A cascaded modular direct current power supply and a dynamic redundancy control method thereof

Abstract

The application relates to a kind of cascaded modular DC power supply and its dynamic redundancy control method, belong to power electronics and DC power supply technical field.Firstly, activate redundancy-main component cooperative initialization mechanism, utilize redundancy component built-in pre-charging structure dynamic matching main component capacitor voltage, simultaneously load redundancy state pre-sensing mechanism parameter correlation model;Based on dynamic parameter real-time sensing mechanism, collect main component key state parameters, realize abnormal hierarchical determination and switching instruction generation through hierarchical redundancy trigger mechanism;Start real-time redundancy scheduling mechanism, control abnormal main component soft-off and redundancy component zero-current access, combined with dynamic power matching compensation mechanism to accurately fill the power gap;Rely on redundancy state real-time monitoring mechanism to activate cooperative recovery process, realize main component smooth access and redundancy component standby recovery.The application effectively guarantees the power supply reliability of cascaded link under different working conditions through whole-process dynamic redundancy control.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power electronics and direct current power supply, and particularly relates to a cascaded modular direct current power supply and a dynamic redundancy control method thereof. BACKGROUND

[0002] With the rapid growth of power generation from new energy, industrial direct current power supply and power supply demand of high-density IT equipment in data centers, the cascaded modular direct current power supply has become the core power supply equipment in the above-mentioned scenarios due to the advantages of wide adjustable range of output voltage, flexible power expansion, and single module failure not affecting the overall link. However, the redundancy control technology of the current cascaded modular direct current power supply still has many deficiencies, which is difficult to adapt to the power supply demand of high reliability and high precision.

[0003] On the one hand, the existing redundancy control is mostly in the mode of "fixed redundancy configuration + passive switching after failure": the redundancy components are in standby state for a long time, and the switching trigger only depends on the threshold hard decision of a single parameter such as overcurrent and overtemperature, which cannot predict the gradual parameter faults such as slow voltage drift and temperature gradient rise of the main components in advance. This hysteresis may cause instantaneous power failure of servers due to delay of redundancy switching in the data center scenario; in the new energy storage system, it is easy to cause power interruption of the cascaded link, affecting the stability of power grid connection.

[0004] On the other hand, there are obvious defects in the system initialization and fault recovery process: during initialization, there is a large initial deviation between the capacitance voltages of the redundancy components and the main components, and direct access will produce instantaneous impact current, accelerating the aging of core components such as switch tubes and capacitors; when the main component fails, the existing technology uses fixed current pre-charging, which cannot dynamically adjust the rate according to the parameter deviation of the main component and the link, and is easy to cause access fluctuation due to insufficient pre-charging or component overheating due to over-charging. At the same time, in the face of dynamic power fluctuation of the load, the existing compensation mechanism compensates according to the rated power of the main component, which cannot match the real-time demand of the link, resulting in output voltage fluctuation exceeding the standard; when multiple main components are abnormal, there is a lack of priority scheduling based on the emergency degree of the fault and the criticality of the load, which is easy to expand the influence of the fault due to disordered compensation. SUMMARY

[0005] In order to solve the above problems in the prior art, the application provides a cascaded modular direct current power supply and a dynamic redundancy control method thereof.

[0006] The purpose of the application can be achieved by the following technical scheme:

[0007] A cascaded modular direct current power supply and a dynamic redundancy control method thereof, characterized by comprising:

[0008] S1: The main controller activates the redundancy-main component cooperative initialization mechanism, the redundancy component has a pre-charge structure, and the main component capacitor voltage real-time tracking is the core logic. The main controller loads the redundancy state pre-awareness mechanism to establish the association model of the main component state parameters and the wake-up timing of the redundancy component.

[0009] S2: Based on the dynamic parameter real-time awareness mechanism, the input voltage stability, output current fluctuation, and temperature change rate of the main component are collected in real time, and transmitted to the main controller. The main controller calls the hierarchical redundancy trigger mechanism, and based on the parameter deviation-redundancy urgency dynamic mapping relationship, the main component state parameters are processed hierarchically to generate a redundancy switching instruction.

[0010] S3: The redundancy switching instruction starts the real-time redundancy scheduling mechanism: the abnormal main component reduces the output power according to the dynamic slope, synchronously disconnects the output isolation structure, and controls the redundancy component's bidirectional thyristor switch to close based on the zero-current dynamic capture strategy. The dynamic power matching compensation mechanism is activated: based on the total output power demand of the cascade link and the remaining main component output capacity, the gap value of the redundancy component output power is calculated, and the output voltage and switching frequency of the redundancy component are adjusted.

[0011] S4: The redundancy state monitoring and recovery trigger linkage mechanism activates the redundancy-main component cooperative recovery mechanism based on the redundancy component output power, temperature, and switch tube working condition data: the output power gradient transfer is based on the deviation of the main component's current capacitor voltage, temperature, and cascade link parameters to adjust the main component's pre-charge current and output voltage; when the main component output power and parameter matching degree reach the preset standard, the cascade link is accessed, and the redundancy component is disconnected and restored to standby.

[0012] Specifically, the capacitor voltage deviation control of the redundancy component and the main component needs to collect the real-time voltage of both through a voltage sensor, calculate the capacitor voltage deviation, and then start the pre-charge current adjustment. The pre-charge current is adjusted according to the deviation-current ratio: the main controller retrieves the rated charging current of the redundancy component, sets the reference current input current adjustment structure, and increases the current by the reference ratio when the deviation increases; the current adjustment loop has an overcurrent protection structure, which monitors the current in real time and triggers current limiting when the rated charging current is reached. The voltage deviation is continuously collected, and when it stabilizes within the proportional range, the adjustment is stopped and the current is locked.

[0013] Specifically, the correlation model of the redundancy state pre-sensing mechanism first starts the load power collection structure, collects load power at a fixed period, and after a continuous period, calculates the power change rate and inputs the state parameter deviation of the main component into the correlation model; calculates the redundancy wake-up pre-judgment value and compares it with the threshold value, and if the threshold value is reached, the main controller sends a pre-wake-up instruction, and after the self-checking of the redundant component is passed, it switches to the quasi-working state and feeds back the completion signal, and the main controller records the pre-wake-up time and reserves a preparation period for subsequent switching.

[0014] Specifically, the dynamic parameter real-time sensing mechanism first starts the load fluctuation calculation structure, collects the total output power of the cascaded link, calculates the fluctuation at a fixed period, and compares it with two preset threshold values; when the fluctuation exceeds the first threshold value, the collection frequency is increased and the cache capacity is increased, and when it is between the two threshold values, the collection frequency and cache are maintained, and when it is below the second threshold value, the collection frequency is reduced and the cache is reduced; after each adjustment of the collection frequency, the parameter collection structure outputs the transition data, which is averaged by sliding average filtering to generate effective data, and then the effective data integrity is confirmed by the verification structure.

[0015] Specifically, the parameter deviation-redundancy urgency dynamic mapping process of the hierarchical redundancy triggering mechanism is as follows: first, calculate the state parameter deviation of the main component, calculate the deviation of each parameter for input voltage, output current, and temperature, and then calculate the comprehensive deviation according to the preset weight; compare the comprehensive deviation with the preset slight deviation threshold and moderate deviation threshold; if the comprehensive deviation exceeds the moderate deviation threshold, or if the state parameter exceeds the rated value, a serious deviation warning is generated, the main controller generates a redundancy switching instruction, including the number of abnormal main components, the switching trigger reason, and the switching time limit requirement, which is transmitted to the redundant component and the abnormal main component through the communication bus, and the switching instruction generation time is recorded.

[0016] Specifically, in the redundancy dynamic allocation process of the real-time redundancy scheduling mechanism, the main controller obtains the number, rated power, and current fault parameter of all abnormal main components through the communication bus, and calculates the power proportion of the abnormal main components; at the same time, the emergency degree of the fault is evaluated, the temperature over-temperature amplitude and the current overload multiple are calculated as the emergency degree index, and based on the power proportion and the emergency degree index, a priority matrix is constructed, the rows of the matrix represent the abnormal main components, the columns represent the evaluation indexes, and each cell is filled with the normalized value of the corresponding index; the two indexes are assigned weights by the analytic hierarchy process, the priority scores of the abnormal main components are calculated, and a priority list is generated.

[0017] Specifically, the abnormal main component dynamically reduces the output power first by the power collection structure to collect the total output power and the total power of the remaining normal main component, and to calculate the current power demand; the main controller calls the rated power, combines the current power demand to calculate the output power reduction slope, sends the reduction power instruction including the output power reduction slope and the target power to the conversion structure, receives the output power reduction instruction, and reduces the output power by adjusting the switch tube duty cycle, monitors the actual output power reduction rate in real time, and if the actual power reduction rate deviates from the set slope by more than the allowed range, immediately feeds back to the main controller, which recalculates the slope and corrects the adjustment amplitude of the duty cycle.

[0018] Specifically, the zero-current capture strategy is based on the current sensor of the input and output of the bidirectional thyristor switch, the main controller pre-sets the current threshold, and configures the continuous comparison period number; the current sensor collects real-time current data, compares the real-time current value with the current threshold, records the near-zero current state in the continuous collection period, and determines that the bidirectional thyristor switch is currently in a closed state; sends a closing instruction to the switch drive, the switch drive outputs a trigger voltage to make the bidirectional thyristor switch conductive, and the switch drive collects the switch conduction current through the current sensor and feeds back the closing success signal of the bidirectional thyristor switch to the main controller.

[0019] Specifically, the dynamic power matching compensation mechanism first starts the power gap analysis structure, collects the total output power demand and the total power of the remaining main component, and calculates the power gap; the power gap, the redundant component rated voltage and the switch frequency are input into the particle swarm optimization algorithm structure, in each iteration process, each particle represents a combination of output voltage and switch frequency parameters of a group of redundant components, the fitness value of the particle is calculated, the fitness value is the weighted result of the comprehensive error and loss, the global optimal particle and the local optimal particle are tracked, the particle position is adjusted by particle speed update, the optimal parameter combination is generated, and after the iteration is completed, the optimal output voltage and switch frequency parameters are output.

[0020] Specifically, the redundant state monitoring and recovery trigger linkage mechanism integrates a multi-parameter fusion abnormality judgment algorithm: by collecting the output power, core temperature, switch tube conduction voltage drop and switch speed of the redundant component, the actual value is proportionally converted with the rated value, and the processed parameters are transmitted to the state analysis structure; the state analysis structure is based on the multi-parameter fusion abnormality judgment algorithm, and the weighted sum of the weighted parameters is generated based on the weight, and the overload warning is generated by comparing the preset safety threshold.

[0021] Specifically, the output power gradient transfer first collects the voltage and current of the main assembly and the cascade link, calculates the parameter deviation of the main assembly and the cascade link, and when the deviation exceeds a preset threshold, the main controller sends a power transfer instruction to the redundant assembly.

[0022] Specifically, the main assembly pre-charge current adjustment first retrieves the rated input and pre-charge current, sets the upper and lower limit input pre-charge control structure, stops pre-charge based on the capacitance of the main assembly and the voltage deviation of the link, and generates a pre-charge completion signal.

[0023] The beneficial effects of the present application are:

[0024] The present application has the following advantages: on the one hand, the redundant state pre-sensing mechanism predicts potential failures of the main assembly, the hierarchical prediction type redundant switching trigger mechanism dynamically determines the redundancy urgency, avoids the hysteresis of traditional passive switching, greatly reduces the risk of cascade link power interruption, effectively adapts to scenarios such as data centers and precision manufacturing that require high continuity of power supply, and significantly improves the reliability of power supply; on the other hand, the redundant assembly adjusts the pre-charge in real time by tracking the capacitance voltage of the main assembly, controls the switch closing through the zero current dynamic capture strategy, adjusts the pre-charge current of the main assembly based on the parameter deviation, can eliminate the voltage and current impact in the initialization and recovery phase, prevent overcharging and undercharging problems, protect core components such as switch tubes and capacitors, and prolong the overall life of the power supply system; at the same time, the dynamic power matching compensation mechanism combines the total power demand of the link and the output capacity of the remaining main assembly, accurately calculates the redundant power gap and adjusts the output parameters, can adapt to dynamic scenarios such as new energy generation light fluctuation and industrial load start-stop in real time, avoids the problem of power surplus or deficiency of traditional fixed compensation, and ensures the stability of output voltage; in addition, the real-time redundant scheduling mechanism realizes the cooperative control of abnormal main assembly and redundant assembly, the redundant state real-time monitoring and recovery trigger linkage mechanism activates the smooth recovery process, and cooperates with the hierarchical processing and dynamic scheduling logic, so that the system can flexibly cope with single component or multiple component abnormalities, improve multi-working-condition adaptability and reduce the risk of fault expansion. BRIEF DESCRIPTION OF DRAWINGS

[0025] In order to facilitate the understanding of those skilled in the art, the present application will be further described below with reference to the accompanying drawings.

[0026] Figure 1 A flowchart of the cascaded modular DC power supply and the dynamic redundant control method thereof of the present application;

[0027] Figure 2 A timing diagram of the cascaded modular DC power supply and the dynamic redundant control method thereof of the present application. DETAILED DESCRIPTION

[0028] To further illustrate the technical means and effects taken by the present application to achieve the predetermined inventive purpose, the specific embodiments, structures, features and effects according to the present application are described in detail below in combination with the drawings and preferred embodiments.

[0029] Please refer to Figure 1 , 2 A cascaded modular DC power supply and its dynamic redundancy control method, comprising:

[0030] A cascaded modular DC power supply and its dynamic redundancy control method, characterized by comprising:

[0031] S1: the main controller activates the redundancy-main component cooperative initialization mechanism, the redundancy component has a pre-charging structure, and the real-time tracking of the main component capacitor voltage is the core logic, and the redundancy component capacitor voltage is dynamically adjusted, the main controller loads the redundancy state pre-perception mechanism, and establishes the association model of the main component state parameters and the redundancy component wake-up timing;

[0032] S2: based on the dynamic parameter real-time perception mechanism, the input voltage stability, output current fluctuation degree and temperature change rate of the main component are collected in real time, and are transmitted to the main controller; the main controller calls the hierarchical redundancy triggering mechanism, and based on the parameter deviation-redundancy urgency dynamic mapping relationship, the main component state parameters are processed in stages, and a redundancy switching instruction is generated;

[0033] S3: the redundancy switching instruction starts the real-time redundancy scheduling mechanism: the abnormal main component reduces the output power according to the dynamic slope, synchronously disconnects the output isolation structure, and simultaneously controls the bidirectional thyristor switch of the redundancy component to be closed based on the zero-current dynamic capture strategy; the dynamic power matching compensation mechanism is activated: based on the total output power demand of the cascaded link and the remaining main component output capacity, the gap value of the redundancy component output power is calculated, and the output voltage and switching frequency of the redundancy component are adjusted;

[0034] S4: the redundancy state monitoring and recovery triggering linkage mechanism activates the redundancy-main component cooperative recovery mechanism based on the redundancy component output power, temperature and switch tube working condition data: the output power gradient transfer is based on the deviation of the current capacitor voltage, temperature and cascaded link parameters of the main component, and the pre-charging current and output voltage of the main component are adjusted; when the main component output power and parameter matching degree reach the preset standard, the cascaded link is accessed, and the redundancy component is disconnected and restored to standby.

[0035] Specifically, the voltage deviation between the redundant component and the main component needs to be controlled by first collecting the real-time voltages of both through a voltage sensor, calculating the voltage deviation, and then starting the pre-charge current adjustment. The pre-charge current is adjusted in proportion to the deviation. The main controller retrieves the rated charging current of the redundant component, sets a reference current input current adjustment structure, and increases the current in proportion to the reference when the deviation increases. An overcurrent protection structure is set in the current adjustment loop to monitor the current in real time and trigger current limiting when the rated charging current is reached. The voltage deviation is continuously collected, and the current is locked when the deviation stabilizes within a certain proportion of the rated voltage.

[0036] The voltage deviation control process between the redundant component and the main component is as follows: First, the real-time voltages of the redundant component and the main component are collected by a voltage sensor, and the collection interval is set to a very short period to ensure data timeliness. The voltage values collected twice are substituted into the deviation calculation formula (deviation value = |redundant component voltage - main component voltage| / rated voltage x 100%), and if the deviation value exceeds a certain proportion of the rated voltage, the pre-charge current adjustment is immediately started. The dynamic adjustment of the pre-charge current uses a "deviation-current" proportional adjustment logic: first, the main controller retrieves the rated charging current parameters of the redundant component, sets a reference charging current (a lower proportion of the rated charging current) and inputs it into the current adjustment module. The current voltage deviation value is compared with the preset deviation threshold in real time. If the deviation value increases by a certain proportion, the charging current increases by a fixed proportion of the reference current. If the deviation value decreases by a certain proportion, the charging current decreases by a fixed proportion of the reference current. At the same time, an overcurrent protection module is set in the current adjustment loop to monitor the charging current in real time. When the current reaches a higher proportion of the rated charging current of the redundant component, the current limiting action is automatically triggered to keep the current stable below this proportion, preventing excessive current from damaging the capacitor or the charging circuit components. The voltage deviation value is continuously collected during the adjustment process until the deviation stabilizes within a certain proportion of the rated voltage, and the current dynamic adjustment is stopped and the current is locked.

[0037] Specifically, the association model of the redundancy state pre-perception mechanism first starts the load power collection structure, collects the load power at a fixed period, and then inputs the power change rate and the main component state parameter deviation degree into the association model through differential calculation after a continuous period. The main controller sends a pre-wakeup instruction when the pre-wakeup value reaches the threshold. After the self-check of the redundant component is passed, it switches to the standby state and feeds back the completion signal. The main controller records the pre-wakeup time and reserves a preparation period for subsequent switching.

[0038] The correlation model of the redundancy state pre-sensing mechanism runs as follows: firstly, the load power collection is started, the real-time power data of the load is collected at a fixed period, after continuous collection for multiple periods, the load power change rate is calculated through difference calculation (change rate = |current period power - last period power| / last period power x 100%); the load power change rate and the main component state parameter deviation degree (calculated by voltage, current, temperature parameters) are input into the correlation model; during model initialization, the weight coefficients k1 and k2 are configured according to the system factory test data (k1 corresponds to the main component state parameter deviation degree, k2 corresponds to the load power change rate, and k1+k2=1), if the system is in heavy load working condition, k2 proportion can be increased manually or automatically by the main controller, if it is in light load working condition, k1 proportion is increased; the pre-judgment value is calculated according to the formula "pre-judgment value = k1 x main component state parameter deviation degree + k2 x load power change rate", and then compared with the preset pre-judgment threshold value in real time; if the pre-judgment value reaches the threshold value, the main controller immediately sends a pre-wakeup instruction to the redundant component, after receiving the instruction, the redundant component first starts internal self-checking (checks whether the power supply and control loop are normal), after the self-checking is passed, it switches from standby state to quasi-working state (at this time, the core circuit of the redundant component is powered on, only the output end is not connected to the cascade link), and sends a pre-wakeup completion signal to the main controller; after receiving the feedback, the main controller records the pre-wakeup time, ensures that the possible redundancy switching action is ahead of time for a short period of time, and reserves a parameter adjustment preparation period for the redundant component.

[0039] Specifically, the dynamic parameter real-time sensing mechanism first starts the load fluctuation degree calculation structure, collects the total output power of the cascade link, calculates the fluctuation degree at a fixed period, and compares it with two preset threshold values; when the fluctuation degree exceeds the first threshold value, the collection frequency is increased and the buffer capacity is increased, between the two threshold values, the collection frequency and the buffer are maintained, and below the second threshold value, the collection frequency is reduced and the buffer is reduced; after each collection frequency adjustment, the parameter collection structure outputs the transition data, the average value is generated after the sliding average filtering, and the integrity of the effective data is confirmed by the verification structure.

[0040] Specifically, the parameter deviation-redundancy urgency dynamic mapping process of the hierarchical redundancy triggering mechanism is as follows: firstly, the state parameter deviation degree of the main component is calculated, the deviation degrees of input voltage, output current and temperature are calculated respectively, and the comprehensive deviation degree is calculated according to the preset weight; the comprehensive deviation degree is compared with the preset slight deviation threshold value and moderate deviation threshold value; if the comprehensive deviation degree exceeds the moderate deviation threshold value, or the state parameter exceeds the rated value, a serious deviation warning is generated, the main controller generates a redundancy switching instruction, which includes the number of abnormal main components, the switching trigger reason and the switching time limit requirement, and is transmitted to the redundant component and the abnormal main component through the communication bus, and the switching instruction generation time is recorded.

[0041] The "parameter deviation - redundancy urgency" dynamic mapping process of the hierarchical redundancy triggering mechanism is as follows: first, the state parameter deviation of the main component is calculated by the parameter processing module, and the deviation of each parameter is calculated for the input voltage, output current, and temperature, and then the comprehensive deviation is calculated according to the preset weight; The comprehensive deviation is compared with the preset "slight deviation threshold" and "moderate deviation threshold": if the comprehensive deviation is in the slight deviation range, it is determined as "slight deviation", the main controller sends a pre-wakeup instruction to the redundant component, and starts the parameter trend recording module to store the comprehensive deviation data at a fixed period, and generates a deviation trend curve by linear fitting, if the curve slope is positive and continuously increases, trigger the second warning to prompt the parameter deterioration risk; if the comprehensive deviation is in the moderate deviation range, it is determined as "moderate deviation", the main controller first starts the main component-redundant component parameter synchronization module, and real-time transmits the input voltage and output current parameters of the main component to the redundant component, and the control loop of the redundant component adjusts the PFC rectifier component and DC-DC conversion component parameters according to the main component parameters, so that the output voltage of the redundant component gradually approaches the current output voltage of the main component. Compare the voltage difference between the two every short period during the adjustment process, until the difference is stable in a small range; if the comprehensive deviation exceeds the moderate deviation threshold, or a single key parameter exceeds the rated value, it is immediately determined as "serious deviation", the main controller generates a redundancy switching instruction, which contains the abnormal main component number, switching trigger reason, and switching time limit requirement, and is transmitted to the control module of the redundant component and the abnormal main component through the high-speed communication bus, and the switching instruction generation time is recorded for subsequent tracing.

[0042] Specifically, in the redundancy dynamic allocation process of the real-time redundancy scheduling mechanism, the main controller obtains the number, rated power, and current fault parameter of all abnormal main components through the communication bus, and calculates the power ratio of the abnormal main components; At the same time, the emergency degree of the fault is evaluated, the temperature over-temperature amplitude and the current overload multiple are calculated as the emergency degree index, and the priority matrix is constructed based on the power ratio and the emergency degree index, the matrix row represents the abnormal main component, the column represents the evaluation index, and each cell is filled with the normalized value of the corresponding index; The two indexes are assigned weights by the analytic hierarchy process, the priority score of the abnormal main component is calculated, and a priority list is generated.

[0043] Specifically, the abnormal main component dynamically reduces the output power first by the power collection structure to collect the total output power and the total power of the remaining normal main component, calculate the current power demand; the main controller calls the rated power, combines the current power demand to calculate the output power reduction slope, sends the reduction power instruction including the output power reduction slope and the target power to the conversion structure, receives the output power reduction instruction, reduces the output power by adjusting the switch tube duty cycle, monitors the actual output power reduction rate in real time, if the actual power reduction rate deviates from the set slope beyond the allowed range, immediately feedback to the main controller, the main controller recalculates the slope and corrects the adjustment amplitude of the duty cycle.

[0044] Specifically, the zero-current capture strategy is based on the current sensor of the input and output of the bidirectional thyristor switch, the main controller sets the current threshold in advance, and configures the continuous comparison period number; the current sensor collects real-time current data, compares the real-time current value with the current threshold, records the near-zero current state in the continuous collection period, determines that the bidirectional thyristor switch is currently in the closed state; sends a closing instruction to the switch drive, the switch drive outputs a trigger voltage to make the bidirectional thyristor switch conductive, and the switch drive collects the switch conduction current through the current sensor and feeds back the closing success signal of the bidirectional thyristor switch to the main controller.

[0045] Specifically, the dynamic power matching compensation mechanism first starts the power gap analysis structure, collects the total output power demand and the total power of the remaining main component, calculates the power gap; input the power gap, the rated voltage of the redundant component and the switch frequency into the particle swarm optimization algorithm structure, in each iteration process, each particle represents a combination of output voltage and switch frequency parameters of a group of redundant components, calculates the fitness value of the particle, the fitness value is the weighted result of the comprehensive error and loss, tracks the global optimal particle and the local optimal particle, adjusts the particle position by updating the particle speed, generates the optimal parameter combination, after the iteration is completed, outputs the optimal output voltage and switch frequency parameters.

[0046] The adaptive algorithm of the dynamic power matching compensation mechanism runs as follows: first, start the power gap analysis module, collect the total output power demand of the cascade link and the total output power of the remaining main components, and calculate the power gap; input the power gap, the rated output voltage of the redundant components, the rated switching frequency and other parameters into the particle swarm optimization algorithm module, set the particle population size and the number of iterations during algorithm initialization, and take "minimum power gap matching error" and "lowest redundant component loss" as the double objective functions; in each iteration process, each particle represents a combination of the output voltage and switching frequency parameters of a group of redundant components, the algorithm calculates the fitness value of each particle, which is the weighted result of the comprehensive error and loss, tracks the global optimal particle and local optimal particle, adjusts the particle position through particle speed update, and gradually approaches the optimal parameter combination; after iteration, output the optimal output voltage and switching frequency parameters, and transmit them to the PFC rectifier components and DC-DC conversion components of the redundant components; at the same time, start the load fluctuation prediction module, retrieve the load power data of multiple collection periods in the past, fit the power change straight line through the linear regression algorithm, predict the power demand change amount in the short term in the future according to the slope of the straight line, add the change amount to the current power gap, and adjust the output parameters of the redundant components in advance; after adjustment, the actual output power of the redundant components is collected in real time through the power monitoring module, and compared with the predicted power demand, if the error exceeds a small range, the particle swarm optimization algorithm is triggered again to correct the parameters, until the error is stable within a small range.

[0047] Specifically, the redundancy state monitoring and recovery trigger linkage mechanism integrates a multi-parameter fusion abnormality judgment algorithm: by collecting the output power, core temperature, switch tube conduction voltage drop and switching speed of the redundant components, the actual values are proportionally converted with the rated values, and are uniformly mapped to the same numerical interval, and the processed parameters are transmitted to the state analysis structure; the state analysis structure generates an abnormality judgment value by weighting and summing the weighted standardized parameters based on the multi-parameter fusion abnormality judgment algorithm, and generates an overload warning by comparing with the preset safety threshold.

[0048] Specifically, the output power gradient transfer first collects the voltage and current of the main components and the cascade link, calculates the parameter deviation of the main components and the cascade link, and when the deviation exceeds the preset threshold, the main controller sends a power transfer instruction to the redundant components.

[0049] The output power gradient transfer process is as follows: first, the current output voltage and current of the main component and the current output voltage and current of the cascade link are collected by the parameter collection module, the parameter deviation of the two is calculated, and the larger value of the voltage deviation and the current deviation is taken as the basis for judgment; if the parameter deviation exceeds the preset deviation threshold, the main controller sends a power transfer instruction to the redundant component, and sets the power transfer step to be a smaller proportion of the current output power of the redundant component; the redundant component reduces the output power by the step, keeps it for a very short time to stabilize the cascade link power, and monitors the link power fluctuation at the same time, if the fluctuation is lower than the allowed range, continue to reduce the power by the step; if the parameter deviation is lower than the preset deviation threshold, the main controller re-sets the power transfer step to be a larger proportion of the current output power of the redundant component, and speeds up the power transfer speed; before each step adjustment, the main controller first sends a step change notification to the redundant component and the main component to ensure that the control logic of the two is coordinated; during the power transfer process, the parameter deviation of the main component and the link is continuously calculated, when the output power of the redundant component is close to zero and the parameter deviation of the main component and the link is lower than the minimum range, the power gradient transfer is stopped, and the subsequent operation of the main component accessing the link is prepared.

[0050] Specifically, the main component pre-charge current adjustment first retrieves the rated input and pre-charge current, sets the upper and lower limits of the current input pre-charge control structure, and stops pre-charging based on the capacitance of the main component and the voltage deviation of the link to generate a pre-charge completion signal.

[0051] The adjustment process of the pre-charge current of the main component is as follows: first, the main controller retrieves the rated input current and rated pre-charge current parameters of the main component, sets the upper and lower limits of the pre-charge current, and inputs the pre-charge control module; the current capacitance voltage of the main component and the cascade link voltage are collected by the voltage sensor, and the parameter deviation is calculated; the pre-charge control module adjusts the current according to the "deviation - current" closed-loop control logic: if the deviation exceeds the first deviation threshold, the pre-charge current is set to the lower limit value, and the capacitance is slowly charged to avoid a large current causing a sudden voltage rise under a large deviation; if the deviation is between the first deviation threshold and the second deviation threshold, the current is adjusted in the opposite direction according to the deviation size, so that the larger the deviation, the smaller the current; if the deviation is lower than the second deviation threshold, the pre-charge current is set to the rated pre-charge current, and the remaining charging is quickly completed; during the charging process, the current sensor monitors the pre-charge current in real time, if the current exceeds the upper limit value, the overcurrent protection is triggered and the current is reduced to the upper limit value; if the current is lower than the lower limit value, it is determined that the charging is insufficient, the charging loop is checked and the current is increased to the lower limit value; at the same time, the voltage sensor continuously collects the capacitance voltage of the main component, when the voltage deviation and the link voltage deviation are lower than the minimum range and stable for multiple cycles, the pre-charge is stopped, a pre-charge completion signal is sent to the main controller, and it is confirmed that the main component has the conditions to access the link.

[0052] In this embodiment, a cascaded modular DC power supply system of a 100 kW new energy energy storage power station is taken as an example. The system includes 8 main components (single module rated power 12.5 kW, rated output voltage 600 V), 2 redundant components (parameters consistent with the main components), and a main controller using an STM32H743 microprocessor. The specific execution process is as follows:

[0053] The main controller activates the redundant-main component cooperative initialization mechanism: the redundant component built-in pre-charge structure collects the main component capacitor voltage (initial value 300 V) at a period of 50 ms, dynamically adjusts the pre-charge current according to the adjustment coefficient of "voltage deviation x 0.8" (the current is set to 40 A when the deviation is 50 V, and the current is set to 16 A when the deviation is 20 V), and makes the redundant component capacitor voltage stable to the main component voltage ±2V within 30s. At the same time, the main controller loads the redundant state pre-perception mechanism, establishes an associated model of "main component temperature change rate (weight 0.6) + output current fluctuation degree (weight 0.4)", and sets when the model calculation value > 0.7 (temperature change rate > 0.5℃ / min and current fluctuation degree > 5%), the redundant component pre-wakes up.

[0054] The running dynamic parameter real-time perception mechanism collects the main component input voltage stability (monitors ±5V fluctuation), output current fluctuation degree (monitors ±10A fluctuation), and temperature change rate (monitors ±0.3℃ / min change) at a period of 20ms, and transmits them to the main controller. When a main component appears a temperature change rate of 0.6℃ / min and an output current fluctuation degree of 8% due to heat dissipation failure, the main controller calls the hierarchical redundancy trigger mechanism: the parameter deviation degree is 0.8 (exceeding the moderate deviation threshold 0.6), based on the "parameter deviation degree-redundancy urgency" mapping relationship (deviation degree 0.6-0.8 corresponds to "high urgency"), the hierarchical pre-judgment type redundancy switching trigger mechanism is activated, and the redundancy switching instruction (target: enable No. 1 redundant component to replace the failed main component) is generated.

[0055] After the redundancy switching instruction is triggered, the real-time redundancy scheduling mechanism starts: the control fault master component reduces the output power at a dynamic slope of 0.5 kW / s (from 12.5 kW to 0 kW, which takes 25 s), and synchronously opens the output isolation relay (response time < 10 ms); at the same time, the master controller collects the bidirectional thyristor switch current of the redundancy component at a period of 10 μs, and when the current < 0.5 A (zero current threshold) is detected, the switch is triggered to be closed (closure time < 20 μs), so as to avoid the impact current. Subsequently, the dynamic power matching compensation mechanism is activated: the total output power demand of the cascade link (currently 80 kW) and the total output capacity of the remaining 7 master components (87.5 kW) are collected, and the power gap is calculated to be 0 (because the remaining master components have compensated when the fault master component is not completely disconnected), and when the load power rises to 90 kW, the gap is calculated to be 2.5 kW, the redundancy component output voltage is adjusted to 602 V, the switching frequency is adjusted to 0 kHz, the redundancy component output power is stabilized to 2.5 kW, and the total output voltage fluctuation of the link is ensured to be < ±1 V.

[0056] Before the fault master component is reconnected after maintenance, the redundancy state real-time monitoring and recovery triggering mechanism collects the redundancy component output power (stabilized at 2.5 kW), the core temperature (45℃, lower than the rated upper limit of 60℃), and the switch tube conduction voltage drop (1.2 V, normal range 1-1.5 V) at a period of 15 ms. When the fault master component is repaired, the redundancy-master component cooperative recovery mechanism is activated: the output power gradient transfer reduces the redundancy component power at a step of 500 W (each step maintains the link for 1 s), while the deviation between the master component capacitor voltage (598 V) and the link voltage (600 V) is 2 V, the master component pre-charge current is adjusted to 12 A, and the deviation between the master component voltage and the link voltage is < 1 V within 15 s. When the master component output power reaches 12.5 kW and the parameter matching degree is > 0.95, the control master component is connected to the cascade link, the redundancy component is disconnected and returns to standby, and the link power fluctuation during the whole recovery process is < 3%.

[0057] In the embodiment, the system realizes the fault switching time < 500 ms, the power supply interruption risk is reduced by more than 90%, and the core element life is extended by 30% through the above process, which fully meets the demand of new energy storage power station for power supply continuity and stability.

[0058] The above is only a preferred embodiment of the present application, and is not intended to limit the present application in any form. Although the present application has been disclosed as above with a preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the technical solution of the present application. Any simple modification, equivalent change and modification of the above embodiments made according to the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. A cascaded modular DC power supply and its dynamic redundancy control method, characterized in that, include: S1: The main controller activates the redundancy-main component collaborative initialization mechanism. The redundant component has a built-in pre-charging structure. The core logic is to dynamically adjust the capacitor voltage of the redundant component by tracking the capacitor voltage of the main component in real time. The main controller loads the redundancy state pre-sensing mechanism to establish a correlation model between the state parameters of the main component and the wake-up time of the redundant component. S2: Based on the dynamic parameter real-time sensing mechanism, the main component's input voltage stability, output current fluctuation, and temperature change rate are collected in real time and transmitted to the main controller; the main controller calls the hierarchical redundancy triggering mechanism, and based on the parameter deviation-redundancy urgency dynamic mapping relationship, it processes the main component's state parameters in a hierarchical manner and generates redundancy switching instructions. The parameter deviation-redundancy urgency dynamic mapping process of the hierarchical redundancy triggering mechanism is as follows: First, the deviation of the state parameters of the main component is calculated. The deviation of each of the three types of parameters, namely input voltage, output current and temperature, is calculated separately. Then, the comprehensive deviation is calculated according to the preset weight. The comprehensive deviation is compared with the preset slight deviation threshold and moderate deviation threshold. If the comprehensive deviation exceeds the moderate deviation threshold or the state parameter exceeds the rated value, a serious deviation warning is generated. The main controller generates a redundancy switching command, which includes the abnormal main component number, the switching trigger reason and the switching time limit requirement. It is transmitted to the redundant component and the abnormal main component through the communication bus. At the same time, the time of the switching command generation is recorded. S3: The redundancy switching instruction activates the real-time redundancy scheduling mechanism: controls the abnormal main component to reduce the output power at a dynamic slope, synchronously disconnects the output isolation structure, and simultaneously controls the bidirectional thyristor switch of the redundant component to close based on the zero-current dynamic capture strategy. Activate dynamic power matching compensation mechanism: Based on the total output power requirement of the cascaded link and the output capacity of the remaining main components, calculate the power gap value of the redundant components and adjust the output voltage and switching frequency of the redundant components. S4: Redundancy status monitoring and recovery trigger linkage mechanism: Based on the output power, temperature and switching transistor operating data of the redundant components, activate the redundant-main component collaborative recovery mechanism: The output power gradient transfer, based on the deviation between the current capacitor voltage and temperature of the main component and the cascade link parameters, adjusts the pre-charging current and output voltage of the main component; When the output power and parameter matching degree of the main component reach the preset standard, the cascaded link is connected, and the redundant component is disconnected and returns to standby.

2. The method according to claim 1, characterized in that, In S1, the capacitor voltage deviation control between the redundant component and the main component requires first acquiring the real-time voltage of both via a voltage sensor, calculating the capacitor voltage deviation, and then initiating pre-charging current adjustment. The pre-charging current is adjusted according to the deviation-current ratio: the main controller retrieves the rated charging current of the redundant component, sets a reference current input current adjustment structure, and the current increases synchronously according to the reference ratio when the deviation increases; the current adjustment circuit is equipped with an overcurrent protection structure, monitors the current in real time, triggers current limiting when the rated charging current is reached, continuously acquires the voltage deviation, and stops adjustment and locks the current when it stabilizes within the proportional range.

3. The method according to claim 1, characterized in that, In S1, the correlation model of the redundancy state pre-sensing mechanism first starts the load power acquisition structure, acquires the load power at a fixed period, and after a continuous period, obtains the power change rate through differential calculation and inputs it together with the deviation of the main component state parameter into the correlation model; calculates the redundancy wake-up prediction value and compares it with the threshold. If the threshold is reached, the main controller sends a pre-wake-up command. After the redundant component passes the self-test, it switches to the quasi-working state and feeds back a completion signal. The main controller records the pre-wake-up time to reserve a preparation period for subsequent switching.

4. The method according to claim 1, characterized in that, In S2, the dynamic parameter real-time sensing mechanism first activates the load fluctuation calculation structure, collects the total output power of the cascaded links, calculates the fluctuation at a fixed period, and compares it with two preset thresholds. When the fluctuation exceeds the first threshold, the collection frequency is increased and the buffer capacity is increased. The collection frequency and buffer are maintained between the two thresholds. When the fluctuation is below the second threshold, the collection frequency is reduced and the buffer is reduced. After each adjustment of the collection frequency, the parameter collection structure outputs periodic transition data, which is filtered by a moving average to obtain the average value to generate valid data. The validity of the valid data is then confirmed by a verification structure.

5. The method according to claim 1, characterized in that, In S3, during the redundancy dynamic allocation process of the real-time redundancy scheduling mechanism, the main controller obtains the number, rated power, and current fault parameters of all abnormal main components through the communication bus, and calculates the power ratio of the abnormal main components; at the same time, it assesses the urgency of the fault, calculates the temperature over-temperature range and current overload multiple as urgency indicators, and constructs a priority matrix based on the power ratio and urgency indicators. The matrix rows represent abnormal main components, the columns represent evaluation indicators, and each cell is filled with the normalized value of the corresponding indicator. Weights are assigned to the two indicators using the analytic hierarchy process (AHP), the priority score of the main abnormal component is calculated, and a priority list is generated.

6. The method according to claim 1, characterized in that, In S3, the abnormal control main component reduces its output power according to a dynamic slope. First, it collects the total output power and the total power of the remaining normal main components through the power acquisition structure to calculate the current power demand. The main controller retrieves the rated power and calculates the output power reduction slope in combination with the current power demand. It then sends a power reduction command to the conversion structure, including the power reduction slope and the target power. After receiving the power reduction command, the main controller reduces the output power by adjusting the duty cycle of the switching transistor. It monitors the actual rate of output power reduction in real time. If the actual power reduction rate deviates from the set slope beyond the allowable range, it immediately feeds back to the main controller. The main controller then recalculates the slope and corrects the adjustment range of the duty cycle.

7. The method according to claim 1, characterized in that, In S3, the zero current capture strategy is based on the current sensors at the input and output terminals of the bidirectional thyristor switch. The main controller presets the current threshold and configures the number of consecutive comparison cycles. The current sensor collects current data in real time, compares the real-time current value with the current threshold, and records a near-zero current state for each continuous collection period to determine that the bidirectional thyristor switch is currently in a closed state. A closing command is sent to the switch driver, which outputs a trigger voltage to turn on the bidirectional thyristor switch. The current sensor collects the switch conduction current and sends a signal to the main controller that the bidirectional thyristor switch has been successfully closed.

8. The method according to claim 1, characterized in that, In S3, the dynamic power matching compensation mechanism first activates the power gap analysis structure, collects the total output power demand and the total power of the remaining main components, and calculates the power gap. The power gap, the rated voltage of the redundant components, and the switching frequency are input into the particle swarm optimization algorithm structure. In each iteration, each particle represents a combination of output voltage and switching frequency parameters of a set of redundant components. The fitness value of the particle is calculated. The fitness value is a weighted result of comprehensive error and loss. The globally optimal particle and the locally optimal particle are tracked. The particle position is adjusted by updating the particle velocity to generate the optimal parameter combination. After the iteration ends, the optimal output voltage and switching frequency parameters are output.

9. The method according to claim 1, characterized in that, In S4, the redundant status monitoring and recovery trigger linkage mechanism incorporates a multi-parameter fusion anomaly judgment algorithm: by collecting the output power, core temperature, on-state voltage drop and switching speed of the redundant components, the four types of parameters are standardized one by one, the actual values ​​are converted to the rated values, and uniformly mapped to the same numerical range. The processed parameters are then transmitted to the status analysis structure. The state analysis structure is based on the multi-parameter fusion anomaly determination algorithm. It generates anomaly determination values ​​by weighted summation of standardized parameters according to weights, compares them with preset safety thresholds, and generates overload warnings.

10. The method according to claim 1, characterized in that, In S4, the output power gradient transfer first collects the voltage and current of the main component and the cascaded link, calculates the parameter deviation between the main component and the cascaded link, and when it exceeds a preset threshold, the main controller sends a power transfer command to the redundant component.

11. The method according to claim 1, characterized in that, In S4, the main component pre-charge current adjustment first retrieves the rated input and pre-charge current, sets the upper and lower current limits for the pre-charge control structure, and stops pre-charging based on the main component capacitance and link voltage deviation, and generates a pre-charge completion signal.

Images