC power supply unit are first time-aligned. Through linear interpolation and time synchronization, the virtual state sequences of different power supply units are unified to the same time scale to ensure the comparability of their dynamic responses. Two types of parameters, power phase drift and impedance change slope, are extracted from the aligned time sequence. A time difference sequence is obtained by calculating the difference between adjacent time points. Simultaneously, different power supply units are paired and differentially analyzed at the same time point to form a power phase drift difference sequence and an impedance change slope difference sequence.

[0032] Next, a multi-scale wavelet decomposition method is used to decompose the above difference sequence. Through discrete wavelet transform, each difference sequence is decomposed into low-frequency trend components and high-frequency disturbance components at different time scales. The low-frequency trend component represents the overall steady-state change trend of the system, while the high-frequency disturbance component reveals the transient dynamic characteristics of the power supply unit under short-term disturbances. After completing the multi-scale decomposition, spectral entropy analysis is performed on each scale component. By calculating the power spectral density of each component and normalizing the energy distribution, the spectral entropy value is calculated according to the Shannon entropy formula to characterize the degree of energy dispersion of the signal in the frequency domain. A larger spectral entropy value indicates a more unstable and complex system response. The spectral entropy values ​​of different DC power supply units at the same time scale are paired and compared, their differences are calculated, and matrix processing is performed to form a dynamic spectral difference entropy matrix.

[0033] Subsequently, orthogonal decoupling transformation is used to perform eigenvalue decomposition on the dynamic spectral difference entropy matrix, removing linear correlations and redundant couplings between different power supply units, so that the dynamic characteristics are presented in the form of mutually independent principal components in the orthogonal space. The orthogonalized matrix obtained by this transformation is the relative deviation matrix, and its elements reflect the deviation intensity of the dynamic response between the corresponding power supply units at the same time.

[0034] Finally, spectral characteristic analysis is performed on the relative deviation matrix, and the dynamic stability of the system is quantified by calculating the spectral radius and characteristic entropy. The spectral radius measures the concentration of the maximum dynamic deviation, while the characteristic entropy measures the complexity of the overall system response. The calculation results are compared with a pre-set threshold. When the spectral radius or characteristic entropy exceeds the threshold, a first anomaly factor is constructed. The first anomaly factor serves as an anomaly detection index to identify abnormal power supply behavior between DC power supply units, thereby enabling the identification of dynamic imbalance states of the bus and the detection of abnormal power supply behavior.

[0035] Furthermore, in the system provided in the application embodiment, the anomaly identification module 14, which uses the first anomaly factor to identify abnormal power supply behavior, further includes: The power supply control parameters of each DC power supply unit are read, including the output voltage setpoint, conduction angle, PWM modulation ratio, and feedback gain coefficient. After configuring the control parameter response model, the power supply control parameters are input into the control parameter response model to calculate the theoretical power phase drift and output stability margin of each DC power supply unit under the current power supply control parameters, generating a prediction deviation vector. The prediction deviation vector is mapped to the feature space of the relative deviation matrix to obtain the relative response vector. Anomaly identification based on direction angle and amplitude ratio is performed using the relative response vector and the prediction deviation vector to establish a second anomaly factor. The first anomaly factor and the second anomaly factor are logically fused to identify abnormal power supply behavior.

[0036] In this embodiment, the power supply control parameters of each DC power supply unit are first read. The output voltage setpoint, conduction angle, PWM modulation ratio, and feedback gain coefficient are extracted into the same time reference according to the DC power supply unit and time sequence. Time synchronization and linear interpolation are used to align the missing time points and unify the dimensions so that the power supply control parameters at each time point can directly participate in subsequent calculations to obtain power supply control parameters that can be used for solving.

[0037] Subsequently, the control parameter response model is configured. In this process, based on the relationship between power supply control parameters and output voltage and current, the phase relationship between voltage and current is determined by analyzing the output voltage response curve under varying modulation ratios. Then, the overshoot amplitude and recovery time of the output voltage are recorded when the voltage setpoint changes, used to describe the stabilization process. After completing these parameter tests, a mapping relationship between power supply control parameters and response changes is established. Through this process, the control parameter response model is configured. Then, the read power supply control parameters are input item by item into the control parameter response model to calculate the theoretical power phase drift and output stability margin of each DC power supply unit under the current power supply control parameters. These two results are compared with the reference state to obtain the deviation value reflecting the difference between theory and reality, and are combined sequentially into a prediction deviation vector.

[0038] Next, the predicted bias vector is mapped to the feature space of the relative bias matrix. Specifically, the relative bias matrix is ​​first decomposed into eigenvalues ​​to extract eigenvectors as reference directions. Then, the predicted bias vectors are projected onto these reference directions one by one to obtain the response components in each principal direction, and rearranged according to the order of the feature directions to form the relative response vector.

[0039] Then, anomaly identification is performed by comparing the directional angle and the amplitude ratio. The directional angle measures the consistency of the trend between the prediction deviation and the relative response vector, while the amplitude ratio assesses the degree of deviation in response intensity. During calculation, the squares of the differences between the two vectors at corresponding components are summed, and then the differences are normalized to reflect the actual degree of deviation. If the directional angle exceeds the tolerance range or the amplitude ratio deviates from the normal proportional range, the degree of deviation is quantified proportionally and used as the value of the second anomaly factor to represent the degree of abnormal response of the power supply unit under the current control parameters.

[0040] Finally, the first and second abnormal factors are logically fused. During the fusion process, the changing trends of the two types of factors are compared under the same time coordinate and DC power supply unit identifier. When both exceed their respective thresholds simultaneously within the same time interval, an abnormality identifier is output, and the corresponding time interval and power supply unit number are marked to achieve accurate identification and location of abnormal power supply behavior.

[0041] Management module 15 is used to manage dynamic power supply switching based on the identification results.

[0042] Furthermore, in the system provided in the application embodiment, the management module 15, which uses the identification results for dynamic power supply switching management, also includes: The anomaly level is output based on the identification results; the battery internal resistance of each DC power supply unit is measured in real time using a battery voltage monitor; a health priority sequence of the DC power supply units is configured based on the measurement results and the power data of the DC power supply units; the anomaly level is used to perform adaptation analysis of the health priority sequence to generate a dynamic switching strategy; and dynamic power supply switching management is executed according to the dynamic switching strategy.

[0043] In this embodiment, when the management module 15 performs dynamic power supply switching management using the identification results, it first outputs the anomaly level based on the identification results. After completing the anomaly identification, a first anomaly factor and a second anomaly factor corresponding to each DC power supply unit are obtained. These two types of factors reflect the abnormal characteristics of the power supply unit in terms of power transmission stability and control response consistency, respectively. The management module 15 uses the numerical amplitude, duration, and frequency of occurrence of these two types of factors as input parameters and compares them with a preset anomaly level threshold. For example, when the amplitude of the anomaly factor is less than 0.1 and the duration is less than 2 seconds, the unit is determined to be a level 1 anomaly; when the amplitude is between 0.1 and 0.3 and the duration is between 2 and 5 seconds, it is determined to be a level 2 anomaly; when the amplitude exceeds 0.3 or the duration exceeds 5 seconds, it is defined as a level 3 anomaly. Through this quantitative grading process, a list containing the anomaly levels of each DC power supply unit is output, which is used to characterize the risk level of each unit in the current operating cycle.

[0044] Next, a battery voltage monitor is used to measure the internal resistance of each DC power supply unit in real time, and a health priority sequence is configured based on the power data. During this process, the battery terminal voltage and output current are collected every 10 seconds, and the current internal resistance is calculated according to Ohm's law. The real-time measured internal resistance is compared with the battery's rated internal resistance to obtain the internal resistance deviation rate. For example, if the rated internal resistance is 50 milliohms, and the measured internal resistance is 60 milliohms, the deviation rate is 20%. A higher deviation rate indicates a more severe degree of battery aging. Subsequently, a comprehensive health index is calculated based on the battery's remaining power (SOC), with the internal resistance deviation rate weighted at 0.7 and the SOC weighted at 0.3. Battery units with a high health index indicate strong output capability and sufficient energy. The DC power supply units are sorted from high to low based on the health index to form a health priority sequence; for example, a unit with a health index of 0.95 participates in power supply before a unit with a health index of 0.80.

[0045] Next, an adaptation analysis is performed on the health priority sequence using anomaly levels. The anomaly level list is compared with the health priority sequence, and the final operating priority of each power supply unit is determined through comprehensive evaluation. When a unit has a high health index but an anomaly level of three, its operating priority is reduced, for example, from the first to the third position in the sequence, to prevent high-risk units from continuing to bear the main load. If a unit has a medium health index (e.g., 0.85) but an anomaly level of one, its priority can be appropriately increased when the load is stable. In this process, a weighted decision matrix is ​​used to integrate the anomaly level and the health index for calculation. For example, the final priority score = health index × 0.6 + (1 - anomaly level coefficient) × 0.4. Through this calculation, a dynamic switching strategy is generated that includes the priority of DC power supply units, switchability conditions, and load thresholds.

[0046] Finally, dynamic power supply switching management is executed according to the dynamic switching strategy. During this stage, the operating status of each unit is continuously monitored. When the anomaly level of the main power supply unit reaches the preset switching conditions, such as a sustained Level 3 anomaly for more than 3 seconds or an output voltage lower than the nominal value by 2%, the next priority unit is selected for switching operation according to the dynamic switching strategy. The switching process includes voltage matching, current paralleling, and load transfer. In this process, firstly, the output voltage of the target unit is detected, and the difference between it and the bus voltage is controlled within 0.05 volts through a pre-charging circuit; secondly, the output current of the target unit is gradually increased until it shares more than 30% of the load; finally, a smooth transition control achieves complete takeover, and the original main unit exits power supply and enters standby mode. For example, when the output voltage of main power supply unit A drops from 48V to 46.8V for 5 seconds, its anomaly level is detected as Level 3, and the standby unit B with the highest health index is automatically selected for switching. Unit B completes voltage alignment within 1 second, and then smoothly takes over all loads within 2 seconds. This method achieves a dynamic switching process with uninterrupted power supply and stable bus voltage.

[0047] Furthermore, the system provided in the application embodiments also includes: The capacity verification module is used to control the discharge voltage of each DC power supply unit to be higher than the preset voltage upper limit. It constructs a real-time capacity response curve by using the voltage perturbation feedback on the bus side and the current response signal at the DC power supply unit end, and performs full capacity verification based on the real-time capacity response curve.

[0048] In this embodiment, the capacity verification module first controls the discharge voltage of the DC power supply unit to be higher than a preset voltage upper limit, so that the unit operates in the rated voltage floating range, thereby stimulating the high-stress characteristics of the electrochemical reaction and thus exhibiting the true capacity release capability within a limited discharge cycle. For example, in a power supply unit with a rated voltage of 48V, the discharge voltage is increased to 50V, so that it exhibits more obvious voltage recovery and capacity decay characteristics under high potential conditions.

[0049] Subsequently, voltage sensors and Hall current sensors located at both ends of the busbar are used to collect voltage perturbation feedback signals from the busbar side and current response signals from the power supply unit side, respectively. The voltage perturbation feedback reflects the dynamic voltage fluctuation characteristics of the busbar during discharge, while the current response signal characterizes the output changes of the power supply unit under external disturbances.

[0050] After data acquisition, a real-time capacity response curve is constructed by correlating the changes in bus voltage perturbations with the current response of the power supply unit. In this process, the current response signal is integrated over time to obtain the output charge. The curve is plotted with the change in bus voltage on the x-axis and the output charge on the y-axis. The slope of the curve characterizes the capacity release rate, and the area under integration corresponds to the total available capacity. For example, when the voltage perturbation amplitude is 0.2V, if the corresponding output charge is 95% of the nominal value, the unit's capacity can be considered to be well maintained; if the response charge drops to 80% of the nominal value, it indicates a significant capacity decay.

[0051] Finally, the obtained real-time capacity response curve is compared with the preset standard capacity reference curve to analyze the differences between the two in terms of release rate, total capacity, and voltage recovery characteristics. When the real-time capacity curve matches the standard curve within the allowable error range, the power supply unit capacity is determined to be normal; if the curve area or slope is lower than the reference value, the capacity retention rate and attenuation rate are calculated to determine the capacity decrease. For example, when the integral area of ​​the real-time curve is 87% of the standard value, the capacity retention rate is recorded as 87%, and the attenuation rate as 13%.

[0052] Through the above process, the real-time capacity response characteristics and capacity retention rate of each DC power supply unit are obtained, and the full capacity verification is completed.

[0053] Furthermore, the system provided in the application embodiments also includes: The charging control module is used to execute reverse three-stage charging control if the full capacity verification is passed. The reverse three-stage charging control includes dynamic voltage regulation charging in the first stage, phase coordination charging in the second stage, and inertial compensation charging in the third stage.

[0054] In this embodiment, when the capacity verification module determines, through the real-time capacity response curve, that the capacity retention rate of the power supply unit is not less than 90% of the nominal capacity, and the capacity release rate deviates from the standard curve by no more than 5%, the full capacity verification is deemed successful. At this time, the charging control module executes reverse three-stage charging control.

[0055] In the reverse three-stage charging control process, the first stage is dynamic voltage stabilization charging, which gradually increases the charging voltage to near the upper limit of the rated voltage and maintains a constant voltage output. During this stage, the rate of change of the charging current is monitored in real time. When the current decline slows down and stabilizes, it indicates that the electrochemical absorption process is balanced, completing dynamic voltage stabilization charging and achieving initial energy replenishment. The second stage is phase coordination charging, which adjusts the charging pulse triggering sequence by detecting the phase difference between the bus voltage and the power supply unit current, ensuring they are in phase or slightly ahead, thereby reducing reactive power loss and improving energy transfer efficiency. For example, when the current lags the voltage by more than a preset phase deviation angle, phase synchronization is achieved by triggering the PWM signal in advance, ensuring efficient energy transfer. The third stage is inertia compensation charging. When the voltage is close to full charge and the current drops too quickly, the rate of current decline is monitored, and a short-time pulse current is injected to compensate for electrochemical inertia, allowing for full absorption of internal energy and preventing overcharging or undercharging.

[0056] Through continuous control of three stages—dynamic voltage regulation charging, phase coordination charging, and inertia compensation charging—the stability, efficiency, and balance of the power supply unit's charging process are ultimately achieved, ensuring that the goal of complete charging and capacity matching is achieved during energy recovery, thus completing the reverse three-stage charging control process.

[0057] Furthermore, the system provided in the application embodiments also includes: The anomaly reporting module is used to capture anomaly features based on the real-time capacity response curve if the full capacity verification fails. The anomaly feature capture includes capacity change rate capture and total capacity decay capture. Based on the anomaly feature capture results, the module performs anomaly level matching for the DC power supply unit and executes anomaly reporting.

[0058] In this embodiment, when the full capacity verification fails, the anomaly reporting module performs an anomaly feature capture process, identifying the performance degradation characteristics of the DC power supply unit based on the real-time capacity response curve and matching the anomaly level. In this process, the capacity change rate is first calculated based on the real-time capacity response curve, where the capacity change rate represents the rate of capacity change per unit time. By extracting capacity data from adjacent time points on the real-time capacity response curve, the ratio of capacity increment to time interval is calculated to form the capacity change rate curve. If the capacity change rate experiences a sudden drop, continuous decrease, or remains close to zero for a long time during discharge, it indicates that the DC power supply unit experiences internal chemical reaction lag, polarization, or increased resistance during energy release, thus capturing the abnormal capacity change rate characteristic.

[0059] Next, capacity decay detection is performed. The capacity retention rate is calculated by comparing the total capacity of the current discharge cycle with the rated nominal capacity. If the capacity retention rate is below 90%, and the capacity cannot be recovered to above 95% of the rated capacity during consecutive discharge cycles, the power supply unit is deemed to have an abnormal capacity decay. Simultaneously, the voltage recovery segment of the capacity response curve is analyzed. When voltage recovery lags behind capacity recovery and exhibits a significant non-linear trend, it indicates a decrease in internal reaction efficiency and an increase in polarization, verifying the existence of capacity decay.

[0060] After capturing the rate of change in capacity and the total capacity decay, anomaly level matching is performed. During this process, anomalies are classified into different levels by comparing the magnitude of the deviation in the rate of change in capacity with the percentage decrease in the capacity retention rate. A level 1 anomaly is defined as a deviation in the rate of change in capacity between 5% and 10% and a capacity retention rate below 95%; a level 2 anomaly is defined as a deviation between 10% and 20% and a capacity retention rate below 90%; and a level 3 anomaly is defined as a deviation exceeding 20% ​​and a capacity retention rate below 85%.

[0061] Finally, anomaly reporting is performed based on the anomaly level. The identified anomaly level is integrated with the corresponding capacity change rate, capacity retention rate, detection time, and power supply unit number to form anomaly reporting data, which is then transmitted to the monitoring terminal.

[0062] Furthermore, in the system provided in the application embodiment, the system also includes a self-maintenance management module, used to perform the following steps: Configure a set of simulated high-current scenarios and set an impulse discharge strategy that maps to the set of simulated high-current scenarios; use the impulse discharge strategy to perform scenario simulation tests on each DC power supply unit and establish simulation test results; perform predictive maintenance management of the DC power supply unit based on the simulation test results.

[0063] In this embodiment, the self-maintenance management module first configures a set of simulated high-current scenarios and sets an impulse discharge strategy mapped to the set. During this process, based on the historical operating data of each DC power supply unit and the changes in bus load, typical high-current operating states are selected as scenario samples. Key parameters such as current rise rate, peak current, and duration are determined and integrated to form the set of simulated high-current scenarios. The current rise rate is determined by the ratio of current change to time change; the peak current represents the maximum current intensity in the scenario; and the duration of the current at its peak is the duration of the current. Next, an impulse discharge strategy mapped to the set of simulated high-current scenarios is set. Using the bus voltage offset threshold as the trigger condition, the discharge process is divided into a ramp-up phase, a constant current phase, and a decay phase. The current rise, hold, and fall processes are controlled respectively, and corresponding voltage fluctuation ranges and sampling periods are set to ensure the stability and repeatability of the discharge process.

[0064] Subsequently, scenario simulation tests were performed on each DC power supply unit using an impulse discharge strategy. During the test, a discharge current corresponding to the scenario parameters was applied to each power supply unit. Bus voltage, unit output current, and temperature signals were synchronously acquired using voltage and current sensors, and data was recorded according to a set sampling interval. Based on the recorded data, the voltage drop amplitude was calculated as the difference between the steady-state voltage and the lowest instantaneous voltage; the voltage recovery time was the time required for the voltage to recover from the lowest point to the steady-state tolerance range; the current response time was the time from the start of discharge to the current reaching its peak value; the energy release was obtained by integrating the product of voltage and current over time; the temperature rise rate was calculated by dividing the temperature change by the time change; and the equivalent impedance of the ramp-up segment was determined by the ratio of the voltage change to the current change. The calculation results were then organized according to the power supply unit number and scenario index to form the simulation test results.

[0065] Finally, predictive maintenance management of the DC power supply unit is carried out based on the simulation test results. In this process, using reference data from the factory or last maintenance as a benchmark, the changes in voltage recovery time, temperature rise rate, energy release, and current response time relative to the benchmark values ​​are calculated. The performance status of the power supply unit is judged based on the change ratio and set thresholds. When the change of any indicator exceeds the threshold, it is marked as an object requiring inspection; when multiple indicators exceed limits simultaneously, they are marked as priority maintenance objects. Combining the marking results, an operating status table and maintenance priority sequence for the DC power supply unit are established, the order of inspection and maintenance is determined, and operations such as repair, retesting, and component replacement are arranged accordingly, achieving predictive maintenance management.

[0066] Furthermore, the system provided in the application embodiment also includes a charging control module for monitoring the charging and discharging records of each DC power supply unit and performing periodic alternating charging and discharging management.

[0067] In this embodiment, the charging control module continuously monitors and records the charging and discharging records of each DC power supply unit, including the number of charging and discharging cycles and their duration. Based on this, the charging control module performs periodic rotational charging and discharging management. According to the aforementioned charging and discharging records, it periodically and automatically adjusts the working state of each DC power supply unit, causing units that are in a long-term discharging or idle state to enter charging mode, while simultaneously putting fully charged units into or preparing to enter the power supply cycle. Through this orderly rotation, balanced use of the battery pack is achieved, delaying the aging of individual units.

[0068] In summary, the embodiments of this application have at least the following technical effects: This application acquires real-time voltage perturbations, instantaneous current fluctuations, and ripple spectrum characteristics of the DC bus to establish a dynamic feature vector for the bus. This dynamic feature vector characterizes the dynamic response of the bus under load disturbances. The dynamic feature vector is reconstructed in both the time and frequency domains to extract the corresponding inertia coefficient, impedance change slope, and power phase drift. A bus state mapping model is established based on the extraction results. The bus state mapping model is used to perform inverse mapping of the bus dynamic response, generating virtual state vectors for each DC power supply unit. These virtual state vectors include a unity power gain response coefficient, instantaneous current reflection characteristics, and output stability margin. Based on the temporal evolution differences of the virtual state vectors, a relative deviation matrix between DC power supply units is constructed to identify abnormal power supply behavior. The identification results are used for dynamic power supply switching management. This invention addresses the technical problems of the inability to identify the dynamic response of DC buses in real time and the lag in the detection and switching management of power supply unit anomalies in the prior art. By establishing a dynamic feature vector of the bus, constructing a bus state mapping model, and realizing the linkage management of virtual state inversion and anomaly identification of power supply units, it achieves the technical effects of improving the reliability of system power supply and realizing adaptive mutual backup and dynamic switching among multiple units.

[0069] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0070] 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 combined and mutually redundant DC power supply system, characterized in that, The system includes: The monitoring module is used to collect voltage disturbances, instantaneous current fluctuations and ripple spectrum characteristics of the DC bus in real time, and establish a dynamic feature vector of the bus, which characterizes the dynamic response of the bus under load disturbance. The dynamic feature reconstruction module is used to perform time-domain and frequency-domain joint reconstruction of the dynamic feature vector of the bus, extract the corresponding inertia coefficient, impedance change slope and power phase drift of the bus, and establish a bus state mapping model based on the extraction results; The reverse mapping module is used to perform reverse mapping of the bus dynamic response using the bus state mapping model, and generate virtual state vectors for each DC power supply unit. The virtual state vectors include unity power gain response coefficient, instantaneous current reflection characteristics and output stability margin. An anomaly identification module is used to construct a relative deviation matrix between DC power supply units based on the temporal evolution differences of the virtual state vector, and to identify abnormal power supply behavior. The management module is used to manage dynamic power supply switching based on the identification results.

2. The DC power supply system with mutual backup as described in claim 1, characterized in that, In the anomaly identification module, a relative deviation matrix between DC power supply units is constructed based on the temporal evolution differences of the virtual state vectors, including: After aligning the virtual state vectors of each DC power supply unit in time, calculate the difference sequence between the power phase drift and the impedance change slope. Multi-scale wavelet decomposition and spectral entropy analysis are performed on the differential sequence to generate a dynamic spectral difference entropy matrix between each DC power supply unit pair; After performing an orthogonal decoupling transformation on the dynamic spectral difference entropy matrix, a relative deviation matrix is ​​constructed. The spectral radius or characteristic entropy of the relative deviation matrix is ​​then thresholded to identify a first anomaly factor. This first anomaly factor is then used to identify abnormal power supply behavior.

3. The DC power supply system with mutual backup as described in claim 2, characterized in that, In the anomaly identification module, abnormal power supply behavior is identified using the first anomaly factor, including: Read the power supply control parameters of each DC power supply unit, including the output voltage setting value, conduction angle, PWM modulation ratio and feedback gain coefficient; After configuring the control parameter response model, the power supply control parameters are input into the control parameter response model to calculate the theoretical power phase drift and output stability margin of each DC power supply unit under the current power supply control parameters, and generate a prediction deviation vector. The prediction deviation vector is mapped to the feature space of the relative deviation matrix to obtain the relative response vector; Anomaly identification based on the direction angle and amplitude ratio is performed using the relative response vector and the prediction deviation vector to establish a second anomaly factor; The first and second abnormal factors are logically fused to identify abnormal power supply behavior.

4. The combined backup DC power supply system as described in claim 1, characterized in that, In the reverse mapping module, the bus state mapping model is used to perform reverse mapping of the bus dynamic response, generating virtual state vectors for each DC power supply unit, including: Based on the mutual information relationship between the bus state mapping model and the bus dynamic feature vector, a nonlinear reversible subset of the bus dynamic response is constructed. The nonlinear reversible subset is constructed by performing polynomial perturbation mapping training in the three-dimensional parameter space of inertia coefficient, impedance change rate and power phase drift. The bus dynamic feature vector is input into the nonlinear reversible operator subset, and a virtual state candidate set for each DC power supply unit is generated through multi-operator collaborative iteration; The virtual state candidate set is optimized and screened using constraints based on global power conservation of the bus, node current consistency and transient energy balance, and the virtual state vector is output.

5. A combined and mutually redundant DC power supply system as described in claim 1, characterized in that, The management module utilizes the identification results for dynamic power supply switching management, including: Output the anomaly level based on the identification results; The battery internal resistance of each DC power supply unit is measured in real time using a battery voltage monitor. Based on the measurement results and the power data of the DC power supply unit, a health priority sequence of the DC power supply unit is configured. The abnormality level is used to perform health priority sequence adaptation analysis to generate a dynamic switching strategy; Dynamic power supply switching management is performed according to the dynamic switching strategy.

6. The DC power supply system with mutual backup as described in claim 1, characterized in that, The system also includes: The capacity verification module is used to control the discharge voltage of each DC power supply unit to be higher than the preset voltage upper limit. It constructs a real-time capacity response curve by using the voltage perturbation feedback on the bus side and the current response signal at the DC power supply unit end, and performs full capacity verification based on the real-time capacity response curve.

7. A combined and mutually redundant DC power supply system as described in claim 6, characterized in that, The system also includes: The charging control module is used to execute reverse three-stage charging control if the full capacity verification is passed. The reverse three-stage charging control includes dynamic voltage regulation charging in the first stage, phase coordination charging in the second stage, and inertial compensation charging in the third stage.

8. A combined and mutually redundant DC power supply system as described in claim 6, characterized in that, The system also includes: The anomaly reporting module is used to capture anomaly features based on the real-time capacity response curve if the full capacity verification fails. The anomaly feature capture includes capacity change rate capture and total capacity decay capture. Based on the anomaly feature capture results, the module performs anomaly level matching for the DC power supply unit and executes anomaly reporting.

9. A combined and mutually backup DC power supply system as described in claim 1, characterized in that, The system also includes a self-maintenance management module, used to perform the following steps: Configure a set of simulated high-current scenarios and set an impulse discharge strategy that maps to the set of simulated high-current scenarios; The aforementioned impulse discharge strategy was used to perform scenario simulation tests on each DC power supply unit, and the simulation test results were established. Predictive maintenance and management of the DC power supply unit are performed based on the simulation test results.

10. A combined and mutually redundant DC power supply system as described in claim 1, characterized in that, The system also includes a charging control module for monitoring the charging and discharging records of each DC power supply unit and performing periodic alternating charging and discharging management.

Images