Cooperative scheduling operation method and system for uninterruptible power supply module and computing load of data center

Abstract

The invention relates to a data center uninterruptible power supply module and calculation load cooperative scheduling operation method and system, and the method comprises the following steps: (1) collecting real-time operation data in a data center, and forming an energy consumption characteristic curve of a UPS module; (2) constructing an optimization problem model of UPS module start-stop and load calculation cooperative scheduling, including an upper layer optimization problem and a lower layer optimization problem, and setting the start-stop state of the UPS module as a decision variable of the upper layer optimization problem, setting a distribution mode of a calculation task, a charging and discharging plan of energy storage equipment, and a start-stop state and output power of a generator as decision variables of a lower-layer optimization problem, wherein an optimization target is to minimize the overall energy consumption and operation cost of the data center within a specific time; and (3) applying the strategy output in the step (2) to operation control of the data center. The technical problems that in the prior art, the energy consumption is increased and the operation cost is increased due to the fact that the calculation load of the data center and the UPS module scheduling are not coordinated are solved.

Description

Technical Field

[0001] The invention relates to a data center uninterruptible power supply module and a computing load coordinated scheduling operation method and system, belonging to the technical field of energy scheduling. Background Art

[0002] With the rapid development of cloud computing and big data technologies, the scale and number of data centers are constantly expanding, and their energy consumption issues are becoming increasingly prominent. In traditional data centers, uninterruptible power supply (UPS) modules are mainly used to provide uninterruptible power protection for key equipment such as servers. However, in actual operation, the start and stop strategies of UPS modules and the scheduling strategies of computing loads are often carried out independently, lacking coordinated optimization. This lack of coordination may cause the UPS module to operate inefficiently for a long time, increasing energy consumption and operating costs.

[0003] In the prior art, the research on the energy consumption characteristics of UPS modules is relatively limited, and the energy consumption relationship of UPS modules under different load conditions cannot be accurately described. In addition, the scheduling of computational loads often only considers the performance and resource utilization of the server, and fails to incorporate the operating status of the UPS module into the scheduling decision, resulting in the inability of the data center to optimize the overall energy consumption and reducing energy utilization efficiency.

[0004] Therefore, there is an urgent need for a method that can collaboratively optimize the UPS module start-stop strategy and the calculation load scheduling strategy to reduce the overall energy consumption of the data center and improve the operating efficiency. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method and system for the coordinated scheduling and operation of uninterruptible power supply modules and computing loads in a data center, which solves the technical problem in the prior art that the computing load of the data center and the UPS module scheduling cannot be coordinated, resulting in increased energy consumption and higher operating costs.

[0006] The technical solution of the present invention is as follows:

[0007] A method for coordinating and operating an uninterruptible power supply module and a computing load in a data center, comprising the following steps:

[0008] (1) Collect real-time operating data in the data center to form the energy consumption characteristic curve of the UPS module;

[0009] (2) Construct an optimization problem model for the coordinated scheduling of UPS module start-stop and computing load, including the upper-level optimization problem and the lower-level optimization problem. The start-stop state of the UPS module is set as the decision variable of the upper-level optimization problem, and the allocation method of computing tasks, the charging and discharging plan of the energy storage device, the start-stop state and output power of the generator are set as the decision variables of the lower-level optimization problem. The optimization goal is to minimize the overall energy consumption and operating cost of the data center within a specific time.

[0010] (3) Apply the strategy output from step (2) to the operation control of the data center. Through the centralized communication system, dynamic allocation of computing tasks is realized. According to the optimization results, computing tasks are reasonably allocated to different servers to balance the server load and improve the utilization of computing resources. UPS module start and stop control. According to the optimized start and stop strategy, the UPS module is enabled and dormant in real time to reduce the energy consumption of the UPS. Energy storage equipment management. According to the optimized charging and discharging plan, the charging and discharging process of the energy storage equipment is controlled to smooth load fluctuations and improve energy utilization efficiency. Generator operation control. The generator is started and stopped as needed to adjust the output power to ensure power supply reliability and economy.

[0011] Preferably, according to the present invention, in step (1), the real-time operating data includes the real-time input power of the UPS module, the real-time output power of the UPS module, the real-time power of the server powered by the UPS module, and the real-time computing load processed by the server.

[0012] Preferably, in step (1), the energy consumption characteristic curve of the UPS module is formed by analyzing the real-time operation data, obtaining the real-time output power and the real-time input power of the UPS module, calculating the difference between the two, obtaining the energy consumption of the UPS module, collecting and selecting multiple groups of parameters and substituting them into the following formula for parameter fitting, obtaining a quadratic function relationship model between the energy consumption of the UPS module and the real-time output power of the UPS module, and forming the energy consumption characteristic curve of the UPS module, such as Figure 3 As shown:

[0013]

[0014] in, is the UPS module energy consumption, It is the rated power marked on the nameplate of the UPS module. a0, a1, and a2 are related parameters. The related parameters are determined by experimentally measuring the characteristic curves of different types of UPS modules.

[0015] Preferably, in step (2) of the present invention, the upper optimization problem is solved by a genetic algorithm, the start and stop status of the UPS module is encoded, and a binary encoding method is adopted, wherein 1 indicates that the UPS module is enabled and 0 indicates that the UPS module is dormant. Appropriate genetic algorithm parameters are set, including population size, crossover probability, mutation probability, etc., and iterative optimization is performed through genetic operations (selection, crossover, mutation) to seek the optimal UPS start and stop strategy.

[0016] Preferably, in step (2), the lower optimization problem is solved by using a branch and cut algorithm. Under a given UPS start and stop strategy, the branch and cut algorithm is used to solve the allocation mode of computing tasks, the charging and discharging plan of the energy storage device, the start and stop state and output power of the generator. By branching possible decision plans and pruning when the constraints are not met, the amount of calculation is reduced and the solution efficiency is improved.

[0017] The genetic algorithm of the upper optimization problem affects the constraints and parameters of the lower optimization problem by changing the on-state of the UPS module. The optimization results of the lower optimization problem are fed back to the upper optimization problem as the fitness value to guide the evolution of the genetic algorithm. The two-layer optimization method coordinates to obtain the optimal solution to the problem.

[0018] Preferably, according to the present invention, in step (2), the lower optimization problem model includes the following constraints:

[0019] Server load rate constraints: set the minimum and maximum load rates of the server to ensure that the server operates within a safe and efficient load range;

[0020] UPS module output power constraint: set the maximum output power limit of the UPS module to prevent the UPS module from overloading;

[0021] Energy storage device constraints, including energy storage state constraints, charge / discharge power constraints, and energy storage capacity constraints, ensure that the energy storage device operates within its physical limitations;

[0022] Generator operation constraints, including generator power output constraints, minimum start and stop time constraints, and ramp rate constraints, to avoid frequent start and stop and power mutations;

[0023] Power balance constraints ensure that the total power supply and total power consumption of the data center are balanced in each time period.

[0024] Preferably, according to the present invention, the generator power output constraint is:

[0025]

[0026] in: is the actual output power of the generator, and are the upper and lower limits of the generator output power, It is an identification variable that represents the start and stop status of the generator;

[0027] The minimum start and stop time constraints are:

[0028]

[0029]

[0030]

[0031]

[0032] in, and is the minimum start time and minimum shutdown time of the generator, and It is an identification variable that characterizes the start-up and shutdown actions of the generator;

[0033] The climbing rate constraint is:

[0034]

[0035]

[0036] in, is the maximum power rise rate when the generator starts, It is the maximum power reduction rate when the generator is shut down.

[0037] Preferably, according to the present invention, the energy storage device is an energy storage system (ESS), and the energy storage state constraint is:

[0038]

[0039] Among them, ES i,t is the energy storage state of the energy storage device, and is the charging power and discharging power of the energy storage device, and is the charging efficiency and discharging efficiency of the energy storage device. The energy storage state constraint indicates that the energy storage state at a certain moment is determined by the energy storage state at the previous moment and the charging and discharging power.

[0040] The charge / discharge power constraints are:

[0041]

[0042]

[0043] in: and is the maximum charging power and maximum discharging power of the energy storage device;

[0044] The energy storage capacity constraint is:

[0045]

[0046] Among them, ES min，i and ES max，i It is the minimum energy storage capacity and maximum energy storage capacity of the energy storage device.

[0047] Preferably, according to the present invention, the power balance constraint is:

[0048]

[0049]

[0050] in, The real-time power of electricity purchased from the power grid for the data center. is the energy consumption of the UPS module, is the output power of the data center generator, The output power of photovoltaic power generation to support the data center.

[0051] Preferably, according to the present invention, in step (2), the overall energy consumption and operating costs include server energy consumption, UPS module energy consumption and operating costs of supporting equipment.

[0052] A data center uninterruptible power supply module and computing load coordinated scheduling operation system, comprising:

[0053] Data acquisition module: used to collect real-time operation data in the data center and form the energy consumption characteristic curve of the UPS module;

[0054] Optimization problem model construction module: used to construct an optimization problem model for coordinated scheduling of UPS module start-stop and computing load, including upper-level optimization problems and lower-level optimization problems. The start-stop status of the UPS module is set as the decision variable of the upper-level optimization problem, and the allocation method of computing tasks, the charging and discharging plan of the energy storage equipment, the start-stop status and output power of the generator are set as the decision variables of the lower-level optimization problem. The optimization goal is to minimize the overall energy consumption and operating cost of the data center within a specific time.

[0055] Application module: used to apply the strategy output from step (2) to the operation control of the data center, realize dynamic allocation of computing tasks through the centralized communication system, and reasonably allocate computing tasks to different servers according to the optimization results to balance the server load; UPS module start and stop control, according to the optimized start and stop strategy, real-time control of the activation and sleep of the UPS module; energy storage device management, according to the optimized charging and discharging plan, control the charging and discharging process of the energy storage device; generator operation control, start and stop the generator as needed to adjust the output power.

[0056] The beneficial effects of the present invention are:

[0057] 1. The present invention proposes a method for collaborative optimization of UPS module power on / off and data center load scheduling, which collaboratively schedules computing tasks and the operating status of power supply modules within the framework of energy consumption and cost.

[0058] 2. In the modeling of the scheduling problem, the present invention introduces a nonlinear energy consumption model of the UPS module and models the optimization problem as an upper and lower layer optimization problem, thereby realizing the decomposition and solution of the high-order planning problem. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic diagram of the process of the present invention;

[0060] Figure 2 It is a structural diagram of the data center microgrid of the present invention;

[0061] Figure 3 It is an energy consumption characteristic curve diagram of the UPS module of the present invention, wherein the abscissa is the energy consumption characteristic curve of the UPS module, and the ordinate is the energy consumption of the UPS module;

[0062] Figure 4 The switch status of the UPS module in the data center computer room where the present invention is not adopted;

[0063] Figure 5 The switch status of the UPS module in the computer room of a data center after adopting the present invention. DETAILED DESCRIPTION

[0064] The present invention will be further described below by way of embodiments in conjunction with the accompanying drawings, but is not limited thereto.

[0065] Embodiment 1:

[0066] This embodiment provides a method for coordinated scheduling and operation of an uninterruptible power supply module and a computing load in a data center, and the steps are as follows:

[0067] (1) Collecting real-time operation data in the data center to form an energy consumption characteristic curve of the UPS module. The process of forming the energy consumption characteristic curve of the UPS module is to analyze the real-time operation data, obtain the real-time output power and real-time input power of the UPS module, calculate the difference between the two, obtain the energy consumption of the UPS module, collect and select multiple groups of parameters and substitute them into the following formula for parameter fitting, obtain the quadratic function relationship model between the energy consumption of the UPS module and the real-time output power of the UPS module, and form the energy consumption characteristic curve of the UPS module, such as Figure 3 As shown:

[0068]

[0069] in, is the UPS module energy consumption, It is the rated power marked on the nameplate of the UPS module. a0, a1, and a2 are related parameters. The related parameters are determined by experimentally measuring the characteristic curves of different types of UPS modules.

[0070] (2) Construct an optimization problem model for the coordinated scheduling of UPS module start-stop and computing load, including an upper-level optimization problem and a lower-level optimization problem. The start-stop state of the UPS module is set as the decision variable of the upper-level optimization problem, and the allocation method of computing tasks, the charging and discharging plan of the energy storage device, the start-stop state and output power of the generator are set as the decision variables of the lower-level optimization problem. The optimization goal is to minimize the overall energy consumption and operating cost of the data center within a specific time. The overall energy consumption and operating cost include server energy consumption, UPS module energy consumption and the operating cost of supporting equipment.

[0071] The following operations are performed when building the optimization problem model: Load type: The workload of the data center is divided into interactive load and batch load. Interactive load requires immediate processing and is highly time-sensitive. Therefore, it cannot be scheduled and executed at different times of the day and can only be scheduled between different devices in the data center. Batch load allows processing to be completed within a certain delay. You can choose its execution time and schedule it between different devices in the data center. The workload is represented as follows:

[0072]

[0073] Where: i，t is the total computing load the server is handling, is the amount of interactive load the server is handling, is the total batch load that the server is processing;

[0074] Time discretization: Divide the entire operation cycle into a series of time periods of equal length (such as every hour) so that the problem can be modeled as a mixed integer programming problem, which is convenient for modeling and solving the problem and reduces the difficulty of calculation. At the same time, in the process of time discretization, the effective division of time periods can also achieve the effect of refined scheduling and management of load and power demand;

[0075] Load response of UPS modules: Each UPS module can be turned on or off independently at different times, and its status is determined by the dynamic changes of the workload; when the computing load is distributed more, the power consumption of the equipment increases, the output power of the UPS module increases, and the number of UPS modules turned on needs to be increased to meet the power supply demand; when the computing load is distributed less, the power consumption of the equipment decreases, the UPS output power decreases, and some UPS modules can be turned off to reduce UPS energy consumption.

[0076] The optimization problem model includes the following constraints:

[0077] Server load rate constraints: set the minimum and maximum load rates of the server to ensure that the server operates within a safe and efficient load range;

[0078] UPS module output power constraint: set the maximum output power limit of the UPS module to prevent the UPS module from overloading;

[0079] Energy storage device constraints, including energy storage state constraints, charge / discharge power constraints, and energy storage capacity constraints, ensure that the energy storage device operates within its physical limitations;

[0080] The energy storage device is an energy storage system (ESS), and the energy storage state constraints are:

[0081]

[0082] Among them, ES i,t is the energy storage state of the energy storage device, and is the charging power and discharging power of the energy storage device, and is the charging efficiency and discharging efficiency of the energy storage device. The energy storage state constraint indicates that the energy storage state at a certain moment is determined by the energy storage state at the previous moment and the charging and discharging power.

[0083] The charge / discharge power constraints are:

[0084]

[0085]

[0086] in: and is the maximum charging power and maximum discharging power of the energy storage device;

[0087] The energy storage capacity constraint is:

[0088]

[0089] Among them, ES min，i and ES max，i It is the minimum energy storage capacity and maximum energy storage capacity of the energy storage device.

[0090] Generator operation constraints, including generator power output constraints, minimum start and stop time constraints, and ramp rate constraints, to avoid frequent start and stop and power mutations;

[0091] The generator power output constraint is:

[0092]

[0093] in: is the actual output power of the generator, and are the upper and lower limits of the generator output power, It is an identification variable that represents the start and stop status of the generator;

[0094] The minimum start and stop time constraints are:

[0095]

[0096]

[0097]

[0098]

[0099] in, and is the minimum start time and minimum shutdown time of the generator, and It is an identification variable that characterizes the start-up and shutdown actions of the generator;

[0100] The climbing rate constraint is:

[0101]

[0102]

[0103] in, is the maximum power rise rate when the generator starts, It is the maximum power reduction rate when the generator is shut down.

[0104] Power balance constraints keep the total power supply and total power consumption of the data center balanced in each time period:

[0105]

[0106] in, The real-time power of electricity purchased from the power grid for the data center. is the energy consumption of the UPS module, is the output power of the data center generator, The output power of photovoltaic power generation to support the data center.

[0107] The upper optimization problem is solved using genetic algorithm, specifically:

[0108] Encoding,Using integer encoding, each individual represents the start-up state sequence of the UPS module in the entire scheduling cycle, and each bit in the gene sequence represents the state of a UPS module in a specific time period;

[0109] Initial population generation: randomly generate a certain number of individuals to form the initial population.

[0110] A fitness function is established, and the optimal operating cost calculated by the lower-level optimization problem is used as the fitness value. The smaller the fitness value, the better the solution.

[0111] The selection operation uses methods such as roulette or tournament selection to select excellent individuals for the next generation based on the fitness value. The crossover operation performs a crossover operation on the selected individuals to exchange some genes with a certain probability to generate new individuals. The mutation operation mutates the genes of individuals with a certain probability to avoid falling into the local optimum.

[0112] The iteration is terminated after reaching the preset number of iterations or the fitness value converges.

[0113] After the lower-level problem is solved to determine the on-state of the UPS module, the lower-level problem is transformed into a mixed integer linear programming (MILP) problem and solved using the Branch-and-Cut algorithm, which can be divided into several steps: initialization, node selection and branching, cutting plane generation, pruning, and iteration;

[0114] Initialization: Establish the initial relaxed model of the problem as the root node, solve the initial relaxed problem, and obtain the initial solution;

[0115] Node selection and branching operations: select the variable to be branched according to the value of the variable, generate child nodes, select a fractional variable, create two sub-problems, take the supremum and infimum of the variable respectively, and repeat the branching and cutting plane operations for the newly generated sub-problems;

[0116] Cutting plane generation and processing,At each node, check whether the current solution satisfies the integer constraints.,If not, a cutting plane is generated and added to the model.

[0117] The pruning strategy uses the upper and lower bounds to compare. If the lower bound of the current node is greater than the known optimal upper bound, pruning is performed.

[0118] Iterate the solution and repeat the above steps. When all nodes have been processed or pruned, the algorithm terminates, the search tree is traversed, or the termination condition is reached, the iteration stops.

[0119] The upper-layer genetic algorithm changes the on-state of the UPS module to affect the constraints and parameters of the lower-layer problem. The optimization results (operating costs) of the lower layer are fed back to the upper layer as fitness values ​​to guide the evolution of the genetic algorithm. The two-layer optimization methods coordinate to obtain the optimal solution to the problem. Figure 4 and Figure 5 The difference in the on-state of the UPS modules can be found, and this optimization can reduce the total cost by 0.44% to 6.48%.

[0120] (3) Apply the strategy output from step (2) to the operation control of the data center. Through the centralized communication system, dynamic allocation of computing tasks is realized. According to the optimization results, computing tasks are reasonably allocated to different servers to balance the server load and improve the utilization of computing resources. UPS module start and stop control. According to the optimized start and stop strategy, the UPS module is enabled and dormant in real time to reduce the energy consumption of the UPS. Energy storage equipment management. According to the optimized charging and discharging plan, the charging and discharging process of the energy storage equipment is controlled to smooth load fluctuations and improve energy utilization efficiency. Generator operation control. The generator is started and stopped as needed to adjust the output power to ensure power supply reliability and economy.

[0121] Embodiment 2:

[0122] This embodiment provides a data center uninterruptible power supply module and computing load collaborative scheduling operation system, including:

[0123] Data acquisition module: used to collect real-time operation data in the data center and form the energy consumption characteristic curve of the UPS module;

[0124] Optimization problem model construction module: used to construct an optimization problem model for coordinated scheduling of UPS module start-stop and computing load, including upper-level optimization problems and lower-level optimization problems. The start-stop status of the UPS module is set as the decision variable of the upper-level optimization problem, and the allocation method of computing tasks, the charging and discharging plan of the energy storage equipment, the start-stop status and output power of the generator are set as the decision variables of the lower-level optimization problem. The optimization goal is to minimize the overall energy consumption and operating cost of the data center within a specific time.

[0125] Application module: used to apply the strategy output from step (2) to the operation control of the data center, realize dynamic allocation of computing tasks through the centralized communication system, and reasonably allocate computing tasks to different servers according to the optimization results to balance the server load; UPS module start and stop control, according to the optimized start and stop strategy, real-time control of the activation and sleep of the UPS module; energy storage device management, according to the optimized charging and discharging plan, control the charging and discharging process of the energy storage device; generator operation control, start and stop the generator as needed to adjust the output power.

Claims

1. A method for coordinated scheduling and operation of an uninterruptible power supply module and computing load in a data center, characterized in that: Here are the steps: (1) Collect real-time operating data in the data center to form the energy consumption characteristic curve of the UPS module; (2) Construct an optimization problem model for the coordinated scheduling of UPS module start-stop and computing load, including an upper-level optimization problem and a lower-level optimization problem. The start-stop state of the UPS module is set as the decision variable of the upper-level optimization problem, and the allocation method of computing tasks, the charging and discharging plan of the energy storage device, the start-stop state and output power of the generator are set as the decision variables of the lower-level optimization problem. The optimization goal is to minimize the overall energy consumption and operating cost of the data center within a specific time. (3) Applying the strategy output from step (2) to the operation control of the data center, dynamically allocating computing tasks through a centralized communication system, and reasonably allocating computing tasks to different servers based on the optimization results to balance the server load; UPS module start and stop control, according to the optimized start and stop strategy, real-time control of the UPS module activation and sleep; Energy storage equipment management: controls the charging and discharging process of energy storage equipment according to the optimized charging and discharging plan; generator operation control: starts and stops the generator as needed to adjust the output power.

2. The method for coordinated scheduling and operation of uninterruptible power supply modules and computing loads in a data center according to claim 1, characterized in that: In step (1), the real-time operation data includes the real-time input power of the UPS module, the real-time output power of the UPS module, the real-time power of the server powered by the UPS module, and the real-time computing load processed by the server.

3. The method for coordinated scheduling and operation of uninterruptible power supply modules and computing loads in a data center according to claim 2, characterized in that: In step (1), the process of forming the energy consumption characteristic curve of the UPS module is to analyze the real-time operation data, obtain the real-time output power and real-time input power of the UPS module, calculate the difference between the two, obtain the energy consumption of the UPS module, collect and select multiple groups of parameters and substitute them into the following formula for parameter fitting, obtain the quadratic function relationship model between the energy consumption of the UPS module and the real-time output power of the UPS module, and form the energy consumption characteristic curve of the UPS module: in, is the UPS module energy consumption, It is the rated power marked on the UPS module nameplate, and a0, a1, and a2 are related parameters.

4. The method for coordinated scheduling and operation of uninterruptible power supply modules and computing loads in a data center according to claim 3, characterized in that: In step (2), the upper optimization problem is solved by using a genetic algorithm to encode the start and stop status of the UPS module using a binary encoding method, where 1 indicates that the UPS module is enabled and 0 indicates that the UPS module is dormant. The genetic algorithm parameters are set, including population size, crossover probability, and mutation probability. The optimal UPS start and stop strategy is sought through iterative optimization through genetic operations.

5. The method for coordinated scheduling and operation of uninterruptible power supply modules and computing loads in a data center according to claim 4, characterized in that: In step (2), the lower optimization problem is solved by using the branch and cut algorithm. Under a given UPS start and stop strategy, the branch and cut algorithm is used to solve the allocation of computing tasks, the charging and discharging plan of the energy storage device, the start and stop status and output power of the generator. The possible decision plans are branched and pruned when the constraints are not met. The genetic algorithm of the upper optimization problem affects the constraints and parameters of the lower optimization problem by changing the on-state of the UPS module. The optimization results of the lower optimization problem are fed back to the upper optimization problem as the fitness value to guide the evolution of the genetic algorithm. The two-layer optimization method coordinates to obtain the optimal solution to the problem.

6. The method for coordinated scheduling and operation of uninterruptible power supply modules and computing loads in a data center according to claim 5, characterized in that: In step (2), the lower optimization problem model includes the following constraints: Server load rate constraint, setting the minimum and maximum load rates of the server; UPS module output power constraint, set the maximum output power limit of the UPS module; Energy storage device constraints, including energy storage state constraints, charge / discharge power constraints, and energy storage capacity constraints; Generator operation constraints, including generator power output constraints, minimum start and stop time constraints, and ramp rate constraints; Power balance constraints ensure that the total power supply and total power consumption of the data center are balanced in each time period; Preferably, the generator power output constraint is: in: is the actual output power of the generator, and are the upper and lower limits of the generator output power, It is an identification variable that represents the start and stop status of the generator; The minimum start and stop time constraints are: in, and is the minimum start time and minimum shutdown time of the generator, and It is an identification variable that characterizes the start-up and shutdown actions of the generator; The climbing rate constraint is: in, is the maximum power rise rate when the generator starts, It is the maximum power reduction rate when the generator is shut down.

7. The method for coordinated scheduling and operation of uninterruptible power supply modules and computing loads in a data center according to claim 6, characterized in that: The energy storage device is an energy storage system, and the energy storage state constraints are: Among them, ES i,t is the energy storage state of the energy storage device, and is the charging power and discharging power of the energy storage device, and The charging efficiency and discharging efficiency of the energy storage device; The charge / discharge power constraints are: in: and is the maximum charging power and maximum discharging power of the energy storage device; The energy storage capacity constraint is: Among them, ES min，i and ES max，i It is the minimum energy storage capacity and maximum energy storage capacity of the energy storage device.

8. The method for coordinated scheduling and operation of uninterruptible power supply modules and computing loads in a data center according to claim 7, characterized in that: The power balance constraint is: in, The real-time power of electricity purchased from the power grid for the data center. is the energy consumption of the UPS module, is the output power of the data center generator, The output power of photovoltaic power generation to support the data center.

9. The method for coordinated scheduling and operation of uninterruptible power supply modules and computing loads in a data center according to claim 8, characterized in that: In step (2), the overall energy consumption and operating costs include server energy consumption, UPS module energy consumption and operating costs of supporting equipment.

10. A data center uninterruptible power supply module and computing load coordinated scheduling operation system, characterized in that: include: Data acquisition module: used to collect real-time operation data in the data center and form the energy consumption characteristic curve of the UPS module; Optimization problem model construction module: used to construct an optimization problem model for coordinated scheduling of UPS module start-stop and computing load, including upper-level optimization problems and lower-level optimization problems. The start-stop status of the UPS module is set as the decision variable of the upper-level optimization problem, and the allocation method of computing tasks, the charging and discharging plan of the energy storage equipment, the start-stop status and output power of the generator are set as the decision variables of the lower-level optimization problem. The optimization goal is to minimize the overall energy consumption and operating cost of the data center within a specific time. Application module: used to apply the strategy output from step (2) to the operation control of the data center, realize dynamic allocation of computing tasks through the centralized communication system, and reasonably allocate computing tasks to different servers according to the optimization results to balance the server load; UPS module start and stop control, according to the optimized start and stop strategy, real-time control of the UPS module startup and sleep; Energy storage equipment management: controls the charging and discharging process of energy storage equipment according to the optimized charging and discharging plan; generator operation control: starts and stops the generator as needed to adjust the output power.

Images