A standby power supply system for building fire emergency equipment

Abstract

The present application relates to the technical field of fire emergency power supply and DC micro-grid control, in particular to a standby power supply system for building fire emergency equipment, comprising a plurality of distributed power supply nodes, a DC power supply bus, a topology initialization module, a state data sensing module, a virtual impedance solving engine and a bottom circuit cooperative control module; upon power-up, the basic droop control coefficient is allocated according to the access equipment nameplate power and static importance level, the local environmental state data time rate of change, energy storage state of charge parameters and local output current are collected, the environmental state data time rate of change is mapped to the environmental dangerous potential gradient and the virtual impedance adjustment parameter is calculated, the pulse width modulation control signal is generated to adjust the power conversion unit output reference voltage, and the response, retreat, maintenance or block isolation control is executed based on the bus voltage fluctuation; the present application can complete the standby power redistribution and form the extreme fire area physical isolation capability under the condition of weak communication or no communication.

Description

Technical Field

[0001] This invention relates to the field of fire emergency power supply and DC microgrid control technology, specifically a backup power supply system for building fire emergency equipment. Background Technology

[0002] In the field of emergency power supply for fire protection in buildings, backup power systems play an important role in providing continuous power to key equipment such as smoke exhaust fans, emergency lighting, fire broadcasting and roller shutter controllers under fire and mains power outage conditions. With the increase in smoke control zones, the complexity of load types and the speed of fire evolution in super high-rise buildings, emergency power supply control mechanisms oriented towards multi-node collaboration have gradually become an important technical direction for improving escape protection capabilities.

[0003] Currently, fire emergency backup power systems generally adopt centralized control or relatively independent power supply configuration by floor or area. If it is necessary to realize the power distribution between different zones during the dynamic development of a fire, it is usually necessary to rely on the central controller, building communication links and preset load switching strategies to complete the power distribution and branch control of each backup power node.

[0004] However, when providing emergency power supply for fire and communication disruption scenarios using the above methods, it relies on upper-level scheduling to collect, judge, and issue commands for the status of each node. This makes it difficult to complete millisecond-level power redistribution in a timely manner based on local environmental changes and energy storage status when communication links are interrupted, local sensors fail, or the fire spreads rapidly. This results in problems such as delayed coordinated response and insufficient power supply tilting capacity in critical areas. At the same time, this approach usually cannot utilize the physical coupling relationship of the DC bus to achieve self-organized power rebalancing between nodes. Furthermore, in extreme cases such as local short circuits and continuous deterioration of strong combustion zones, it lacks the underlying closed-loop control capability that balances dynamic support and rapid isolation, resulting in low system resilience and a high risk of fault propagation. Summary of the Invention

[0005] The purpose of this invention is to provide a backup power supply system for building fire emergency equipment, addressing the following technical problems: Existing building fire emergency backup power supply systems lack the ability to perform millisecond-level power redistribution based on local environmental changes and energy storage status when facing dynamic fire development and communication disruption scenarios. Furthermore, they have significant shortcomings in the underlying closed-loop control that balances dynamic support and rapid isolation. There is an urgent need for a backup power supply system for building fire emergency equipment that can achieve self-organized power rebalancing without communication dependence based on the physical coupling relationship of the DC bus, while also considering power tilting towards critical escape areas and physical isolation of extreme fire zones. The purpose of this invention can be achieved through the following technical solutions:

[0006] A backup power supply system for building fire emergency equipment includes multiple distributed power supply nodes connected to access devices and containing local sensors, local energy storage units and local power conversion units, a DC power supply bus interconnecting the nodes, a topology initialization module, a status data sensing module, a virtual impedance calculation engine and a low-level circuit collaborative control module.

[0007] When the topology initialization module is powered on, the base droop control coefficient is allocated according to the nameplate power and static importance level of the access device;

[0008] The status data sensing module obtains the time-series change rate of local environmental status data based on the environmental status data collected by the local sensors, obtains the energy storage state of charge parameters based on the local energy storage unit, and obtains the local output current of the local power conversion unit.

[0009] The virtual impedance calculation engine maps the time-series rate of change to the environmental hazard potential energy gradient based on the droop control reference curve with impedance bias characteristics, and calculates the virtual impedance adjustment parameters by combining the energy storage state of charge parameters with the basic droop control coefficient as the starting point.

[0010] The underlying circuit collaborative control module generates a pulse width modulation control signal to adjust the output reference voltage of the local power conversion unit, and presets a preset response threshold for the environmental hazard potential energy gradient, a preset limit blocking threshold greater than the preset response threshold, a preset bus voltage reference value, and a preset blocking extreme value for performing physical isolation.

[0011] Optionally, the state data perception module includes an environmental drastic change feature extraction unit to obtain the temperature rise rate and smoke concentration change rate output by the local sensor;

[0012] The environmental change feature extraction unit performs a weighted summation feature fusion operation on the temperature rise rate and the smoke concentration change rate to generate the local environmental state data time-series change rate.

[0013] The status data sensing module relies solely on data collected by the local sensors and the local energy storage unit.

[0014] Optionally, the local power conversion unit is a bidirectional DC-DC converter;

[0015] Each bidirectional DC-DC converter is interconnected through the DC power supply bus to construct a multi-source, multi-load DC microgrid topology;

[0016] The pulse width modulation control signal is directly input into the voltage and current dual-loop control circuit of the bidirectional DC-DC converter.

[0017] Optionally, when calculating the virtual impedance adjustment parameters, the virtual impedance calculation engine performs an equivalent internal resistance bias calculation based on the droop control equation.

[0018] The equivalent internal resistance bias calculation is based on a preset rated reference voltage, a preset base impedance, and a preset proportionality coefficient with impedance dimensions.

[0019] The logic of the equivalent internal resistance bias operation is as follows: make the output reference voltage equal to the preset rated reference voltage minus the product of the dynamic compensation impedance and the local output current.

[0020] The dynamic compensation impedance is obtained by subtracting the product of the environmental hazard potential energy gradient and the preset proportional coefficient from the preset basic impedance.

[0021] Optionally, the underlying circuit collaborative control module includes a residual fitting analysis unit;

[0022] The residual fitting analysis unit continuously monitors the voltage of the DC power supply bus when the environmental hazard potential energy gradient is lower than the preset response threshold.

[0023] In response to the environmental hazard potential energy gradient being lower than a preset response threshold and the DC power supply bus voltage being continuously monitored to be lower than the preset bus voltage reference value, the residual fitting analysis unit determines that there is a high-risk node in the bus drawing current. Based on Kirchhoff's current law, the residual fitting analysis unit triggers the local power conversion unit to reduce the output current along the droop control reference curve. The specific logic is as follows: calculate the bus voltage residual by subtracting the current actual DC power supply bus voltage from the preset bus voltage reference value, multiply the bus voltage residual by a preset yield gain constant to obtain the yield current correction amount, and superimpose the yield current correction amount as a negative current bias into the control loop of the bidirectional DC-DC converter.

[0024] Optionally, the underlying circuit collaborative control module includes a flexible electrical isolation unit;

[0025] The flexible electrical isolation unit has a preset short-circuit threshold including a current rise rate threshold and a current amplitude threshold.

[0026] In response to the environmental hazard potential energy gradient being equal to or higher than the preset limit blocking threshold, or the detection that the rise rate or amplitude of the local output current reaches or exceeds the corresponding current rise rate threshold or current amplitude threshold in the preset short-circuit threshold, the flexible electrical isolation unit sets the virtual impedance adjustment parameter to the preset blocking extreme value, or controls the duty cycle of the pulse width modulation control signal to zero, thereby limiting the current output to the DC power supply bus.

[0027] Optionally, the state data sensing module, the virtual impedance calculation engine, and the underlying circuit collaborative control module are integrated and deployed in a field-programmable gate array or a digital signal processing device.

[0028] The field-programmable gate arrays or digital signal processing devices are distributed and installed in the node controllers of each distributed power supply node.

[0029] Optionally, the virtual impedance adjustment parameter is equal to the base droop control coefficient minus the product obtained by multiplying the environmental hazard potential energy gradient, the state of charge support, and the preset droop adjustment ratio, wherein the state of charge support is a piecewise function mapping of the energy storage state of charge parameter.

[0030] Optionally, the underlying circuit collaborative control module performs the following collaborative control based on the environmental hazard potential energy gradient:

[0031] When the environmental hazard potential energy gradient is higher than the preset response threshold but lower than the preset limit blocking threshold, the virtual impedance adjustment parameter is reduced to increase the current drawn from the DC power supply bus.

[0032] When the environmental hazard potential energy gradient is lower than the preset response threshold, if the DC power supply bus voltage is lower than the preset bus voltage reference value, the output power is reduced along the droop control reference curve; otherwise, the basic droop control coefficient is maintained.

[0033] When the environmental hazard potential energy gradient equals the preset response threshold, the output reference voltage is maintained.

[0034] When the environmental hazard potential energy gradient is equal to or higher than the preset limit blocking threshold, the virtual impedance adjustment parameter is set to the preset blocking extreme value, or the duty cycle of the pulse width modulation control signal is controlled to be zero, and physical isolation is performed.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. To address the issue of existing systems relying on central controller commands, which leads to delayed responses during fire spread or communication disruptions, this system utilizes the DC power supply bus interconnecting each node to construct a low-level physical coupling. The status data sensing module acquires data solely through local sensors and energy storage units, and a virtual impedance calculation engine, combined with a droop control baseline curve, maps the time-series rate of change to an environmental hazard potential energy gradient. When the hazard potential energy gradient in a high-risk area exceeds a preset response threshold, the low-level circuit collaborative control module automatically reduces the virtual impedance adjustment parameter to increase the current drawn from the DC power supply bus, directly generating a pulse width modulation control signal to adjust the bidirectional DC-DC converter. This mechanism enables the system to complete the millisecond-level transfer of backup power to critical escape areas even under weak or no communication conditions, improving the redundant power supply capability of the power supply system in the event of local node failure.

[0037] 2. To address the issue of susceptibility to local environmental interference leading to misjudgments or slow responses when relying on single sensor data, this system introduces an environmental change feature extraction unit in the state data perception module. This unit does not depend on absolute static environmental values ​​but directly acquires the temperature rise rate and smoke concentration change rate output by local sensors, and performs a weighted summation feature fusion operation on the two to generate a stable and reliable time-series change rate of local environmental state data. This fusion extraction mechanism based on multi-source change rates can identify the fire spread front earlier and more accurately, providing accurate data support for subsequent virtual impedance adjustment and reducing the probability of power supply system mis-switching or missed switching. Attached Figure Description

[0038] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0039] Figure 1 This is a structural diagram of a backup power supply system for a building fire emergency equipment according to the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0041] like Figure 1 As shown, a backup power supply system for a building's fire emergency equipment includes multiple distributed power supply nodes connected to access devices and containing local sensors, local energy storage units, and local power conversion units; a DC power supply bus interconnecting the nodes; a topology initialization module; a status data sensing module; a virtual impedance calculation engine; and a low-level circuit collaborative control module.

[0042] When the topology initialization module is powered on, the foundation droop control coefficient is assigned according to the nameplate power and static importance level of the connected device;

[0043] The status data sensing module obtains the time-series change rate of local environmental status data based on the environmental status data collected by local sensors, obtains the energy storage charge state parameters based on the local energy storage unit, and obtains the local output current of the local power conversion unit.

[0044] The virtual impedance calculation engine maps the time-series rate of change to the environmental hazard potential energy gradient based on the droop control reference curve with impedance bias characteristics. It then calculates the virtual impedance adjustment parameter by combining the energy storage state of charge parameter with the basic droop control coefficient. The virtual impedance adjustment parameter is equal to the basic droop control coefficient minus the product obtained by multiplying the environmental hazard potential energy gradient, the state of charge support, and the preset droop adjustment ratio. The state of charge support is a piecewise function mapping of the energy storage state of charge parameter.

[0045] The underlying circuit collaborative control module generates pulse width modulation control signals to adjust the output reference voltage of the local power conversion unit, and presets a preset response threshold for the environmental hazardous potential energy gradient, a preset limit blocking threshold greater than the preset response threshold, a preset bus voltage reference value, and a preset blocking extreme value for performing physical isolation.

[0046] When the environmental hazard potential energy gradient is higher than the preset response threshold but lower than the preset limit blocking threshold, the virtual impedance adjustment parameter is reduced to increase the current drawn from the DC power supply bus.

[0047] When the gradient is lower than the preset response threshold, if the DC power supply bus voltage is lower than the preset bus voltage reference value, the output power is reduced along the droop control reference curve; otherwise, the basic droop control coefficient is maintained.

[0048] When the gradient equals the preset response threshold, the output reference voltage is maintained;

[0049] When the gradient is equal to or higher than the preset blocking threshold, the virtual impedance adjustment parameter is set to the preset blocking extreme value, or the duty cycle of the pulse width modulation control signal is controlled to be zero, and physical isolation is performed.

[0050] This embodiment provides a distributed DC collaborative power supply mechanism for fire emergency power supply in super high-rise office buildings. Specifically, taking a 36-story building above ground and three-story building below ground as the main scenario, after a fire occurs in the building at night and the mains power is interrupted, the distributed power supply nodes set up in each smoke control zone no longer wait for the central dispatcher to issue instructions one by one. Instead, they rely on the physical coupling relationship between local sensors, local energy storage units, power conversion units and DC power supply buses to directly redistribute backup power at the circuit layer.

[0051] In this scenario, each distributed power supply node can serve one or more fire emergency devices. For example, a node may be connected to a smoke exhaust fan, an emergency lighting device, a fire shutter controller, or a fire broadcast system. Each node is equipped with local temperature and smoke sensors. The local energy storage unit can be a lithium battery pack or a combination of supercapacitors. The local power conversion unit can supply power to the DC power supply bus or draw power from the bus. Multiple nodes are interconnected through the same DC power supply bus, so that the power required by a certain area can be indirectly supported by nodes in adjacent areas or on other floors. During the power-on initialization phase, the topology initialization module first reads the nameplate power and static importance level of the connected devices, and assigns a base droop control coefficient to each node according to the preset parameter allocation table.

[0052] Specific parameter allocation examples are as follows: If the equipment is a fire pump with a nameplate power of 30kW and a static importance level of 5, the corresponding foundation sag control coefficient is 0.15; if the equipment is a smoke exhaust fan with a nameplate power of 10kW and a static importance level of 4, the corresponding coefficient is 0.20; if the equipment is an emergency lighting device with a nameplate power of 2kW and a static importance level of 3, the corresponding coefficient is 0.28. The foundation sag control coefficient can be understood as the initial power output state of each node under stable conditions. The smaller the value, the more likely it is that the node has the characteristic of maintaining an output power higher than the preset benchmark when the bus voltage changes.

[0053] Through this initialization action, even if subsequent communication is interrupted, each node still has an independent starting control baseline; the status data sensing module continuously collects three basic input quantities: the time-series change rate of local environmental status data, the energy storage charge state parameter, and the local output current; for ease of explanation, it can be assumed that within a certain sampling period, the temperature change rate of the smoke prevention zone on the east side of the 18th floor exceeds the first set threshold, and the smoke concentration also reaches the second set increase, which, after processing by the local sensor, forms an environmental status data time-series change rate of 0.72; at the same time, the energy storage charge state of this node is 0.81, and the local output current is 18A;

[0054] Meanwhile, the environmental parameter fluctuations in the smoke control zone on the west side of the 16th floor are within the set safety thresholds. The time-series change rate of the environmental status data is only 0.18, the energy storage state of charge is 0.77, and the local output current is 12A. These quantities are all collected locally by each node and do not rely on the building bus to return global data. The virtual impedance calculation engine calls the preset droop control reference curve to map the aforementioned time-series change rate of environmental status data to the environmental hazard potential energy gradient. This mapping can be done by looking up a table or by piecewise linear method. Taking the data mapping process of this time-series change rate as an example, the droop control reference curve is set as follows: a time-series change rate below 0.2 is mapped to a hazard potential energy gradient of 0.15, 0.2 to 0.5 is mapped to 0.4, and above 0.5 is mapped to 0.8.

[0055] Therefore, 0.72 on the east side of the 18th floor is mapped to a dangerous potential energy gradient of 0.8, and 0.18 on the west side of the 16th floor is mapped to 0.15. The engine combines the energy storage state of charge to correct the virtual impedance adjustment parameter. To clarify the specific calculation process of the comprehensive correction, the virtual impedance calculation engine executes the calculation logic of the embodiment: that is, the virtual impedance adjustment parameter is equal to the basic droop control coefficient minus the product obtained by multiplying the environmental dangerous potential energy gradient, the state of charge support, and the preset droop adjustment ratio. Among them, the state of charge support is a piecewise function mapping of the energy storage state of charge parameter: when the state of charge is higher than 0.5, the support is 1.0; when it is lower than 0.5 but higher than 0.2, the support decreases proportionally with the state of charge; when it is lower than 0.2, the support is forced to zero. Based on this clear data flow rule, the system can be rigidly constrained by the actual discharge capacity of the local energy storage when performing virtual impedance reduction.

[0056] For example, under the same level of danger, if the local energy storage state of charge is lower than the set normal operating limit, the virtual impedance will not be reduced without constraint due to the weakening of support, but a portion of the energy will be reserved for subsequent stages. The base droop control coefficient is used as the starting point for iteration. For example, the original starting point on the east side of the 18th floor is 0.20. After the above rules are combined with the dangerous potential energy and the state of charge for comprehensive deduction and correction, a new virtual impedance adjustment parameter of 0.12 is obtained. The west side of the 16th floor is maintained at around 0.28 or slightly adjusted to 0.30. After receiving the adjustment parameter, the bottom circuit collaborative control module generates a pulse width modulation control signal, thereby adjusting the output reference voltage of the power conversion unit.

[0057] Specifically: when the virtual impedance adjustment parameter of a node decreases, the node will act as a power source with a smaller equivalent internal resistance or a larger equivalent load in the bus, causing more current to flow to the fire-fighting equipment branch coupled to it; when the adjustment parameter of a node increases, its output power will decrease along the droop control reference curve; since all nodes are connected to the same DC power supply bus, explicit communication negotiation is not required, and global rebalancing can be achieved based on Kirchhoff's current law and bus voltage fluctuations.

[0058] Furthermore, the underlying circuit collaborative control module executes a set of physical closed-loop logic and presets the ultimate blocking threshold, response threshold, and bus voltage reference value. The specific collaborative control logic is as shown in the embodiment. According to the different intervals of the environmental hazard potential energy gradient mapped by local acquisition, corresponding power adjustment or isolation actions are performed: To facilitate the display of the status of each branch, the response threshold can be set to 0.50, the ultimate blocking threshold to 0.90, and the bus voltage reference value to 750V. If the hazard potential energy gradient of a node is higher than 0.50 and lower than 0.90, for example, the value is 0.80 on the east side of the 18th floor, then the node reduces the virtual impedance adjustment parameter to make it draw more current from the DC power supply bus and prioritize maintain the smoke exhaust fan and evacuation indicator in this area. If the hazard potential energy gradient of a node is lower than 0.50 and the bus voltage is detected to be lower than 750V, for example, the hazard potential energy gradient is 0.15 on the west side of the 16th floor and the bus voltage drops to 742V, then the node actively reduces the output power to the bus or reduces the power supply level of local equipment along the droop control reference curve, thereby transferring power to more dangerous areas.

[0059] If the dangerous potential energy gradient is below 0.50 and the bus voltage is not below 750V, the node maintains the basic droop control coefficient of the initialization phase without additional intervention. If the dangerous potential energy gradient is exactly equal to 0.50, the node maintains the current output reference voltage to prevent control jitter. If the dangerous potential energy gradient reaches or exceeds 0.90, for example, if a fully ignited strong combustion zone reaches 0.95, the virtual impedance adjustment parameter can be directly set to the preset blocking extreme value, or the duty cycle of the pulse width modulation control signal can be set to zero, so that the branch performs a physical circuit isolation operation to prevent the fault in this area from causing the bus voltage of the entire building to drop. This operation is achieved through the flexible electrical isolation unit in the underlying circuit collaborative control module, thereby completing the real-time physical blocking of the extreme danger zone.

[0060] As a backup protection mechanism, when instantaneous sensor jitter causes environmental state data to fluctuate between high and low values ​​within two sampling periods, a short-term holding window can be set for the dangerous potential energy gradient. For example, a significant impedance reduction action is only performed if the gradient is higher than the response threshold for two consecutive sampling periods. If the energy storage state of charge is lower than the minimum threshold, such as below 0.10, the maximum current draw can be limited even if the environmental dangerous potential energy gradient is high, preventing individual nodes from failing due to deep discharge. If the bus voltage is abnormally low and the local output current has reached the hardware safety limit, current limiting is prioritized instead of continuing to increase power, thereby protecting the power devices.

[0061] After a fire broke out in the office area on the east side of the 18th floor of the building, the smoke exhaust fan node of the smoke control zone on the east side detected a significant increase in temperature rise and smoke changes, and its dangerous potential energy gradient quickly exceeded the response threshold. At the same time, the adjacent zones on the 17th and 16th floors had not yet caught fire directly, but the local voltage of the busbar dropped due to the increased current draw on the east side of the 18th floor. Therefore, the node on the east side of the 18th floor automatically reduced its virtual impedance and obtained more current, while the nodes on the 17th and 16th floors, due to their lower local danger level and the detection of the busbar voltage drop, automatically reduced their power along the droop curve. The entire process did not require the central controller to make explicit decisions on the start and stop of the nodes on each floor, but the underlying circuit completed the power routing in milliseconds.

[0062] The purpose of this step is to directly convert the dynamic changes in the fire environment into circuit control quantities that can drive the power conversion unit, so that the backup power system can still rely on the physical closed loop of the DC power supply bus to complete the power tilt towards the key escape area under weak or even no communication conditions, and form physical isolation capability in extreme fire zones.

[0063] In this embodiment, the state data perception module includes an environmental drastic change feature extraction unit to acquire the temperature rise rate and smoke concentration change rate output by the local sensor.

[0064] The environmental change feature extraction unit performs a weighted summation feature fusion operation on the rate of temperature rise and the rate of change of smoke concentration to generate the time-series change rate of local environmental state data.

[0065] The status data sensing module relies solely on data collected by local sensors and local energy storage units.

[0066] This embodiment provides a mechanism for extracting features of dramatic changes in the local environment. Specifically, in the continuous evolution process after the fire started in the smoke control zone on the east side of the 18th floor of the aforementioned building, it is easy to make a misjudgment if only a single sensor value is used for judgment. For example, if only the temperature value is looked at, the node near the hot air duct may be mistaken for the fire area. If only the smoke value is looked at, it may be affected by cleaning agent aerosols or construction dust.

[0067] Therefore, this embodiment introduces an environmental change feature extraction unit to fuse the temperature rise rate and the smoke concentration change rate to form a more stable local environmental state data time-series change rate. In specific implementation, the local sensor does not directly use absolute temperature and absolute smoke concentration, but processes the change in adjacent sampling times. Assuming the sampling interval is 1 second, if the temperature of a certain node is 42℃ at time t and 46℃ at time t+1, the temperature rise rate can be simplified to 4℃ / second. Similarly, if the smoke concentration rises from 0.30 to 0.45, the smoke concentration change rate is 0.15 / second.

[0068] The environmental change feature extraction unit normalizes these two rates of change separately. For example, the temperature rise rate is mapped to the interval between 0 and 1 to obtain 0.70, and the smoke concentration change rate is mapped to 0.60. To ensure the consistency and uniqueness of the feature fusion operation in the underlying hardware, the mathematical formula for its weighted summation is defined as: the time-series change rate of the generated local environmental state data. Equal to the preset temperature weight With normalized rate of temperature rise The product of these, plus the preset smoke weight. Change rate of normalized smoke concentration The product of, i.e. And the preset temperature weight With smoke weight Must meet Physical constraints;

[0069] If the formula is executed, Set to 0.55. If we set it to 0.45, then It can be obtained as This result will serve as the direct input for subsequent hazardous potential gradient mapping. This formalized definition eliminates the divergent ambiguity in the feature extraction calculation weights. To demonstrate a simplified calculation example, two more nodes are set up for comparison. Node A is located outside the smoke exhaust fan room on the east side of the 18th floor, with a normalized temperature rise rate of 0.80 and a normalized smoke change rate of 0.75, resulting in a fused value of 0.7775. Node B is located in the west corridor on the 16th floor, with a normalized temperature rise rate of 0.20 and a normalized smoke change rate of 0.10, resulting in a fused value of 0.155.

[0070] Therefore, node A reaches the response threshold earlier than node B. Because it uses a rate of change rather than a purely static quantity, the system can identify the fire spread front more quickly, without waiting for the absolute temperature to accumulate to the preset trigger threshold. Furthermore, this embodiment emphasizes that the status data sensing module relies solely on data collected by local sensors and local energy storage units. This means that node A does not need to read data from the 20th or 15th floors or the central control room, does not need to rely on the building network to broadcast fire alarm levels, and does not require continuous communication with the fire alarm control panel. Judgment of drastic environmental changes is completed locally immediately, and the energy storage charge status is also provided in real time by the local battery management unit. This avoids sensing failure caused by communication link damage after a fire.

[0071] As a backup protection mechanism, if the temperature sensor fails while the smoke sensor is normal, the system can temporarily use the smoke concentration change rate and the temperature estimate from the most recent few cycles for degradation fusion; if the smoke sensor fails, the process is reversed; if both sensors fail simultaneously, the time-series change rate of the local environmental state data can be set to a conservative value, for example, maintaining the result from the previous cycle and shortening the retest cycle, until the sensors recover or the node degradation is triggered by the upper-level security policy; if the weighted summation result exceeds the preset upper limit, it is pruned to the upper limit of 1; if both inputs are close to zero, it indicates that the local environment is stable, and subsequent control maintains the basic state.

[0072] The fire on the east side of the 18th floor did not initially start as a full-scale fire, but rather a localized smoldering fire in a document room. The absolute temperature increase was not significant, but the smoke concentration rose noticeably first. A few seconds later, the flames spread to adjacent areas, and the rate of temperature increase increased simultaneously. After the environmental change feature extraction unit merged these two quantities, it could identify the area as a rapidly rising area of ​​dangerous potential energy earlier than by looking at only one of them. As a result, the local power supply node lowered the virtual impedance in advance to provide priority power supply for smoke extraction and evacuation equipment. The purpose of this step is to extract the intensity of environmental change using a simple fusion method that can be calculated locally in real time, reduce false triggering or missed triggering caused by a single sensor, and ensure that the entire power supply control link can still operate independently in a closed loop without global communication support.

[0073] In this embodiment, the local power conversion unit is a bidirectional DC-DC converter;

[0074] Each bidirectional DC-DC converter is interconnected through a DC power supply bus to form a multi-source, multi-load DC microgrid topology;

[0075] The pulse width modulation control signal is directly input into the voltage and current dual-loop control circuit of the bidirectional DC-DC converter.

[0076] This embodiment provides a mechanism for constructing a multi-source, multi-load DC microgrid using a bidirectional DC-DC converter. Specifically, in the aforementioned scenario, if each power supply node uses only a unidirectional converter, the node can only output to local equipment and cannot flexibly transfer energy between floors. When the danger level of a certain smoke control zone suddenly increases, the excess power in other zones cannot provide timely support. Therefore, this embodiment uses a bidirectional DC-DC converter as a power conversion unit, so that each node can act as both a source node that supplies energy to the bus and a load node that draws energy from the bus.

[0077] Specifically, the east node of the 18th floor, the middle node of the 17th floor, and the west node of the 16th floor can be connected to the same 750V DC power supply bus. Each node is equipped with a local energy storage unit, such as a battery pack with a rated voltage of 96V, and is connected to the bus through a bidirectional DC-DC converter. Under stable conditions, if the local equipment load is lower than the set low load threshold and the energy storage state of charge is higher than the set sufficient judgment benchmark, the node can send a portion of the power to the bus after boosting the voltage. When the local equipment load suddenly increases or the local environmental hazard level increases rapidly, additional current can be drawn from the bus through the same converter to ensure that equipment such as smoke exhaust fans and emergency lighting are not degraded due to insufficient local energy storage capacity in a short period of time.

[0078] To better illustrate the relationship between the pulse width modulation control signal and the converter, we can assume that a bidirectional DC-DC converter operates at a frequency of 50kHz. After the underlying circuit and the control module calculate the target output reference voltage, they directly provide the duty cycle. If the node needs to enhance the support obtained from the bus side, the duty cycle is adjusted from 0.42 to 0.50, shifting the equivalent output characteristics of the converter towards higher support. If the node is in the safe zone and the bus voltage is detected to be lower than the reference value, the duty cycle is adjusted from 0.42 to 0.36, reducing its output power.

[0079] Direct input here means that instead of translating commands from the host computer into multi-level switching actions, the pulse width modulation control signal generated at the bottom layer is directly sent to the power device driver stage. The advantages of the multi-source, multi-load topology are also reflected in the natural redundancy that can be formed between branches. For example, when a fire breaks out on the east side of the 18th floor, the local energy storage on the 18th floor will initially provide power. If the local energy storage discharges quickly in a short time, the nodes in the middle of the 17th floor and the west side of the 16th floor can supplement power through the bus. Conversely, when the fire on the 18th floor is brought under control, its dangerous potential energy gradient decreases, the nodes on the 18th floor return to their drooping state, and the energy flow of the bus can gradually return to equilibrium. This topology is not a tree structure of a single main power source plus multiple branches, but a mesh power supply relationship with multiple interconnected points.

[0080] As a backup protection mechanism, if a bidirectional DC-DC converter temporarily shuts down due to device overheating, that node can switch to local energy storage to maintain only the minimum load, while the other nodes on the bus continue to operate. If the interface between a node and the bus is disconnected, it will not affect the bidirectional energy exchange of other nodes. If multiple nodes request high power support at the same time, the shunting result in the bus is determined by the virtual impedance adjustment parameters of each node, rather than relying on the central processor to queue them one by one.

[0081] After the smoke exhaust fan on the east side of the 18th floor is put into full speed, its local battery pack can maintain high-power operation for about 12 minutes on its own. However, since the nodes in the middle of the 17th floor and the west side of the 16th floor are connected to the bus through bidirectional DC-DC converters, and there is no fire in this area, these two nodes can transfer the excess local energy storage to the corresponding branch on the east side of the 18th floor. The bottom controller only needs to modify the pulse width modulation duty cycle, without the need for frequent switching of mechanical relays, to complete the smooth power migration. The purpose of this mechanism is to integrate the originally fragmented floor backup power supplies into a mutually supportive energy network through bidirectional DC-DC converters and DC microgrid topology, so that the pulse width modulation control signal can directly affect the power flow direction and power magnitude, thereby improving the continuity and dispatchability of power supply during the dynamic stage of a fire.

[0082] In this embodiment, the virtual impedance calculation engine performs an equivalent internal resistance bias calculation based on the droop control equation when calculating the virtual impedance adjustment parameters.

[0083] The equivalent internal resistance bias calculation is based on the preset rated reference voltage, preset base impedance, and preset proportional coefficient with impedance dimensions.

[0084] The logic of the equivalent internal resistance bias operation is: make the output reference voltage equal to the preset rated reference voltage minus the product of the dynamic compensation impedance and the local output current.

[0085] The dynamic compensation impedance is obtained by subtracting the product of the environmental hazard potential energy gradient and the preset proportional coefficient from the preset basic impedance.

[0086] This embodiment provides an equivalent internal resistance bias calculation mechanism based on the droop control equation. Specifically, in the aforementioned system, if it is only generally described that more power is allocated to the dangerous area, the underlying control cannot be directly implemented to the converter reference voltage. This embodiment converts the environmental danger potential energy gradient into a dynamic compensation impedance, and then uses this to generate the output reference voltage, so that the control chain forms a calculable closed loop from environmental perception to circuit execution.

[0087] In specific calculations, the preset rated reference voltage can be set to 760V, the preset base impedance can be set to 1.0Ω, and the preset proportional coefficient can be set to 0.6Ω. For a certain node, if the local output current is known to be 20A and the environmental hazardous potential energy gradient is 0.8 within one sampling period, then the dynamic compensation impedance is equal to 1.0 - 0.6 × 0.8 = 0.52Ω; correspondingly, the output reference voltage is equal to 760 - 0.52 × 20 = 749.6V. If the hazardous potential energy gradient of another node in the safe zone is only 0.2 and the local output current is 20A, then the dynamic compensation impedance is 1.0 - 0.6 × 0.2 = 0.88Ω, and the output reference voltage is equal to 760 - 0.88 × 20 = 742.4V.

[0088] It is evident that high-risk nodes, due to their smaller dynamic compensation impedance, tend to have an output reference voltage approaching the preset rated reference voltage under the same current conditions, thus making it easier to maintain or attract power support. To further illustrate the circuit meaning of equivalent internal resistance bias, a simplified calculation example can be provided using three nodes: Node N1 is located on the east side of the 18th floor with a dangerous potential energy gradient of 0.8; Node N2 is located in the middle of the 17th floor with a dangerous potential energy gradient of 0.4; and Node N3 is located on the west side of the 16th floor with a dangerous potential energy gradient of 0.1.

[0089] If the output current of all three is temporarily set to 15A, then the dynamic compensation impedance of N1 is 0.52Ω, and the output reference voltage is 752.2V; the dynamic compensation impedance of N2 is 0.76Ω, and the output reference voltage is 748.6V; the dynamic compensation impedance of N3 is 0.94Ω, and the output reference voltage is 745.9V. In this way, under the coupling effect of the common bus, the branch corresponding to N1 will more actively maintain power supply, while N3 will more easily yield power. This process is not based on artificially setting which layer must be disconnected, but on continuously achieving power stratification at the bottom layer through equivalent internal resistance bias. If energy storage state of charge is considered, a limiting effect can be added on the basis of the above.

[0090] For example, with the same dangerous potential gradient of 0.8, if the local energy storage state of charge is only 0.12, the dynamic compensation impedance should be reduced, but it can be limited to a minimum of 0.60Ω to prevent over-discharge of the battery; conversely, if the state of charge is 0.90, the dynamic compensation impedance is allowed to drop to 0.45Ω; in this way, the environmental hazard level and the remaining energy storage capacity are simultaneously incorporated into the same voltage reference generation process.

[0091] As a backup protection mechanism, if the dynamic compensation impedance calculated by the above formula is negative, it means that the combination of the dangerous potential energy gradient and the proportional coefficient exceeds the preset upper limit of the product. At this time, the dynamic compensation impedance should be cut to the preset minimum positive value, such as 0.10Ω, to prevent control instability. If the local output current is close to zero, the output reference voltage is close to the rated reference voltage. At this time, the node is in standby or light load state. If transient spike interference occurs in the sampling current, the short window average current can be used to replace the instantaneous value in the calculation to reduce the jitter of the output reference voltage.

[0092] After the fire on the east side of the 18th floor changed from smoldering to open flame, the danger potential energy gradient increased from 0.55 to 0.82. The local calculation engine immediately mapped this change to a decrease in dynamic compensation impedance from 0.67Ω to 0.51Ω. As a result, the output reference voltage of the smoke exhaust fan node increased, and the converter tended to maintain its high power operation. Meanwhile, the environmental danger potential energy of the evacuation corridor on the west side of the 16th floor remained within the preset safety range, with a danger potential energy gradient of only 0.12. Its dynamic compensation impedance remained at a high level, causing the power supply capacity of non-critical equipment in this area to naturally shrink.

[0093] To further reveal the underlying closed-loop logic that the tendency of a high output reference voltage to translate into increased current draw needs to be clarified, the response mechanism of droop control under bus voltage fluctuations needs to be defined. In a multi-source, multi-load parallel DC microgrid, the converter uses inner-loop current and outer-loop voltage control to make its actual output voltage approach the output reference voltage. When the high-risk node N1 obtains a higher output reference voltage of, for example, 749.6V due to the dynamic compensation impedance reduction to, for example, 0.52Ω, if the actual DC power supply bus voltage is pulled down to, for example, 745V due to the power shortage of the entire system, then the reference voltage of node N1 at 749.6V is higher than the actual bus voltage of 745V.

[0094] This positive voltage difference, after being processed by the underlying proportional-integral regulator, drives the bidirectional DC-DC converter to increase its duty cycle, thereby increasing the current drawn from the bus at a preset maximum rate of change to compensate for the voltage difference and prioritize maintaining the power supply to the core fire-fighting equipment at this node. Conversely, due to its large dynamic compensation impedance (e.g., 0.94Ω), the 745.9V output reference voltage of the safe zone node N3 drops rapidly to near the actual bus voltage, thus automatically reducing its output proportion in the bus power distribution. This bias operation not only avoids non-transparent direct power command distribution but also cleverly utilizes the following characteristics of the control loop to map the dangerous potential energy gradient of environmental parameters into the slope bias of the droop equation, thus fully constructing the technical correlation path from fire perception to physical circuit reconstruction.

[0095] The purpose of this mechanism is to map the abstract degree of danger into equivalent electrical parameters that can directly drive the droop control, so that changes in environmental conditions can affect the output reference voltage in a deterministic and reproducible manner, thereby achieving continuous and adjustable control of the DC bus power flow direction in fire scenarios.

[0096] In this embodiment, the underlying circuit collaborative control module includes a residual fitting analysis unit;

[0097] The residual fitting analysis unit continuously monitors the voltage of the DC power supply bus when the environmental hazard potential energy gradient is lower than the preset response threshold.

[0098] In response to an environmental hazard potential energy gradient falling below a preset response threshold and the DC power supply bus voltage continuously being monitored to be lower than a preset bus voltage reference value, the residual fitting analysis unit determines that there is a high-risk node in the bus drawing current. Based on Kirchhoff's current law, the residual fitting analysis unit triggers the local power conversion unit to reduce the output current along the droop control reference curve. The specific logic is as follows: calculate the bus voltage residual by subtracting the current actual DC power supply bus voltage from the preset bus voltage reference value, multiply the bus voltage residual by a preset yield gain constant to obtain the yield current correction amount, and add the yield current correction amount as a negative current bias to the control loop of the bidirectional DC-DC converter.

[0099] This embodiment provides a cooperative yielding mechanism based on bus voltage residual fitting. Specifically, in the aforementioned scheme, high-risk nodes will actively reduce virtual impedance and attract more current. However, if other low-risk nodes simply remain unchanged, persistent low voltage may still occur locally on the bus, resulting in slow overall power distribution convergence speed.

[0100] To address this issue, this embodiment introduces a residual fitting analysis unit, enabling low-risk nodes to infer from bus voltage deviation that high-risk nodes in the system are requesting support, and to reduce local output current along the drooping control reference curve accordingly. Specifically, the residual fitting analysis unit mainly operates when the local danger potential energy gradient is below the response threshold; it continuously monitors the difference between the actual bus voltage and the bus voltage reference value.

[0101] For example, if the baseline value is set to 750V, and the measured bus voltage of a branch where a low-risk node is located is 748V, 746V, and 745V at three consecutive sampling times, the corresponding residuals are 2V, 4V, and 5V, respectively. If the local environmental hazard potential gradient is consistently below 0.50 and the residuals continue to expand, it can be reasonably determined that the fire is not local, but rather that other nodes in the system reduce their virtual impedance due to the increased hazard, causing the bus to be locally pulled down.

[0102] At this time, the residual fitting analysis unit triggers the local power conversion unit to reduce the output current according to the preset rules; to illustrate the reasoning relationship based on Kirchhoff's current law, a simplified current balance calculation example can be made; assume that a certain bus is locally connected to three nodes, node A is the high-risk node on the east side of the 18th floor, node B is the low-risk node in the middle of the 17th floor, and node C is the low-risk node on the west side of the 16th floor; at a certain moment, the total current balance of the bus is IA+IB+IC=Iload; where IA is the output current of node A, IB is the output current of node B, IC is the output current of node C, and Iload is the total load current of the bus;

[0103] If A increases its current draw by 8A due to the escalation of the fire, and the change in the total load side is mainly concentrated in the area corresponding to A, then if B and C are not adjusted, the bus voltage will continue to drop. The residual fitting analysis unit does not need to know how many amperes A has increased. As long as it observes that there is no fire in the local area and the bus voltage continues to drop, it can reduce the local output current by 3A and 5A respectively according to the droop curve, thereby reapproaching the balance.

[0104] Residual fitting essentially maps the trend of voltage difference changes to the amount of current adjustment that needs to be yielded, without having to precisely identify the remote node. This mechanism is particularly suitable for communication interruption scenarios because low-risk nodes do not receive a plaintext command to support the east side of the 18th floor, but instead sense system changes through the common physical quantity of bus voltage, forming a self-organizing response. Compared to directly cutting off branches, this yield is continuous and smooth, reducing frequent equipment start-ups and shutdowns.

[0105] As a backup protection mechanism, if a low-risk node detects that the bus voltage is lower than the reference value, but at the same time the local danger potential gradient begins to rise and approaches the response threshold, it will not immediately retreat significantly, but will enter a preset delay judgment period to prevent the power reduction command from being triggered prematurely when it is about to become a new danger zone; if the bus voltage fluctuates briefly and then recovers quickly, for example, if it drops by 2V in one sampling period and recovers in the next period, it can be regarded as an instantaneous disturbance and will not trigger obvious current retreat; if the local output current has dropped to the minimum protection value after multiple retreats, it will maintain the protection value to avoid power loss of the minimum safe loads such as emergency lighting in this area;

[0106] After the exhaust fan on the east side of the 18th floor was activated at full speed, no obvious fire occurred at the central node of the 17th floor, and the local danger potential gradient was only 0.22. However, it detected that the bus voltage gradually decreased from 750V to 744V. Based on this, the residual fitting analysis unit judged that other high-risk nodes were preempting current. Therefore, it controlled the converter corresponding to the non-critical load in the central part of the 17th floor to gradually reduce the output current along the droop curve, for example, from 14A to 10A. The western node of the 16th floor also made a similar response. In this way, the eastern side of the 18th floor can receive power transfer from the adjacent area without communication commands.

[0107] To further clarify the specific implementation logic of the residual fitting analysis unit in triggering the reduction of output current based on Kirchhoff's current law, it is not necessary to introduce a complex global optimization algorithm with undisclosed internal calculation logic. Instead, it relies on the following clearly defined local derivation calculation formula: the bus voltage residual calculated at the low-risk node within a certain control cycle. Strictly defined as the preset bus voltage reference value Subtract the current actual DC power supply bus voltage That is, the operation logic is This unit will With the preset yield gain constant Multiply by the product to calculate the clearance current correction. That is, the computational logic is ;

[0108] For example, for node B, if the measurement is... 5V, constant If pre-configured to 0.6A / V, then The value was determined to be 3A by the formula. By superimposing this correction amount into the current feedback reference channel of the underlying control, it is equivalent to artificially adding a negative current bias with an absolute value of 3A on the original droop curve, forcing the active output current of the bidirectional DC-DC converter to be smoothly reduced from 14A to 11A. According to Kirchhoff's current law, the 3A current actively yielded by node B will naturally flow into the high-risk node along the bus voltage gradient because the bus is physically interconnected and the high-risk node is in a strong current-absorbing state. This design details the micro-circuit mechanism of the residual to current adjustment mapping, ensuring the control stability and reproducibility of the yielding process when there is no global communication handshake.

[0109] The purpose of this mechanism is to enable low-risk nodes to automatically identify systemic power shortages and make concessions by using the common physical feedback of bus voltage residuals, thereby accelerating the power accumulation process in high-risk areas and reducing reliance on centralized coordination links.

[0110] In this embodiment, the underlying circuit collaborative control module includes a flexible electrical isolation unit;

[0111] The flexible electrical isolation unit has preset short-circuit thresholds including current rise rate threshold and current amplitude threshold;

[0112] In response to an environmental hazard potential energy gradient equal to or higher than a preset limit blocking threshold, or the detection that the rate of rise or amplitude of the local output current exceeds the corresponding current rise rate threshold or current amplitude threshold in the preset short-circuit threshold, the flexible electrical isolation unit sets the virtual impedance adjustment parameter to a preset blocking extreme value, or controls the duty cycle of the pulse width modulation control signal to zero, thereby limiting the current output to the DC power supply bus.

[0113] This embodiment provides a flexible electrical isolation mechanism. Specifically, under the aforementioned continuous power tilt strategy, the system usually tends to allocate more power to high-risk areas. However, under extreme conditions, some areas may have changed from critical escape areas that urgently need power supply to areas of intense combustion or electrical faults. If a large current is still continuously supplied to this branch at this time, it may cause the DC power supply bus to collapse or even expand the fault.

[0114] Therefore, this embodiment introduces a flexible electrical isolation unit to implement rapid current limiting or physical isolation of the node when the ultimate blocking condition is reached or short-circuit symptoms appear. In specific implementation, the flexible electrical isolation unit pays attention to two types of triggering sources simultaneously. The first type is when the environmental danger potential energy gradient reaches or exceeds the ultimate blocking threshold. For example, if the threshold is set to 0.90, when a node measures a danger potential energy gradient of 0.93, it can be considered that the area has entered an extreme combustion state that cannot be supported any longer. The second type is when the rate of increase or amplitude of the local output current reaches or exceeds the short-circuit threshold.

[0115] For example, if the short-circuit threshold is set to a current amplitude of 60A or a current rise rate of 20A / millisecond, and a node that originally outputs 15A suddenly jumps to 68A, or rises from 15A to 40A within 1 millisecond, it can be judged as a short circuit or near-short circuit event. Once triggered, the flexible electrical isolation unit can adopt two action modes: one is to directly set the virtual impedance adjustment parameter to the blocking extreme value, so that the node exhibits extremely high equivalent impedance in the bus, thereby significantly compressing the current; the other is to directly set the duty cycle of the pulse width modulation control signal to zero, so that the bidirectional DC-DC converter is turned off, forming physical isolation. Both can be used according to the degree of fault. For example, when the dangerous potential gradient exceeds the threshold but the current has not yet run away from control, high-impedance isolation is used first. If the current continues to rise, it enters the physical disconnection state with the duty cycle set to zero.

[0116] To illustrate the difference between flexible and conventional circuit breaker tripping, a microscopic process can be used as an example: At time t0, the critical potential energy gradient of a certain node is 0.85, which has not yet reached the ultimate blocking threshold. However, the local output current suddenly surges from 18A to 35A, and the rate of current rise approaches the short-circuit threshold. The system first rapidly increases the virtual impedance from 0.5Ω to 5Ω and observes it for two sampling cycles. If the current falls back to below 20A, it indicates that it may be a transient surge, and the high-impedance state can be maintained for continued monitoring. If the current continues to climb to 65A, the next step is to set the duty cycle to zero and directly cut off the node's output to the bus.

[0117] This avoids the slow operation of ordinary mechanical circuit breakers and the false tripping caused by excessive sensitivity to transient disturbances. To explain in detail how to achieve a physical closed loop of current limiting by setting a blocking extreme value, rather than just staying at the level of an abstract concept: combined with the equivalent internal resistance bias logic of droop control, when the system detects a short circuit symptom such as a current surge from 15A to 40A, the flexible electrical isolation unit does not rely on mechanical mechanism action, but forcibly overwrites the dynamic compensation impedance to a maximum blocking extreme value of, for example, 50.0Ω. At this extreme value, even if there is only a 0.5A output tendency to the bus, the output reference voltage calculated by the node will suddenly drop to 760 minus 50.0 multiplied by 0.5, which is 735V.

[0118] Since the reference voltage is momentarily lower than the actual bus voltage, such as 740V, the converter's control loop will be subject to strong reverse constraints, quickly converging to a near-zero output state or even a reverse current isolation state. This design directly embeds the electrical isolation logic into the droop control formula, which not only enables the protection response speed to reach the microsecond level, but also preserves a soft-start channel for adaptive controlled reconnection after the subsequent fire subsides.

[0119] As a backup protection mechanism, if the environmental hazard potential gradient reaches the ultimate blocking threshold, but the local equipment is the minimum backup load on the life evacuation route, different equipment branches can be isolated in layers. For example, high-power auxiliary equipment can be cut off first, and only the minimum emergency lighting can be retained. If the local output current exceeds the amplitude threshold but the duration is extremely short and the current rise rate is not high, current limiting can be performed first and then retested to reduce false isolation caused by surge initiation. If the node has entered the isolation state and the environmental hazard level decreases afterward, controlled reconnection can be allowed after the insulation retest, short circuit disappearance and time delay expiration are met.

[0120] After the fire on the east side of the 18th floor further expanded, a smoke exhaust branch near the core of the fire experienced a near short circuit due to heat damage to the cable insulation. Its local current jumped from 16A to 70A in a very short time. The flexible electrical isolation unit did not wait for the upstream circuit breaker to trip, but instead raised the virtual impedance to the blocking extreme value within its own node. After detecting that the current had not fallen back, it immediately reduced the pulse width modulation duty cycle to zero, decoupling the branch from the DC bus. At the same time, other non-faulty nodes on the east side of the 18th floor, as well as nodes on the 17th and 16th floors, could continue to maintain power supply to critical fire protection equipment.

[0121] The purpose of this mechanism is to achieve faster and finer-grained isolation action than traditional mechanical protection by using underlying power conversion control in extreme cases where fire and electrical faults overlap, so as to prevent a single point fault from causing the voltage of the entire DC power supply bus to collapse and to maintain the continuous power supply capability of other critical zones.

[0122] In this embodiment, the state data sensing module, the virtual impedance calculation engine, and the underlying circuit collaborative control module are integrated and deployed in a field-programmable gate array or a digital signal processing device.

[0123] Field-programmable gate arrays (FPGAs) or digital signal processing devices are distributed and installed within the node controllers of each distributed power supply node.

[0124] This embodiment provides a distributed hardware deployment mechanism. Specifically, in the aforementioned scheme, environmental change perception, virtual impedance calculation, and underlying collaborative control all require a fast response speed. If these functions are still centrally deployed in a single industrial computer in the building's central control room, damage to communication links after a fire, data upload delays, or centralized processing congestion will all weaken the control effect.

[0125] Therefore, in this embodiment, the state data sensing module, the virtual impedance calculation engine, and the underlying circuit collaborative control module are integrated into a field-programmable gate array or digital signal processing device, and distributed and installed in the node controller of each distributed power supply node; specifically, each smoke prevention zone can be configured with a node control board; the control board includes a sensor acquisition interface, an energy storage battery management interface, a bus voltage and current sampling interface, a field-programmable gate array or digital signal processing device, and a pulse width modulation drive interface;

[0126] Sensor data is directly fed into the hardware computing unit after local sampling. The unit performs calculations of the time-series change rate of environmental state data, mapping of dangerous potential energy gradients, updating of virtual impedance adjustment parameters, and generation of pulse width modulation control signals within the board. Since these actions are completed locally in a closed loop, each node controller can continue to operate independently even if the main communication switch of the building is powered off, the fiber optic cable is damaged, or the fire alarm control panel is temporarily offline. When implemented with a field-programmable gate array, data sensing, parameter lookup, drooping operations, and protection logic can be arranged in parallel.

[0127] For example, within the same sampling period, the fusion of temperature and smoke change rate, reading of energy storage state of charge, lookup of dangerous potential energy gradient, calculation of dynamic compensation impedance, and duty cycle refresh can be executed in parallel, thereby compressing the overall response to a shorter time. When implemented with digital signal processing devices, the control program can run according to a fixed interrupt cycle, such as completing a sampling and control update every 1 millisecond. Both implementation methods can achieve the goal of distributed and low-latency operation.

[0128] To illustrate the effect of distributed installation, we can assume that the communication connection of the east node controller on the 18th floor is interrupted due to its proximity to the fire. However, the local field-programmable gate array can still continuously receive temperature, smoke, bus voltage, and output current signals, and independently generate new pulse width modulation control signals. The node controllers on the 17th and 16th floors also complete residual fitting and yielding actions on their respective boards. In this way, although the entire system loses the central main controller, it still retains the local automatic response loop of each partition and forms system-level coordination through the physical coupling of the DC bus.

[0129] As a backup protection mechanism, if the hardware computing unit in a node controller is abnormally reset, its corresponding converter can be allowed to return to the preset safety droop coefficient operation first, and dynamic control can be restored after the local self-test passes; if the field programmable gate array resources are insufficient, some non-real-time diagnostic functions can be transferred to the host computer, while environmental change extraction, virtual impedance calculation and protection logic are retained in the local hardware; if a smoke prevention zone node fails completely, the other zone nodes can continue to operate without relying on that node to forward commands.

[0130] After an open fire broke out on the east side of the 18th floor and the building's electrical shaft was damaged by heat, the central monitoring room could no longer receive the complete status packet of that zone in real time. However, the digital signal processing device in the node controller on the east side of the 18th floor still read the local temperature rise rate, smoke change rate and battery state of charge at a 1-millisecond cycle and continuously adjusted the bidirectional DC-DC converter. The node controllers on the 17th and 16th floors also monitored the bus voltage residual and made concessions. In the end, although the central communication link was interrupted, the smoke exhaust and lighting on the critical escape routes were still powered.

[0131] The purpose of this mechanism is to decentralize key control functions to the local hardware nodes of each distributed power supply node, shorten the control loop, reduce dependence on the central communication and computing platform, and ensure that the system still has distributed autonomous power supply capabilities under extreme fire conditions.

[0132] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A backup power supply system for building fire emergency equipment, characterized in that, It includes multiple distributed power supply nodes connected to access devices and containing local sensors, local energy storage units and local power conversion units, a DC power supply bus interconnecting the nodes, a topology initialization module, a status data sensing module, a virtual impedance calculation engine and a low-level circuit collaborative control module; When the topology initialization module is powered on, the base droop control coefficient is allocated according to the nameplate power and static importance level of the access device; The status data sensing module obtains the time-series change rate of local environmental status data based on the environmental status data collected by the local sensors, obtains the energy storage state of charge parameters based on the local energy storage unit, and obtains the local output current of the local power conversion unit. The virtual impedance calculation engine maps the time-series rate of change to the environmental hazard potential energy gradient based on the droop control reference curve with impedance bias characteristics, and calculates the virtual impedance adjustment parameters by combining the energy storage state of charge parameters with the basic droop control coefficient as the starting point. The underlying circuit collaborative control module generates a pulse width modulation control signal based on the virtual impedance adjustment parameter to adjust the output reference voltage of the local power conversion unit, and presets a preset response threshold for the environmental hazard potential energy gradient, a preset limit blocking threshold greater than the preset response threshold, a preset bus voltage reference value, and a preset blocking extreme value for performing physical isolation. The local power conversion unit is a bidirectional DC-DC converter; Each bidirectional DC-DC converter is interconnected through the DC power supply bus to construct a multi-source, multi-load DC microgrid topology; The pulse width modulation control signal is directly input to the voltage and current dual-loop control circuit of the bidirectional DC-DC converter; The underlying circuit collaborative control module includes a residual fitting analysis unit; The residual fitting analysis unit continuously monitors the voltage of the DC power supply bus when the environmental hazard potential energy gradient is lower than the preset response threshold. In response to the environmental hazard potential energy gradient being lower than a preset response threshold and the DC power supply bus voltage being continuously monitored to be lower than the preset bus voltage reference value, the residual fitting analysis unit determines that there is a high-risk node in the DC power supply bus drawing current. Based on Kirchhoff's current law, the residual fitting analysis unit triggers the local power conversion unit to reduce the output current along the droop control reference curve. The specific logic is as follows: calculate the bus voltage residual by subtracting the current actual DC power supply bus voltage from the preset bus voltage reference value, multiply the bus voltage residual by a preset yield gain constant to obtain the yield current correction amount, and superimpose the yield current correction amount as a negative current bias into the control loop of the bidirectional DC-DC converter.

2. The backup power supply system for a building fire emergency equipment according to claim 1, characterized in that, The state data perception module includes an environmental drastic change feature extraction unit, which acquires the temperature rise rate and smoke concentration change rate output by the local sensor. The environmental change feature extraction unit performs a weighted summation feature fusion operation on the temperature rise rate and the smoke concentration change rate to generate the local environmental state data time-series change rate. The status data sensing module relies solely on data collected by the local sensors and the local energy storage unit.

3. The backup power supply system for a building fire emergency equipment according to claim 1, characterized in that, When calculating the virtual impedance adjustment parameters, the virtual impedance calculation engine performs an equivalent internal resistance bias calculation based on the droop control equation. The equivalent internal resistance bias calculation is based on a preset rated reference voltage, a preset base impedance, and a preset proportionality coefficient with impedance dimensions. The logic of the equivalent internal resistance bias operation is as follows: make the output reference voltage equal to the preset rated reference voltage minus the product of the dynamic compensation impedance and the local output current. The dynamic compensation impedance is obtained by subtracting the product of the environmental hazard potential energy gradient and the preset proportional coefficient from the preset basic impedance.

4. The backup power supply system for a building fire emergency equipment according to claim 1, characterized in that, The underlying circuit collaborative control module includes a flexible electrical isolation unit; The flexible electrical isolation unit has a preset short-circuit threshold including a current rise rate threshold and a current amplitude threshold. In response to the environmental hazard potential energy gradient being equal to or higher than the preset limit blocking threshold, or the detection that the rise rate or amplitude of the local output current reaches or exceeds the corresponding current rise rate threshold or current amplitude threshold in the preset short-circuit threshold, the flexible electrical isolation unit sets the virtual impedance adjustment parameter to the preset blocking extreme value, or controls the duty cycle of the pulse width modulation control signal to zero, thereby limiting the current output to the DC power supply bus.

5. A backup power supply system for a building fire emergency equipment according to claim 1, characterized in that, The state data sensing module, virtual impedance calculation engine, and underlying circuit collaborative control module are integrated and deployed in a field-programmable gate array or digital signal processing device. The field-programmable gate arrays or digital signal processing devices are distributed and installed in the node controllers of each distributed power supply node.

6. The backup power supply system for a building fire emergency equipment according to claim 1, characterized in that, The virtual impedance adjustment parameter is equal to the basic droop control coefficient minus the product obtained by multiplying the environmental hazard potential energy gradient, the state of charge support, and the preset droop adjustment ratio. The state of charge support is a piecewise function mapping of the energy storage state of charge parameter.

7. A backup power supply system for a building fire emergency equipment according to claim 1, characterized in that, The underlying circuit collaborative control module performs the following collaborative control based on the environmental hazard potential energy gradient: When the environmental hazard potential energy gradient is higher than the preset response threshold but lower than the preset limit blocking threshold, the virtual impedance adjustment parameter is reduced to increase the current drawn from the DC power supply bus. When the environmental hazard potential energy gradient is lower than the preset response threshold, if the DC power supply bus voltage is lower than the preset bus voltage reference value, the output power is reduced along the droop control reference curve; otherwise, the basic droop control coefficient is maintained. When the environmental hazard potential energy gradient equals the preset response threshold, the output reference voltage is maintained. When the environmental hazard potential energy gradient is equal to or higher than the preset limit blocking threshold, the virtual impedance adjustment parameter is set to the preset blocking extreme value, or the duty cycle of the pulse width modulation control signal is controlled to be zero, and physical isolation is performed.

Images