A method and system for collaborative scheduling of computing power and electricity based on multiple levels of requirements

Abstract

The application relates to the technical field of cloud computing and servers, and particularly discloses a computing power and power collaborative scheduling optimization method and system based on multiple levels of requirements, wherein the method comprises the following steps: determining a computing power supply-demand gap according to a computing power demand curve and a renewable energy power generation curve; determining a cost function of each period under each scene according to the computing power supply-demand gap, establishing a computing power scheduling model, a power scheduling model and a power scheduling optimization model based on a storage energy source; dividing the cost into a cost saving part and an energy saving and emission reduction part according to the cost function of each period under each scene; and minimizing the cost of the cost saving part and maximizing the income of the energy saving and emission reduction part through an iterative linear programming algorithm, so as to obtain a computing power allocation result of each period; and according to the computing power allocation result, the computing power supply-demand gap is gradually reduced through iteration until the supply-demand gap is smaller than a predetermined threshold, so that a final computing power and power collaborative scheduling scheme is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of cloud computing and servers, and particularly relates to a computing power and electricity collaborative scheduling optimization method and system based on multi-level requirements. BACKGROUND

[0002] With the rapid development of information technology, cloud computing, big data and artificial intelligence technologies are widely used, and the demand for computing power is rising. As the core infrastructure of modern information society, data centers undertake a large amount of computing tasks, and at the same time, a large amount of energy is consumed to maintain high computing power. Due to the limited computing power, data centers face the problem of how to efficiently allocate and use computing power. Computing power scheduling, that is, dynamically allocating computing resources of data centers according to the priority and energy consumption demand of tasks, is particularly important. This involves evaluating the priority of different tasks to ensure that critical tasks (such as emergency computing tasks, real-time data processing) are given priority, while non-critical tasks are appropriately delayed or rescheduled to adapt to changes in power supply. At the same time, in order to meet the energy demand, the traditional power used by data centers is increasing, leading to an increase in carbon emissions. In the context of increasing global environmental protection awareness, the use of green power (such as solar, wind, etc.) has become an important way to solve energy problems. Through smart grid technology and energy management systems, data centers can more flexibly manage their energy supply, preferentially use green power, and switch to traditional power when necessary. This collaborative scheduling not only improves energy use efficiency and reduces operating costs, but also maximizes environmental impact. In summary, in the process of rising global computing power demand and energy transformation, how to allocate computing power of data centers and how to balance the use of traditional power and green power have become key problems to be solved. SUMMARY

[0003] To achieve the purpose of the present application, the present application provides a computing power and electricity collaborative scheduling optimization method based on multi-level requirements, comprising:

[0004] Step S1: determining the computing power supply and demand gap according to the computing power demand curve and the renewable energy generation curve;

[0005] Step S2: determining the cost function of each period under each scenario according to the computing power supply and demand gap, establishing a computing power scheduling model, a power scheduling model and a power scheduling optimization model based on energy storage, wherein the scenarios include: the cost of only purchasing traditional power, the cost of only using renewable energy and the cost of mixed use of renewable energy and traditional power;

[0006] Step S3: According to the cost function of each period under each scenario, the cost is divided into a saving cost part and an energy saving and emission reduction part; and through an iterative linear programming algorithm, the cost of the saving cost part is minimized, and the benefit of the energy saving and emission reduction part is maximized, to obtain the computing power allocation result of each period;

[0007] Step S4: According to the computing power allocation result, steps S2-S4 are repeated to gradually narrow the computing power supply-demand gap until the supply-demand gap tends to 0, to obtain the final computing power and power collaborative scheduling scheme.

[0008] In some embodiments, step S2 includes:

[0009] According to the demand of the data center cluster for computing power, the sum of the traditional power and green power cost is minimized as the objective function, and according to the demand fluctuation of the supply and demand sides, the power balance of the supply and demand sides, the capacity of the battery storage and the maximum transmission power are preset as constraint conditions to establish a computing power scheduling model.

[0010] In some embodiments, step S2 includes:

[0011] On the basis of the computing power scheduling model, a segmented electricity price is introduced, the difference between the electricity purchased in a period and the proportion of green power and traditional power used is minimized as the objective function, and a day-ahead scheduling decision is made, considering market demand, new energy output and battery storage, and a power scheduling optimization model is established by using Pareto optimization.

[0012] In some embodiments, step S2 further includes:

[0013] On the basis of the power scheduling optimization model, a battery energy storage system is introduced, the difference between the electricity purchased in a period and the proportion of green power and traditional power used is minimized as the objective function, a power scheduling optimization model based on stored energy is established, the power scheduling when there is no sudden situation and the power scheduling when there is a sudden shortage of green power are determined.

[0014] In some embodiments, the power scheduling optimization model based on stored energy uses stored green power to replace traditional power, determines the storage power allocation in the sudden situation and the non-sudden situation, and calculates the funds that can be saved after optimization.

[0015] To achieve the same purpose of the invention, the application also provides a computing power and power collaborative scheduling optimization system based on multi-level requirements, which includes:

[0016] Gap determination module: for determining the computing power supply-demand gap according to the computing power demand curve and the renewable energy generation curve;

[0017] a model construction module configured to determine a cost function of each time period under each scenario according to the supply-demand gap of computing power, and to establish a computing power scheduling model, a power scheduling model, and a power scheduling optimization model based on energy storage, wherein each scenario includes a cost of only purchasing traditional power, a cost of only using renewable energy, and a cost of mixed use of renewable energy and traditional power;

[0018] a computing power allocation module configured to divide the cost into a cost saving part and an energy saving and emission reduction part according to the cost function of each time period under each scenario, and to obtain a computing power allocation result of each time period by using an iterative linear programming algorithm to minimize the cost of the cost saving part and maximize the benefit of the energy saving and emission reduction part;

[0019] a scheme determination module configured to gradually narrow the supply-demand gap by repeating steps S2-S4 according to the computing power allocation result, until the supply-demand gap tends to 0, and to obtain a final computing power and power collaborative scheduling scheme.

[0020] In some embodiments, the model construction module is configured to:

[0021] According to the demand of the data center cluster for computing power, the sum of the cost of traditional power and green power is minimized as an objective function, and the power balance on both supply and demand sides, the capacity of battery storage, and the maximum transmission power are preset as constraint conditions to establish a computing power scheduling model.

[0022] In some embodiments, the model construction module is configured to:

[0023] On the basis of the computing power scheduling model, a segmented electricity price is introduced, the difference between the cost of power purchased in a certain period and the proportion of green power and traditional power used is minimized as an objective function, and a day-ahead scheduling decision is made to consider market demand, new energy output, and battery storage, and a power scheduling optimization model is established by using Pareto optimization.

[0024] In some embodiments, the model construction module is further configured to:

[0025] On the basis of the power scheduling optimization model, a battery energy storage system is introduced, the difference between the cost of power purchased in a certain period and the proportion of green power and traditional power used is minimized as an objective function, a power scheduling optimization model based on energy storage is established, and the power scheduling under the condition of no sudden situation and the power scheduling under the condition of sudden shortage of green power are determined.

[0026] In some embodiments, the power dispatch optimization model based on the stored energy determines the storage power distribution in the case of an emergency and in the case of no emergency, and calculates the saved funds after optimization.

[0027] The beneficial effects of the above technical solutions are:

[0028] The present application establishes three models, including: a computing power dispatch model, a power dispatch model and a power dispatch optimization model based on stored energy, which respectively analyzes and optimizes the computing power and power collaborative dispatch of the data center. Through accurate power and computing power collaborative dispatch, not only the efficient operation of the data center can be ensured, but also the operation cost can be significantly reduced and the proportion of green power can be improved. In the case of no storage energy and no emergency, the power dispatch model optimizes the power dispatch to achieve the goals of cost saving and emission reduction. In the case of storage energy, the power dispatch optimization model based on stored energy further optimizes the use of power through the battery energy storage system (BESS), which not only reduces the cost, but also improves the proportion of green power. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0030] Figure 1 A flowchart of a computing power and power collaborative dispatch optimization method based on multi-level requirements provided by an embodiment of the present application is shown in the figure.

[0031] Figure 2 An hourly computing power allocation diagram of a computing power and power collaborative dispatch optimization method based on multi-level requirements provided by an embodiment of the present application is shown in the figure.

[0032] Figure 3 A power dispatch diagram before optimization of a computing power and power collaborative dispatch optimization method based on multi-level requirements provided by an embodiment of the present application is shown in the figure.

[0033] Figure 4 An energy allocation diagram after optimization of a computing power and power collaborative dispatch optimization method based on multi-level requirements provided by an embodiment of the present application is shown in the figure.

[0034] Figure 5 An optimized computing power dispatch scheme diagram of a computing power and power collaborative dispatch optimization method based on multi-level requirements provided by an embodiment of the present application is shown in the figure.

[0035] Figure 6 A schematic diagram of an optimized computing power scheduling scheme of a multi-level requirement-based computing power and power collaborative scheduling optimization method provided for an embodiment of the present application is shown in FIG. 1.

[0036] Figure 7 A schematic diagram of power scheduling in a 6 o'clock burst situation of a multi-level requirement-based computing power and power collaborative scheduling optimization method provided for an embodiment of the present application is shown in FIG. 2.

[0037] Figure 8 A schematic diagram of power scheduling in a 12 o'clock burst situation of a multi-level requirement-based computing power and power collaborative scheduling optimization method provided for an embodiment of the present application is shown in FIG. 3.

[0038] Figure 9 A schematic diagram of power scheduling in an 18 o'clock burst situation of a multi-level requirement-based computing power and power collaborative scheduling optimization method provided for an embodiment of the present application is shown in FIG. 4.

[0039] Figure 10 A structural schematic diagram of a multi-level requirement-based computing power and power collaborative scheduling optimization system provided for an embodiment of the present application is shown in FIG. 5. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments.

[0041] Examples of the embodiments are shown in the drawings, in which the same or similar symbols represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by reference to the drawings are exemplary and are intended to explain the present application, and cannot be understood as limiting the present application.

[0042] Embodiment One

[0043] An embodiment of the present application provides a multi-level requirement-based computing power and power collaborative scheduling optimization method, as shown in FIG. 1, which comprises: Figure 1

[0044] Step S1: determining a computing power supply-demand gap according to a computing power demand curve and a renewable energy generation curve;

[0045] Step S2: determining a cost function of each period under each scenario according to the computing power supply-demand gap, establishing a computing power scheduling model, a power scheduling model and a power scheduling optimization model based on a storage energy, wherein the scenarios include: a cost of only purchasing traditional power, a cost of only using renewable energy and a cost of mixed use of renewable energy and traditional power;

[0046] ​Step S3: According to the cost function of each period under each scenario, the cost is divided into a saving cost part and an energy saving and emission reduction part; and through an iterative linear programming algorithm, the cost of the saving cost part is minimized, and the benefit of the energy saving and emission reduction part is maximized, to obtain the computing power allocation result of each period;

[0047] Step S4: According to the computing power allocation result, steps S2-S4 are repeated to gradually narrow the computing power supply-demand gap until the supply-demand gap tends to 0, to obtain a final computing power and power collaborative scheduling scheme.

[0048] In one specific embodiment of the application, step S2 comprises:

[0049] According to the demand of the data center cluster for computing power, the sum of the traditional power and green power fees is minimized as an objective function, and according to the demand fluctuation of the supply and demand sides, a computing power scheduling model is established with the power balance of the supply and demand sides, the capacity of the battery storage and the maximum transmission power as constraint conditions.

[0050] Specifically, the minimum objective function in the jth hour is is determined according to the following formula:

[0051]

[0052]

[0053]

[0054] In the formula, i represents the task publishing time, j represents the task completion time; x ij , y ij , z ij respectively represent the amount of high, medium and low priority tasks completed in the jth hour and published in the ith hour; , , respectively represent the remaining amount of high, medium and low priority tasks in the jth hour; W x , W y , W z respectively represent the amount of electricity required to complete each high, medium and low priority task; α, β, γ respectively represent the weights of high, medium and low priority tasks.

[0055] By minimizing the objective function, the task procrastination which has negative impact on the quality of service can be reduced. Among them, the weights α, β, γ can reflect the priority of different tasks, that is, the higher the priority, the greater the weight. The setting of the weight is not absolute, and the application considers that the task with large power demand requires large amount of calculation, and the high energy consumption task is generally related to high income, assuming that the required power and income are in linear relationship, and the weight of the simplest integer ratio is used as the priority weighting, that is, the weights of high, medium and low priority tasks are 8, 5 and 3 respectively.(j i+1) is the weight of the procrastination time at j, and the value is set as the procrastination time of the remaining task at the end of the moment. Since the procrastination of the task will reduce the satisfaction of the customer, and has negative impact on the customer loyalty and subscription of the enterprise, the data center needs to complete the task with long procrastination time as soon as possible, so the priority of the task with long procrastination time is higher, and the weight is larger.

[0056] The constraint condition of the objective function is:

[0057]

[0058] The amount of task x completed in the jth hour of the ith hour released ij ,y ij ,z ij will not exceed the respective remaining amount of task in the jth hour 、 and The total power consumption of all tasks completed in this hour will not exceed the upper limit of computing power u per hour j , and the upper limit of computing power is equal in value to the upper limit of power.

[0059] Iterate by the following formula:

[0060]

[0061] The remaining task amount in the 0th hour is the task amount newly added at 0. The remaining task amount in the j+1th hour inherits the remaining task amount in the jth hour and the newly added task amount in the j+1th hour minus the task amount completed in the jth hour.

[0062] The iterative linear programming obtains the task arrangement at each moment, and according to the task arrangement, the ratio of the allocated computing power to the upper limit of computing power of each priority task per hour can be calculated. As follows:

[0063]

[0064]

[0065]

[0066] Based on the task arrangement at each time obtained by iterative linear optimization, the average delay time of each priority task is calculated, that is, the total delay time of all tasks of the priority is divided by the number of all tasks of the priority. The calculation process is as follows:

[0067] After the specific values of xij, yij and zij are obtained:

[0068]

[0069] Based on the computer iteration of the objective function, under the ideal state (the available power can support the upper limit of the computing power used by the data center every hour, 3000kW), the ratio of the computing power required by each priority task completed per hour to the upper limit of the computing power is as follows: Figure 2 As shown in the figure, in the first hour of receiving the task, the model will preferentially allocate computing power to high-priority tasks. However, when the medium-priority task is delayed for too long, the data center will consider the satisfaction of the customer of the medium-priority task, and temporarily suspend the completion of the high-priority task, and so on. Finally, all tasks can be completed in the day, and the average delay time of high, medium and low priority tasks is 1.90, 2.61 and 1.54 hours respectively.

[0070] According to the computing power scheduling scheme one in Figure 2 , the power scheduling scheme one is obtained, and the computing power and power collaborative scheduling under the cost of completing the purchase of traditional power is completed: as much as possible to use green power in the case of green power supply, and use traditional power to make up when green power cannot meet the demand for computing power.

[0071] As shown in Figure 3 , only the power used to complete the task is considered, and the power remaining to ensure that the computing power limit is reached is not considered, and the power scheduling cost is 32510 yuan, and the green power proportion is 64.66%.

[0072] In one specific embodiment of the application, step S2 comprises:

[0073] On the basis of the computing power scheduling model, a segmented electricity price is introduced, a difference between the electricity purchased in a certain period and the proportion of green power and traditional power used is minimized as an objective function, and a day-ahead scheduling decision is used. On the basis of considering market demand, new energy output and battery storage, a power scheduling optimization model is established by using Pareto optimization.

[0074] Specifically, the objective function is determined according to the following formula:

[0075] f(na,na+1,...,nb,ma,ma+1,...,mb):

[0076]

[0077] where a is the start time of the time period, b is the end time of the time period minus 1; n j , m j are the amount of green and traditional energy purchased at time j respectively; pj, qj are the price of green and traditional energy at time j; k is the weight of emission reduction coefficient.

[0078] The objective function represents the difference between the cost of purchased power and the proportion of green and traditional power used in a time period. is the cost of purchased power from time a to b+1, where a is the start time of the time period and b+1 is the end time of the time period. The total cost of purchased power can measure the cost, while the proportion of green and traditional power can measure the use of green power. k is the weight of emission reduction coefficient, which depends on the relative importance of the company to save cost and reduce emissions, and is initially set to 10.

[0079] The constraints of the objective function are:

[0080]

[0081] In the above constraints, j satisfies a≤j≤b, is the ideal state that the available power at each time is equal to the amount of power required to complete the task in the time period with the upper limit of computing power. The constraints ensure the following three requirements:

[0082] The purchased green power does not exceed its supply at that time;

[0083] The sum of purchased green power and traditional power does not exceed the upper limit of computer computing power, in order to avoid power waste;

[0084] The amount of power generated by the purchased power is equal to the amount of power required for task arrangement in the time period. This is because of the requirement for cost saving and emission reduction in problem two, the power at each time does not need to reach the upper limit of computing power, only needs to meet the requirement of this time period, which can guarantee to complete the task amount in the ideal state of each time period. For example, in the ideal state solution obtained in the computing power scheduling model, it can guarantee to complete all tasks in the first two hours in the first time period (0:00-6:00), so that the energy is surplus and wasted from 2:00 to 6:00.

[0085] Since the duration of power purchase is one hour, when calculating the energy generated by the power in the time period, only need to multiply the power value by the duration (1 hour).

[0086] The power scheduling model is solved using SLSQP (Sequential Least Squares Quadratic Programming) to solve nonlinear programming problems. Specifically, this algorithm approximates the original nonlinear optimization problem by iteratively solving a series of quadratic optimization problems. Quadratic programming problems (QP) are easier to solve than nonlinear optimization problems, and the computational complexity of the solution process is smaller. Through this step-by-step approximation, SLSQP can quickly obtain a near-optimal solution, thereby improving overall solution efficiency. For the data size of 24-hour power scheduling of the data center, SLSQP shows excellent solution speed, so it can quickly respond and plan power purchase.

[0087] After the scheduling optimization of traditional power and green power for each period, the power scheduling diagram is as follows Figure 4 As shown, the red part represents the mobilized traditional power, and the green part represents the mobilized green power, in kW. It can be seen from Figure 4 that the optimized scheduling can flexibly adjust the call power based on the task demand of each period, rather than using full power at each moment, causing waste of power. In addition to this, the algorithm will prefer to choose green power within each period. For example, in the first period (0:00-6:00), only 4:00-5:00 and 5:00-6:00 have green power supply, and the power supply is 500kW and 1400kW respectively. Therefore, the algorithm buys all the power of these two moments, generates power of 500kWh and 1400kWh, and meets the requirement of energy saving. In addition to this, for the period where only traditional power is available, the algorithm will automatically buy the traditional power with lower price. For example, in 22:00-23:00 and 23:00-0:00 of the next day, the green power supply is 0, and only traditional power can be purchased. Since the traditional power price of 23:00-0:00 of the next day is lower, more traditional power of 23:00-0:00 of the next day is called. Finally, the proportion of green power in the power scheduling model is 68.52%, and the cost is 32230 yuan, which is reduced by 280 yuan compared with scheme one.

[0088] The power scheduling scheme two given by the power scheduling model changes the upper limit of the computing power usage per hour u j =n j +m j in the model one to obtain the computing power scheduling scheme two. The average delay time of the high, medium and low priority tasks in scheme two is 1.90, 2.78 and 3.09 hours respectively.

[0089] In one specific embodiment of the present application, step S2 further comprises:

[0090] Based on the power dispatch optimization model, a battery energy storage system is introduced. The objective function is to minimize the difference between the electricity cost purchased in a certain period and the ratio of green electricity and traditional electricity used. A power dispatch optimization model based on energy storage is established to determine the power dispatch when the green electricity supply is insufficient to meet the computing power requirements under normal circumstances, and the power dispatch when there is a sudden shortage of green electricity under emergency circumstances.

[0091] In one specific embodiment of the present invention, the power dispatch optimization model based on energy storage uses stored green electricity to replace traditional electricity, determines the allocation of stored electricity under emergency and non-emergency conditions, and calculates the funds that can be saved after optimization.

[0092] Specifically, this model assumes that the data center's batteries consistently store 1000kWh of electricity under normal circumstances. Secondly, considering that green electricity supply exceeds the maximum computing power limit (3000kWh) during the following times each day (11:00-12:00, 12:00-13:00, 13:00-14:00, and 14:00-15:00), it can certainly be stored. Storing all available electricity can reach 1000kWh. Therefore, under normal circumstances, even if stored electricity is sometimes used, the green electricity available during these four periods exceeding the computing power limit can always be used to recharge 1000kWh. Based on the price of this electricity, the average price of a fully charged 1000kWh is 0.35 yuan / kWh.

[0093] Specifically, in the absence of sudden events, the objective function is determined according to the following formula:

[0094]

[0095] In the formula, c j Let q be the amount of stored power mobilized at time j; j Let be the price of traditional electricity at time j; and t be the moment when the supply of green energy drops to zero due to a sudden event. Optimize the energy mobilized from stored energy at each time after time t to maximize the savings from utilizing stored energy.

[0096] The constraints of the objective function are:

[0097]

[0098] Where, m j Let represent the amount of conventional electricity purchased at time j, as calculated in Model 2. Constraints guarantee that the amount of stored energy used does not exceed the storage limit. Only the use of stored energy is considered; storage itself is not considered, otherwise storage would not significantly affect the objective function. The amount of stored energy used does not exceed the original required conventional electricity.

[0099] The results of the power dispatch optimization model based on energy storage are as follows: Figure 6 As shown, the yellow portion represents the BESS power used, the green portion represents the purchased green power, and the red portion represents the purchased conventional power. The results indicate that using BESS power instead of conventional power during the hours with the highest conventional power prices, such as 10:00-11:00, 18:00-19:00, and 19:00-20:00, maximized cost savings. This power dispatching scheme cost 31,290 yuan and increased the proportion of green power to 70.32%.

[0100] Since the total power consumption per hour remains constant and the computing power scheduling remains unchanged, a third computing power-power coordinated scheduling scheme can be given based on the power scheduling optimization model based on energy storage and the second computing power scheduling scheme. The average drag time for each task is consistent with the data in the second computing power-power coordinated scheduling scheme.

[0101] Before an emergency occurs, since the emergency is unforeseen, the scheduling of stored energy is the same as under normal circumstances. Assuming that the green electricity supply suddenly drops to zero at time t, subsequent power scheduling will be optimized as follows. An emergency means that due to natural disasters or weather conditions, the green electricity supply suddenly decreases at a certain moment. For example, the clean energy supply drops to zero at 06:00.

[0102] In the event of an emergency, the objective function is determined according to the following formula:

[0103]

[0104] Optimize the energy allocated from storage at each time step after time t to maximize the cost savings from utilizing storage energy. The constraints are:

[0105]

[0106] The objectives of the constraints are almost identical to those in the no-contingency model, with the following changes: For the mobilized stored energy, the mobilization amount before time t is equal to the mobilization already solved under no-contingency conditions; the mobilization amount after time t is the variable to be optimized. The green electricity used at time t does not include green electricity purchased on time.

[0107] To verify the model's effectiveness under unforeseen circumstances, this group assumed that the green electricity supply would suddenly drop to zero at 6:00, 12:00, and 18:00, respectively. The study compared the cost savings and increased green electricity usage compared to directly purchasing conventional electricity using battery storage systems. The power dispatching scenario under the condition that the green electricity supply suddenly drops to zero at 6:00 is as follows: Figure 7 As shown, the power dispatching in the case where the green electricity supply suddenly drops to 0 at 12:00 is as follows: Figure 8As shown, the power dispatching when the green power supply suddenly becomes 0 at 18:00 is as follows Figure 9 As shown.

[0108] The above results show that the BESS power is still used as much as possible at the time when the traditional power is expensive, instead of filling the gap of green power. However, after using the BESS battery energy storage system, compared with purchasing traditional power when the green power supply is 0, the stored new energy can save 940, 940 and 950 yuan respectively for the cases that the green power supply is 0 at 6:00, 12:00 and 18:00, and the proportion of green power use in all energy use does not change. It can be seen that the storage system has a significant effect on fund saving, but the emission reduction effect is limited.

[0109] In addition, there are schemes with and without emergency conditions, and the computing power scheduling does not change. The average delay time of each task is consistent with the data in the second computing power and power collaborative scheduling scheme.

[0110] Embodiment two

[0111] One embodiment of the present application provides a computing power and power collaborative scheduling optimization system based on multi-level requirements, comprising:

[0112] The gap determination module 10 is configured to determine the computing power supply and demand gap according to the computing power demand curve and the renewable energy generation curve;

[0113] The model construction module 20 is configured to determine the cost function of each period under each scenario according to the computing power supply and demand gap, and establish a computing power scheduling model, a power scheduling model and a renewable energy-based power scheduling optimization model, wherein the scenarios include the cost of only purchasing traditional power, the cost of only using renewable energy, and the cost of mixed use of renewable energy and traditional power;

[0114] The computing power allocation module 30 is configured to divide the cost into a cost saving part and an energy saving and emission reduction part according to the cost function of each period under each scenario, and obtain the computing power allocation result of each period by minimizing the cost of the cost saving part and maximizing the income of the energy saving and emission reduction part through an iterative linear programming algorithm;

[0115] The scheme determination module 40 is configured to gradually reduce the computing power supply and demand gap by repeating steps S2-S4 according to the computing power allocation result, until the supply and demand gap tends to 0, and obtain the final computing power and power collaborative scheduling scheme.

[0116] In some embodiments, the model construction module 20 is configured to:

[0117] According to the demand of the data center cluster for computing power, the sum of the traditional power and green power cost is minimized as the objective function, and the power balance on both supply and demand sides, the capacity of the battery storage and the maximum transmission power are preset as constraint conditions to establish a computing power scheduling model.

[0118] In some embodiments, the model building module 20 is configured to:

[0119] On the basis of the computing power scheduling model, a segmented electricity price is introduced, the difference between the cost of electricity purchased in a certain period and the proportion of green power and traditional power used is minimized as the objective function, and a day-ahead scheduling decision is made, considering market demand, new energy output and battery storage, and a power scheduling optimization model is established by using Pareto optimization.

[0120] In some embodiments, the model building module 20 is further configured to:

[0121] On the basis of the power scheduling optimization model, a battery energy storage system is introduced, the difference between the cost of electricity purchased in a certain period and the proportion of green power and traditional power used is minimized as the objective function, a power scheduling optimization model based on stored energy is established, the power scheduling when the green power supply is insufficient for computing power under no sudden situation and the power scheduling when the green power is suddenly short of under sudden situation are determined.

[0122] In some embodiments, the power scheduling optimization model based on stored energy uses stored green power to replace traditional power, determines the storage power distribution under sudden and no sudden situations, and calculates the funds saved after optimization.

[0123] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0124] The various embodiments in the specification are described in progressive manner, and each embodiment focuses on the difference from other embodiments, and the same and similar parts between embodiments can be referred to each other. The embodiments of the present application are described with reference to flowcharts and / or block diagrams of the method, terminal device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the computer or other programmable data processing terminal device produce a device that implements the functions specified in the flow Figure 1 one or more flows and / or blocks Figure 1 one or more flows and / or blocks Figure 1 one or more flows and / or blocks Figure 1 one or more flows and / or blocks Figure 1 one or more flows and / or blocks Figure 1The steps of the functions specified in one or more boxes. Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the invention. Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0125] The methods and apparatus provided by the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

[0126] In the description of this specification, references to terms such as "an embodiment," "some embodiments," "example," "specific example," or "a specific embodiment" or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A multi-level requirement-based computing power and power collaborative scheduling optimization method, characterized in that, The method comprises the following steps: Step S1: determining a computing power supply-demand gap according to a computing power demand curve and a renewable energy power generation curve; Step S2: determining a cost function of each time period in each scenario according to the computing power supply-demand gap, and establishing a computing power scheduling model, a power scheduling model and a power scheduling optimization model based on a storage energy, wherein each scenario comprises a cost of only purchasing traditional power, a cost of only using renewable energy and a cost of mixed use of renewable energy and traditional power; Step S3: dividing the cost into a cost saving part and an energy saving and emission reduction part according to the cost function of each time period in each scenario; and through an iterative linear programming algorithm, minimizing the cost of the cost saving part while maximizing the benefit of the energy saving and emission reduction part, to obtain a computing power allocation result of each time period; Step S4: repeating steps S1-S3 according to the computing power allocation result, gradually narrowing the computing power supply-demand gap, until the supply-demand gap is less than a predetermined threshold, to obtain a final computing power and power collaborative scheduling scheme; Step S2 comprises: According to the demand of the data center cluster for computing power, taking the minimization of the sum of the traditional power and green power cost as the objective function, and according to the demand fluctuation of the supply and demand sides, presetting constraint conditions of power balance on both sides, capacity and maximum transmission power of the battery storage to establish a computing power scheduling model; Step S2 comprises: Introducing a segmented electricity price based on the computing power scheduling model, taking the minimization of the difference between the power purchased in a certain period and the proportion of green power and traditional power used as the objective function, and adopting a day-ahead scheduling decision, based on market demand, new energy output and battery storage, and using Pareto optimization to establish a power scheduling optimization model; Step S2 further comprises: Introducing a battery energy storage system based on the power scheduling optimization model, taking the minimization of the difference between the power purchased in a certain period and the proportion of green power and traditional power used as the objective function, establishing a power scheduling optimization model based on storage energy, determining the power scheduling when the green power supply is insufficient for the computing power required under the condition of no sudden situation, and the power scheduling when the green power is suddenly short of under the condition of sudden situation.

2. The method of claim 1, wherein, The power scheduling optimization model based on storage energy uses the stored green power to replace the traditional power, determines the storage power distribution under the condition of sudden situation and no sudden situation, and calculates the funds that can be saved after optimization.

3. A multi-level requirement-based computing power and power collaborative scheduling optimization system, characterized in that, The method comprises: A gap determination module configured to determine a computing power supply-demand gap according to a computing power demand curve and a renewable energy power generation curve; A model construction module configured to determine a cost function of each time period in each scenario according to the computing power supply-demand gap, and establish a computing power scheduling model, a power scheduling model and a power scheduling optimization model based on a storage energy, wherein each scenario comprises a cost of only purchasing traditional power, a cost of only using renewable energy and a cost of mixed use of renewable energy and traditional power; A computing power allocation module configured to divide the cost into a cost saving part and an energy saving and emission reduction part according to the cost function of each time period in each scenario; and through an iterative linear programming algorithm, minimize the cost of the cost saving part while maximizing the benefit of the energy saving and emission reduction part, to obtain a computing power allocation result of each time period; The scheme determination module is configured to: according to the computing power allocation result, repeat steps S1-S3, gradually narrow the computing power supply-demand gap, until the supply-demand gap is less than a predetermined threshold, and obtain a final computing power and power cooperative scheduling scheme; The model construction module is configured to: According to the demand of the data center cluster for computing power, taking the sum of the traditional power and green power cost as the objective function, and according to the demand fluctuation of the supply and demand sides, the power balance of the supply and demand sides, the capacity of the battery storage and the maximum transmission power are preset as constraint conditions to establish a computing power scheduling model; The model construction module is configured to: On the basis of the computing power scheduling model, a segmented electricity price is introduced, the difference between the electricity cost purchased in a certain period and the proportion of green power and traditional power used is minimized as the objective function, and a day-ahead scheduling decision is adopted, based on market demand, new energy output and battery storage, a power scheduling optimization model is established by using Pareto optimization; The model construction module is further configured to: On the basis of the power scheduling optimization model, a battery energy storage system is introduced, the difference between the electricity cost purchased in a certain period and the proportion of green power and traditional power used is minimized as the objective function, a power scheduling optimization model based on energy storage is established, the power scheduling under the condition of no sudden situation through green power storage to deal with the insufficient supply of green power for computing power required, and the power scheduling under the condition of sudden shortage of green power is determined.

4. The multi-level requirement-based computing power and power collaborative scheduling optimization system according to claim 3, characterized in that, The power scheduling optimization model based on energy storage uses the stored green power to replace the traditional power, determines the storage power distribution under the condition of sudden situation and no sudden situation, and calculates the funds that can be saved after optimization.

Images