Cooperative scheduling method for computing power and electric power, computer equipment and medium

Abstract

The embodiment of the invention is suitable for the technical field of computing power, and provides a computing power and electric power cooperative scheduling method, computer equipment and a medium, and the method comprises the steps: determining the predicted green power generation amount within a preset time; current electricity consumption information, to-be-executed calculation task information and available calculation power information of the data center are obtained, the current electricity consumption information comprises the current electricity consumption demand, and the calculation task information comprises calculation time information and interaction relation information of a plurality of calculation tasks; on the basis of the calculation time information and the interaction relationship information, sorting the calculation tasks by using a queuing network model to obtain a calculation task sequence; and determining a plurality of calculation tasks from the calculation task sequence as target calculation tasks in the preset time based on the predicted green power generation amount, the current power consumption demand and the available calculation power information. Through the method, the operation cost can be reduced when the cooperative scheduling of the computing power and the electric power is realized.

Description

Technical Field

[0001] The present application belongs to the field of computing power technology, and in particular, relates to a method for coordinated scheduling of computing power and electricity, computer equipment and media. Background Art

[0002] Green power refers to the use of specific power generation equipment, such as wind turbines, solar photovoltaic cells, etc., to convert renewable energy such as wind energy and solar energy into electrical energy. Weather factors such as wind power and sunlight intensity will affect the production and storage of green power, resulting in its intermittent and volatile characteristics, which require the use of energy storage equipment to solve this problem.

[0003] Using green electricity to provide energy support for the computing power network can reduce environmental pollution; since the price of green electricity is generally lower, using green electricity to provide technical support can reduce the operating costs of the computing power network.

[0004] However, the generation of green electricity is volatile, and the demand for electricity in the computing network is also volatile. When the volatility of the power supply side and the demand side does not match, it will bring more operating costs to the computing network. Therefore, it is crucial to reasonably plan the electricity demand of the computing network to reduce the operating costs of the computing network.

[0005] In the existing data center power dispatching method, the consumption of clean energy is increased by dispatching based on energy. However, the intermittent, random and sudden nature of green electricity is not considered, resulting in poor economic benefits. Summary of the invention

[0006] In view of this, an embodiment of the present application provides a method for coordinated scheduling of computing power and electricity, a computer device and a medium, so as to reduce the operating costs of a data center when realizing the coordinated scheduling of computing power and electricity.

[0007] A first aspect of an embodiment of the present application provides a method for coordinated scheduling of computing power and electricity, including:

[0008] Determine the forecasted green electricity generation within a preset time period;

[0009] Obtaining current power usage information, computing task information to be executed, and available computing power information of the data center, wherein the current power usage information includes current power demand, and the computing task information includes computing time information and interaction relationship information of multiple computing tasks;

[0010] Based on the computing time information and the interaction relationship information, each of the computing tasks is sorted using a queuing network model to obtain a computing task sequence;

[0011] Based on the predicted green electricity generation, the current electricity demand and the available computing power information, a plurality of the computing tasks are determined from the computing task sequence as target computing tasks within the preset time.

[0012] A second aspect of an embodiment of the present application provides a coordinated scheduling device for computing power and electricity, including:

[0013] A power generation prediction module is used to determine the predicted green power generation within a preset time;

[0014] An information acquisition module, used to acquire the current power consumption information of the data center, the computing task information to be executed and the available computing power information, wherein the current power consumption information includes the current power demand, and the computing task information includes the computing time information and interaction relationship information of multiple computing tasks;

[0015] A task sorting module, used to sort each of the computing tasks using a queuing network model based on the computing time information and the interaction relationship information to obtain a computing task sequence;

[0016] A task scheduling module is used to determine a plurality of computing tasks from the computing task sequence as target computing tasks within the preset time based on the predicted green electricity generation, the current electricity demand and the available computing power information.

[0017] A third aspect of an embodiment of the present 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 implements the method described in the first aspect when executing the computer program.

[0018] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect above is implemented.

[0019] A fifth aspect of the embodiments of the present application provides a computer program product. When the computer program product is run on a computer device, the computer device executes the method described in the first aspect.

[0020] Compared with the prior art, the embodiments of the present application have the following advantages:

[0021] The method in the embodiment of the present application can be used to coordinate the scheduling of computing power and electricity for the data center. Providing power support for the equipment in the data center based on green electricity can save energy and protect the environment and reduce the electricity price of the data center. However, the power generation of green electricity is volatile. Ideally, if the volatility of green electricity generation can match the fluctuation of the demand for electricity in the computing power network, the use and maintenance of energy storage equipment and the use of power grid electricity can be reduced, thereby reducing the operating cost of the energy storage equipment in the data center and reducing the use of power grid electricity with higher electricity prices, thereby reducing the operating cost of the data center. In the embodiment of the present application, in order to more reasonably coordinate the scheduling of computing power and electricity, the predicted green electricity generation within the preset time can be determined, and the current electricity consumption information of the data center, the computing task information to be executed and the available computing power information can be obtained. The current electricity consumption information includes the current electricity demand, and the computing task information includes the computing time information and the interaction relationship information of multiple computing tasks; the computing tasks have a priority level of execution, so based on the computing time information and the interaction relationship information, the queuing network model can be used to sort each computing task to obtain a computing task sequence; then based on the predicted green electricity generation, the current electricity demand and the available computing power information, multiple computing tasks are determined from the computing task sequence as the target computing tasks within the preset time. In the embodiment of the present application, the computing tasks executed within the preset time can be scheduled based on the predicted green electricity generation within the preset time, and when scheduling the computing tasks, the computing resources of the computing power network and the priority of the computing tasks can be considered, so that when the method in the present application realizes the coordinated scheduling of computing power and electricity, the operating cost of the energy storage equipment of the data center and the use of power grid electricity with higher electricity prices are reduced, thereby reducing the operating cost of the data center. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art.

[0023] Figure 1 It is a schematic diagram of the steps of a method for coordinated scheduling of computing power and electricity provided in an embodiment of the present application;

[0024] Figure 2 It is a schematic diagram of the steps of another method for coordinated scheduling of computing power and electricity provided in an embodiment of the present application;

[0025] Figure 3 is a schematic diagram of a coordinated scheduling device for computing power and electricity provided in an embodiment of the present application;

[0026] Figure 4 It is a schematic diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0027] In the following description, specific details such as specific system structures, technologies, etc. are proposed for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from hindering the description of the present application.

[0028] Data centers consume a lot of electricity and the amount is increasing year by year. Electricity expenditure accounts for more than 60% of the total operating costs. Optimizing the electricity cost of data centers has become an important research direction.

[0029] Green electricity refers to the use of specific power generation equipment, such as wind turbines, solar photovoltaic cells, etc., to convert renewable energy such as wind energy and solar energy into electrical energy. Weather factors such as wind power and sunlight intensity will affect the production and storage of green electricity, resulting in its intermittent and volatile characteristics. Energy storage equipment is needed to solve this problem.

[0030] Using green electricity to provide energy support for the computing network can save energy and protect the environment. And because the price of green electricity is generally lower, using green electricity to provide technical support can reduce the operating costs of the computing network.

[0031] However, due to the volatility and intermittency of green electricity generation, the demand for electricity in the computing power network is also volatile. When the volatility of the power supply side and the demand side does not match, it will bring more operating costs to the computing power network. For example, when the power generation is too much and the power consumption is too little, the green electricity needs to be stored in the energy storage device, which brings the operation and maintenance costs of the energy storage device; when the power generation is too little and the power consumption is too much, the thermal power of the power grid needs to be used to provide power support, and the price of thermal power is higher than that of green power, which brings higher operating costs. In an ideal state, when the power generation and power consumption match each other, the operation and maintenance costs of the energy storage equipment can be reduced, and the purchase of thermal power with higher electricity prices can be reduced, thereby reducing the operating costs of the data center.

[0032] Therefore, applying green power resources and energy storage to data centers, rationally planning computing resources and effectively scheduling data center electricity consumption while ensuring the operational reliability of the data center are crucial to the safe and green operation of the entire data center. Reasonable solutions to green power resource scheduling and electricity management are the key to optimizing data center operating costs.

[0033] The existing data center power dispatching optimizes the overall dispatching of computing power, electricity, heat and other energy sources, which can reduce energy waste and increase the consumption of clean energy. However, it does not take into account the intermittent, random and sudden changes of green electricity, resulting in poor economic benefits.

[0034] In addition, existing data centers perform real-time energy management on power loads operating in an uncertain green power resource environment without considering the intermittent, random, and sudden characteristics of clean energy in the energy system. This results in poor service stability in the data center and only considers the economic feasibility of the data center from the perspective of energy consumption costs.

[0035] In addition, existing data centers can improve the utilization rate of green electricity resources by using data center computing power time shifting and cooling, energy storage and diesel generators alternately, and use energy storage and data center computing power scheduling strategies to improve the utilization rate of green electricity in data centers, but the electricity cost is not considered.

[0036] Based on this, an embodiment of the present application proposes a method for coordinated scheduling of computing power and electricity, which can reduce the operating costs of a data center when coordinated scheduling of computing power and electricity.

[0037] The technical solution of the present application is described below through specific embodiments.

[0038] Reference Figure 1 , shows a schematic flow chart of the steps of a method for coordinated scheduling of computing power and electricity provided in an embodiment of the present application, which may specifically include the following steps:

[0039] S101, determining the predicted green power generation within a preset time.

[0040] The execution subject of the embodiment of the present application may be a computer device, and the present application does not limit the specific type of the computer device. The method in the present application can be applied to a data center in a computing network, and the computer device may be a scheduling server of the data center. The scheduling server may receive computing tasks and distribute the computing tasks to various computing devices in the data center for execution. During the operation of the data center, computing devices and cooling equipment, etc., all need to consume electricity.

[0041] The above-mentioned green electricity generation can be the generation of green electricity, that is, the green electricity generation obtained by converting renewable resources such as wind energy and solar energy. The generation methods of green electricity can include photovoltaic power generation and wind power generation. Photovoltaic power generation refers to a power generation method that directly converts solar radiation into electrical energy. When photovoltaic power generation is carried out, when the photovoltaic cell is determined, the power generation is related to the light intensity. Wind power generation is the conversion of wind energy into electrical energy. When the blade area, wind wheel efficiency, etc. of the wind turbine are determined, the power generation is related to the wind speed. For example, when the wind speed doubles, the capacity of wind power generation will increase by 8 times. Therefore, wind energy can be better predicted by accurately measuring wind speed. Green electricity generation is related to weather data such as wind, sunlight, and pressure. Due to the influence of weather conditions, green electricity generation is not fixed, but has a certain volatility.

[0042] Therefore, when predicting green power generation, it can be predicted based on weather data. When predicting green power generation within a preset time, the predicted weather data within the preset time can be obtained first. The prediction of weather data can be based on numerical weather forecast (Numerical Weather Prediction, NWP). Numerical weather forecast is a way to predict future weather conditions by using computers to simulate atmospheric movement and changes based on meteorological principles and mathematical models.

[0043] After obtaining the forecast weather data, the forecast weather data can be input into the forecast network model to obtain the forecast power generation. The forecast network model is obtained by training the bidirectional long short-term memory network (BiLSTM) with historical weather data and corresponding historical green power generation as sample data.

[0044] Bidirectional long short-term memory network is a network model improved based on LSTM (Long Short-Term Memory Network). LSTM is a recursive neural network that can effectively process long-term dependencies in sequence data. However, LSTM can only consider the context information before the current moment and cannot capture subsequent context information. Therefore, BiLSTM introduces a reverse network that can consider the previous and subsequent context information at the same time, so as to better process sequence data.

[0045] The computer device can obtain the historical power generation information of green electricity and the corresponding historical weather data, and then use the historical power generation information and the corresponding historical weather data to train the bidirectional long short-term memory network to obtain a prediction network model. For example, the historical power generation information and the corresponding historical weather data can be preprocessed to obtain multidimensional sample data including time, irradiance, wind speed, wind direction, temperature, humidity, pressure, actual power, etc., and then the multidimensional sample data is input into the bidirectional long short-term memory network for model training, so as to obtain a prediction network model. The input data of the prediction network model may include time and weather data, and the output data may include the actual power generated. The weather data may include irradiance, wind speed, wind direction, temperature, humidity, pressure, etc.

[0046] By inputting the acquired forecast weather data into the pre-trained forecast network model, the green electricity generation at the preset time can be obtained.

[0047] S102, obtaining the current power consumption information, the computing task information to be executed and the available computing power information of the data center, wherein the current power consumption information includes the current power demand, and the computing task information includes the computing time information and the interaction relationship information of multiple computing tasks.

[0048] The above-mentioned preset time can use the current time as the starting time. For example, the preset time can be one hour after the current time. At the current time, there are computing tasks currently being executed in the data center. The computing power equipment and cooling equipment in the data center all require power supply. The above-mentioned current power demand can be the amount of electricity consumed by the computing tasks currently being executed by the data center. Exemplarily, the computer equipment can capture the real-time power demand data of subsystems such as the computing power equipment and cooling equipment in the data center through the power environment monitoring system of the data center, thereby determining the power demand of the data center, that is, the current power demand.

[0049] The data center may also include multiple computing tasks that are not currently executed and are waiting to be executed. Each computing task may have information such as task arrival time, execution time information, computing resource requirements and / or task interaction relationships. The task interaction information may include execution order dependency information between computing tasks. For example, computing task A needs to use the calculation results of computing task B, or computing task B must be executed before computing task A. The execution time information may include information such as the latest execution time and the earliest execution time, based on the execution time. Based on the execution time information of the computing task and the interaction relationship information, the execution priority of the computing task can be determined.

[0050] The above-mentioned available computing power information may include the remaining computing resource information of each computing power device in the data center. Each computing power device in the data center may include multiple computing resources, for example, computing resources may include multiple types such as central processing unit (CPU), graphics processing unit (GPU), memory, disk capacity, bandwidth, etc. When executing computing tasks, computing power support is required from the remaining computing resources in the data center.

[0051] S103: Based on the computing time information and the interaction relationship information, use a queuing network model to sort the computing tasks to obtain a computing task sequence.

[0052] Queuing network, also known as queuing graphical review technology, is a network model that combines random service system theory with GERT network technology. It is used to solve network problems that require queuing and are difficult to accurately describe with GERT network models. That is, the implementation condition of a node requires not only the completion of the previous activity, but also a certain amount of traffic, or in other words, it is necessary to wait in line before entering a certain activity.

[0053] Based on the queuing network, the priority of computing tasks can be queued to obtain a computing task sequence. The order of each computing task in the computing task sequence can be used to characterize the execution priority. For example, the earlier the computing task, the higher the execution priority.

[0054] S104, based on the predicted green electricity generation, the current electricity demand and the available computing power information, determining a plurality of computing tasks from the computing task sequence as target computing tasks within the preset time.

[0055] The execution time information of the computing tasks may include the latest execution time information. Based on the latest execution time information of each computing task in the computing task sequence, the first computing task that must be executed within a preset time may be determined.

[0056] The first computing task has first computing resource demand information, which may include the demand for various types of computing resources. Based on the demand for various types of computing resources, the number of times each computing resource, such as a processor and memory, needs to work can be determined, thereby calculating the first amount of electricity consumed in executing the first computing task.

[0057] The computer device can calculate the sum of the first power amount and the current power demand amount, and the sum of the power amount is the power that the data center must consume to execute the first computing task within a preset time.

[0058] The computer device can calculate the power difference value obtained by subtracting the power sum from the predicted green power generation, and the power difference value can reflect whether the green power generation can provide sufficient power to perform the first computing task.

[0059] Based on the first computing resource demand information and the available computing power information, a computing resource difference is determined, which can reflect whether the data center has sufficient computing power to execute the first computing task.

[0060] Based on the first computing task, the power difference and the computing resource difference, the computer device can determine the target computing task to be performed by the data center within a preset time.

[0061] For example, if both the power difference and the computing resource difference are positive values, it can be shown that the data center can provide enough power to perform the first computing task and has enough computing power to perform the first computing task. At the same time, the data center may also have remaining available computing resources and remaining green power. At this time, the second computing task can be determined from the computing task sequence based on the power difference and the computing resource difference in the order of arrangement.

[0062] Each computing task may have a computing resource requirement, and the power consumption of the computing task may be calculated based on the computing resource requirement. Thus, each computing task may have a corresponding computing resource requirement and power consumption.

[0063] The computing resource requirements and power consumption of each computing task arranged after the first computing task can be obtained from the computing task sequence. Then, based on the power difference and computing resource difference, that is, based on the remaining available computing resources and remaining green power of the data center, the second computing task that can be performed by the data center is determined. Then the first computing and the second computing tasks are used as the above-mentioned target computing tasks.

[0064] If the power difference is 0 or the computing resource difference is 0, then the data center has no remaining available computing resources or no remaining green power, that is, the data center can only execute the first computing task within the preset time, and the first computing task can be used as the above target computing task.

[0065] If the power difference is negative and the computing resource difference is positive, it means that the data center has sufficient computing power to execute the first computing task within the preset time, but cannot provide enough electricity to execute the first computing task. When green electricity is insufficient, it can be obtained from energy storage devices or purchased from the power grid. Therefore, the first computing task can be used as the target computing task.

[0066] If the difference in computing resources is negative, it means that the data center does not have enough computing power to perform the first computing task within the preset time, but the first computing task must be performed within the preset time, so the computer device can split the first computing task into a third computing task performed in the current data center and a fourth computing task performed in other data centers. Data centers can exist in the computing power network. The computing power network is a key infrastructure to support the high-quality development of the digital economy. It can connect multi-source heterogeneous and massive ubiquitous computing power through the network to achieve efficient resource scheduling, green and low-carbon facilities, flexible computing power supply, and intelligent on-demand services. The computing power network can provide computing power support for technologies such as privacy computing, federated learning, and blockchain. The computing power network can include multiple data centers. Therefore, when the computing power of a data center is insufficient, the computing task can be sent to other data centers in the computing power network.

[0067] The computer device may use the third computing task as the target computing task; and send the fourth computing task to other data centers for execution.

[0068] In the embodiments of the present application, in view of the intermittent and random characteristics of green electricity resources, a large amount of historical data can be used to obtain a green electricity generation prediction network model, so that the green electricity generation can be predicted. The scheduling of electricity and computing power based on the predicted green electricity generation can reduce the operating costs of the data center while considering the execution characteristics of the computing tasks. The exchange between power grids or the storage of excess electricity in energy storage devices. According to the average response time of different types of computing tasks, the amount of electricity exchanged between the data center and the power grid, the dynamic operation of the energy storage device charging / discharging, and the electricity price are used as factors affecting the operating costs of the data center. A dynamic optimization scheduling model that minimizes the operating costs is constructed to obtain the optimal resource configuration of the data center, and determine the computing resources that need to be scheduled, as well as the operating status of the computing equipment and cooling equipment in the data center.

[0069] Reference Figure 2 , shows a schematic flow chart of another method for collaboratively scheduling computing power and electricity provided in an embodiment of the present application, which may specifically include the following steps:

[0070] S201, determining the predicted green electricity generation within a preset time.

[0071] The execution subject of this embodiment may be the computer device of the previous embodiment.

[0072] S202, obtaining the current power consumption information, the computing task information to be executed and the available computing power information of the data center, wherein the current power consumption information includes the current power demand, and the computing task information includes the computing time information and the interaction relationship information of multiple computing tasks.

[0073] S203: Based on the computing time information and the interaction relationship information, use a queuing network model to sort the computing tasks to obtain a computing task sequence.

[0074] S204: Based on the predicted green electricity generation, the current electricity demand and the available computing power information, determine a plurality of computing tasks from the computing task sequence as target computing tasks within the preset time.

[0075] S201 - S204 of this embodiment are similar to S101 - S104 of the previous embodiment, and they can be referenced to each other and are not described in detail here.

[0076] S205: Based on the predicted green power generation, the target computing task and the reserve power, perform power dispatch for the target computing task.

[0077] Under ideal conditions, the predicted green electricity generation can just meet the electricity demand within the preset time. At this time, the data center can only use the predicted power generation to provide power support for the data center's computing equipment, cooling equipment, etc.

[0078] However, it is predicted that green electricity generation may be excessive or insufficient.

[0079] The required power within the preset time can be the sum of the power consumed by executing the target computing task and the current power demand. Therefore, the target power consumed by executing the target computing task can be determined; then the target power and the current power demand are added to obtain a second power, which is the total power consumed by the equipment in the data center within the preset time.

[0080] By comparing the predicted green electricity generation with the second electricity quantity, it can be determined whether the predicted green electricity generation meets the requirement.

[0081] If the predicted green power generation is equal to the second power, then there is no need to perform power dispatch. If the predicted green power generation is greater than the second power, then the excess power can be stored in the energy storage device.

[0082] If the predicted green electricity generation is less than the second amount of electricity, additional electricity needs to be obtained from the energy storage device and / or the power grid. The electricity in the power grid has a grid electricity price, and the green electricity in the energy storage device has a green electricity price. The grid electricity price is generally higher than the green electricity price, but the energy storage device needs to be operated and maintained. The energy storage device has an operation and maintenance cost, and the number of operations and maintenance of the energy storage device is related to its charging and discharging frequency. When performing power dispatching, if you want to reduce operating costs, you can determine the power dispatching plan based on the grid electricity price, green electricity price, and the operation and maintenance price of the energy storage device, so as to minimize the cost of the operation center.

[0083] The computer device can obtain the grid electricity price, green electricity price and operation and maintenance price of the energy storage device within a preset time; then based on the grid electricity price, operation and maintenance price and green electricity price, determine the grid dispatched electricity dispatched from the grid and the energy storage dispatched electricity dispatched from the energy storage device, where the grid dispatched electricity and the energy storage dispatched electricity are selected based on the lowest operating cost.

[0084] Exemplarily, the grid dispatching power dispatched from the grid and the energy storage dispatching power dispatched from the energy storage device are determined based on the grid electricity price, operation and maintenance price, and green electricity price by the following formula:

[0085] A=x+y

[0086]

[0087] Among them, A is the difference between the second power and the predicted green power generation, x is the power dispatched by the power grid, y is the power dispatched by the energy storage, and B is the operating cost. B1 is the minimum value of B, x1 is the power dispatched by the power grid when the operating cost is the lowest, y1 is the power dispatched by the energy storage when the operating cost is the lowest, v1 is the power grid price, v2 is the green power price, v3 is the operation and maintenance price of a single energy storage device, and n is the reserve power of a single energy storage device. At this time, the power dispatched by the power grid dispatched from the power grid can be x1, and the power dispatched by the energy storage device can be y1.

[0088] The present application proposes a collaborative scheduling method based on data center computing power and green electricity resources. In view of the intermittent and random characteristics of green electricity resources, a large amount of historical data can be used to obtain a green electricity generation prediction network model, so that the green electricity generation can be predicted. The scheduling of electricity and computing power based on the predicted green electricity generation can reduce the operating costs of the data center while considering the execution characteristics of the computing tasks. In the embodiment of the present application, the optimization scheduling model constructed based on the predicted green electricity output power can be used to derive the allocation of computing power resources and power consumption plans, considering the response time of the data center computing power task execution and the real-time changes in the power generation power and electricity prices of green electricity resources in many aspects to ensure the reliability and economy of the data center operation. The method in the embodiment of the present application can effectively manage the power of the data center on the premise of reasonably meeting the computing power task arrangement, and reasonably solve the resource scheduling to minimize the operating costs.

[0089] It should be noted that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0090] Reference Figure 3 , shows a schematic diagram of a computing power and power coordinated scheduling device provided in an embodiment of the present application, which may specifically include a power generation prediction module 31, an information acquisition module 32, a task sorting module 33 and a task scheduling module 34, wherein:

[0091] The power generation prediction module 31 is used to determine the predicted green power generation within a preset time;

[0092] An information acquisition module 32 is used to acquire current power consumption information of the data center, information on computing tasks to be executed, and available computing power information, wherein the current power consumption information includes current power demand, and the computing task information includes computing time information and interaction relationship information of multiple computing tasks;

[0093] A task sorting module 33 is used to sort each of the computing tasks using a queuing network model based on the computing time information and the interaction relationship information to obtain a computing task sequence;

[0094] The task scheduling module 34 is used to determine a plurality of the computing tasks from the computing task sequence as target computing tasks within the preset time based on the predicted green electricity generation, the current electricity demand and the available computing power information.

[0095] In a possible implementation, the power generation prediction module 31 includes:

[0096] A forecast weather data acquisition submodule is used to acquire the forecast weather data within the preset time;

[0097] The predicted power generation determination submodule is used to input the predicted weather data into the prediction network model to obtain the predicted power generation. The prediction network model is obtained by training a bidirectional long short-term memory network with historical weather data and corresponding historical green power generation as sample data.

[0098] In a possible implementation, the task scheduling module 34 includes:

[0099] A first computing task determination submodule is used to determine a first computing task in the computing task sequence that must be executed within the preset time, wherein the first computing task has first computing resource requirement information;

[0100] A first power calculation submodule, configured to determine a first power consumed by executing the first computing task based on the first computing resource requirement information;

[0101] An electric quantity sum calculation submodule, used for calculating the electric quantity sum of the first electric quantity and the current electric power demand;

[0102] An electric quantity difference calculation submodule, used for calculating the electric quantity difference of the predicted green electric power generation minus the electric quantity sum;

[0103] A computing resource difference calculation submodule, used to determine a computing resource difference based on the first computing resource requirement information and the available computing power information;

[0104] A target computing task determination submodule is used to determine the target computing task based on the first computing task, the power difference and the computing resource difference.

[0105] In a possible implementation, the target computing task determination submodule includes:

[0106] A first determining unit is configured to determine, if both the power difference and the computing resource difference are positive values, a second computing task from the computing task sequence based on the power difference and the computing resource difference and in an arrangement order; and use the first computing task and the second computing task as the target computing task;

[0107] The second determining unit is configured to use the first computing task as the target computing task if the power difference is a negative value and the computing resource difference is a positive value.

[0108] In a possible implementation, the target computing task determination submodule further includes:

[0109] a splitting unit, configured to split the first computing task into a third computing task executed in the current data center and a fourth computing task executed in other data centers if the computing resource difference is a negative value;

[0110] A third determining unit, configured to use the third computing task as the target computing task;

[0111] A sending unit is used to send the fourth computing task to other data centers for execution.

[0112] In a possible implementation, the current power usage information further includes the reserve power of the energy storage device of the data center, and the above-mentioned device further includes:

[0113] The power dispatching module is used to perform power dispatching for the target computing task based on the predicted green power generation, the target computing task and the reserve power.

[0114] In a possible implementation, the power dispatching module includes:

[0115] A target power determination submodule, used to determine the target power consumed by executing the target computing task;

[0116] A second power determination submodule, configured to add the target power and the current power demand to obtain a second power;

[0117] a price information acquisition submodule, configured to acquire the power grid price, the green power price and the operation and maintenance price of the energy storage device within the preset time if the predicted green power generation is less than the second power amount;

[0118] The dispatching power determination submodule is used to determine the grid dispatching power dispatched from the grid and the energy storage dispatching power dispatched from the energy storage device based on the grid electricity price, the operation and maintenance price and the green electricity price.

[0119] In a possible implementation, the grid dispatched electricity amount dispatched from the grid and the energy storage dispatched electricity amount dispatched from the energy storage device are determined based on the grid electricity price, the operation and maintenance price, and the green electricity price by the following formula:

[0120] A=x+y

[0121]

[0122]

[0123] Among them, A is the difference between the second power and the predicted green electricity generation, B1 is the minimum value of B, x1 is the grid dispatched power, y1 is the energy storage dispatched power, v1 is the grid electricity price, v2 is the green electricity price, v3 is the operation and maintenance price of a single energy storage device, and n is the reserve power of a single energy storage device.

[0124] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiment part.

[0125] Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present application. Figure 4 As shown, the computer device 4 of this embodiment includes: at least one processor 40 ( Figure 4 Only one is shown in the figure), a memory 41, and a computer program 42 stored in the memory 41 and executable on the at least one processor 40, wherein the processor 40 implements the steps of any of the above-mentioned method embodiments when executing the computer program 42.

[0126] The computer device 4 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud computing device. The computer device may include, but is not limited to, a processor 40 and a memory 41. Those skilled in the art will appreciate that Figure 4 It is only an example of the computer device 4 and does not constitute a limitation on the computer device 4. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0127] The processor 40 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0128] In some embodiments, the memory 41 may be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 4. Further, the memory 41 may also include both an internal storage unit and an external storage device of the computer device 4. The memory 41 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory 41 may also be used to temporarily store data that has been output or is to be output.

[0129] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0130] An embodiment of the present application provides a computer program product. When the computer program product is run on a computer device, the computer device can implement the steps in the above-mentioned method embodiments when executing the computer device.

[0131] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application is described in detail with reference to the above-mentioned embodiments, a person skilled in the art should understand that the technical solutions described in the above-mentioned embodiments can still be modified, or some of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for coordinated scheduling of computing power and electricity, characterized in that: include: Determine the forecasted green electricity generation within a preset time period; Obtaining current power usage information, computing task information to be executed, and available computing power information of the data center, wherein the current power usage information includes current power demand, and the computing task information includes computing time information and interaction relationship information of multiple computing tasks; Based on the computing time information and the interaction relationship information, each of the computing tasks is sorted using a queuing network model to obtain a computing task sequence; Based on the predicted green electricity generation, the current electricity demand and the available computing power information, a plurality of the computing tasks are determined from the computing task sequence as target computing tasks within the preset time.

2. The method according to claim 1, characterized in that The step of determining the predicted green electricity generation within a preset time period includes: Obtaining forecast weather data within the preset time; The predicted weather data is input into a prediction network model to obtain the predicted green electricity generation. The prediction network model is obtained by training a bidirectional long short-term memory network using historical weather data and corresponding historical green electricity generation as sample data.

3. The method according to claim 1 or 2, characterized in that The determining, based on the predicted green electricity generation, the current electricity demand and the available computing power information, a plurality of computing tasks as the target computing tasks from the computing task sequence includes: Determine a first computing task in the computing task sequence that must be executed within the preset time, the first computing task having first computing resource requirement information; Determine a first amount of power consumed by executing the first computing task based on the first computing resource demand information; Calculating the sum of the first power quantity and the current power demand; Calculate the difference between the predicted green electricity generation and the sum of the electricity; Determining a computing resource difference based on the first computing resource requirement information and the available computing power information; The target computing task is determined based on the first computing task, the power difference and the computing resource difference.

4. The method according to claim 3, characterized in that The determining the target computing task based on the first computing task, the power difference and the computing resource difference includes: If both the power difference and the computing resource difference are positive values, then based on the power difference and the computing resource difference, in order of arrangement, determine a second computing task from the computing task sequence; and use the first computing task and the second computing task as the target computing task; If the power difference is a negative value and the computing resource difference is a positive value, the first computing task is used as the target computing task.

5. The method according to claim 4, characterized in that The method further comprises: If the computing resource difference is a negative value, splitting the first computing task into a third computing task executed in the current data center and a fourth computing task executed in other data centers; Using the third computing task as the target computing task; The fourth computing task is sent to other data centers for execution.

6. The method according to any one of claims 1-2 or 4-5, characterized in that: The current power consumption information also includes the reserve power of the energy storage device of the data center, and the method further includes: Based on the predicted green electricity generation, the target computing task and the reserve electricity, power scheduling is performed for the target computing task.

7. The method according to claim 6, characterized in that The performing power dispatching on the data center based on the predicted green power generation, the target computing task and the reserve power includes: Determining a target amount of power consumed by executing the target computing task; Adding the target power and the current power demand to obtain a second power; If the predicted green electricity generation is less than the second electricity amount, obtaining the power grid electricity price, green electricity price and operation and maintenance price of the energy storage device within the preset time; Based on the grid electricity price, the operation and maintenance price and the green electricity price, the grid dispatching power dispatched from the grid and the energy storage dispatching power dispatched from the energy storage device are determined.

8. The method according to claim 7, characterized in that The grid dispatched electricity quantity dispatched from the grid and the energy storage dispatched electricity quantity dispatched from the energy storage device are determined based on the grid electricity price, the operation and maintenance price and the green electricity price by the following formula: A=x+y Among them, A is the difference between the second electricity amount and the predicted green electricity generation, B is the operating cost, x is the grid dispatched electricity, and y is the energy storage dispatched electricity; B1 is the minimum value of B, x1 is the grid dispatched electricity corresponding to the minimum operating cost, y1 is the energy storage dispatched electricity corresponding to the minimum operating cost, v1 is the grid electricity price, v2 is the green electricity price, v3 is the operation and maintenance price of a single energy storage device, and n is the reserve electricity of a single energy storage device.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images