Dynamic hierarchical scheduling method and system for multi-priority load nodes

Abstract

The application relates to the technical field of power dispatch, and discloses a dynamic hierarchical dispatching method and system for multi-priority load nodes, wherein the dynamic hierarchical dispatching method for multi-priority load nodes comprises the following steps: acquiring real-time dynamic information and preconfigured static information of load nodes; identifying a disturbance event through a multi-disturbance detection model; calculating a dynamic priority of the load nodes and executing a local resource immediate dispatching strategy; establishing a cross-region collaborative optimization objective function, and solving the collaborative optimization objective function to obtain a cross-region mutual aid scheme; and dynamically adjusting a storage energy state of charge target value and a cooling system power plan; through the dynamic hierarchical dispatching method and a distributed collaborative control architecture, the application effectively solves the problems of rapid response and accurate dispatching of a multi-priority load node system under complex disturbance conditions, and provides important technical support for safe and stable operation of a modern power system.

Description

Technical Field

[0001] This invention relates to the field of power dispatching technology, and more specifically, to a dynamic hierarchical dispatching method and system for multi-priority load nodes. Background Technology

[0002] Data centers have extremely high requirements for power supply continuity and power quality, with very small allowable interruptions and power quality deviation windows. A typical data center consists of IT load and a cooling system. The former cannot be interrupted, while the latter has a certain degree of thermal inertia and can gradually reduce some power in a short period of time, forming a tiered demand of prioritizing computing and reducing cooling.

[0003] A Chinese patent with authorization announcement number CN107292516B discloses a load reliability assessment method considering load classification and energy dispatch. The method includes: 1) acquiring raw data of the power distribution system and raw meteorological data over a period of time; 2) establishing a reliability model of a distributed generation and energy storage combined generation system; 3) equating the local network composed of distributed generation and energy storage to a single node, denoted as the PCC node, and using the Go algorithm to calculate the equivalent state model and parameters of external components relative to the PCC node; 4) classifying the loads within the PCC node, setting weight coefficients, and using an improved particle swarm optimization algorithm to solve for the optimal dispatch method; and 5) calculating the load reliability index using Monte Carlo simulation. This invention can more accurately calculate the reliability of loads with different characteristics and can improve load reliability by proposing energy dispatch schemes through optimization algorithms.

[0004] However, under rapid disturbances such as sudden drops in main grid frequency, short-term voltage drops, bus disturbances, and sudden power flow changes, millisecond or second-level stable support and rapid and feasible load response are required. Existing technologies divide loads into two categories, critical and interruptible, and set fixed weights. They lack dynamic priority mapping that changes with state, events, SLA, and operating conditions, making it difficult to design differentiated control strategies that can both ensure power quality and uninterrupted operation for critical loads such as IT, and implement safe and controllable throttling for thermally inert loads such as cooling, thus transferring active and reactive power support capabilities. Summary of the Invention

[0005] The purpose of this invention is to provide a dynamic hierarchical scheduling method and system for multi-priority load nodes in order to solve the above-mentioned problems.

[0006] This invention provides a dynamic hierarchical scheduling method for multi-priority load nodes, comprising the following steps:

[0007] Obtain real-time dynamic information and pre-configured static information of load nodes;

[0008] Based on the real-time dynamic information, disturbance events are identified through a multi-disturbance detection model;

[0009] Based on the disturbance events and pre-configured static information, the dynamic priority of the load nodes is calculated and the local resource real-time scheduling strategy is executed.

[0010] Based on real-time dynamic information after executing the local resource real-time scheduling strategy, a cross-regional collaborative optimization objective function is established, and the cross-regional mutual assistance scheme is obtained by solving the collaborative optimization objective function.

[0011] Based on the execution results of the cross-regional mutual assistance scheme, the target value of the energy storage state of charge and the power plan of the cooling system are dynamically adjusted.

[0012] Furthermore, disturbance events include frequency disturbance events and voltage dip events, wherein:

[0013] A frequency disturbance event is determined when the frequency change rate is greater than a preset frequency change rate threshold or when the absolute value of the difference between the instantaneous frequency value and the rated frequency is greater than a preset frequency deviation threshold. Frequency disturbance events include mild frequency disturbances, moderate frequency disturbances, and severe frequency disturbances, which are determined based on the magnitude of the absolute value of the difference between the instantaneous frequency value and the rated frequency.

[0014] A voltage drop event is defined as a voltage drop event when the voltage drop depth exceeds a preset voltage drop threshold and the duration exceeds a preset duration threshold. The voltage drop depth is the difference between the rated voltage and the real-time voltage, divided by the rated voltage. Voltage drop events include first-level voltage drop events and second-level voltage drop events, which are determined based on the magnitude of the voltage drop depth.

[0015] Furthermore, calculating the dynamic priority of load nodes and executing local resource real-time scheduling strategies includes:

[0016] Based on the disturbance event, the weight parameters for dynamic priority calculation are adjusted. The weight parameters include a preset first weight, a preset second weight, and a preset third weight.

[0017] The dynamic priority of the load node is calculated as follows: the dynamic priority is the sum of the preset first weight multiplied by the SLA level and the preset second weight multiplied by the temperature margin coefficient, and then the power deviation term is subtracted. The power deviation term is the preset third weight multiplied by the power deviation coefficient. The temperature margin coefficient is the difference between the preset maximum allowable temperature and the real-time temperature, divided by the preset maximum allowable temperature. The power deviation coefficient is equal to the absolute value of the current power minus the minimum operating power, divided by the rated power. The SLA level is determined according to the load node level, which includes P1, P2, P3, P4, and P5 levels.

[0018] Based on dynamic priorities and disturbance events, local resource real-time scheduling strategies are implemented for load nodes of different levels.

[0019] Furthermore, local resource real-time scheduling strategies include:

[0020] For P1 level IT loads, the UPS control mode is locked to double conversion mode, and the energy storage system is called to provide virtual inertia. The virtual inertia is the product of the preset reference virtual inertia coefficient and the energy storage state of charge adjustment coefficient.

[0021] For cooling loads of P2 to P4 levels, power descent control is implemented in order of priority from low to high. The change in cooling power must not exceed the cooling equipment ramp rate multiplied by the preset control time interval. At the same time, the real-time temperature is continuously monitored to ensure that it does not exceed the preset maximum allowable temperature.

[0022] For non-essential auxiliary loads at the P5 level, a tiered shelving strategy is implemented based on the severity of the disturbance. For mild and moderate frequency disturbance events, loads at the P5 level with a preset P5 load shelving ratio are shelved. For severe frequency disturbance events or voltage drop events of any level, all loads at the P5 level are shelved.

[0023] Furthermore, the objective function is to minimize the comprehensive deviation, which is the sum of the preset first weight multiplied by the load power deviation term, the preset second weight multiplied by the frequency deviation term, and the preset third weight multiplied by the voltage deviation term.

[0024] Constraints include power flow constraints and voltage constraints;

[0025] Based on the constraints, a distributed optimization algorithm is used to solve the objective function to obtain the cross-regional mutual assistance scheme. The cross-regional mutual assistance scheme includes active power transfer across the common connection point, reactive power coordinated regulation scheme, cross-regional coordinated charging and discharging strategy of energy storage system, and timing arrangement of tie switch operation.

[0026] Furthermore, the load power deviation term is the sum of the squares of the load power deviations of multiple IT loads, the frequency deviation term is the sum of the absolute values ​​of the frequency deviations of multiple load nodes, the voltage deviation term is the sum of the squares of the voltage deviations of multiple load nodes, the load power deviation is the actual power of the i-th IT load minus the minimum guaranteed power of the i-th IT load, the frequency deviation is the instantaneous frequency value of the j-th load node minus the absolute value of the preset rated frequency, and the voltage deviation is the voltage of the k-th load node minus the rated voltage.

[0027] Furthermore, the power flow constraint is that the active power flow of the line is less than the active power capacity limit of the line, and the reactive power flow of the line is less than the reactive power capacity limit of the line. The voltage constraint is that the voltage of the load node should be between the preset lower voltage limit and the preset upper voltage limit.

[0028] Furthermore, the target value for the energy storage state of charge at the next moment is the current energy storage state of charge minus the state of charge consumption, plus the product of the energy storage charging power and the preset optimization time interval.

[0029] The current cooling system power plan is the sum of the products of the cooling power baseline, the preset recovery ramp rate, and the preset optimization time interval.

[0030] Furthermore, the real-time dynamic information includes electrical parameter data and equipment status data; wherein, the electrical parameter data includes the instantaneous frequency value, frequency change rate, three-phase voltage, active power flow and reactive power flow of each load node, and the equipment status data includes the load power, real-time temperature, UPS control mode, energy storage charge status and charging and discharging power of the IT load.

[0031] The pre-configured static information includes load constraints and SLA parameters; wherein, the load constraint parameters include the minimum guaranteed power of IT load, the ramp rate of cooling equipment, and the minimum operating power, and the SLA parameters include the SLA level of each load node.

[0032] This invention provides a dynamic hierarchical scheduling system for multi-priority load nodes, which stores computer-readable instructions and, when read, can execute the aforementioned dynamic hierarchical scheduling method for multi-priority load nodes; the system includes:

[0033] The data acquisition module obtains real-time dynamic information and pre-configured static information of the load nodes;

[0034] The disturbance detection module identifies disturbance events based on the real-time dynamic information using a multi-disturbance detection model;

[0035] The local control module calculates the dynamic priority of load nodes and executes local resource real-time scheduling strategies based on the disturbance events and pre-configured static information.

[0036] The regional collaboration module establishes a cross-regional collaborative optimization objective function based on real-time dynamic information after executing the local resource real-time scheduling strategy, and solves the collaborative optimization objective function to obtain the cross-regional mutual assistance scheme.

[0037] The energy optimization module dynamically adjusts the target value of the energy storage state of charge and the power plan of the cooling system based on the execution results of the cross-regional mutual assistance scheme.

[0038] The beneficial effects of this invention are as follows: Through a three-level distributed control architecture, this invention achieves millisecond-level response throughout the entire process from disturbance detection to control execution. The edge controller can immediately initiate dynamic priority calculation and local real-time control upon detecting a disturbance, with a response time reaching the millisecond level, which is 1-2 orders of magnitude faster than the second-level response time of traditional centralized control systems. This rapid response capability ensures that critical loads are effectively protected in the early stages of disturbances, significantly improving the power supply reliability of the system.

[0039] By establishing a disturbance detection model encompassing three dimensions—frequency disturbance, voltage disturbance, and topology disturbance—the system can accurately identify disturbance events of varying severity, ranging from mild to severe. Based on the disturbance type and severity, the system adaptively adjusts the weight parameters for dynamic priority calculation, achieving intelligent hierarchical scheduling of disturbance perception. Compared to traditional single-parameter judgment methods, this multi-dimensional disturbance perception mechanism improves the accuracy of disturbance identification and effectively avoids misjudgment and missed detection.

[0040] By comprehensively considering multiple real-time status parameters such as SLA level, temperature margin coefficient, and power deviation coefficient, a dynamic priority calculation model was established, which can dynamically adjust the priority ranking according to the real-time operating status of load nodes. Compared with the traditional static classification method, this refined management approach can more accurately reflect the actual importance and urgency of loads, improving the rationality of scheduling decisions.

[0041] A three-tiered distributed control architecture enables multi-timescale coordinated control, achieving millisecond-level edge autonomy, second-level regional collaboration, and minute-level energy optimization. The edge controller handles rapid local response, the regional collaboration controller facilitates cross-regional resource sharing, and the upper-level energy optimization controller manages long-term energy usage, forming a complete control loop. This hierarchical collaborative mechanism ensures both rapid response capabilities and global optimization, improving overall system scheduling efficiency.

[0042] This invention designs differentiated control strategies for loads of different levels, P1 to P5, based on their characteristics and importance: P1 core IT loads are protected with the highest level of protection, locking the UPS in double-conversion mode and providing virtual inertia support; P2 to P4 cooling loads are subject to priority-based power descent control, strictly adhering to ramp-up constraints; and P5 auxiliary loads are subject to tiered shelving based on the severity of disturbances. This differentiated protection strategy ensures the absolute safety of critical loads while maximizing the overall system operational capacity.

[0043] Through a disturbance-aware weight adjustment mechanism and an adaptive communication strategy, the system can automatically adjust its control strategy according to different disturbance scenarios and communication conditions. When communication is restricted, the system automatically switches to a local autonomous mode; when the severity of the disturbance changes, the system automatically adjusts the priority weight configuration. This adaptive capability significantly enhances the system's robustness and reliability in complex operating environments and shortens system fault recovery time.

[0044] This invention achieves a coordinated balance between power supply reliability, power quality, and economy by establishing a multi-objective optimization function that comprehensively considers load power guarantee, system frequency stability, and voltage quality maintenance. Compared with traditional single-objective optimization methods, this invention can maximize the overall operating efficiency of the system and improve comprehensive performance indicators while ensuring the power supply to critical loads.

[0045] In summary, this invention effectively solves the problem of rapid response and accurate scheduling of multi-priority load node systems under complex disturbance conditions through innovative dynamic hierarchical scheduling methods and distributed cooperative control architecture, providing important technical support for the safe and stable operation of modern power systems. Attached Figure Description

[0046] Figure 1 This is a flowchart illustrating the dynamic hierarchical scheduling method for multi-priority load nodes according to the present invention.

[0047] Figure 2 This is an example diagram illustrating the execution of a local resource real-time scheduling strategy for the dynamic hierarchical scheduling method for multi-priority load nodes according to the present invention.

[0048] Figure 3 This is an example diagram illustrating the cross-regional mutual assistance scheme obtained from the dynamic hierarchical scheduling method for multi-priority load nodes of the present invention.

[0049] Figure 4 This is a module example diagram of the dynamic hierarchical scheduling system for multi-priority load nodes of the present invention. Detailed Implementation

[0050] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0051] A dynamic hierarchical scheduling method and system for multi-priority load nodes, including the following embodiments:

[0052] Example 1:

[0053] This embodiment adopts a three-level distributed architecture consisting of an edge controller, a regional collaborative controller, and an upper-level energy optimization controller.

[0054] Load node edge controllers are deployed in individual data centers or key urban load sites, responsible for local disturbance detection, dynamic load classification, and real-time device control. An edge controller includes a data acquisition module, a disturbance detection module, a priority calculation module, and a local control module.

[0055] The regional collaborative controller covers distribution networks and microgrid areas interconnected by multiple points of common connection (PCCs). It is responsible for collecting the status of each edge controller, solving inter-regional mutual assistance schemes, verifying power flow feasibility, and generating switching operation sequences. As the middle layer of a three-tiered distributed control architecture, the regional collaborative controller plays a crucial role in connecting the upper and lower layers. It receives and processes real-time status information from multiple edge controllers and interacts with the upper-level energy optimization controller for information exchange and command coordination. This controller adopts a distributed computing architecture, possessing powerful data processing and optimization capabilities, enabling it to complete complex multi-objective optimization calculations and inter-regional resource coordination and scheduling within a second-level timescale. The control range of the regional collaborative controller typically covers a complete distribution network area or multiple interconnected microgrid systems, managing dozens of edge controllers and involving various important load types, including data centers, industrial parks, and commercial complexes. The regional collaborative controller includes a data aggregation module, an optimization solution module, and a command issuance module.

[0056] The upper-level energy optimization controller performs rolling energy optimization at the minute and hour level, receives second-level scheduling results from the regional controller, and updates energy storage state of charge targets, diesel engine start-stop plans, etc. The upper-level controller includes an energy management module, a long-term optimization module, and a plan generation module.

[0057] Dynamic hierarchical scheduling methods for multi-priority load nodes, such as... Figure 1 As shown, it includes the following steps:

[0058] Step 100: Obtain real-time dynamic information and pre-configured static information of the load nodes.

[0059] The edge controller acquires real-time dynamic information and pre-configured static information through dual channels: edge sensing and regional collaboration.

[0060] Real-time dynamic information includes electrical parameter data, equipment status data, and communication status. Specifically, electrical parameter data is collected via local sensors on the edge controller at a sampling frequency of 1000 Hz, including instantaneous frequency values ​​at each load node, frequency change rate, three-phase voltage, active power flow, and reactive power flow. Equipment status data is collected via the regional SCADA system at a refresh frequency of 50 Hz, including IT load power, real-time temperature, UPS control mode and load rate, energy storage status of charge and charging / discharging power, and reactive power compensation device output. Communication status is monitored, including communication latency between the edge controller and the regional controller, and data packet loss rate.

[0061] Pre-configured static information includes load constraints, SLA parameters, system topology, and mutual assistance rules. Specifically, load constraint parameters are obtained from the system initialization configuration database, including minimum guaranteed power for IT loads, cooling equipment ramp-up rate, and minimum operating power; SLA parameters are retrieved in real-time from the SLA protocol database, including SLA levels for each load node, interruption tolerance time, and power quality thresholds; and topology and mutual assistance rules are obtained, including multi-PCC tie-line operation sequence constraints, inter-regional transfer line capacity, and a pre-stored edge autonomy strategy library.

[0062] Step 200: Based on the real-time dynamic information, identify disturbance events using a multi-disturbance detection model.

[0063] Based on the real-time electrical parameter data collected in step 100, the edge controller uses a disturbance detection model to quickly identify and accurately classify abnormal system states. This detection mechanism employs a hierarchical and progressive criterion system. First, frequency disturbance detection is performed as a core indicator of system stability. Then, voltage disturbance detection is combined to ensure the integrity of power quality monitoring. Finally, topology disturbance detection ensures the reliability of the network structure, forming an interconnected disturbance identification closed loop. The purpose of disturbance detection and classification is to provide accurate disturbance type information for the dynamic priority calculation in step 300 and the regional collaborative optimization in step 400, enabling differentiated control strategies based on disturbance severity and ensuring that the most appropriate countermeasures are taken under different disturbance scenarios. The disturbance detection model includes frequency disturbance detection, voltage disturbance detection, and topology disturbance detection.

[0064] In the frequency disturbance detection phase, the edge controller continuously monitors the dynamic changes in frequency. When the rate of frequency change exceeds a preset threshold or the absolute value of the instantaneous frequency deviation from the rated frequency exceeds a preset frequency deviation threshold, it is immediately identified as a frequency disturbance event. The preset rate of frequency change threshold is determined based on the load type and system inertia characteristics, obtained through statistical analysis of historical disturbance data; in a data center scenario, the default value is 0.5 Hz per second. The preset frequency deviation threshold is set according to the power system frequency stability requirements, determined through system simulation verification; the default value is 0.8 Hz. Based on the absolute value of the instantaneous frequency deviation from the rated frequency, frequency disturbance events are classified into different levels: mild, moderate, and severe. Mild frequency disturbance events correspond to an absolute frequency deviation greater than 0.2 Hz and less than or equal to 0.8 Hz; moderate frequency disturbance events correspond to an absolute frequency deviation greater than 0.8 Hz and less than or equal to 1.5 Hz; and severe frequency disturbance events correspond to an absolute frequency deviation greater than 1.5 Hz. This hierarchical mechanism based on the absolute value of frequency deviation can uniformly handle high-frequency and low-frequency disturbances, providing a precise criterion basis for subsequent differentiated control strategies.

[0065] In addition to frequency disturbance detection, voltage disturbance detection is performed simultaneously to ensure comprehensive power quality monitoring. A voltage sag event is identified when the voltage drop depth at a target node exceeds a preset voltage sag threshold and the duration exceeds a preset duration threshold. The voltage sag depth is calculated by dividing the difference between the rated voltage and the real-time voltage by the rated voltage. The preset voltage sag threshold is determined based on load sensitivity and power quality standards, obtained through load withstand tests, with a default value of 10% to 30%. The preset duration threshold is set based on load response characteristics, determined through experimental testing, with a default value of 20 milliseconds. Voltage sag events are classified into two levels based on their sag depth: Level 1 voltage sag events correspond to sag depths greater than 20% and less than or equal to 30%, while Level 2 voltage sag events correspond to sag depths greater than 30% and less than or equal to 50%. Voltage disturbance detection and frequency disturbance detection complement each other, forming a dual guarantee for power system stability monitoring.

[0066] To ensure the structural integrity of the multi-PCC interconnection network, topology disturbance detection is further implemented. A topology disturbance event is identified when any tie switch in the multi-PCC interconnection network malfunctions, the line power flow exceeds a preset multiple of the line capacity, or the communication delay exceeds a preset communication delay threshold. The preset line capacity multiple is determined based on the line's thermal stability limit and safety margin, obtained through power flow calculation and thermal stability analysis, with a default value of 1.1 times the line's rated capacity. The preset communication delay threshold is set according to the control system's real-time requirements, determined through communication network performance testing, with a default value of 50 milliseconds. Topology disturbance detection not only focuses on changes in the physical network but also monitors the status of the communication network, providing crucial information for the system's adaptive control.

[0067] After completing three-dimensional disturbance detection, the edge controller executes corresponding communication strategies based on the detection results. When operation is normal and the communication link is stable, the edge controller sends disturbance event notifications to the regional collaborative controller and synchronously uploads its local real-time status, ensuring that regional-level collaborative optimization can make decisions based on accurate information. However, when the communication latency exceeds a preset communication latency threshold, the upload process is automatically skipped, and the controller directly enters local autonomous mode, ensuring that basic control functions can still be maintained under communication constraints. This adaptive communication mechanism guarantees reliability and real-time performance under various operating conditions.

[0068] Step 300: Based on the disturbance event and pre-configured static information, calculate the dynamic priority of the load nodes and execute the local resource real-time scheduling strategy, specifically as follows: Figure 2 As shown.

[0069] After receiving the disturbance type information identified in step 200, the edge controller immediately initiates a collaborative response mechanism combining dynamic priority calculation and local real-time control. This mechanism achieves millisecond-level load tiered scheduling response through dynamic priority adjustment based on disturbance perception and a priority-based differentiated control strategy. The entire process consists of three closely related stages: adaptive adjustment of disturbance response priority weights, load node priority calculation based on real-time status, and local real-time resource scheduling based on priority levels.

[0070] First, the edge controller adaptively adjusts the weight parameters of the dynamic priority calculation based on the disturbance event and its severity. These weight parameters include preset first weight, preset second weight, and preset third weight. This disturbance-aware weight adjustment mechanism ensures that the most critical protection targets are highlighted under different disturbance scenarios. For frequency disturbance events, the weight impact of the SLA level gradually increases with the severity of the disturbance. The preset first weight is 0.5, preset second weight is 0.3, and preset third weight is 0.2 for mild frequency disturbance events; this is adjusted to 0.6, 0.3, and 0.1 for moderate frequency disturbance events; and finally to 0.7, 0.2, and 0.1 for severe frequency disturbance events, maximizing the power supply continuity of core loads. For voltage dip events, the protection effect of temperature margin is strengthened based on the corresponding frequency disturbance weight. For first-level voltage dip events, the preset second weight is increased by 0.1, and for second-level voltage dip events, it is increased by 0.15, further improving the equipment's thermal safety margin. In response to topology disturbance events, a weight configuration of 0.8 for the first weight, 0.15 for the second weight, and 0.05 for the third weight is adopted to prioritize the power supply reliability of critical loads during network reconfiguration.

[0071] After the weighting parameters are determined, the edge controller generates the dynamic priority of each load node based on the SLA parameters and real-time dynamic information obtained in step 100 through a weighting function. The dynamic priority is the sum of the preset first weight multiplied by the SLA level and the preset second weight multiplied by the temperature margin coefficient, minus the power deviation term. The power deviation term is the preset third weight multiplied by the power deviation coefficient. The temperature margin coefficient is the difference between the preset maximum allowable temperature and the real-time temperature, divided by the preset maximum allowable temperature. The power deviation coefficient is equal to the absolute value of the current power minus the preset minimum operating power, divided by the preset rated power. The SLA level is determined according to the load node's level: the preset SLA level for P1 level core IT loads is 5; the preset SLA level for P2 level high-priority cooling loads is 4; the preset SLA level for P3 level medium-priority cooling loads is 3; the preset SLA level for P4 level low-priority cooling loads is 2; and the preset SLA level for P5 level non-essential auxiliary loads is 1. The preset SLA level is determined based on load importance and business continuity requirements, obtained through system reliability analysis and business impact assessment; a higher value indicates greater load importance. The preset maximum allowable temperature is determined based on the equipment's thermal stability and safe operation requirements, obtained through the equipment manufacturer's technical specifications and on-site testing, with a default value between 24 and 27 degrees Celsius.

[0072] Load nodes are pre-classified into different levels based on equipment type and importance. These levels are P1, P2, P3, P4, and P5. P1 level represents core IT loads, including critical computing resources such as servers, network equipment, and storage devices. P2 level represents high-priority cooling loads, including precision air conditioning units and chilled water circulation pumps in data centers. P3 level represents medium-priority cooling loads, including cooling tower fans and cooling water circulation pumps. P4 level represents low-priority cooling loads, including fresh air systems and exhaust systems. P5 level represents non-essential auxiliary loads, including interruptible loads such as lighting systems, monitoring systems, and office equipment. Dynamic priority calculation results are used for fine-grained sorting within the same level to determine the specific scheduling execution order without changing the basic level classification of the load nodes.

[0073] Based on dynamic priority calculations and disturbance events, the edge controller immediately implements differentiated local resource real-time scheduling strategies for loads of different priorities. The core objective of this strategy is to achieve millisecond-level rapid response and local autonomous control at the edge controller level. By implementing differentiated emergency control over various devices within a single data center or load site, it ensures immediate protection and initial stabilization of critical loads. This priority-based hierarchical control strategy serves as the first line of defense in disturbance response, providing a stable edge controller foundation for subsequent regional cross-regional collaborative optimization steps 400. Together, they constitute a complete control system from local emergency response to regional collaborative optimization, ensuring the rational allocation of resources and priority protection of critical loads under disturbance conditions. Local resource real-time scheduling strategies include:

[0074] For P1 load, i.e., the core IT load, the highest level of protection measures are adopted, locking the UPS in double-conversion mode, and simultaneously utilizing the energy storage system to provide virtual inertia support to enhance system stability. Virtual inertia refers to the control parameter of the energy storage system simulating the rotational inertia characteristics of a traditional synchronous generator. It is used to provide inertial response support during frequency disturbances, and its value directly affects frequency stability. The energy storage state of charge (SBC) refers to the percentage of the energy storage system's current charge relative to its rated capacity, reflecting the system's available energy level and serving as a crucial constraint for the energy storage system's participation in system regulation. Virtual inertia equals the product of a preset baseline virtual inertia and the energy storage SBC adjustment coefficient. The preset baseline virtual inertia is dynamically adjusted according to the severity of the disturbance, while the energy storage SBC adjustment coefficient is determined based on the energy storage SBC. For minor frequency disturbances, when the energy storage state of charge (SBC) exceeds the preset SBC threshold, the default preset baseline virtual inertia is 8 ms / Hz. For moderate frequency disturbances, the default preset baseline virtual inertia is 10 ms / Hz. For severe frequency disturbances, the default preset baseline virtual inertia is 12 ms / Hz. The energy storage SBC adjustment coefficient is determined according to a piecewise function based on the energy storage SBC. The adjustment coefficient is 1.0 when the energy storage SBC is greater than 50%, 0.8 when the energy storage SBC is between 30% and 50%, 0.6 when the energy storage SBC is between 20% and 30%, and 0.4 when the energy storage SBC is less than 20%. Simultaneously, the virtual inertia is guaranteed to be no less than the preset minimum virtual inertia of 5 ms / Hz. The preset SBC threshold is determined based on the requirements for safe operation and regulation of the energy storage system, obtained through energy storage system technical specifications and operational experience. The default value is 30%, ensuring that the energy storage system has sufficient regulation margin. The preset minimum virtual inertia is determined based on the system frequency stability requirements and the energy storage system response characteristics. It is obtained through system simulation analysis and field testing, with a default value of 5 megawatts per hertz (MW / Hz) to ensure basic inertial support capability under severe disturbance conditions. This disturbance-aware virtual inertia adjustment mechanism ensures the power supply quality and stability of core IT loads under various disturbance conditions.

[0075] For loads P2 to P4, i.e., the cooling system, power descent control is implemented in ascending order of dynamic priority, with the descent rate strictly adhering to preset ramp-up constraints to avoid equipment damage. Specifically, P2 level cooling loads include the precision air conditioning unit and chilled water circulation pumps in the computer room; these devices directly affect the operating temperature and stability of IT equipment and have the highest cooling priority. P3 level cooling loads include cooling tower fans and cooling water circulation pumps, responsible for providing cooling support to the precision air conditioning system; their operating status affects overall cooling efficiency. P4 level cooling loads include the fresh air system and exhaust system, primarily responsible for computer room environmental regulation and air quality maintenance; they can be prioritized for adjustment in emergency situations. The change in cooling power must not exceed the cooling equipment ramp-up rate multiplied by the preset control time interval. Depending on the severity of the disturbance, different descent rates are adopted. For mild frequency disturbances, the cooling power change must not exceed 0.01 per unit; for moderate frequency disturbances, the change must not exceed 0.02 per unit; and for severe frequency disturbances, the change must not exceed 0.03 per unit. Simultaneously, real-time temperature is continuously monitored to ensure it does not exceed the preset maximum allowable temperature. The preset control time interval is determined based on response speed and control accuracy requirements, obtained through control system design and on-site debugging. The default value is 100 milliseconds, ensuring the real-time performance and stability of the control system. The cooling equipment ramp rate is determined based on the mechanical characteristics and thermal inertia of the cooling equipment, obtained through equipment technical specifications and on-site testing, to prevent equipment damage due to excessively rapid power changes. For voltage dips, the descent rate is reduced by 20% from the corresponding frequency disturbance rate to prevent further voltage deterioration. This gradual power regulation strategy ensures stability while maintaining the basic functions of the cooling system to the greatest extent possible.

[0076] For P5 loads, i.e., non-essential auxiliary loads, a tiered load shelving strategy is implemented based on the severity of the disturbance to quickly release capacity to support critical loads. For minor and moderate frequency disturbances, a preset P5 load shelving ratio is shelved; for severe frequency disturbances or voltage dips of any level, all P5 loads are shelved directly, releasing active power equal to the current power of the P5 loads. The preset P5 load shelving ratio is determined based on stability requirements and load importance analysis, obtained through system simulation verification and operational experience. The default value is 50% for minor disturbances and 80% for moderate disturbances, ensuring appropriate system capacity is released under different disturbance severity levels. For topology disturbances, P5 loads related to the fault area are prioritized for shelving, reserving operational margin for subsequent network reconfiguration. This rapid-response load shelving strategy provides crucial assurance for stable operation under severe disturbance conditions.

[0077] Through the dynamic priority calculation of disturbance perception and the priority-based differentiated control strategy described above, the edge controller realizes a coordinated response throughout the entire process from disturbance detection to control execution, ensuring the rapid stabilization of the system and reliable power supply to critical loads under various disturbance scenarios.

[0078] Step 400: Based on the real-time dynamic information after executing the local resource real-time scheduling strategy, establish a cross-regional collaborative optimization objective function, and solve the collaborative optimization objective function to obtain the cross-regional mutual assistance scheme, as detailed below. Figure 3 As shown.

[0079] Based on the real-time dynamic information received from each edge controller after completing local immediate control in step 300, the regional collaborative controller initiates a cross-regional collaborative optimization mechanism. This mechanism takes the dynamic priority matrix generated by the edge controllers in step 300 and the local control execution results, and combines this with the disturbance type characteristics identified in step 200. Through global resource coordination at the regional level and cross-connection point collaborative scheduling, it achieves a smooth transition from local emergency response to regional collaborative optimization. The core of regional cross-regional collaborative optimization lies in fully utilizing the load priority order established in step 300. Through cross-regional resource sharing and network reconfiguration, it further eliminates system deviations that still exist after local control in step 300, ensuring the optimal operating state of all multi-priority load nodes under disturbance conditions.

[0080] The regional collaborative controller first performs a comprehensive analysis and status assessment of the information uploaded by each edge controller. This information includes the dynamic priority ranking results of each load node after step 300, the actual power allocation status of loads from P1 to P5, the change in the state of charge of the energy storage system, the power regulation range of the cooling system, and the operation mode switching status of the UPS system. Simultaneously, the regional collaborative controller also needs to obtain information on the available regulation resources of each edge controller, including the remaining regulation capacity of the energy storage system, the further regulation space of the cooling system, the available output range of the reactive power compensation device (SVG), and the transmission capacity margin of each common connection point. Through in-depth analysis of this information, the regional collaborative controller can accurately grasp the execution effect of the local control in step 300 and the current resource configuration status of each node, providing a reliable decision-making basis for subsequent cross-regional collaborative optimization.

[0081] Based on the completed state assessment, the regional collaborative controller establishes a targeted optimization objective function for cross-regional collaboration, using the disturbance types identified in step 200 and real-time dynamic information after executing the local resource real-time scheduling strategy. The optimization objective function focuses on minimizing the overall deviation, comprehensively considering three key indicators: IT load power assurance, frequency stability, and voltage quality maintenance. The overall deviation is equal to the sum of the preset first weight multiplied by the load power deviation term, the preset second weight multiplied by the frequency deviation term, and the preset third weight multiplied by the voltage deviation term. Specifically, the load power deviation term is the sum of the squares of the load power deviations of multiple IT loads; the frequency deviation term is the sum of the absolute values ​​of the frequency deviations of multiple load nodes; the voltage deviation term is the sum of the squares of the voltage deviations of multiple load nodes; the load power deviation is equal to the actual power of the i-th IT load minus the minimum guaranteed power of the i-th IT load; the frequency deviation is equal to the instantaneous frequency value of the j-th load node minus the absolute value of the preset rated frequency; and the voltage deviation is equal to the voltage of the k-th load node minus the rated voltage. The design of this multi-objective optimization function fully reflects the load priority concept established in step 300, ensuring that the power supply quality of P1 level core IT loads is always at the highest level of protection.

[0082] The weight parameter configuration in the optimized objective function directly inherits the weight adjustment mechanism for disturbance perception in step 300, and is dynamically set according to the disturbance type and severity detected in step 200. The preset rated frequency is determined based on the power system frequency stability standard and obtained through frequency characteristic analysis, with a default value of 50 Hz. The preset weights are determined based on the disturbance type and operational priority detected in step 200, obtained through multi-objective optimization theory and operational experience. For frequency disturbance events, the emphasis on frequency stability in step 300 is maintained, with the second weight preset to a default value of 0.5 when frequency stability is prioritized, ensuring the priority of frequency recovery during cross-regional collaborative optimization. For voltage drop events, the guarantee of voltage quality is strengthened, with the third weight preset to a default value of 0.5 when voltage quality is prioritized, achieving rapid voltage recovery through cross-regional reactive power support. For topology disturbance events, the importance of load power supply continuity is highlighted, with the first weight preset to a default value of 0.6 when load protection is prioritized, ensuring uninterrupted power supply to critical loads through cross-regional load transfer and network reconfiguration.

[0083] The constraints for cross-regional collaborative optimization fully consider the operational boundaries and actual operational limitations after local control in step 300. Power flow constraints ensure that the power transmission of each line does not exceed safety limits; the active power flow of a line must not exceed its active power capacity limit, and the reactive power flow must not exceed its reactive power capacity limit. This constraint is particularly important during topology disturbance events, as the line capacity limit needs to be dynamically updated according to the actual topology after network reconfiguration to ensure transmission security during cross-regional power transfer. Voltage constraints ensure that the voltage quality of each node meets operational standards; the voltage of each load node should be between the preset lower voltage limit and the preset upper voltage limit, meaning that the voltage of each node must not be lower than the preset lower voltage limit and must not be higher than the preset upper voltage limit. The preset lower voltage limit and preset upper voltage limit are determined based on power quality standards and equipment safety operation requirements, obtained through power system technical regulations and equipment technical specifications. The default value for the preset lower voltage limit is 0.9 per unit, and the default value for the preset upper voltage limit is 1.05 per unit. In response to voltage dip events, the preset lower voltage limit is temporarily adjusted to 0.85 per unit, providing greater operational flexibility for recovery. This dynamic adjustment mechanism is consistent with the disturbance response strategy in step 300. Switching operation constraints ensure the safety and reliability of the network reconfiguration process, and the switching sequence of tie switches satisfies the no-circulating current principle. For topology disturbance events, switching operation timing constraints are added to ensure power supply continuity during network reconfiguration and avoid secondary disturbances caused by improper operation.

[0084] After establishing the objective function and setting constraints, the regional collaborative controller employs a distributed optimization algorithm to solve for the inter-regional mutual assistance scheme. The inter-regional mutual assistance scheme refers to a comprehensive scheduling strategy within a distribution network / microgrid area interconnected by multiple common connection points, achieving load power supply guarantee and stable operation through resource coordination and power mutual support among different edge controllers. The inter-regional mutual assistance scheme includes core elements such as active power transfer across common connection points, reactive power collaborative regulation schemes, inter-regional collaborative charging and discharging strategies for energy storage systems, timing arrangements for tie switch operations, and load transfer path planning. The organic combination of these elements forms a complete regional resource optimization allocation system. The solution process fully considers the control execution results and resource consumption of each edge controller in step 300, achieving optimal resource allocation within the region through iterative optimization. For different disturbance types, the optimization solution strategy exhibits significant differences, which are highly consistent with the disturbance-aware control concept in step 300. In response to frequency disturbance events, the optimization solution prioritizes adjusting the active power balance. Through the coordinated adjustment of active power transfer across common junctions and cross-regional coordination of energy storage systems, frequency stability is quickly restored. At the same time, the virtual inertia support effect provided by the energy storage system in step 300 is fully utilized.

[0085] Cross-point-of-combination (CPC) active power transfer refers to the transfer of active power from power-surplus areas to power-deficient areas via interconnecting lines between multiple CPCs, achieving dynamic balance and mutual support of active power between regions. The implementation of CPC active power transfer fully considers the power surplus / deficit situation of each edge controller after local real-time control in step 300 and the capacity constraints of transmission lines, determining the optimal power transfer path and transfer power amount through optimization calculations. The active power transfer process strictly follows the power system flow distribution law and network security constraints, ensuring that the transferred power does not exceed the active power capacity limit of the interconnecting lines, while avoiding voltage exceedances and equipment overloads caused by power transfer. For frequency disturbance events, CPC active power transfer prioritizes providing active power support to areas with larger frequency deviations, achieving coordinated frequency recovery through rapid power redistribution. For topology disturbance events, active power transfer coordinates with network reconfiguration operations, ensuring the power supply continuity of critical loads by changing the power transmission path, especially guaranteeing uninterrupted power supply to P1-level core IT loads during network topology changes.

[0086] The cross-regional coordinated charging and discharging strategy for energy storage systems refers to the overall optimization of active power balance and the coordinated maintenance of frequency stability by uniformly coordinating the charging and discharging power allocation and active power transmission across the point of common junction within the jurisdiction of multiple edge controllers. The formulation of this strategy fully considers the state of charge and active power regulation capabilities of the energy storage systems after local real-time control by each edge controller in step 300. Optimal calculations determine the optimal charging and discharging power allocation for each energy storage system and the optimal path for cross-regional active power transmission.

[0087] The cross-regional collaborative charging and discharging process of the energy storage system strictly adheres to the operational constraints of the energy storage system and the power balance principle of the power system. It prioritizes the use of energy storage systems with higher states of charge for discharge regulation. When energy storage resources in a single region are insufficient, mutual support between regional energy storage resources is achieved through active power transfer across points of common coupling. In response to frequency disturbance events, the cross-regional collaborative charging and discharging strategy of the energy storage system responds quickly to frequency deviation signals by increasing the discharge power of the energy storage system and decreasing the charging power, thereby rapidly restoring the frequency to the normal operating range. In response to voltage dip events, the cross-regional collaborative charging and discharging strategy of the energy storage system, in conjunction with reactive power regulation, provides a stable power base for voltage regulation by maintaining active power balance, ensuring the stable operation of the reactive power compensation device (SVG) during voltage regulation. In response to topology disturbance events, the cross-regional collaborative charging and discharging strategy of the energy storage system, in conjunction with network reconfiguration operations, adapts to network topology changes by adjusting the charging and discharging power allocation of the energy storage system, ensuring the power supply continuity of critical loads during network reconfiguration, especially guaranteeing the uninterrupted power supply needs of P1-level core IT loads.

[0088] To address voltage dips, the optimization solution prioritizes reactive power support adjustment. Through the coordinated output of reactive power compensation devices (SVG) and cross-regional reactive power coordination, rapid voltage recovery is achieved, effectively complementing the cooling system power adjustment in step 300. The reactive power coordination scheme refers to the unified coordination of the output allocation of reactive power compensation devices (SVG) and the transmission of reactive power across the point of common coupling (PCC) within the jurisdiction of multiple edge controllers, thereby achieving overall optimization of system voltage quality and coordinated maintenance of voltage stability. The reactive power coordination scheme fully considers the reactive power surplus / deficit and voltage support requirements of each edge controller after local real-time control in step 300. Optimal calculations determine the optimal output allocation of each reactive power compensation device (SVG) and the optimal path for cross-regional reactive power transmission. The reactive power coordination process strictly adheres to the principle of local reactive power balance and voltage regulation characteristics of the power system, prioritizing the use of local reactive power resources for voltage regulation. When local reactive power resources are insufficient, cross-PCC reactive power transfer is used to achieve mutual support of reactive power between regions. In response to voltage dips, the reactive power coordinated regulation scheme responds quickly to voltage deviation signals by increasing the capacitive output of the SVG (Static Var Generator) and reducing reactive power consumption by inductive loads, thereby rapidly restoring the voltage to its normal operating range. In response to frequency disturbances, the reactive power coordinated regulation scheme works in conjunction with active power regulation to maintain system voltage stability, providing a favorable operating environment for active power regulation and ensuring the stable operation of the energy storage and cooling systems during frequency regulation.

[0089] In response to topology disturbance events, the optimization solution prioritizes load transfer and network reconfiguration. Through tie switch operation timing and load transfer path planning, it ensures the continuity of power supply to critical loads and minimizes the impact of P5 load shedding in step 300 on system operation. Tie switch operation timing refers to the rational arrangement of the opening and closing sequence and time intervals of each tie switch during network topology reconfiguration to achieve a safe and smooth transition of the network topology and continuous power supply to the loads. The tie switch operation timing is formulated by fully considering the network topology state and load power supply requirements after the local real-time control of each edge controller in step 300, and the optimal operation sequence and time intervals of each tie switch are determined through optimization calculations.

[0090] The operation of the tie switch strictly adheres to the power system's non-circulating current operation principle and equipment safety operating procedures to ensure that the switch operation does not generate circulating current impacts or equipment overloads, while also avoiding secondary disturbances and power outages caused by improper operation. In response to topology disturbance events, the tie switch operation sequence prioritizes ensuring the power supply continuity of P1-level core IT loads, achieving seamless load transfer through a "close first, then disconnect" operation strategy, ensuring uninterrupted power supply to critical loads during network reconfiguration. In response to frequency disturbance events, the tie switch operation sequence coordinates with active power regulation actions, optimizing power transmission paths by adjusting the network topology, providing better network support for power regulation of energy storage and cooling systems.

[0091] In response to voltage dip events, the timing of tie switch operations is coordinated with reactive power regulation actions to improve voltage distribution characteristics by altering the network topology, providing better network conditions for the voltage regulation of the SVG (Static Var Generator). Load transfer path planning refers to the rational planning of transfer paths and sequences for loads of different levels during network topology reconfiguration, ensuring a smooth transition of power supply and continuous power quality. The formulation of load transfer path planning fully considers the load distribution status and power supply capacity margin of each edge controller after local real-time control in step 300, determining the optimal transfer path and sequence for each load level through optimization calculations. The load transfer process strictly adheres to load priority principles and power supply capacity constraints, prioritizing the continuity of power supply to high-priority loads. When power supply capacity is insufficient, loads are cut off in reverse order of P5, P4, P3, P2, P1 to ensure the power supply safety of core loads. In response to topology disturbance events, load transfer path planning prioritizes the transfer path of P1-level core IT loads, ensuring reliable power supply to core loads under any network topology change through multi-path redundancy design.

[0092] For frequency disturbance events, load transfer path planning, combined with cross-regional coordinated charging and discharging strategies for energy storage systems, reduces the regulation pressure on energy storage systems by optimizing load distribution, providing a better load foundation for frequency recovery. For voltage dip events, load transfer path planning, combined with reactive power coordinated regulation schemes, improves voltage distribution by adjusting load distribution, providing a better load configuration for voltage recovery.

[0093] After the cross-regional mutual assistance solution is completed, the regional collaborative controller immediately issues collaborative control commands to the relevant edge controllers. These commands include specific values ​​of power transferred across the point of common coupling, output adjustment commands for the reactive power compensation device (SVG), charging and discharging power adjustment commands for the energy storage system, and necessary handover switch operation sequences. The command issuance process adopts a priority transmission mechanism to ensure that control commands related to P1-level core IT loads are transmitted and executed first, reflecting the continuation of the load priority order established in step 300 at the regional collaborative level. After receiving the collaborative control commands, each edge controller, while maintaining the local control effect of step 300, executes the corresponding cross-regional collaborative adjustment actions, achieving a seamless transition from local optimization to regional collaborative optimization, and ensuring the coordinated and stable operation of the entire multi-priority load node under disturbance conditions.

[0094] Step 500: Based on the execution results of the cross-regional mutual assistance scheme, dynamically adjust the target value of the energy storage state of charge and the power plan of the cooling system.

[0095] After completing step 400 of cross-regional collaborative optimization, the regional collaborative controller immediately uploads the second-level scheduling execution results and status change information to the upper-level energy optimization controller. This key information includes the state of charge consumption of each edge controller energy storage system during the virtual inertia support process, the actual operating power baseline value of the P2 to P4 level cooling systems after power descent control, the amount of unnecessary auxiliary loads cut off at the P5 level, and the power transmission changes of each common connection point during cross-regional mutual assistance.

[0096] After receiving real-time dynamic information, the upper-level energy optimization controller initiates minute-level and hour-level long-term energy balance optimization mechanisms based on the cross-regional collaborative optimization results established in step 400, ensuring a smooth transition from second-level disturbance response to long-term stable operation.

[0097] The upper-level energy optimization controller first dynamically adjusts the target trajectory of the energy storage state of charge (SOC) based on the actual energy consumption of the energy storage system participating in virtual inertia support in step 400. The SOC update mechanism fully considers the energy consumption characteristics of the energy storage system when providing virtual inertia support in step 300 and the power demand changes of cross-regional coordinated regulation in step 400. The target value of the energy storage SOC at the next moment is determined by subtracting the SOC consumption during the execution of step 400 from the current energy storage SOC, and adding the product of the energy storage charging power and the preset optimization time interval. The SOC consumption refers to the percentage of electricity actually consumed by the energy storage system during the disturbance response process relative to its rated capacity. This consumption includes the discharge consumption of the energy storage system when providing virtual inertia support in step 300, the net discharge consumption when participating in cross-regional coordinated charging and discharging regulation in step 400, and additional consumption such as internal losses of the energy storage system and converter losses.

[0098] The calculation of state of charge (SCC) consumption fully considers the power output characteristics and duration differences of the energy storage system under different disturbance severity levels. SCC consumption under mild frequency disturbances is typically 2% to 5% of the rated capacity, under moderate frequency disturbances it is 5% to 10%, and under severe frequency disturbances it can reach 10% to 20%. Accurate calculation of SCC consumption provides crucial information for energy management and long-term operation planning of the energy storage system, ensuring that the system still has sufficient adjustment margin to cope with subsequent disturbances after participating in disturbance response. The preset optimization time interval is determined based on the energy management system's scheduling cycle and energy storage response characteristics, obtained through system operation experience and optimization algorithm analysis; the default value is 15 minutes. This SCC target update mechanism based on actual consumption ensures that the energy storage system can promptly recover to a reasonable energy level after participating in disturbance response, reserving sufficient adjustment margin for subsequent disturbances.

[0099] Based on the updated energy storage state of charge target, the upper-level energy optimization controller synchronously adjusts the long-term power operation plan of the cooling system. This plan fully incorporates the execution results of the cooling system power descent control in step 400, and avoids drastic fluctuations in system temperature through a gradual power recovery strategy. The cooling system power plan update process strictly follows the cooling equipment ramp-up rate constraints and preset maximum allowable temperature limits established in step 300, ensuring equipment safety during the power recovery process for P2-level high-priority cooling loads, P3-level medium-priority cooling loads, and P4-level low-priority cooling loads.

[0100] The current cooling system power plan is determined by summing the products of the cooling power baseline value after step 400, the preset recovery ramp rate, and the preset optimization time interval. The cooling power baseline value refers to the actual operating power level of the cooling system after the cross-zone collaborative optimization is completed in step 400. This baseline value includes the stable operating power of the P2 level high-priority cooling load after power descent control, the actual output power of the P3 level medium-priority cooling load after collaborative adjustment, and the final power state of the P4 level low-priority cooling load during the disturbance response process. The determination of the cooling power baseline value fully considers the power adjustment range and equipment operating boundaries of each level of cooling load during the disturbance response process. The power baseline value for the P2 level cooling load is typically 70% to 90% of its rated power, the power baseline value for the P3 level cooling load is 50% to 80% of its rated power, and the power baseline value for the P4 level cooling load is 30% to 70% of its rated power.

[0101] The cooling power baseline value serves as the starting reference point for long-term power operation planning, providing accurate basic data for the gradual power recovery of the cooling system. This ensures that the power recovery process not only meets the equipment's safe operation requirements but also gradually restores the cooling system's full support capability for P1-level core IT loads. The preset recovery ramp rate is determined based on the equipment's thermal inertia and temperature control requirements, obtained through thermodynamic analysis and on-site testing, to avoid temperature rebound during the power recovery process.

[0102] After updating the energy storage state-of-charge target and cooling system power plan, the upper-level energy optimization controller generates a comprehensive rolling energy optimization plan and feeds it back to the regional coordination controller. This optimization plan includes not only the charging and discharging power scheduling instructions for the energy storage system and the power recovery timing arrangement for the cooling system, but also the gradual recovery strategy for P5 level non-essential auxiliary loads and the long-term balancing scheme for power transmission at each point of common coupling. Upon receiving the rolling energy optimization plan, the regional coordination controller immediately translates it into specific coordinated control instructions and issues them to each edge controller for execution.

[0103] These coordinated control commands include long-term charging and discharging power setpoints for the energy storage system, phased target values ​​for cooling system power recovery, timing schedules for the activation of P5-level non-essential auxiliary loads, and adjustment commands for power transmission at each common coupling point, ensuring the precise execution of long-term energy optimization plans at the edge controller level. Through this multi-level energy optimization coordination mechanism, the system achieves complete timescale coverage from millisecond-level local real-time control in step 300, second-level regional cross-regional coordinated optimization in step 400, to minute-level long-term energy balance in step 500, ensuring comprehensive coordinated response and long-term stable operation of the multi-priority load node system under disturbance conditions.

[0104] In step 600, during the execution of the long-term energy optimization plan in step 500, the edge controller and the regional collaborative controller simultaneously initiate a comprehensive system status monitoring and closed-loop adjustment mechanism. This mechanism, as the final link in the entire multi-priority load node dynamic hierarchical scheduling method, ensures the robustness and reliability of the system under complex disturbance environments by continuously monitoring and dynamically correcting the control execution effects at each stage from steps 300 to 500. The system status monitoring and closed-loop adjustment mechanism encompasses real-time tracking of equipment operating status, continuous evaluation of communication network status, and cyclical verification of the disturbance recovery process, forming a multi-dimensional monitoring system from the local equipment level to the regional network level, providing comprehensive protection for the system's adaptive adjustment and fault prevention.

[0105] While executing the energy optimization plan in step 500, the edge controller continuously monitors the operational status changes of various key devices, paying particular attention to the operational boundaries of the UPS system and energy storage system, which perform core protection functions in step 300. When the UPS load rate exceeds the preset UPS load rate threshold, the edge controller immediately determines that the P1-level core IT load faces a power supply risk and automatically triggers the backup energy storage system to operate, alleviating the load pressure on the UPS system by increasing the discharge power of the energy storage system. The backup energy storage power is set according to the system safety margin and load guarantee requirements, with an increase of 0.2 per unit value to ensure that the power supply continuity of the P1-level core IT load is not affected. At the same time, the edge controller closely monitors the temperature changes of the cooling system where power derating control is implemented in step 300. When the cooling temperature exceeds the preset maximum allowable temperature, the ongoing power derating operation is immediately suspended, and the cooling system power is restored to a safe operating level of 0.1 per unit value above the minimum operating power, preventing equipment damage due to overheating and ensuring the cooling system's continuous support capability for the P1-level core IT load.

[0106] Based on monitoring the status of each edge controller device, the regional collaborative controller focuses on the reliability of the communication network and the quality of data transmission during the cross-regional collaborative optimization process in step 400. The communication status monitoring mechanism assesses the communication network's support level for collaborative control capabilities by tracking key indicators such as data packet loss rate, transmission latency, and signal strength between each edge controller and the regional collaborative controller in real time. When the data packet loss rate exceeds a preset threshold, the regional collaborative controller determines that the communication network has suffered severe degradation and immediately issues an autonomous switching command to the relevant edge controllers, activating the pre-stored communication degradation strategy library. The communication degradation strategy library contains local autonomous control schemes for different communication failure scenarios, ensuring that the edge controllers can maintain the local real-time control functions established in step 300 even under communication constraints, avoiding the loss of system control capabilities due to communication failures.

[0107] The core of the system status monitoring and closed-loop adjustment mechanism lies in the cyclic verification and dynamic optimization of the disturbance recovery process. Every second, the regional collaborative controller performs a comprehensive verification of key operating indicators, including the degree to which the instantaneous frequency values ​​of each node deviate from the rated frequency, the deviation of the voltage levels of each node from the rated voltage, and the margin between the temperature of each edge controller device and the preset maximum allowable temperature. The cyclic verification process is strictly executed according to the disturbance detection criterion system established in step 200 to ensure the accuracy and consistency of the disturbance recovery assessment. When the absolute value of the instantaneous frequency value deviating from the rated frequency is less than the preset frequency recovery threshold and the voltage of each node is not less than the preset voltage recovery threshold, the disturbance event is determined to be effectively controlled, the disturbance response process officially ends, and the system transitions to steady-state monitoring mode. In steady-state monitoring mode, the long-term energy optimization plan in step 500 continues to be executed, gradually restoring the normal operation of P5 level non-essential auxiliary loads, achieving a smooth transition from the disturbance response state to the normal operation state.

[0108] The preset UPS load rate threshold is determined based on UPS safe operation requirements, obtained through UPS technical specifications and safety margin analysis, with a default value of 110%, ensuring reliable operation of the UPS system under load fluctuation conditions. The preset data packet loss rate threshold is determined based on communication system reliability requirements, obtained through communication network performance testing and reliability analysis, with a default value of 30%, ensuring communication quality for regional collaborative control. The preset frequency recovery threshold is determined based on power system frequency stability standards, obtained through system frequency characteristic analysis, with a default value of 0.2 Hz, consistent with the judgment criteria for minor frequency disturbance events in step 200. The preset voltage recovery threshold is determined based on power quality standards and equipment normal operation requirements, obtained through power system technical regulations, with a default value of 0.95 per unit, ensuring that system voltage quality meets the normal operation requirements of various loads. Through this cyclic verification mechanism based on multiple threshold criteria, precise control and dynamic optimization of the entire disturbance response process are achieved, ensuring the effectiveness and robustness of the multi-priority load node dynamic hierarchical scheduling method under various complex disturbance scenarios.

[0109] Example 2

[0110] See Figure 4 As shown, a dynamic hierarchical scheduling system for multi-priority load nodes is provided. This system stores computer-readable instructions, which, when read, can execute the aforementioned dynamic hierarchical scheduling method for multi-priority load nodes. The system includes:

[0111] The data acquisition module 101 acquires real-time dynamic information and pre-configured static information of the load nodes;

[0112] The disturbance detection module 102 identifies disturbance events based on the real-time dynamic information using a multi-disturbance detection model.

[0113] The local control module 103 calculates the dynamic priority of the load nodes and executes the local resource real-time scheduling strategy based on the disturbance event and pre-configured static information.

[0114] The regional collaboration module 104 establishes a cross-regional collaborative optimization objective function based on real-time dynamic information after executing the local resource real-time scheduling strategy, and solves the collaborative optimization objective function to obtain a cross-regional mutual assistance scheme.

[0115] The energy optimization module 105 dynamically adjusts the target value of the energy storage state of charge and the power plan of the cooling system based on the execution results of the cross-regional mutual assistance scheme.

[0116] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A method for dynamic hierarchical scheduling of multi-priority load nodes, characterized in that, The method comprises the following steps: obtaining real-time dynamic information and pre-configured static information of the load node; based on the real-time dynamic information, identifying a disturbance event through a multi-disturbance detection model; the disturbance event includes a frequency disturbance event and a voltage sag event, wherein: when the frequency change rate is greater than a preset frequency change rate threshold or the absolute value of the difference between the frequency instantaneous value and the rated frequency is greater than a preset frequency deviation threshold, it is determined as a frequency disturbance event; the frequency disturbance event includes a mild frequency disturbance, a moderate frequency disturbance and a severe frequency disturbance, and is determined based on the absolute value of the difference between the frequency instantaneous value and the rated frequency; when the voltage sag depth exceeds a preset voltage sag threshold and the duration exceeds a preset duration threshold, it is determined as a voltage sag event; the voltage sag depth is the difference between the rated voltage and the real-time voltage divided by the rated voltage, and the voltage sag event includes a first-level voltage sag and a second-level voltage sag event, and is determined based on the voltage sag depth; based on the disturbance event and the pre-configured static information, calculating the dynamic priority of the load node and executing a local resource real-time scheduling strategy; comprising: based on the disturbance event, adjusting the weight parameters of the dynamic priority calculation, the weight parameters including a preset first weight, a preset second weight and a preset third weight; calculating the dynamic priority of the load node, the dynamic priority being the sum of the preset first weight multiplied by the SLA level and the preset second weight multiplied by the temperature margin coefficient, and then subtracting the power deviation term, the power deviation term being the preset third weight multiplied by the power deviation coefficient; wherein the temperature margin coefficient is the difference between the preset maximum allowed temperature and the real-time temperature divided by the preset maximum allowed temperature, and the power deviation coefficient is equal to the absolute value of the difference between the current power and the minimum operating power divided by the rated power; the SLA level is determined according to the level of the load node, and the level of the load node includes P1 level, P2 level, P3 level, P4 level and P5 level; based on the dynamic priority and the disturbance event, implementing a local resource real-time scheduling strategy for load nodes of different levels; based on the real-time dynamic information after executing the local resource real-time scheduling strategy, establishing a cross-region collaborative optimization objective function, and solving the collaborative optimization objective function to obtain a cross-region mutual aid scheme; based on the execution result of the cross-region mutual aid scheme, dynamically adjusting the energy storage state of charge target value and the cooling system power plan.

2. The method of claim 1, wherein, The local resource real-time scheduling strategy comprises: for the P1 level IT load, locking the UPS control mode to the double conversion mode, and calling the energy storage system to provide virtual inertia, the virtual inertia being the product of a preset reference virtual inertia coefficient and an energy storage state of charge adjustment coefficient; for the P2 to P4 level cooling load, implementing power ramp-down control in the order of priority from low to high, the cooling power change amount not being allowed to exceed the climbing rate of the cooling equipment multiplied by a preset control time interval, while continuously monitoring the real-time temperature to ensure that it does not exceed the preset maximum allowed temperature; For non-essential auxiliary load of P5 level, hierarchical cut-off strategy is implemented according to disturbance severity, for mild frequency disturbance event and moderate frequency disturbance event, P5 level load of preset P5 load cut-off ratio is cut off, for severe frequency disturbance event or voltage drop event of any level, all P5 level load is cut off.

3. The method of claim 1, wherein, The optimization objective function is to minimize the comprehensive deviation, which is the sum of the preset first weight multiplied by the load power deviation term, the preset second weight multiplied by the frequency deviation term, and the preset third weight multiplied by the voltage deviation term; The constraint conditions include power flow constraints and voltage constraints; The cross-regional interconnection scheme is obtained by solving the optimization objective function using a distributed optimization algorithm based on the constraint conditions, and the cross-regional interconnection scheme includes cross-point active power transfer, reactive power collaborative regulation scheme, energy storage system cross-regional collaborative charging and discharging strategy, and tie switch operation time sequence arrangement.

4. The method of claim 1, wherein, The load power deviation term is the cumulative sum of the squares of the load power deviations of the plurality of IT loads, the frequency deviation term is the cumulative sum of the absolute values of the frequency deviations of the plurality of load nodes, and the voltage deviation term is the cumulative sum of the squares of the voltage deviations of the plurality of load nodes. The load power deviation is the actual power of the ith IT load minus the minimum guaranteed power of the ith IT load, the frequency deviation is the absolute value of the frequency instantaneous value of the jth load node minus the preset rated frequency, and the voltage deviation is the voltage of the kth load node minus the rated voltage.

5. The method of claim 4, wherein, The power flow constraint is that the line active power flow is less than the line active power capacity limit, and the line reactive power flow is less than the line reactive power capacity limit. The voltage constraint is that the voltage of the load node should be between the preset lower voltage limit and the preset upper voltage limit.

6. The method of claim 1, wherein, The energy storage state of charge target value at the next time is the current energy storage state of charge minus the state of charge consumption, plus the product of the energy storage charging power and the preset optimization time interval; The cooling system power plan at the current time is the sum of the cooling power reference value, the preset recovery ramp rate, and the preset optimization time interval.

7. The method of claim 1, wherein, The real-time dynamic information includes electrical parameter data and equipment state data; wherein, the electrical parameter data includes the frequency instantaneous value, the frequency change rate, the three-phase voltage, the line active power flow and the reactive power flow of each load node, and the equipment state data includes the load power, the real-time temperature, the UPS control mode, the energy storage state of charge and the charging and discharging power of the IT load; The preconfigured static information includes load constraints and SLA parameters; wherein, the load constraint parameters include the minimum guaranteed power of the IT load, the cooling device ramp rate, and the minimum operating power, and the SLA parameters include the SLA level of each load node.

8. A dynamic hierarchical scheduling system for multi-priority load nodes, characterized by, It is used for storing computer readable instructions, which can execute the dynamic hierarchical scheduling method of multi-priority load nodes as claimed in any one of claims 1-7 when read; the system comprises: A data acquisition module acquires real-time dynamic information and preconfigured static information of the load node; A disturbance detection module identifies a disturbance event through a multi-disturbance detection model based on the real-time dynamic information; A local control module calculates a dynamic priority of the load node and executes a local resource immediate scheduling strategy based on the disturbance event and the preconfigured static information. The regional coordination module establishes a cross-region coordination optimization objective function based on real-time dynamic information after executing the local resource instant scheduling strategy, and solves the coordination optimization objective function to obtain a cross-region mutual aid scheme; The energy optimization module dynamically adjusts the storage energy state of charge target value and the cooling system power plan based on the execution result of the cross-region mutual aid scheme.

Images