A distributed energy storage power supply system for building water pumps

Abstract

This invention relates to the field of distributed energy storage and power distribution technology, specifically a distributed energy storage power supply system for building water pumps; it includes variable frequency water pump equipment, distributed energy storage equipment, mains power access interface, power conversion module, and energy management system; the system collects equipment operation and power quality data, and calculates the cumulative amount of thermoelectric stress fatigue of the distributed energy storage equipment accordingly; its core is to compare the cumulative amount with light and heavy fatigue danger thresholds, and generate optimal discharge, suboptimal discharge, or system protection strategy commands in a graded manner to dynamically adjust power distribution and charging and discharging state; this invention overcomes the shortcomings of traditional methods that solely pursue absolute smoothness of bus voltage, and achieves hierarchical power distribution under strong disturbances based on a health state graded response mechanism, which not only meets the transient voltage support requirements of water pumps, but also avoids energy storage degradation due to high-frequency calls, effectively taking into account the long-term resilience of the power supply system.

Description

Technical Field

[0001] This invention relates to the field of distributed energy storage and power supply technology, specifically to a distributed energy storage power supply system for use with building water pumps. Background Technology

[0002] For the secondary water supply and power distribution scenarios in super high-rise complexes, variable frequency pumps, as high-power inductive loads, cause frequent transient disturbances to the power supply bus during their start-up, acceleration, and shutdown processes. To maintain system power quality, distributed energy storage devices are typically connected to the distribution network to work in conjunction with the mains power to achieve bidirectional power regulation. To cope with voltage fluctuations caused by large loads, existing power supply control schemes generally adopt a single transient voltage support strategy, that is, to pursue smooth bus voltage as the sole control objective. Although this scheme has a certain voltage improvement capability during single load changes, when faced with repeated high-current starts of pumps and continuous harmonic surges in the local power grid, the controller will continuously drive the energy storage device to perform high-frequency charging and discharging switching. Due to the high dependence on the full-power response of the energy storage and the lack of consideration for the physical health boundaries of the device, this high-frequency call will lead to rapid heating, increased polarization, and increased internal resistance inside the energy storage, quickly depleting the energy storage system's ability to withstand the next impact start-up, and easily inducing the system's undervoltage shutdown protection at critical moments.

[0003] Therefore, how to balance the transient voltage support of the busbar and the thermoelectric physical health of the energy storage equipment under strong disturbance conditions, and maintain the long-term power supply resilience of the water supply system, has become an urgent technical problem to be solved. Summary of the Invention

[0004] To solve the above-mentioned technical problems, the present invention provides a distributed energy storage power supply system for building water pumps. Specifically, the technical solution of the present invention includes:

[0005] Variable frequency water pump equipment, distributed energy storage equipment, mains power access interface, power conversion module and energy management system;

[0006] The distributed energy storage device and the mains power access interface are respectively electrically connected to the power supply bus of the variable frequency water pump device through the power conversion module.

[0007] The energy management system is equipped with:

[0008] The data acquisition module is used to collect power quality data of the variable frequency water pump equipment, current water supply network pressure value, peak load records during historical startup phases, and operating status data of the distributed energy storage equipment.

[0009] The status assessment module is used to filter the power quality data and the operating status data to obtain effective operating data, and to calculate the cumulative thermoelectric stress fatigue of the distributed energy storage device based on the effective operating data.

[0010] The decision control module is used to generate power allocation instructions based on the accumulated thermoelectric stress fatigue and send them to the variable frequency water pump, the distributed energy storage device and the power conversion module to adjust the operating power status of the variable frequency water pump and the charging and discharging status and power output ratio of the distributed energy storage device.

[0011] The present invention has the following beneficial effects:

[0012] 1. This invention provides a power supply control mechanism based on a health status-based hierarchical response. The state assessment module calculates the cumulative thermoelectric stress fatigue of the distributed energy storage device and compares it with the first and second danger thresholds. Based on this, the decision control module issues optimal discharge, suboptimal discharge, or system protection strategy commands. This design overcomes the shortcomings of traditional methods that solely pursue absolute smoothness of bus voltage. It achieves hierarchical power distribution under strong disturbance conditions, which not only meets the transient voltage support requirements of the variable frequency pump equipment but also avoids the degradation of energy storage due to high-frequency calls, thus taking into account the long-term resilience of the power supply system.

[0013] 2. This invention constructs a dynamic optimization protection mechanism that is both anti-interference and adaptive. The state assessment module uses a noise filtering model to remove harmonic noise data, accurately calculates the fatigue accumulation by combining internal temperature and charge / discharge switching frequency, and predicts the failure risk contribution using a preset failure boundary model. The strategy optimization module dynamically updates the danger threshold based on actual voltage response data and failure risk contribution using a deep reinforcement learning algorithm. This design enables the system to break free from dependence on fixed empirical values, actively reduce failure risk within the preset voltage safety range, and effectively extend equipment life. Attached Figure Description

[0014] Figure 1 This is a structural diagram of a distributed energy storage power supply system for building water pumps according to the present invention. Detailed Implementation

[0015] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. 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.

[0016] Example 1:

[0017] Please see Figure 1 A distributed energy storage power supply system for use with building water pumps, comprising:

[0018] Variable frequency water pump equipment, distributed energy storage equipment, mains power access interface, power conversion module and energy management system;

[0019] The distributed energy storage device and the mains power access interface are electrically connected to the power supply bus of the variable frequency water pump equipment through the power conversion module, respectively.

[0020] The energy management system includes:

[0021] The data acquisition module is used to collect power quality data of variable frequency water pump equipment, current water supply network pressure value, historical peak load records during startup, and operating status data of distributed energy storage equipment.

[0022] The status assessment module is used to filter the power quality data and operating status data to obtain effective operating data, and to calculate the cumulative amount of thermoelectric stress fatigue of the distributed energy storage device based on the effective operating data.

[0023] The decision control module is used to generate power distribution instructions based on the cumulative amount of thermoelectric stress fatigue and send them to the variable frequency water pump equipment, distributed energy storage equipment and power conversion module to adjust the operating power status of the variable frequency water pump equipment and the charging and discharging status and power output ratio of the distributed energy storage equipment.

[0024] The decision control module is preset with a first danger threshold for characterizing mild fatigue and a second danger threshold for characterizing severe fatigue, and the first danger threshold is less than the second danger threshold.

[0025] The power allocation instructions include: the optimal discharge strategy instruction generated when the accumulated thermoelectric stress fatigue is less than the first danger threshold, the suboptimal discharge strategy instruction generated when it is greater than or equal to the first danger threshold and less than the second danger threshold, and the system protection strategy instruction generated when it is greater than or equal to the second danger threshold.

[0026] This embodiment provides a distributed energy storage power supply mechanism for secondary water supply scenarios in super high-rise complexes. Specifically, this embodiment takes a high-rise building with zoned water supply as a continuous main line scenario. Multiple variable frequency water pumps are installed in the underground machine room of the building. The power supply bus is connected to distributed energy storage devices and mains power at the same time, and the power is adjusted bidirectionally through the power conversion module.

[0027] The energy management system continuously collects information such as bus voltage fluctuations, frequency offsets, harmonic distortions, reactive power disturbances, and the state of charge, temperature, and charge / discharge switching status of distributed energy storage devices to determine whether the current control objective is to maintain transient voltage support or to ensure the physical health margin of the stored energy.

[0028] Specifically, the data acquisition module does not only sample a single voltage value, but simultaneously observes the electrical characteristics of the pump during four stages: startup, acceleration, steady-state operation, and shutdown and reversal. For high-power inductive loads like building pumps, what truly makes the system vulnerable is not the single power change itself, but the internal heating, polarization, and increased internal resistance caused by repeated high-current startups and frequent compensation.

[0029] Therefore, after obtaining the collected data, the condition assessment module first filters out high-frequency disturbances that significantly deviate from physical continuity, and then extracts the characterization quantities that can reflect the pressure and heat dissipation burden of the battery cell from the remaining effective operating data, thereby forming the cumulative amount of thermoelectric stress fatigue.

[0030] This cumulative amount does not simply represent the remaining power, but rather is a resilience indicator used to characterize whether the energy storage device can continue to reliably provide the instantaneous power required for the next pump start-up under continuous impacts;

[0031] At the decision-making level, the system adopts a hierarchical strategy rather than a single optimal strategy. The reason is that when there are continuous harmonic surges and temporary drops in the mains power in the local distribution network of the building, if the only goal is to pursue absolute smoothness of the bus voltage, the controller is likely to continuously drive the distributed energy storage device to switch between high-frequency charging and discharging. In the short term, it may seem to improve the voltage quality, but in the long term, it will quickly deplete the distributed energy storage device's ability to withstand the next startup impact.

[0032] Therefore, this embodiment sets a first danger threshold and a second danger threshold: when the fatigue level is low, energy storage is allowed to take priority in providing rapid support to the bus; when the fatigue level enters the intermediate risk zone, the system actively yields and no longer pursues the highest tracking accuracy, but executes the power allocation method corresponding to the suboptimal discharge strategy; when the fatigue level reaches the severe risk zone, the protection of energy storage availability is given higher priority to prevent it from being called up at high frequency in the critical state.

[0033] As a protective alternative, if a local sensor loses connection for a short time, data timestamp is misaligned, or a single temperature probe drifts in the acquisition link, the state assessment module will prioritize calling the steady-state reference data within the most recent reliable time window to maintain the missing items for a short time.

[0034] If the deficiency persists or key quantities such as bus voltage and internal temperature of energy storage are simultaneously unavailable, the decision control module will no longer output aggressive dynamic compensation commands, but will directly enter a conservative power supply mode to prevent the distributed energy storage device from over-discharging or exceeding the power output limit due to incorrect judgment.

[0035] If the system detects that the bus can no longer meet the minimum undervoltage protection boundary of the water pump, it will immediately cut off the high-frequency regulation chain and only retain the protective power supply logic;

[0036] For example, on a hot summer afternoon, the building's air conditioning control system malfunctioned, and some loads in the building switched on and off in a disorderly manner. The remaining available capacity of the distributed energy storage device was within the preset safety margin that could meet the theoretical lower limit of the next frequency converter pump start-up.

[0037] At this time, the acquisition module found that the water pump had not completely stopped, but the bus had experienced several rapid fluctuations. Based on this, the status assessment module determined that the current disturbance was not a single transient event, but a disturbance phase that could continuously consume the resilience of the energy storage. Therefore, the decision control module no longer mechanically pursued that every voltage drop be immediately compensated by the energy storage, but switched between three strategies according to the degree of fatigue accumulation, so that the power supply system shifted from a control approach that only maintained transient stability to one that also took into account the survivability of the next startup.

[0038] The purpose of this step is to expand the traditional control framework that only focuses on voltage stability and basic energy consumption into a hierarchical power allocation mechanism that also considers the physical health boundary of energy storage, so as to achieve the resilience maintenance and failure delay of the building's water supply system under strong disturbance conditions.

[0039] The condition assessment module stores a noise filtering model; the condition assessment module is used to input power quality data and operating status data into the noise filtering model, filter out harmonic noise data, and output valid operating data.

[0040] This embodiment provides a data purification mechanism for environments with strong harmonic pollution. Specifically, in the main scenario where the computer room of the aforementioned building continuously encounters disordered harmonic surges, relying solely on the original sampled data to directly drive the control can easily lead to mistaking harmonic spikes, sampling jitter, and reactive power oscillations for actual load demand, thereby inducing the energy storage equipment to charge and discharge repeatedly for no reason.

[0041] Therefore, in this embodiment, a noise filtering model is preset in the state assessment module to identify and remove high-frequency noise components that do not conform to the physical inertia of the water pump load;

[0042] Specifically, the noise filtering model can adopt a processing method that combines time continuity constraints and frequency domain anomaly identification; since the mechanical rotation system of the water pump has inherent inertia, its actual power change usually has an interpretable rise, maintenance and fall process, and will not randomly flip positive and negative in a very short time; similarly, the internal temperature of the battery also has a thermal capacity effect and will not keep in sync with the voltage spike at a high speed.

[0043] Therefore, when processing bus voltage, current waveforms, energy storage power commands and temperature sequences, the model prioritizes retaining slow-changing components that are consistent with the physical response of the equipment, and regards components that appear suddenly and have a very short duration and lack coordination with adjacent channels as harmonic noise or acquisition disturbances.

[0044] In practice, the noise filtering model can adopt any one of the following algorithms: Kalman filtering algorithm, wavelet transform denoising algorithm, or low-pass limiting filtering algorithm with a cutoff frequency. By setting a time constant that matches the mechanical inertia of the water pump, the effective separation and elimination of high-frequency spikes and random jitter can be achieved.

[0045] For ease of understanding, three consecutive sampling windows can be denoted as W1, W2, and W3: If W1 and W3 both show the same operating trend, while W2 alone shows a sharp reverse jump, and this jump is not confirmed in the pump speed feedback and energy storage temperature response, then this part is judged as noise and suppressed.

[0046] Furthermore, the noise filtering model not only processes power quality data but also processes operating status data simultaneously. The reason is that if only the bus-side data is purified without purifying the energy storage-side data, the controller may still make biased judgments based on the noisy battery status.

[0047] For example, if a temperature sampling point of a distributed energy storage device experiences a short-term abnormal rise due to the hot air recirculation in the data center, and if it is not verified for consistency with the temperature and power changes of adjacent modules, the localized incidental value may be mistaken for an overall thermal imbalance, leading to unnecessary derating.

[0048] If the intensity of harmonic pollution is higher than the range that the model can stably identify, causing multiple sampling channels to be distorted at the same time, the system will no longer forcibly output fine compensation results, but will mark the time window as a low confidence interval and hand it over to the subsequent decision-making to adopt conservative control.

[0049] If the amount of effective data after noise filtering is insufficient to support the state assessment, the short-term control parameters of the previous reliable state are maintained, and the proportion of grid power is temporarily increased to avoid continuing to rely on energy storage for high-frequency support during the period of information distortion.

[0050] For example, on the afternoon of the aforementioned high temperature, the bus voltage waveform collected in the computer room showed continuous high-frequency transient spikes, characterized by high-frequency fluctuations in the water pump load, but the speed feedback and pipeline pressure changes remained slow.

[0051] Based on this, the noise filtering model determined that the waveform mainly originated from the local harmonic injection caused by the instability of the air conditioning group control, rather than the actual changes in the pumping conditions. Therefore, the corresponding peak components were removed, and only the data segments reflecting the actual start-up demand of the water pump and the actual heat load of the energy storage were retained as input for subsequent fatigue assessment.

[0052] The purpose of this step is to prevent the control system from generating high-frequency and invalid power compensation outputs due to abnormal disturbance data, and to provide a reliable data basis for subsequent risk assessment.

[0053] The operational status data includes charge / discharge switching frequency data and internal temperature data of distributed energy storage devices. The effective operational data includes the filtered charge / discharge switching frequency data and internal temperature data.

[0054] The condition assessment module is also used to determine the electrical stress index based on the filtered charge-discharge switching frequency data in the effective operating data and the preset electrical stress conversion coefficient, and to determine the thermal stress index based on the internal temperature data and the preset thermal stress conversion coefficient; the condition assessment module is also used to determine the cumulative amount of thermoelectric stress fatigue based on the electrical stress index and the thermal stress index.

[0055] This embodiment provides a concrete mechanism for constructing the cumulative amount of thermoelectric stress fatigue. Specifically, after noise filtering, the system determines the electric stress index based on the filtered charge-discharge switching frequency data in the effective operating data and the preset electric stress conversion coefficient, determines the thermal stress index based on the internal temperature data in the effective operating data and the preset thermal stress conversion coefficient, and determines the cumulative amount of thermoelectric stress fatigue by combining the electric stress index and the thermal stress index.

[0056] There are obvious flaws in measuring risk using only the state of charge. This is because even if a distributed energy storage device has sufficient remaining power, if it experiences too many charge-discharge reversals in a short period of time, the electrode interface and conductive path may enter a high-load state. At the same time, if the internal temperature remains high, the cell reaction rate, internal resistance change and aging speed will also be amplified.

[0057] The combination of these two factors will directly weaken its ability to provide a large current during the next strong start of the water pump; therefore, in this embodiment, the charging and discharging switching frequency is regarded as the external manifestation of electrical stress, and the internal temperature is regarded as the external manifestation of thermal stress, and they are respectively mapped to stress indices that can be uniformly processed through preset conversion relationships.

[0058] For ease of explanation, a simplified theoretical model can be used: If the power direction of the distributed energy storage device is discharge-charge-discharge within three consecutive control cycles, the switching behavior of this time window is more drastic than continuous discharge-continuous discharge-slow decline, and the former corresponds to a higher level of electrical stress; if the module temperature changes from the normal temperature zone to the continuous high temperature zone within the same time window, it indicates that the drastic switching has been transformed into a substantial thermal burden.

[0059] The condition assessment module does not focus on a single instantaneous peak, but rather on the combination of frequent reverse scheduling and temperature rise that is not easy to subside, because this combination is more in line with the engineering law of energy storage resilience degradation.

[0060] The reason for using weighted integration of the two types of stress is that the failure mechanisms of different distributed energy storage devices may be different. Some are more susceptible to high-frequency switching, while others are more susceptible to heat dissipation limitations. Therefore, by presetting weights, the evaluation results can be closer to the actual characteristics of the devices. Furthermore, the conversion coefficients and weights can be calibrated through factory tests, bench impact tests, or historical operation and maintenance data, and stored in advance in the condition assessment module.

[0061] The specific quantitative deduction logic is as follows: The charge / discharge switching frequency data is specifically represented as the total frequency of charge / discharge state reversals of distributed energy storage devices within a preset fixed-length evaluation time window. Its value is equivalent to the ratio of the total number of charge / discharge switches within that time window to the length of the time window; let the length of the evaluation time window be... ;

[0062] Let the total number of charge / discharge switching frequencies obtained within this evaluation time window be . The preset electrical stress conversion factor is Let the electric stress index be... The electric stress index is then expressed as:

[0063]

[0064] Let the current internal temperature represented by the currently acquired internal temperature data be... The preset reference temperature is Let the thermal stress index be... Only when The heat load is calculated from the time of the heat load, that is, if but Let the thermal stress conversion coefficient be... The thermal stress index is then expressed as:

[0065]

[0066] It should be noted that the electrical stress conversion factor With thermal stress conversion coefficient It contains a dimensionless normalization factor to ensure that the calculated electrical stress index is accurate. With thermal stress index After being converted into dimensionless scalar values, they are then included in the weighted summation; let the weight of electric stress be... Thermal stress weight is and constraints Let the cumulative amount of thermoelectric stress fatigue be... The cumulative amount of thermoelectric stress fatigue is then calculated as follows:

[0067]

[0068] This explicit linear computational structure avoids complex implicit operations, which can not only intuitively reflect stress growth, but also ensure the transparency and reproducibility of real-time computation at the system's underlying level.

[0069] Its essence is not to pursue precise theoretical calculations, but to transform empirical judgments such as the degree of damage to the battery caused by frequent switching and the danger level of the current temperature rise into a stable and quantifiable basis, so that the controller can use consistent criteria under different equipment and different computer room heat dissipation conditions.

[0070] If the switching frequency is normal during a certain period but the internal temperature rises abnormally, the system will prioritize determining that the heat dissipation link may be blocked to avoid mistakenly attributing all risks to the scheduling strategy; if the temperature is normal but the switching frequency increases abnormally, the system will prioritize restricting the issuance of high-frequency commands to prevent the potential electrical stress that has not yet manifested as a temperature rise from continuing to accumulate; if both types of data are not reliable enough, the time window will not participate in aggressive regulation, but will be taken over by a conservative threshold.

[0071] For example, after the aforementioned building suffered continuous harmonic interference, although the distributed energy storage device was still able to maintain the bus voltage, the background records showed that its charging and discharging direction frequently repeated within several short cycles, and the module temperature on the side closer to the power conversion module began to be higher than the normal nighttime power replenishment conditions.

[0072] The status assessment module thus determines that the current situation is not simply an increase in voltage support tasks, but rather that the energy storage is under the dual impact of frequent electrical switching and accumulated thermal burden. Therefore, it maps this status to an increase in the cumulative amount of thermoelectric stress fatigue, providing a basis for subsequent control degradation.

[0073] The purpose of this step is to transform the hidden degradation process of energy storage devices from invisible internal aging into a resilience indicator that can be continuously monitored and used for decision-making, thereby enabling early identification of failure precursors.

[0074] The condition assessment module is also used to obtain the impact demand of the variable frequency pump equipment calculated based on the rated power of the variable frequency pump equipment, the current water supply network pressure value, and the peak load record of the historical start-up phase; input the cumulative amount of thermoelectric stress fatigue and the impact demand into the preset failure boundary model to output the failure risk contribution.

[0075] This embodiment provides a risk mapping mechanism oriented towards the water supply shutdown boundary; specifically, based on the aforementioned fatigue accumulation amount which can already characterize the degree of energy storage damage, this embodiment further introduces a preset failure boundary model to transform the current fatigue accumulation degree of the distributed energy storage device into a quantitative assessment result characterizing its contribution to the risk of the next water pump undervoltage shutdown.

[0076] In detail, simply knowing that the fatigue accumulation is increasing is not enough to guide power supply decisions; the core is to assess whether the current fatigue accumulation state has posed a substantial threat to the building's next water supply startup; to this end, a pre-set failure boundary model establishes a mapping relationship between the fatigue accumulation and failure risk.

[0077] This mapping relationship is derived from historical startup tests, long-term operation and maintenance records, and the calibration of the impact power supply capability under different temperature conditions. It reflects a real engineering fact: distributed energy storage devices do not fail only when the capacity is exhausted. In many cases, they fail when there is still a capacity margin, because the internal resistance increases, the thermal state deteriorates, and the transient output capability decreases, which can no longer meet the voltage support requirements for pump startup.

[0078] In other words, the risk contribution describes how much the current energy storage state increases the vulnerability of the entire water supply system; this mapping can be understood in a simplified way: if a distributed energy storage device can still stably support a startup under low temperature and low switching frequency while under the same medium state of charge, then its risk contribution is low.

[0079] Another distributed energy storage device, after high temperature and frequent adjustments, although showing similar remaining power, is more likely to cause a deep drop in the busbar once the water pump starts, thus contributing more risk; the preset failure boundary model makes this difference between similar surface capacity and different actual start-up support capabilities explicit.

[0080] To ensure that the prediction process of the preset failure boundary model can be clearly traced, the mapping relationship is specifically decomposed into a preset two-dimensional lookup table matrix structure: the horizontal index of this matrix is ​​the discretized thermoelectric stress fatigue accumulation ladder, and the vertical index is the predicted impact demand ladder of the variable frequency water pump equipment during current operation. The predicted impact demand of the variable frequency water pump equipment is calculated in real time by the system based on the rated power of the variable frequency water pump equipment, the current water supply network pressure value, and the peak load records of the historical start-up phase, using a weighted average prediction algorithm.

[0081] Specifically, the weighted average prediction algorithm is as follows: the rated power of the variable frequency pump equipment, the current water supply network pressure value, and the peak load records of the historical start-up phase are assigned corresponding preset weight coefficients and linearly weighted summed to obtain the impact demand that represents the instantaneous peak power gap; the impact demand represents the instantaneous peak power gap expected to be required during the pump start-up phase; each intersection node of the matrix stores the corresponding failure probability value, which is directly equivalent to the failure risk contribution degree, and its value is between 0 and 1;

[0082] This structured lookup table mapping method avoids complex implicit calculations and ensures that every degradation decision is supported by definite addressing data; the output risk contribution can be used by the downstream controller for sorting and limiting.

[0083] For example, when there are multiple distributed energy storage devices in the system, distributed energy storage devices with higher risk contributions can be prioritized for load reduction or postponed, while distributed energy storage devices with lower risk contributions can continue to undertake limited rapid support tasks; for a single distributed energy storage device scenario, the risk contribution directly determines whether optimal voltage compensation should be abandoned and conservative operation should be adopted.

[0084] If the input data of the preset failure boundary model is outside the coverage of historical samples, such as when the ambient temperature in the computer room is abnormally high and the mains power sag time is significantly longer than normal operating conditions, the system can adjust the risk contribution level to a conservative level to avoid underestimating the risk due to overly optimistic extrapolation of the model.

[0085] If the historical mapping data is insufficient to support fine-grained classification, at least three risk levels—high, medium, and low—should be output to ensure that the control link has a clear basis for action.

[0086] For example, during the process of the aforementioned building's high temperature disturbance continuing to expand, the condition assessment module found that although the cumulative amount of thermoelectric stress fatigue had not yet reached the highest danger zone, the preset failure boundary model predicted that its risk contribution to the next joint start-up scenario of fire protection and domestic water supply had significantly increased.

[0087] This means that continuing to use energy storage as an ideal fast power source will significantly increase the probability of bus undervoltage shutdown, thus causing the system to enter a more conservative power allocation range earlier.

[0088] The purpose of this step is to transform abstract health decline into risk quantification indicators that address actual failure consequences, thereby establishing a direct link between control strategies and water supply security objectives.

[0089] The optimal discharge strategy commands include high-frequency power tracking commands and voltage compensation commands;

[0090] This embodiment provides an optimal discharge control mechanism suitable for the stage of sufficient energy storage resilience. Specifically, when the state assessment results show that the fatigue accumulation is below the first danger threshold, the system believes that the energy storage device still has the transient output capability and thermal buffer margin that meet the preset standards. At this time, the high-frequency power tracking command and voltage compensation command can be activated first to ensure the bus stability during the pump start-up and operation stages.

[0091] Specifically, the high-frequency power tracking command is mainly used to enable energy storage devices to respond quickly to changes in pump load, especially during the initial stage of frequency converter frequency increase and pipeline pressure recovery stage, where energy storage can take on the task of rapidly increasing or decreasing power in a short period of time; the voltage compensation command is used to inject compensation power into the bus through the power conversion module when there is a recoverable short-term voltage drop or local fluctuation on the bus, so as to maintain the DC or AC side of the frequency converter working in a stable range.

[0092] The reason it is defined as optimal is that when the energy storage is in good health, this strategy can simultaneously take into account transient performance and voltage quality, enabling the variable frequency pump equipment to reach the target operating condition within a preset time threshold.

[0093] However, the optimality here is a conditional engineering optimality, not one that must be executed under all circumstances; the premise is that the energy storage device has not yet entered a significant fatigue range and the current sampling data has sufficient reliability; high-frequency tracking and rapid compensation are only truly beneficial when both of these premises are met simultaneously.

[0094] If, although the system is in a low fatigue range, a decrease in data reliability, abnormal amplification of harmonic noise, or a continuous expansion trend of disturbances on the mains side are detected, the system can temporarily postpone entering the strongest compensation mode and adopt a milder tracking intensity. If the water pump is in a non-critical water supply period, the aggressiveness of the compensation can also be reduced to avoid unnecessary energy storage wear.

[0095] For example, during the morning water replenishment phase before the onset of hot weather, the load on the building's computer room is relatively stable, the temperature of the distributed energy storage equipment is moderate, the switching frequency is low, and the system judges that its health margin is sufficient.

[0096] At this time, when the water level in the high-zone water tank drops and the main pump needs to be started to replenish water, the decision control module issues high-frequency power tracking commands and voltage compensation commands, enabling the energy storage to quickly fill the transient gap of the bus during the pump frequency conversion speed-up, reducing the disturbance of voltage drop to the frequency converter;

[0097] The purpose of this step is to fully leverage the rapid response advantage of the energy storage equipment when it is within acceptable limits, thereby achieving smooth pump start-up, stable bus voltage, and improved water supply continuity.

[0098] Example 2:

[0099] Based on Example 1, this embodiment further defines the specific execution mechanism of the optimal discharge strategy and the suboptimal discharge strategy; the optimal discharge strategy instruction includes a high-frequency power tracking instruction and a voltage compensation instruction; the distributed energy storage device is used to perform charge and discharge switching at a preset charge and discharge switching frequency after receiving the high-frequency power tracking instruction and the voltage compensation instruction, so as to maintain the voltage stability of the power supply bus.

[0100] This embodiment provides an execution mechanism that complements the optimal discharge strategy. Specifically, when the energy storage device receives the high-frequency power tracking command and the voltage compensation command, it does not switch the charging and discharging state at an arbitrary speed, but completes the execution under the preset charging and discharging switching frequency constraint, so as to balance the compensation sensitivity and the device's tolerance.

[0101] Simply proposing a high-frequency tracking target is not enough to form a control that can be implemented. If the switching rhythm is not restricted, the controller may become oversensitive due to the instantaneous ripple of the bus, causing the power command to reverse frequently in a very short time. This will turn the control action that is intended to improve voltage quality into a source of stress interference that accelerates the battery aging process.

[0102] Therefore, this embodiment sets a preset switching frequency, which is essentially setting an engineering boundary between rapid support and physical protection: the charging and discharging response rate of distributed energy storage devices must strictly meet the thermal and electrical safety constraints of batteries and power devices;

[0103] For ease of understanding, the continuous control period can be divided into several adjustment segments. If the bus experiences a voltage drop in segment A, recovers in segment B, and experiences another small voltage drop in segment C, the energy storage can respond with a discharge-hold-mild discharge pattern according to a preset rhythm, rather than performing a completely reverse action of discharge-charge-discharge on every tiny ripple. This execution method is closer to the stable adjustment logic of real power electronic equipment in industrial fields.

[0104] Meanwhile, the preset switching frequency can be set according to the battery type, heat dissipation capacity and water pump level; for water supply systems that undertake fire linkage protection tasks, priority can be given to ensuring the start-up success rate, and the response speed can be appropriately increased within the safety boundary; for non-emergency scenarios such as domestic water replenishment, the upper limit of the switching frequency can be reduced to extend the energy storage life.

[0105] If the local temperature of the distributed energy storage device is higher than the normal adjustment range, even if the upper layer still issues a high-frequency tracking request, the execution layer can trigger frequency clamping to prevent further accumulation of internal heat. If the bus disturbance amplitude is less than the set dead zone, no substantial charging and discharging switch will be triggered, and the power conversion module will maintain static balance.

[0106] In the aforementioned morning water replenishment operation, the start of the main pump causes the bus to dip briefly. The energy storage performs continuous but not overly dense discharge compensation according to the preset switching frequency to keep the bus voltage within the allowable range of the frequency converter. When a slight rebound occurs later, the energy storage does not immediately and violently reverse charge, but maintains a short-term buffer to avoid self-excited oscillation due to excessive sensitivity.

[0107] The purpose of this step is to provide the optimal discharge strategy with an engineering-feasible execution boundary, thereby achieving coordination between voltage stability control and energy storage lifetime protection.

[0108] Suboptimal discharge strategy instructions include acceleration slope limitation instructions and mains power access increase instructions;

[0109] This embodiment provides a suboptimal discharge mechanism for the medium-risk stage; specifically, when the accumulated thermoelectric stress fatigue has reached the first danger threshold but has not yet reached the second danger threshold, the system performs shock reduction action through the coordinated cooperation of variable frequency water pump equipment and distributed energy storage equipment.

[0110] In practice, the variable frequency pump equipment is used to reduce its variable frequency acceleration slope during the startup phase after receiving the acceleration slope limit command; the distributed energy storage equipment is used to reduce the discharge power ratio after receiving the mains power access increase command, and to increase the mains power supply ratio from the mains power access interface through the power conversion module.

[0111] In detail, at this stage, the optimal strategy of the previous layer reveals obvious defects: although it can provide better bus transient performance, continuous use will further increase the number of charge and discharge switching and thermal burden, which will rapidly erode the energy storage's ability to support the next critical start-up.

[0112] Therefore, this embodiment adjusts the control objective from eliminating voltage fluctuations as much as possible to ensuring that the water supply function remains available without excessively damaging the energy storage; the specific control logic of the acceleration slope limit command is to allow the variable frequency water pump equipment to operate at a rate lower than the preset standard start-up slope, thereby reducing the instantaneous inrush current and the depth of bus voltage drop;

[0113] The specific control logic of the grid power access increase command is to transfer part of the rapid power support originally mainly undertaken by energy storage to the grid power side. Even if this increases the peak power extraction power of the grid power side and causes the steady-state error of the bus voltage to increase, the available capacity margin of distributed energy storage equipment is reserved first.

[0114] This suboptimal strategy is not a decrease in control capability, but rather a proactive shift in control objectives; especially when the building's water supply system is close to its start-up capacity boundary, maintaining reliable energy storage impact support capability within preset limits has a higher control priority than maintaining the absolute smoothness of the bus voltage waveform.

[0115] If the mains power supply is already severely sluggish and cannot provide the increased capacity, the system will simultaneously tighten the pump speed-up request when outputting the suboptimal strategy, and may adopt time-sharing start-up or delay the commissioning of non-critical pumps if necessary. If the current situation is a forced water supply condition such as fire fighting, the slope limit can be appropriately relaxed, but the upper limit of energy storage call-up is still retained to avoid completely depleting it in a single task.

[0116] On the aforementioned hot afternoon, the malfunction of the air conditioning group control caused continuous disturbances in the building's load, and the fatigue level of the distributed energy storage equipment was approaching the danger zone. At this time, the system judged that continuing to allow the energy storage to follow the slight fluctuations of each busbar for high-frequency compensation would significantly increase the probability of the next high-temperature main pump restart failure. Therefore, it began to issue a suboptimal strategy: on the one hand, it required the acceleration slope of the variable frequency water pump equipment to be reduced, and on the other hand, it increased the proportion of mains power input, so that the busbar mainly relied on the external power grid to maintain energy supply, while the energy storage only retained the necessary auxiliary role.

[0117] The purpose of this step is to prioritize maintaining the long-term survivability of the system when there is a conflict between energy storage health and transient performance, thereby proactively moving away from the failure boundary.

[0118] The variable frequency pump equipment is used to reduce the variable frequency acceleration slope of the variable frequency pump equipment during the startup phase after receiving the acceleration slope limit command; the distributed energy storage equipment is used to reduce the discharge power ratio after receiving the mains power access increase command, and to increase the mains power supply ratio from the mains power access interface through the power conversion module.

[0119] This embodiment provides a device-side implementation mechanism for a suboptimal discharge strategy. Specifically, after entering the medium-risk stage, the variable frequency pump and the energy storage device perform different but coordinated shock reduction actions: the former reduces the load ramp-up speed by lowering the variable frequency acceleration slope during the startup phase, while the latter releases battery pressure by reducing its own discharge power ratio and increasing the mains power supply ratio.

[0120] Specifically, the variable frequency acceleration slope of the water pump directly affects the current rise rate and mechanical hydraulic impact during startup; if the acceleration is too fast, the busbar will be subjected to a large energy absorption pressure in a short period of time, and even if the stored energy can make up for it in a short time, it will still form a significant high current impact and internal heating.

[0121] By limiting the acceleration slope, the process of the water pump from standstill to the target speed is lengthened, the bus power demand curve is flatter, and the energy storage no longer needs to perform transient full power output. Correspondingly, after the energy storage device reduces the proportion of discharge power, it no longer plays the main power supply role, but instead takes a backseat to auxiliary stabilization function. The power conversion module simultaneously increases the proportion of mains power input, so that the whole system shifts to an architecture of main external grid supply and energy storage buffer.

[0122] From an engineering perspective, the value of this coordinated action lies in simultaneously reducing the impact on the power supply side and the internal losses of energy storage. Simply limiting the acceleration of the water pump, if the proportion of mains power access remains unchanged, may still be insufficient to alleviate the pressure on energy storage. Conversely, simply increasing the proportion of mains power, if the water pump continues to accelerate at the original aggressive rate, may still result in a deep voltage drop on the bus. Therefore, only by combining the two can an effective conservative power supply state be formed.

[0123] If the mains power capacity is limited by the margin of the distribution transformer and cannot be increased indefinitely, the power conversion module will gradually increase the input according to the allowable range of the transformer, while further reducing the pump speed-up slope. If the current water tank level is close to the lower limit and excessive delay in water supply is not allowed, the system can limit the slope only in the initial stage of startup, and then restore it appropriately after the speed enters a more stable range, so as to take into account the water supply timeliness.

[0124] For example, in the main scenario, the main pump for domestic water supply in the high-rise area needs to be put back into operation, but the distributed energy storage device has already shown obvious fatigue due to the previous disturbance; after receiving the slope limit command, the frequency converter no longer starts according to the conventional rapid acceleration curve, but adopts a smoother frequency acceleration process.

[0125] Meanwhile, the energy storage device shifted most of the start-up compensation originally planned to the mains power input interface, and the power conversion module automatically increased the share of mains power input; ultimately, although the time required for the water pump to reach its rated operating condition increased slightly, the internal temperature rise and switching frequency of the energy storage were significantly mitigated.

[0126] The purpose of this step is to transform abstract suboptimal decisions into specific equipment actions, thereby achieving substantial peak shaving of startup impacts and preserving energy storage resilience.

[0127] Example 3:

[0128] System protection strategy commands include energy storage lockout commands and full mains power supply commands;

[0129] This embodiment provides a system protection mechanism for high-risk areas; specifically, when the accumulated amount of thermoelectric stress fatigue reaches the second danger threshold or above, the system considers that the energy storage device is close to the danger boundary of being unable to reliably support the next power supply impact. At this time, it is no longer allowed to continue to participate in high-frequency regulation, but instead outputs an energy storage lockout command and a full mains power supply command.

[0130] In summary, during the moderate risk phase, energy storage can still play a limited auxiliary role. However, once fatigue has entered the severe range, continuing to call up energy storage often triggers two problems: the internal temperature rise and polarization state of the battery cells may cause a severe decrease in transient output capability, and even if a discharge command is issued, it may not be able to effectively support the bus. Under this state, forcibly calling up energy storage may cause more severe voltage instability, and may even trigger the superposition of energy storage protection action and water pump undervoltage protection, causing the water supply system to lose two support points at the same time.

[0131] Based on this, this embodiment adopts the system degradation isolation protection approach, that is, to remove the energy storage from the active power supply link to prevent it from being consumed at the most vulnerable moment.

[0132] Energy storage lockout does not necessarily cut off all hardware connections; it can also prohibit its output of rapid compensation power, freeze the charge and discharge switching channel, or limit it to only retain monitoring and standby functions. Full grid power supply means that at the current stage, the external power grid undertakes all the main power supply tasks, and energy storage no longer participates in load sharing.

[0133] For buildings requiring continuous water supply, this strategy, while leading to greater reliance on external power grids and poorer resistance to transient drops, can prevent irreversible physical damage to energy storage devices.

[0134] As a protective backup plan, if the mains power is also severely abnormal and cannot support the entire load on its own, the system can further implement load grading, prioritizing the water supply to fire protection, refuge floors and high-rise foundations, and postponing non-critical water replenishment tasks; if the temperature recovers and the switching stress decreases after the energy storage is locked, and it is confirmed by continuous observation that it has returned to the safe range, the lockout can be lifted and the limited support function can be gradually restored, instead of directly returning to the aggressive mode.

[0135] For example, after the aforementioned extreme high-temperature disturbance lasted for several hours, the temperature rise of the distributed energy storage device had not completely subsided, and the repeated and frequent adjustments had brought its failure risk close to the boundary.

[0136] Based on this, the system issues an energy storage interlock command, prohibiting it from continuing to undertake bus compensation and switching the computer room water supply system to full mains power supply; for non-emergency water pump start / stop requests, the system can also appropriately postpone them to avoid new impact superposition before the power grid is restored;

[0137] The purpose of this step is to decisively cut off the high-load participation path of energy storage when it is about to lose reliability, thereby blocking the risk of irreversible shutdown and protecting critical water supply functions.

[0138] The energy management system also includes a strategy optimization module; used to collect actual voltage response data of variable frequency pump equipment after executing power allocation commands; constructing a state vector based on the actual voltage deviation rate, failure risk contribution, and thermoelectric stress fatigue accumulation extracted from the actual voltage response data; constructing an action space with the adjustment amount of the first danger threshold and the second danger threshold; and dynamically updating the first danger threshold and the second danger threshold through a deep reinforcement learning algorithm with the objective function of maintaining the actual voltage deviation rate within the preset voltage safety range and reducing the failure risk contribution.

[0139] This embodiment provides a threshold adaptive optimization mechanism that combines deep reinforcement learning. Specifically, based on the aforementioned risk mapping which can output the failure risk contribution, this embodiment further sets up a strategy optimization module to continuously collect the actual voltage response after the execution of various power allocation commands, and dynamically update the first danger threshold and the second danger threshold by combining the risk contribution and the cumulative amount of thermoelectric stress fatigue, so that the system no longer depends on long-term fixed empirical thresholds.

[0140] While fixed threshold schemes are easy to implement, they often reveal significant flaws under conditions of continuous high temperatures, deteriorating heat dissipation in computer rooms, changes in the aging of energy storage, or alterations in the building's load structure. Overly optimistic thresholds can delay the system's entry into conservative mode, leading to excessive consumption of energy storage. Conversely, overly conservative thresholds can cause the system to prematurely abandon its energy storage support capabilities, resulting in unnecessary dependence on the external power grid and slow water supply response.

[0141] Therefore, this embodiment introduces a deep reinforcement learning algorithm to extract optimized triggering conditions for switching strategies under different fatigue and risk levels from long-term operational feedback;

[0142] Optionally, considering the discrete characteristics of the action space, the deep reinforcement learning algorithm specifically adopts a deep action value network algorithm or a proximal policy optimization algorithm, which maps the state vector to the discrete action value function by constructing a multilayer perceptron network.

[0143] To clarify the logical operation process of the algorithm and achieve a clear flow of control commands, this embodiment structurally decomposes the operational framework of deep reinforcement learning, specifically as follows: The state space is defined by the current actual voltage deviation rate extracted from the actual voltage response data in real time feedback from the environment. The actual voltage deviation rate is specifically the percentage value of the current actual sampled voltage of the bus deviating from the rated operating voltage of the system; the failure risk contribution value and the cumulative thermoelectric stress fatigue value together form the state vector.

[0144] Define the action space, which is limited to controlled discrete fine-tuning of the first danger threshold and the second danger threshold, such as an adjustment step size of +0.02, 0 or -0.02, and forcibly constrain the first danger threshold to always be less than the second danger threshold to prevent logic reversal;

[0145] The preset voltage safety range is specifically set to a continuous range of 90% to 110% of the rated voltage of the power supply bus of the variable frequency water pump equipment, that is, the maximum allowable deviation is ±10%. The upper and lower limits of this range are pre-calibrated based on the undervoltage protection threshold and overvoltage alarm threshold set by the inverter at the factory.

[0146] The reward function is defined, and its evaluation logic is as follows: when the actual voltage response data after adjustment remains within the preset voltage safety range and the contribution of failure risk shows a downward trend, a positive reward is given.

[0147] Once the actual voltage response data exceeds the preset voltage safety range, a preset negative reward feedback is immediately given; thus, the algorithm transforms from unbounded trial and error to directional optimization with safety as the absolute constraint.

[0148] This optimization does not simply pursue absolute smoothness of the voltage waveform, but is driven by two objectives: on the one hand, it requires the actual voltage response to be maintained within a preset safe range to avoid the water pump from shutting down due to undervoltage or excessive fluctuations; on the other hand, it requires the contribution of failure risk to continue to decrease or at least no longer increase too rapidly, thereby incorporating the protection of energy storage health into the long-term goal.

[0149] A simplified example can be used to understand its iterative optimization process: if a certain round of threshold setting allows the system to still adhere to optimal compensation during midday disturbances, although the bus voltage is relatively smooth in the short term, the contribution of energy storage risk is significantly increased, and the strategy of that round receives a lower evaluation.

[0150] If another round of threshold setting causes the system to switch to a suboptimal strategy earlier, the voltage remains within the safety boundary, and the energy storage fatigue growth slows down, then this round of strategy is highly rated; after multiple rounds of iteration, the optimization module will gradually converge to a threshold range that better reflects the actual working conditions of the building.

[0151] To prevent instability caused by the iterative optimization process itself, this embodiment introduces robust constraints into the threshold update process; that is, any new threshold is only allowed to replace the old threshold if it does not exceed the voltage safety boundary in both the offline verification window and the online small-scale trial.

[0152] If the new threshold causes a deterioration in power supply response, an abnormal amplification of risk contribution, or a significant deviation of training samples from the current operating conditions, it immediately reverts to the previous stable version; this design enables the optimization module to have the engineering properties of continuous optimization but controlled constraints.

[0153] As a protective alternative, if key feedback data is missing for a certain period of time, such as incomplete sampling of actual voltage response or insufficient reliability of risk contribution estimation, the strategy optimization module will pause threshold updates and only continue the most recent stable threshold.

[0154] If the building's operating mode changes abruptly, such as during a fire drill, a full nighttime water replenishment, or a large-scale equipment maintenance, the data from that phase can be marked as special samples and not directly used to update the threshold, so as to avoid misfitting abnormal operating conditions as normal rules.

[0155] For example, in the main scenario, after the system experienced several days of high temperature and multiple harmonic disturbances, the strategy optimization module reviewed the actual bus response after each optimal, suboptimal, and protection switch and found that under a certain old setting, the system often delayed switching to the suboptimal strategy even when the energy storage had already heated up significantly.

[0156] The optimization module gradually moves the trigger range of the first danger threshold forward during subsequent operation, and fine-tunes the second danger threshold based on the nighttime recovery effect and the daytime risk increase. After several operating cycles, the system enters the energy storage-friendly conservative mode earlier without breaking the bus safety boundary, significantly reducing the probability of hitting the failure boundary when the water pump starts again.

[0157] The purpose of this step is to ensure that the threshold is no longer set based on static experience, but is continuously adjusted according to the actual operating environment of the building, the energy storage degradation status, and the bus response results, thereby achieving adaptive control that balances short-term power supply security and long-term system resilience.

[0158] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A distributed energy storage power supply system for use with building water pumps, characterized in that, include: Variable frequency water pump equipment, distributed energy storage equipment, mains power access interface, power conversion module and energy management system; The distributed energy storage device and the mains power access interface are respectively electrically connected to the power supply bus of the variable frequency water pump device through the power conversion module. The energy management system is equipped with: The data acquisition module is used to collect power quality data of the variable frequency water pump equipment, current water supply network pressure value, peak load records during historical startup phases, and operating status data of the distributed energy storage equipment. The status assessment module is used to filter the power quality data and the operating status data to obtain effective operating data, and to calculate the cumulative thermoelectric stress fatigue of the distributed energy storage device based on the effective operating data. The decision control module is used to generate power allocation instructions based on the accumulated thermoelectric stress fatigue and send them to the variable frequency water pump, the distributed energy storage device and the power conversion module to adjust the operating power status of the variable frequency water pump and the charging and discharging status and power output ratio of the distributed energy storage device.

2. The distributed energy storage power supply system for building water pumps as described in claim 1, characterized in that, The decision control module is preset with a first danger threshold for characterizing mild fatigue and a second danger threshold for characterizing severe fatigue, wherein the first danger threshold is less than the second danger threshold; the power allocation instruction includes: an optimal discharge strategy instruction generated when the accumulated thermoelectric stress fatigue is less than the first danger threshold, a suboptimal discharge strategy instruction generated when it is greater than or equal to the first danger threshold and less than the second danger threshold, and a system protection strategy instruction generated when it is greater than or equal to the second danger threshold.

3. The distributed energy storage power supply system for building water pumps as described in claim 1, characterized in that, The status assessment module stores a noise filtering model; the status assessment module is used to input the power quality data and the operating status data into the noise filtering model, filter out harmonic noise data, and output the effective operating data.

4. The distributed energy storage power supply system for building water pumps as described in claim 3, characterized in that, The operating status data includes charge / discharge switching frequency data and internal temperature data. The effective operating data includes the charge / discharge switching frequency data and the internal temperature data after filtering. The status evaluation module is also used to determine the electrical stress index based on the filtered charge / discharge switching frequency data and the preset electrical stress conversion coefficient in the effective operating data, to determine the thermal stress index based on the filtered internal temperature data and the preset thermal stress conversion coefficient in the effective operating data, and to determine the thermoelectric stress fatigue accumulation by combining the electrical stress index and the thermal stress index.

5. The distributed energy storage power supply system for building water pumps as described in claim 2, characterized in that, The condition assessment module is also used to obtain the impact demand of the variable frequency water pump equipment calculated based on the rated power of the variable frequency water pump equipment, the current water supply network pressure value, and the peak load record of the historical start-up phase; and input the accumulated thermoelectric stress fatigue amount and the impact demand into a preset failure boundary model to output the failure risk contribution.

6. The distributed energy storage power supply system for building water pumps as described in claim 2, characterized in that, The optimal discharge strategy command includes a high-frequency power tracking command and a voltage compensation command; the distributed energy storage device is used to perform charge-discharge switching at a preset charge-discharge switching frequency after receiving the high-frequency power tracking command and the voltage compensation command, so as to maintain the voltage stability of the power supply bus.

7. The distributed energy storage power supply system for building water pumps as described in claim 2, characterized in that, The suboptimal discharge strategy instructions include an acceleration slope limit instruction and a mains power access increase instruction; the variable frequency pump device is used to reduce the variable frequency acceleration slope of the variable frequency pump device during the startup phase after receiving the acceleration slope limit instruction; the distributed energy storage device is used to reduce the discharge power ratio of the distributed energy storage device in the power supply bus after receiving the mains power access increase instruction, and increase the mains power supply ratio from the mains power access interface through the power conversion module.

8. The distributed energy storage power supply system for building water pumps as described in claim 2, characterized in that, The system protection strategy commands include energy storage lockout commands and full mains power supply commands.

9. The distributed energy storage power supply system for building water pumps as described in claim 5, characterized in that, The energy management system further includes a strategy optimization module, which is used to collect the actual voltage response data of the variable frequency water pump after executing the power allocation command; construct a state vector based on the actual voltage deviation rate, the failure risk contribution, and the thermoelectric stress fatigue accumulation extracted from the actual voltage response data; construct an action space with the adjustment amount of the first danger threshold and the second danger threshold; and dynamically update the first danger threshold and the second danger threshold using a deep reinforcement learning algorithm with the objective function of maintaining the actual voltage deviation rate within a preset voltage safety range and reducing the failure risk contribution.

Images