A data flow-energy flow coupled scheduling method and device for integrated energy systems

Abstract

This application discloses a data flow-energy flow coupled scheduling method and device for integrated energy systems, relating to the field of intelligent scheduling technology for integrated energy systems. The method is applied to an integrated energy system for an internet data center. The method includes: constructing an integrated energy system model for the internet data center; based on the integrated energy system model, constructing a multi-objective optimization function and constraints; based on the constraints, using a quantum-enhanced multi-objective gray wolf optimization algorithm to obtain a set of non-dominated solutions to the multi-objective optimization function, obtaining the Pareto optimal front of the optimization objective; for the non-dominated solutions, using a membership function to obtain the global optimal solution as the decision basis for the data flow-energy flow coupled scheduling of the integrated energy system for the internet data center. This application enhances the resilience of the integrated energy system for internet data centers under disaster conditions.

Description

Technical Field

[0001] This application relates to the field of intelligent scheduling technology for integrated energy systems, and in particular to a data flow-energy flow coupled scheduling method and device for integrated energy systems. Background Technology

[0002] With the surge in demand for data services, internet data centers are increasingly becoming a huge electricity load, projected to account for 8% of global electricity consumption by 2030. Meanwhile, the waste heat generated during data processing in internet data centers can be captured through advanced heat recovery technologies. Therefore, from the perspective of integrated energy systems, internet data centers can be considered as prosumers, enabling coordinated scheduling of data and energy flows and improving overall energy efficiency.

[0003] However, in recent years, the frequency and severity of natural disasters have increased due to the impacts of climate change and human activities. Disasters can lead to power outages and subsequent data loss. In 2011, the earthquake and tsunami in eastern Japan reduced Tokyo Electric Power Company's power supply by 21 GW and caused 30% of customers to lose access to the Sendai Internet Data Center. In 2012, Hurricane Sandy caused power outages for more than 8 million customers in 21 states in New York, as well as temporary service disruptions and permanent loss of user data for many Internet Data Centers. Therefore, solutions to improve the resilience of the integrated energy systems of Internet Data Centers must be explored. Summary of the Invention

[0004] The purpose of this application is to provide a data flow-energy flow coupled scheduling method and device for integrated energy systems, which can enhance the resilience of integrated energy systems for Internet data centers under disaster conditions.

[0005] To achieve the above objectives, this application provides the following solution:

[0006] In a first aspect, this application provides a data flow-energy flow coupled scheduling method for an integrated energy system, characterized in that the method is applied to an integrated energy system for an Internet data center, which integrates an Internet data center, photovoltaic, wind turbines and cogeneration units;

[0007] The integrated energy system data flow-energy flow coupled scheduling method includes:

[0008] Constructing an integrated energy system model for internet data centers; the integrated energy system model for internet data centers includes: an energy consumption model for internet data centers, a cross-seasonal thermal storage model for integrated energy systems for internet data centers, and a data flow-energy flow coordination mechanism.

[0009] Based on the aforementioned integrated energy system model for Internet data centers, a multi-objective optimization function and constraints are constructed.

[0010] Based on the aforementioned constraints, a set of non-dominated solutions to the multi-objective optimization function is obtained using the quantum-enhanced multi-objective gray wolf optimization algorithm, thus obtaining the Pareto optimal front of the optimization objective. For the non-dominated solutions, the membership function is used to obtain the global optimal solution as the decision basis for the data flow-energy flow coupled scheduling of the Internet data center integrated energy system.

[0011] Secondly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described integrated energy system data flow-energy flow coupled scheduling method.

[0012] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0013] This application provides a data flow-energy flow coupled scheduling method and device for integrated energy systems. It constructs a model combining the energy consumption model and the cross-seasonal thermal storage model of an Internet data center, thereby establishing an integrated energy system model for an Internet data center with a data flow-energy flow collaborative mechanism. The proposed strategy aims to minimize workload loss and scheduling costs, electrical and thermal load loss costs of the integrated energy system for the Internet data center, and maintenance personnel scheduling costs. Furthermore, a quantum-enhanced multi-objective gray wolf optimization algorithm is designed to solve this strategy, thereby enhancing the resilience of the integrated energy system for the Internet data center under disaster conditions.

[0014] Furthermore, this application constructs a multi-dimensional resilience evaluation index for integrated energy systems of Internet Data Centers (IDC-IES). Based on Boolean relations, it quantitatively assesses the resilience of IDC-IES from four dimensions: robustness, speed, resilience, and survivability. The structure of IDC-IES, the energy consumption model of IDC, the CSHS model, and the data flow-energy flow coordination mechanism are introduced. A resilience-oriented IDC-IES data flow-energy flow coupled scheduling strategy considering CSHS is proposed. This strategy minimizes workload loss and total workload scheduling cost, IDC-IES electrical and thermal load loss costs, and maintenance personnel scheduling costs through multi-objective optimization. A quantum-enhanced multi-objective gray wolf optimization algorithm is designed, introducing quantum computing theory to enhance population diversity and global search capabilities. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of a data flow-energy flow coupled scheduling method for an integrated energy system according to one embodiment of this application;

[0017] Figure 2 This is a performance curve of the system encountering a disaster in one embodiment of this application;

[0018] Figure 3 This is a graph showing the annual thermal energy storage variation of CSHS in one embodiment of this application;

[0019] Figure 4 This is a raw workload arrival rate graph in one embodiment of this application, assuming no disaster occurs in the IDC.

[0020] Figure 5 This is a workload arrival rate graph that ignores workload scheduling in one embodiment of this application;

[0021] Figure 6 This is a workload arrival rate graph considering workload scheduling in one embodiment of this application;

[0022] Figure 7 This is a Sankey diagram of data-energy flow distribution in different scenarios in one embodiment of this application;

[0023] Figure 8 This is a heat map showing the changes in workload, energy consumption, and waste heat in Example 4 of one embodiment of this application;

[0024] Figure 9 This is a comparison chart of the convergence curves of the algorithm in one embodiment of this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] In one exemplary embodiment, such as Figure 1 As shown, a data flow-energy flow coupled scheduling method for an integrated energy system is provided. The method is applied to an integrated energy system for an Internet data center, which integrates an Internet data center, photovoltaic, wind turbines and cogeneration units.

[0028] Integrated energy system data flow-energy flow coupled scheduling method, including:

[0029] Determine the multidimensional resilience assessment indicators for Internet Data Center-Integrated Energy Systems (IDC-IES). Quantify these indicators into Boolean relationships between robustness, speed, resilience, and survivability.

[0030] RQAM=R∨(S∧E)∨D.

[0031] RQAM is a multidimensional resilience assessment index for the system, where R represents robustness, S represents speed, E represents recovery, and D represents survivability. IDC-IES undergoes normalization, damage, degradation, and recovery phases during a disaster. Robustness reflects the ability to withstand the impact of a disaster while maintaining high-load energy supply levels. Systems with lower R exhibit more severe damage during the damage phase. During the degradation phase, IDC-IES have the lowest operational performance. The system behaves like it's in "idle mode," without any mitigation measures. Systems with higher S experience shorter degradation phases. During the recovery phase, the performance of IDC-IES gradually recovers to its original state or a new steady state. Systems with higher E recover faster. D reflects the ability of the power supply network to maintain connectivity throughout the disaster.

[0032] Since R, S, E, and D have different dimensions, RQAM is reformulated based on min-max normalization as follows:

[0033] RQAM = R norm +S norm ·E norm +D norm .

[0034] Among them, R norm This is the normalized value of the robustness index. S norm • This is the normalized value for the speed index. E norm This is the normalized value of the restorative index. (D) norm This is the normalized value of the survivability index.

[0035] (1) Robustness.

[0036] R represents the percentage of remaining load in IDC-IES, indicating the system's ability to withstand extreme natural disasters.

[0037]

[0038] R is the robustness index. P represents the frequency of occurrence of fault state i. N represents the total number of extracted scenarios. e This represents the electrical load under normal conditions. P h This represents the heat load under normal conditions. This represents the maximum loss of the electrical load. This represents the maximum loss of heat load.

[0039] (2) Speed.

[0040] S can be expressed as the complement of the ratio of the duration of the system operating at its lowest performance to the time from the occurrence of a disaster to recovery to a new steady state:

[0041]

[0042] S is the speed indicator. t1 is the time when the system performance begins to degrade. t2 is the time when the system performance reaches its lowest point. t3 is the time when the system performance begins to recover. final This is the start time of the stable state after recovery.

[0043] (3) Restorative.

[0044] E represents the recovery rate and level of IDC-IES. Considering the average rate of load recovery and the energy supply level achieved after recovery to a new steady state, recoverability is described as:

[0045]

[0046] E is a recovery indicator. P final The start time t of the stable state after recovery final The load level is P3, which is the load level at time t3 when system performance begins to recover. This refers to the electrical load level reached after the system recovers. This represents the heat load level that the system will reach after recovery.

[0047] (4) Damage resistance.

[0048] The network connectivity of IDC-IES reflects its resilience and can characterize the extent of damage and recovery difficulty caused by extreme disasters. Using the principles of complex network theory, the IDC-IES network can be defined as a weighted directed graph G consisting of vertices and edges. The edge weight adjacency matrix A of G is calculated using the following formula:

[0049]

[0050] The network redundancy of IDC-IES is:

[0051]

[0052] The resilience of a network is quantified by the probability that the network remains connected after a random failure.

[0053]

[0054] D is the resilience metric, used to describe the network connectivity of an integrated power system for an internet data center. d is the network redundancy of the integrated power system for an internet data center. max λ represents the redundancy of the initial network state. i Let A be the i-th eigenvalue of the adjacency matrix A of the weighted directed graph G, which is constructed based on an integrated energy system for an internet data center. n represents the total number of nodes in the system. ij ω is the element in the i-th row and j-th column of the adjacency matrix A of the weighted directed graph G; ij Let be the weight of the direct edge between node i and node j.

[0055] Multidimensional resilience assessment metrics not only serve as the design basis for disaster response strategies but also act as a core link throughout the entire process of model building, algorithm optimization, and performance verification. The four dimensions of resilience characteristics are not independent but interact with each other. Robustness determines the initial state of the system when it enters the degradation phase; speed affects the duration of the performance degradation phase; recoverability determines the efficiency of establishing a new steady state; and survivability provides structural guarantees for the entire process. By comprehensively quantifying resilience requirements across multiple dimensions, abstract resilience is transformed into concrete optimization objectives and constraints, ultimately achieving efficient and stable operation of IDC-IES under extreme conditions.

[0056] Step 101: Construct an integrated energy system model for internet data centers. The integrated energy system model for internet data centers includes: an internet data center energy consumption model, a cross-seasonal thermal storage model for the integrated energy system of internet data centers, and a data flow-energy flow coordination mechanism.

[0057] (1) Structure of integrated energy system for Internet data centers.

[0058] IDC-IES integrates Internet Data Centers (IDCs), photovoltaic (PV) systems, wind turbines (WTs), and combined heat and power (CHP) units. IDCs consume significant amounts of electricity and cooling energy as energy consumers, while also generating waste heat as energy producers. This waste heat is upgraded via heat pumps, stored seasonally, and integrated into the heating network. In addition, IDCs are equipped with uninterrupted power supplies (UPSs) and diesel generators. The UPS is used for the transition to the backup generator. The backup generator can maintain IDC operation for a period during a power outage. The total backup power during this period depends on the amount of diesel fuel stored in the diesel generator.

[0059] (2) IDC energy consumption model.

[0060] Servers in an IDC (Internet Data Center) are the primary computing devices for workloads and also the most power-consuming. Server energy consumption models consider the correlation between the power consumption of server functional units (CPU, memory, etc.) and performance counters. This model captures both fixed and dynamic power consumption, making it particularly suitable for energy management and optimization in IDCs. Compared to traditional linear performance energy consumption models for servers, nonlinear energy consumption regression models based on CPU utilization can more accurately estimate the total power consumption of a single server. Therefore, a calibration parameter r is introduced. The computing power consumption of a single server in an IDC is:

[0061]

[0062] By utilizing dynamic voltage-frequency scaling technology, the server's dynamic power consumption is proportional to the cube of the CPU's operating frequency. For all servers under different operating conditions k, the computational power consumption of IDCi is:

[0063]

[0064] Where i is the Internet data center index; t is the time index; sv is the server type index; k is the server operating status index; P i,sv,t The computing power consumption of the SV-class server in the Internet Data Center (IDCi) at time t; The upper limit of computing power consumption for SV-class servers in the Internet Data Center (IDCi) at time t; The computing power consumption of the Internet Data Center (IDCi) at time t; Ω sv A collection of server types; Let t be the number of SV-class servers powered on in the Internet Data Center (IDCi) at time t. Ω represents the fixed power consumption of an SV-class server in an Internet Data Center (IDCi) when it is idle at time t. k A set of server operating conditions; C i,sv For the dynamic power factor of SV-class servers in Internet Data Center (IDCi); M i,sv,k,t f represents the number of SV-class servers in the Internet Data Center (IDCi) at time t that are in operating condition k; i,sv,k The CPU operating frequency of a Class V server operating at runtime condition k in an Internet Data Center (IDCi); u i,sv,t The CPU utilization of SV-class servers in the Internet Data Center (IDCi) at time t; For u i,sv,t The power of r, where r is the calibration parameter; λ i,sv,t μ represents the workload arrival rate of SV-class servers in the Internet Data Center (IDCi) at time t. i,sv,t The service rate of SV-class servers in the Internet Data Center (IDCi) at time t. i,sv,t The service rate of SV-class servers in the Internet Data Center (IDCi) at time t.

[0065] (3) Cross-seasonal thermal storage of integrated energy systems for Internet data centers.

[0066] During the computation process, the servers within the data center generate a significant amount of waste heat, causing the indoor temperature to rise. The waste heat generated during the computation process is represented as:

[0067]

[0068] A heat pump captures waste heat from the server through its cooling end, lowering the room temperature to maintain equipment stability. The captured waste heat is then upgraded by the heat pump and used for heating purposes. The power consumption and thermal power output of the heat pump are expressed as follows:

[0069]

[0070] IDCs continuously generate waste heat throughout the year. However, due to fluctuations in workload and seasonal variations in heat demand, a mismatch often occurs between heat generation and demand within IDC-IES. The integration of Cross Seasonal Heat Storage (CSHS) technology achieves a balance between heat supply and demand, ensuring the efficient utilization of waste heat. The heat storage process is as follows:

[0071] H t =H t-π +H char,t ·η char -H dis,t / η dis -τ·H loss .

[0072] in, The heat generated during computing in Internet data centers; PUE is the ratio of total power consumption in Internet data centers to the power consumption of IT equipment; η cold For cooling efficiency; P t DC The computing power consumption of Internet data centers; P t HP The power consumption of the heat pump; α represents the heat power output of the heat pump; H represents the heat efficiency of the heat pump; t H represents the amount of heat stored in the thermal storage tank at time t. t-π H represents the amount of heat stored in the thermal storage tank at time t-τ. char,t η represents the heat storage power of the thermal storage tank at time t; char For heat storage efficiency; H dis,t η is the heat release power of the thermal storage tank at time t; dis τ is the heat release efficiency; H is the scheduling interval; τ is the heat release efficiency; τ is the scheduling interval; H is the heat release efficiency. loss This represents the heat loss rate.

[0073] (4) Data flow-energy flow coordination mechanism.

[0074] According to the Service Level Agreement (SLA), computing tasks received by an IDC are categorized into latency-sensitive and latency-tolerant tasks. During data processing, the latency tolerance of certain tasks allows workloads to be spatially migrated between geographically distributed IDCs, or temporally migrated within a single IDC. Workload scheduling allows data flows to indirectly influence energy flows by shifting power loads, cooling demands, and waste heat without altering the physical infrastructure of the IDC-IES, highlighting the coupling between data and energy flows. Conversely, during disasters, limited energy supply reduces the data processing capacity of an IDC, prompting workloads to migrate to other locations to maintain service continuity. However, workload scheduling incurs bandwidth costs due to potential cloud congestion and switching failures. This process illustrates the inverse coupling between data and energy flows, where energy fluctuations affect workload distribution within the IDC-IES.

[0075] The workload request λ from IDCi arrives at time t. i,t The actual quantities are shown below:

[0076]

[0077] λ i,j,t +λ j,i,t =0.

[0078]

[0079]

[0080] Where, λ i,t The actual workload arrival rate of Internet Data Center (IDCi) at time t; Let λ be the initial workload arrival rate of the SV-class servers in the Internet Data Center (IDCi) at time t; i,j,t Let λ be the workload arrival rate exchanged between Internet Data Center IDCi and Internet Data Center IDCj at time t. i,j,t >0 indicates that the workload has been moved from Internet Data Center IDCi to Internet Data Center IDCj, λ i,j,t <0 indicates that workloads are being moved from Internet Data Center i to Internet Data Center j; λ j,i,t The workload arrival rate exchanged between Internet Data Center IDCj and Internet Data Center IDCi at time t; This represents the upper limit of the switching workload arrival rate between Internet Data Center IDCi and Internet Data Center IDCj due to fiber optic cable capacity limitations. Let t be the initial workload arrival rate of the SV-class servers in the Internet Data Center (IDCj) at time t.

[0081] Step 102: Based on the integrated energy system model of the Internet data center, construct a multi-objective optimization function and constraints.

[0082] The optimization objective is to minimize workload loss and workload scheduling costs, IDC-IES electrical and thermal load loss costs, and maintenance personnel scheduling costs. The objective function W is:

[0083] W = min(F1, F2).

[0084]

[0085] Where W is the multi-objective optimization function, and F1 is the data flow cost. d The scheduling period is ζ. l This is the workload loss cost coefficient. The cost is the amount of workload-related losses. s ζ represents the workload scheduling cost coefficient. F2 represents the energy flow cost. e This is the cost coefficient for electrical load loss. This refers to the cost of electrical load loss. h This is the cost coefficient for heat load loss. Cost of heat load loss. RC This is the cost coefficient for dispatching and maintaining maintenance personnel. Schedule time for maintenance personnel.

[0086] The constraints include: Internet data center energy consumption constraints, cross-seasonal thermal storage constraints, and workload scheduling constraints.

[0087] The energy consumption constraint of Internet data centers is the energy consumption model of Internet data centers.

[0088] The cross-seasonal thermal storage constraint is a cross-seasonal thermal storage model for the integrated energy system of Internet data centers.

[0089] Workload scheduling constraints include data flow-energy flow coordination mechanisms and workload storage constraints.

[0090] Workload storage constraints are:

[0091]

[0092] λ sv,1,t =λ sv,t -λ sv,0,t .

[0093] 0≤S t ≤S max .

[0094]

[0095] S T =0.

[0096] Among them, S t S represents the storage capacity for a latency-tolerant workload at time t; t-1 λ represents the storage capacity for latency-tolerant workloads at time t-1. sv,1,t The latency-tolerant workload arrival rate of the SV class server at time t; λ represents the delay-tolerant workload processed at time t. sv,t Let λ be the workload arrival rate of the SV class server at time t; sv,0,t S represents the arrival rate of latency-sensitive workloads on server sv at time t. max This represents the upper limit of the workload's storage capacity. S is the maximum workload with high latency calculated at time t; T This represents the amount of workload storage at the end of the scheduling cycle.

[0097] The energy consumption constraints of IDC are:

[0098]

[0099] The CSHS constraint is:

[0100]

[0101] H t =Ht-π +H char,t ·η char -H dis,t / η dis -τ·H loss .

[0102] The workload scheduling constraints are:

[0103]

[0104] λ i,j,t +λ j,i,t =0.

[0105]

[0106] λ sv,1,t =λ sv,t -λ sv,0,t .

[0107] 0≤S t ≤S max .

[0108]

[0109] S T =0.

[0110] Step 103: Based on the constraints, use the quantum-enhanced multi-objective gray wolf optimization algorithm to obtain a set of non-dominated solutions to the multi-objective optimization function, and obtain the Pareto optimal front of the optimization objective. For the non-dominated solutions, use the membership function to obtain the global optimal solution as the decision basis for the data flow-energy flow coupled scheduling of the Internet data center integrated energy system.

[0111] The quantum-enhanced multi-objective gray wolf optimization algorithm is as follows:

[0112] The multi-objective optimization function is determined to be the fitness function.

[0113] Initialize the wolf population based on the target population size.

[0114] Use the initialized wolf population as the current population.

[0115] Calculate the fitness function values ​​for all individuals in the current population.

[0116] Based on the fitness function value, the current population is updated using the quantum rotation gate-population update formula, and the process returns to the step "calculate the fitness function value of all individuals in the current population" until the iteration termination condition is met. The Pareto front corresponding to the optimal fitness function value is then determined as the Pareto optimal front.

[0117] To address nonlinear multi-objective optimization problems, a quantum-enhanced multi-objective gray wolf optimization (QMOGWO) algorithm is designed, combining quantum computing theory with traditional intelligent algorithms. This algorithm leverages quantum parallelism to enhance population diversity and global search capabilities. The QMOGWO algorithm includes wolf population initialization, fitness calculation, position updates via quantum rotation gate operations, and final scheme selection.

[0118] In the QMOGWO algorithm, the position of the population is represented by the probability magnitude of the qubit. The qubit exists in a superposition state |W> of state |P> and state |Q>, as shown in the following equation:

[0119] |W>=γ|P>+υ|Q>.

[0120] Where γ and υ are arbitrary complex numbers, representing the probability amplitude of the corresponding state of the qubit, γ 2 and υ 2 These represent the probabilities of a qubit being in the states |P> and |Q>, respectively.

[0121] The initial population q of the wolf pack ij As shown below:

[0122]

[0123] γ ij =cos(θ) ij ).

[0124] υ ij =sin(θ) ij ).

[0125] Where θ ij = 2π·r. r represents a random value uniformly distributed in [0,1]. i = 1,2,...,m. j = 1,2,...,n. m is the population size, representing the number of wolves. n is the dimension of the solution space, corresponding to the number of decision variables. The position of the population represents 2n solutions in the solution space.

[0126] The population position in the solution space is obtained through the transformation described by the following equation:

[0127]

[0128] Where x ij and y ij This describes the position of the i-th individual in the j-th dimension of the solution space along different directions. The upper and lower bounds of the decision variables are denoted as K. j,max and K j,min .

[0129] To evaluate the fitness of wolves, the QMOGWO algorithm uses W = min(F1, F2) to represent the multi-objective function of the optimal solution retained during the iteration process. Non-dominated solutions are stored in an external archive, and the trade-offs are maintained by dynamically updating to retain only non-dominated solutions. At the same time, crowding distance sorting prioritizes sparsely distributed solutions to obtain the optimal Pareto front.

[0130] The population position is updated through a quantum rotation gate. The quantum rotation gate changes the value of a qubit towards the optimal individual, allowing the gray wolves to gradually approach their prey and refresh the population. The update equation for each qubit (quantum rotation gate - population update formula) is shown below:

[0131]

[0132] θ ij =d(γ) ij ,υ ij )·Δθ ij .

[0133]

[0134] in, Let θ be the new quantum position of the i-th individual in the j-th dimension after the quantum rotation gate operation; ij Let q be the quantum rotation angle, representing the rotation angle of the i-th individual in the j-th dimension; ij This represents the quantum position before the update. d(γ) ij ,υ ij ) represents the direction of rotation. γ ij υ represents the probability that a qubit is in state |P>. ij Let Δθ be the probability that the qubit is in the state |Q>. ij The quantum rotation angle is predetermined according to the strategy.

[0135] To enhance the algorithm's global search capability and avoid premature convergence, a quantum NOT gate is introduced to perform variational operations. Let the mutation probability be ξ. A random number r between 0 and 1 is generated for the i-th variable. If r < ξ, then a mutation operation is applied to the i-th variable. The execution process is shown in the following equation:

[0136]

[0137] To select a suitable final solution from the Pareto optimal solution set and eliminate the influence of human factors, fuzzy logic rules are employed. The normalized membership function is defined as:

[0138]

[0139] in is the membership function value of the k-th objective of the l-th non-dominated solution. This represents the maximum fitness value for the k-th objective. Let be the fitness value of the k-th objective in the l-th non-dominated solution. φ represents the minimum fitness value for the k-th objective. l N is the normalized comprehensive membership value of the l-th non-dominated solution. obj Let N be the target number. d It is the sum of non-dominated solutions.

[0140] This invention validates the proposed strategy based on the IEEE 33-node power system and 14-node thermal system in Northeast China. The IDC-IES system comprises two 0.5MW cogeneration units, a 1.5MW wind farm, and a 0.8MW photovoltaic power station. Two data centers (IDC1 and IDC2), deployed at power nodes 2 and 28 respectively, each have 10,000 servers and are equipped with UPS and diesel generators to ensure continuous operation. A CSHS device is deployed at thermal node 1. The case study uses a severe ice storm that occurred in Northeast China in November 2021 as the background. All meteorological data, power load data, and equipment failure information are derived from this disaster event. The system has four emergency repair teams and one de-icing team, capable of repairing a maximum of two faulty lines per day. Wind and solar power outputs are modeled using Weibull and Beta distributions, respectively, and a Monte Carlo simulation is used to generate a set of operating scenarios for the IDC-IES under the ice storm scenario. To verify the effectiveness of the strategy, four comparative scenarios are set up:

[0141] Case 1: Data flow and energy flow are scheduled independently and CSHS is not enabled.

[0142] Case 2: Data flow and energy flow are scheduled independently, but CSHS is enabled.

[0143] Case 3: Co-scheduling of data flow and energy flow without enabling CSHS.

[0144] Case 4: Co-scheduling of data flow and energy flow with CSHS enabled.

[0145] All simulations were performed on a laptop equipped with a 3.8GHz AMD 8000 series Riptide 7 processor and 32GB of memory.

[0146] Table 1. Cost Comparison of Four Cases

[0147]

[0148] Table 2 Comparison of resilience assessment results in four cases

[0149]

[0150] Table 1 illustrates the cost structure of the IDC-IES system under four scenarios, including data flow costs, energy flow costs, and total costs. Case 4 achieved the lowest total cost of $444,752.57, a 19.19% reduction compared to Case 1. Specifically, data flow costs decreased by 15.47%, and energy flow costs were reduced by 31.61%. The reduction in data flow costs stemmed from a workload migration strategy from IDC1 to IDC2 during the ice storm, effectively mitigating computing task interruption losses. The optimization of energy flow costs was achieved by reducing electrical and thermal load losses and improving the efficiency of maintenance team scheduling. The results demonstrate that Case 4, through collaborative workload scheduling and CSHS technology, significantly reduced the overall operating costs of the IDC-IES system under disaster conditions, showcasing a significant economic advantage.

[0151] Table 2 lists the resilience assessment indicators of IDC-IES under four scenarios. Robustness, speed, resilience, and survivability are all normalized using min-max normalization, with each indicator's value range being [0,1]. The overall resilience assessment value (RQAM) is set to [0,3]. The study shows that Case 4 exhibits the best performance with an RQAM value of 2.1132: Robustness R increases from 0.4001 in Case 1 to 0.7453; Speed ​​S significantly improves from 0.5714 to 0.8039, indicating a significant enhancement in system performance under disaster conditions; Survivability D optimizes from 0.7061 to 0.8261, reflecting improved management capabilities for latency-sensitive tasks; and Resilience E also shows significant improvement, increasing from 0.2217 to 0.6739. These results verify the synergistic enhancement effect of CSHS technology and the data flow-energy flow collaborative scheduling mechanism on the overall system resilience.

[0152] Figure 2 This section illustrates the changes in the electrical-thermal load recovery rate of IDC-IES under different scenarios. In the initial stage of the disaster (Day 1), although the ice storm caused line disconnections, the inherent redundancy design of the distribution network prevented load shedding. Load shedding began to appear on Day 2 and reached its lowest point on Day 9. On Day 6, a line fault caused a power outage at IDC1, resulting in a significant load drop. As the faulty lines were gradually repaired, the system load recovered to normal levels on Day 14. Comparing Case 1 and Case 3, the workload scheduling strategy reduced the load loss on Day 6 in Case 3 by approximately 15.38% compared to Case 1. This was due to the workload scheduling optimizing power resource allocation and maintaining service continuity. In Case 2 and Case 4, the introduction of CSHS provided additional thermal support, further improving the load recovery rate during the ice storm. By Day 10, the load recovery rate in Case 4 was 18.67% higher than in Case 1, fully demonstrating the enhanced system resilience.

[0153] Figure 3The study demonstrates the annual variation patterns of heat release and total heat storage in the CSHS system, with a focus on analyzing its dynamic characteristics during ice storms. During the disaster, the total heat storage of the system plummeted by 62.55% due to the combined effects of decreased CHP output and a surge in heat load demand. To address this energy shortfall, the CSHS proactively increased its heat release power to achieve a balance between supply and demand in the heating network, effectively ensuring the thermal stability of the IDC-IES under extreme conditions.

[0154] Figures 4-6 The study demonstrates the characteristics of workload arrival rate changes under different scenarios. In Case 3, at t=5 on day 6, IDC1 experienced a power grid outage, triggering UPS and diesel generator backup power. Due to limited diesel reserve capacity, the system migrated most of the workload to IDC2, ensuring service quality through strategic scheduling. By t=14 on day 6, after the power grid connection was restored, the demand for large-scale load migration significantly decreased. Data analysis shows that compared to the scenario without scheduling, the effective scheduling strategy reduced service interruption time by 33.38%, significantly improving system resilience. This result verifies the effectiveness of the collaborative scheduling mechanism in maintaining the service continuity of IDC-IES under extreme conditions, achieving stable operation under disaster conditions through dynamic load balancing and resource optimization.

[0155] Figure 7 The data-energy flow Sankey diagrams are presented for IDC1 under different scenarios during a power outage on day 6. In Case 1, where CSHS (Content Management System) is not deployed, the CHP unit needs to increase its thermal power output to meet the thermal load demand. In Case 2, CSHS provides 82.47% of the thermal energy supply, significantly reducing thermal load loss and improving system resilience. Cases 3 and 4 employ a data-energy flow coordinated scheduling strategy, reallocating some data flow to IDC2, effectively alleviating the load pressure on IDC1 and reducing data loss. Case 4, through deep coupling of energy and data flow, reduces the electrical and thermal power output of the CHP unit by 7.36% and 83.34%, respectively, compared to Case 1.

[0156] Figure 8 The data-energy flow correlation characteristics of IDC-IES throughout the day are further presented in Case 4. The heatmap shows that the color depth of IDC2 is significantly higher than that of IDC1 during the period from t=5 to 14, indicating that during the power outage, some workload was migrated from IDC1 to IDC2 to reduce energy consumption and task loss. This phenomenon verifies that data-energy flow collaborative scheduling plays a crucial role in enhancing system resilience by changing data flow distribution and adjusting energy consumption and waste heat allocation patterns. It not only optimizes energy utilization efficiency but also achieves stable operation under disaster conditions through a dynamic load balancing mechanism.

[0157] Table 3 Comparison of solution time for different algorithms

[0158]

[0159] To verify the performance of the QMOGWO algorithm, this application compares and analyzes it with the traditional Grey Wolf Optimizer (GWO) algorithm and the Non-dominated Sorting Genetic Algorithm II (NSGA-II). The computation time and convergence curves of the algorithms are shown in Table 3 and [Table data would be inserted here]. Figure 9 As shown in the figure. Experimental results show that QMOGWO significantly outperforms the comparative algorithms in both convergence speed and solution set quality. The traditional GWO algorithm achieves optimization search by simulating the social hierarchy (Alpha, Beta, Delta, and Omega) and hunting behavior (encircling, tracking, and attacking prey) of gray wolves. It has a simple structure and few parameters, but suffers from drawbacks such as slow convergence speed and susceptibility to local optima. NSGA-II, as a classic multi-objective optimization algorithm, uses fast non-dominated sorting and crowding calculation to maintain the diversity of the Pareto front, but its computational complexity is high and it relies on elitist strategies and adaptive parameter adjustment to improve performance. Choosing GWO and NSGA-II as comparison benchmarks can verify that QMOGWO, while inheriting the swarm intelligence advantages of the gray wolf algorithm, overcomes the limitations of the traditional GWO through quantum behavior mechanisms and dynamic parameter optimization. On the other hand, the comparison with NSGA-II demonstrates QMOGWO's breakthrough in multi-objective processing efficiency, especially in reducing computational complexity and improving the distribution of Pareto solution sets. Furthermore, QMOGWO integrates the global search characteristics of quantum computing with the social hierarchy guidance mechanism of the Grey Wolf algorithm. Its convergence curve exhibits a faster exponential decay trend, and the hypervolume index of the final solution set is superior to that of the comparison algorithms. This provides a new approach to complex engineering optimization problems that combines efficiency and robustness.

[0160] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a data flow-energy flow coupled scheduling method for an integrated energy system.

[0161] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0162] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0163] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0164] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0165] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0166] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0167] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A data flow-energy flow coupled scheduling method for an integrated energy system, characterized in that, The method is applied to an integrated energy system for an Internet data center, which integrates an Internet data center, photovoltaics, wind turbines, and a combined heat and power unit. The integrated energy system data flow-energy flow coupled scheduling method includes: Constructing an integrated energy system model for internet data centers; the integrated energy system model for internet data centers includes: an energy consumption model for internet data centers, a cross-seasonal thermal storage model for integrated energy systems for internet data centers, and a data flow-energy flow coordination mechanism. The energy consumption model for the Internet data center is as follows: ； ； ； Where i is the Internet data center index; t is the time index; sv is the server type index; and k is the server operating status index. The computing power consumption of the SV-class server in the Internet Data Center (IDCi) at time t; The upper limit of computing power consumption for SV-class servers in the Internet Data Center (IDCi) at time t; Let t be the computing power consumption of Internet Data Center IDCi at time t; A collection of server types; Let t be the number of SV-class servers powered on in the Internet Data Center (IDCi) at time t. The fixed power consumption of an SV-class server in an Internet Data Center (IDCi) at time t when it is idle; This is a collection of server operating conditions. The dynamic power factor for SV-class servers in Internet Data Centers (IDCi); Let t be the number of SV-class servers in Internet Data Center (IDCi) that are in operating condition k at time t. The CPU operating frequency of a Class V server operating in operating condition k within an Internet Data Center (IDCi). The CPU utilization of SV-class servers in the Internet Data Center (IDCi) at time t; for of r Power of 1 r For calibration parameters; The workload arrival rate of SV-class servers in the Internet Data Center (IDCi) at time t; The service rate of SV-class servers in the Internet Data Center (IDCi) at time t; The cross-seasonal thermal storage model for the integrated energy system of the Internet data center is as follows: ； ； ； ； in, Waste heat generated during the computing process in Internet data centers; This is the ratio of total power consumption of an internet data center to the power consumption of IT equipment. For cooling efficiency; The computing power consumption of Internet data centers; The power consumption of the heat pump; This refers to the heat power output of the heat pump; The heat output efficiency of the heat pump; Let be the amount of heat stored in the thermal storage tank at time t; For the thermal storage tank at t- Heat storage at all times; Let be the heat storage power of the heat storage tank at time t; For heat storage efficiency; Let be the heat release power of the thermal storage tank at time t; For heat release efficiency; For scheduling intervals; This refers to the heat loss rate; The data flow-energy flow coordination mechanism is as follows: ； ； ； ； in, The actual workload arrival rate of Internet Data Center (IDCi) at time t; Let t be the initial workload arrival rate of the SV-class servers in the Internet Data Center (IDCi). Let t be the workload arrival rate exchanged between Internet Data Center IDCi and Internet Data Center IDCj at time t. This indicates that workloads are being moved from Internet Data Center IDCi to Internet Data Center IDCj. This indicates that workloads have been moved from Internet Data Center (IDCi) to Internet Data Center (IDCj). The workload arrival rate exchanged between Internet Data Center IDCj and Internet Data Center IDCi at time t; This represents the upper limit of the switching workload arrival rate between Internet Data Center IDCi and Internet Data Center IDCj due to fiber optic cable capacity limitations. Let t be the initial workload arrival rate of the SV-class servers in Internet Data Center (IDCj) at time t. Based on the aforementioned integrated energy system model for Internet data centers, a multi-objective optimization function and constraints are constructed. Based on the aforementioned constraints, a set of non-dominated solutions to the multi-objective optimization function is obtained using the quantum-enhanced multi-objective gray wolf optimization algorithm. This yields the Pareto optimal front for the optimization objective. For the non-dominated solutions, the membership function is used to obtain the global optimal solution, which serves as the decision-making basis for the data flow-energy flow coupled scheduling of the Internet data center integrated energy system. The quantum-enhanced multi-objective gray wolf optimization algorithm is as follows: The multi-objective optimization function is determined to be the fitness function; the wolf population is initialized based on the target population size; the initialized wolf population is used as the current population; the fitness function values ​​of all individuals in the current population are calculated; based on the fitness function values, the current population is updated using the quantum rotation gate-population update formula, and the process returns to the step "calculate the fitness function values ​​of all individuals in the current population" until the iteration termination condition is met. The Pareto front corresponding to the optimal fitness function value is determined as the Pareto optimal front. The quantum rotation gate-population update formula is as follows: ； ； ； in, Let be the new quantum position of the i-th individual in the j-th dimension after the quantum rotation gate operation; Let be the quantum rotation angle, representing the rotation angle of the i-th individual in the j-th dimension; The quantum position before the update; Indicates the direction of rotation; For a quantum bit to be in a state The probability of; For a quantum bit to be in a state The probability of; For quantum rotation angle; The quantum-enhanced multi-objective gray wolf optimization algorithm introduces a quantum NOT gate to perform variational operations; let the mutation probability be... Generate a random number between 0 and 1 for the i-th variable. ,if Then, a mutation operation is applied to the i-th variable; the execution process is as follows: ； The quantum-enhanced multi-objective gray wolf optimization algorithm uses fuzzy logic rules to select a suitable final solution from the Pareto optimal solution set. The normalized membership function is defined as: ； ； in, For the first l The membership function value of the k-th objective in a non-dominated solution; The maximum fitness value for the k-th objective; Let be the fitness value of the k-th objective in the l-th non-dominated solution; Let be the minimum fitness value of the k-th objective; Let be the normalized comprehensive membership value of the l-th non-dominated solution; For the target number, It is the sum of non-dominated solutions.

2. The integrated energy system data flow-energy flow coupled scheduling method according to claim 1, characterized in that, The multi-objective optimization function is: ； ； ； in, W It is a multi-objective optimization function; For data flow costs; The scheduling period; This is the workload loss cost coefficient; Costs incurred due to workload loss; This is the workload scheduling cost coefficient; Cost of energy flow; This is the cost coefficient for electrical load loss; Cost of electrical load loss; This is the cost coefficient for heat load loss; Costs related to heat load loss; This is the cost coefficient for dispatching maintenance personnel; Schedule time for maintenance personnel.

3. The integrated energy system data flow-energy flow coupled scheduling method according to claim 2, characterized in that, The constraints include: Internet data center energy consumption constraints, cross-seasonal thermal storage constraints, and workload scheduling constraints. The energy consumption constraint of the Internet data center is the energy consumption model of the Internet data center; The cross-seasonal thermal storage constraint is the cross-seasonal thermal storage model of the Internet data center integrated energy system. The workload scheduling constraints include a data flow-energy flow coordination mechanism and workload storage constraints. The workload storage constraints are as follows: ； ； ； ； ； in, Storage capacity for latency-tolerant workloads at time t; Storage capacity for latency-tolerant workloads at time t-1; The latency-tolerant workload arrival rate of the SV class server at time t; The amount of delay-tolerant workload processed at time t; The workload arrival rate of the SV class server at time t; Let be the latency-sensitive workload arrival rate of server sv at time t; This represents the upper limit of the workload's storage capacity. The maximum workload with tolerance for delays is calculated at time t. This represents the amount of workload storage at the end of the scheduling cycle.

4. The integrated energy system data flow-energy flow coupled scheduling method according to claim 1, characterized in that, Before constructing the integrated energy system model for Internet data centers, the following is also included: Determine the multidimensional resilience assessment indicators for the integrated energy system of Internet data centers.

5. The integrated energy system data flow-energy flow coupled scheduling method according to claim 4, characterized in that, The multidimensional toughness assessment index is: ； ； ； ； ； ； ； in, RQAM As a multidimensional resilience assessment index; This is the normalized value of the robustness index; This is the normalized value of the speed index; This is the normalized value of the restorative indicator; This is the normalized value of the survivability index; It is a robustness metric used to describe the percentage of remaining load in the integrated energy system of an Internet data center; is the frequency of occurrence of fault state i; N is the total number of extracted scenarios; This represents the electrical load under normal conditions. This represents the heat load under normal conditions. This represents the maximum loss of the electrical load. This represents the maximum loss of heat load; As a speed metric, it is used to describe the complement of the ratio of the duration for which a system operates at its lowest performance to the time from the occurrence of a disaster to recovery to a new steady state; This refers to the time when the system experiences performance degradation. The time it takes for system performance to drop to its lowest point; This is the time when system performance begins to recover; The start time of the stable state after recovery; As a recovery indicator, it is used to describe the recovery speed and level of the integrated energy system of Internet data centers; The start time of the stable state after recovery Load level; The time for system performance to begin recovery The load level; The electrical load level reached after the system is restored; This represents the heat load level achieved after the system recovers. d is a resilience indicator used to describe the network connectivity of an integrated energy system for an Internet data center; d is the network redundancy of the integrated energy system for an Internet data center. Redundancy in the initial state of the network; Let A be the i-th eigenvalue of the adjacency matrix A of the weighted directed graph G, which is constructed based on the Internet data center integrated energy system; n is the total number of system nodes; Let A be the element in the i-th row and j-th column of the adjacency matrix A of the weighted directed graph G; Let be the weight of the direct edge between node i and node j.

6. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the integrated energy system data flow-energy flow coupled scheduling method according to any one of claims 1-5.

Images