Intelligent green energy consumption system and operation state optimization method thereof

Abstract

The invention discloses an intelligent green energy consumption system and an optimization method of an operation state of the intelligent green energy consumption system. Belongs to the field of photovoltaic and intelligent computing power, and comprises photovoltaic power generation equipment, an illumination sensor, a power distribution module, computing equipment and the like, a new optimization algorithm is applied, and the photovoltaic power generation equipment and a power supply alternating current side energy storage system perform power scheduling according to task priority to support computing power distribution; the monitoring system monitors various operation states of the computing equipment in real time so as to maintain the work of the system, and stable operation of the computing equipment is ensured; compared with the prior art, the method has the advantages that key industry computing power requirements can be fully explored, stock computing power resources can be lived, guidance on newly-added computing power construction requirements can be enhanced, the overall utilization rate of the computing resources can be practically improved, green low-carbon technology popularization and application can be enhanced, the green electricity use proportion of a data center can be improved, and the green computing power supply level can be enhanced; the autonomous protection level of the computing power infrastructure is further enhanced, and the security and reliability of the computing power infrastructure are ensured.

Description

Technical Field

[0001] The present invention belongs to the field of photovoltaics and intelligent computing power, and relates to intelligent green energy consumption technology, and in particular, to an optimization method utilizing an intelligent green energy consumption system and its operating status. Background Art

[0002] At present, electricity is usually supplied by traditional power grids, which often come from a variety of energy sources, including fossil fuels, so it is relatively less environmentally friendly. Relying on power grids for electricity may be affected by factors such as electricity price fluctuations, supply and demand, and grid stability. In addition, the cost of traditional electricity may increase due to rising fuel prices. Limited by the coverage and stability of the power grid, stable power supply may not be guaranteed in some remote areas or special environments. In the face of environmental pressure and energy transformation, it may be necessary to seek integration or transformation with clean energy to meet the needs of future sustainable development. Summary of the invention

[0003] In response to the above problems, the purpose of the present invention is to provide an optimization method for utilizing an intelligent green energy consumption system and its operating status, combining a new mathematical model and photovoltaic power generation with flexible computing power to ensure the stable operation of computing equipment and the rational use and allocation of resources.

[0004] The technical solution of the present invention is: the intelligent green energy consumption system described in the present invention includes photovoltaic power generation equipment, light sensors, power distribution modules, power supply AC energy storage systems, maintenance systems, monitoring systems and computing equipment composed of an algorithm optimization system and a computing power distribution module;

[0005] Electricity is generated through photovoltaic power generation equipment, and the operating status of photovoltaic power generation and power supply AC energy storage system is dynamically adjusted according to the computing power requirements of computing equipment to meet the electricity demand in different periods of time; light sensors can monitor the operating status of photovoltaic power generation equipment; power supply AC energy storage system releases stored electric energy during peak electricity demand periods, thereby reducing the load pressure on the power grid and storing electric energy during low electricity demand periods.

[0006] Furthermore, the power supply AC energy storage system is configured on the AC side of the photovoltaic power supply, and uses an independent charge and discharge controller and inverter to charge or invert the battery. It is added to the existing photovoltaic power generation system as an independent energy storage device to form an in-station energy storage system. It can also be built into a completely independently operated energy storage power station according to the different needs of the equipment. When photovoltaic power generation cannot meet the minimum power required for the operation of the computing equipment, the discharge of the power supply AC energy storage system is used to ensure the normal power supply of the computing equipment and the stable operation of the computing equipment.

[0007] Furthermore, the computing device allocates computing power according to the priority of the task, and can program and execute specific logic on the data center;

[0008] First, you need to categorize tasks and set the priority of each task. The priority can be set based on the urgency, importance, customer requirements, and business impact of the task.

[0009] Categorize tasks into three priorities: high, medium, and low;

[0010] Evaluate the currently available computing resources, including processors, memory, storage, and electricity; and develop a computing allocation plan based on task priorities and resource requirements.

[0011] Furthermore, the power distribution module obtains the operating status of the photovoltaic power generation equipment and the supply and demand status of the power of the computing equipment through the monitoring system and the light sensor; through the power distribution module, the supply and demand balance of the power of the computing equipment can be met, and when the photovoltaic power generation generates excess power, it is stored in the power supply AC energy storage system to provide power support when the demand of the computing equipment is at a peak;

[0012] The light sensor is used in monitoring solar photovoltaic power generation systems. In a photovoltaic power station, it is installed on the surface of a solar panel to monitor light intensity in real time, thereby predicting and adjusting the output of the photovoltaic power generation system.

[0013] Furthermore, the monitoring system monitors the operating temperature of the computing equipment and the real-time operating status of the computing equipment in real time, issues a real-time alarm when any of the above is abnormal, and cooperates with the maintenance system to repair it; when the power supply condition is abnormal, the power distribution module and the power supply AC side energy storage system will promptly dispatch and adjust the power distribution.

[0014] Furthermore, in the maintenance system, when the temperature of the computing equipment is abnormal, the operation of the water circulation refrigeration system is adjusted to lower the temperature of the computing equipment and maintain the operating temperature; when various equipment fails, the fault is repaired in time, and hardware maintenance of various equipment is performed regularly.

[0015] Furthermore, an optimization method using the operating status of the intelligent green energy consumption system, the algorithm optimization system and the computing power allocation module work together, the improved sparrow algorithm optimization system is optimized, and the computing power allocation module is adjusted to improve the computing power resources and energy utilization; the optimization steps are:

[0016] Step (1), task priority sorting;

[0017] Step (2), fitness function;

[0018] Step (3), location update rules.

[0019] Furthermore, in step (1), the task priority ranking is specifically:

[0020] After Sparrow assigns a processor to each task, the scheduling results are different due to different priority rankings; when the priority ranking strategy in the current task scheduling algorithm is single, different types of tasks use the same ranking method, and the differences between different types of tasks are not analyzed, resulting in incorrect ranking results; to address this problem, a task priority diversion ranking strategy is adopted; according to the communication-computation ratio of the task graph, the task graph is divided into computing-intensive and communication-intensive tasks to ensure that high-priority tasks get computing resources first, thereby improving the overall operating efficiency and response speed of the system;

[0021] The priority ranking formula for communication-intensive tasks is:

[0022]

[0023] Where, ranku(i) represents the priority value of communication-intensive task i; The average cost of the task is calculated on three processors; son(v i ) represents the subtask set of task i; C ij is the inter-task communication overhead;

[0024] The priority ranking formula for computationally intensive tasks is:

[0025]

[0026] In the formula, rankU(i) represents the priority value of computationally intensive task i, σ i The cost standard deviation is calculated for task i on three processors. The method for sorting computationally intensive tasks can be obtained by calculating the cost standard deviation, and the result of sorting the task sequence values ​​in descending order is used as the final scheduling order.

[0027] Furthermore, in step (2), the fitness function is specifically:

[0028] In order to ensure that different computing power allocation schemes are effectively compared on a unified metric basis, so as to select the optimal solution, the optimization target is set as the processor load balance index for the problem of processor load imbalance, and its calculation formula is:

[0029]

[0030] Where nump(m,i) represents the cumulative number of tasks of processor i in the sparrow m allocation scheme; The average number of tasks assigned to each processor; p is the number of processors; the smaller the pld value, the better;

[0031] The ultimate goal of the task scheduling algorithm is to reduce the overall completion time. Therefore, when the processor load tends to be balanced, the scheduling length is as short as possible. The scheduling length makespan calculation formula is:

[0032] makespan(m)=EFT(m,v exit )

[0033] In the formula, EFT(m,v exit ) represents the completion time of the export task in the sparrow m allocation scheme, and the smaller the scheduling length value, the better;

[0034] The combination of scheduling length and processor load balancing index is used as the final fitness function, but the values ​​of the two are quite different, so they cannot be added directly; in order to make the two at the same measurement level, they are first normalized;

[0035] The normalized processing formula of processor load balancing index (pld) is:

[0036]

[0037] Where max(pld) is the maximum value of the load balancing index in this iteration; normal_pld(m) represents the load balancing index after normalization of sparrow m;

[0038] The normalization formula for scheduling length makespan is:

[0039]

[0040] Where max(makespan) is the maximum scheduling length of this iteration, and normal_makespan(m) represents the normalized scheduling length of sparrow m;

[0041] After normalization, the fitness function calculation formula is:

[0042] fitness(m)=normal_pld(m)+normal_makespan(m)

[0043] After obtaining the fitness function value, it is sorted in descending order as the basis for the division of labor among the sparrow population. The sparrows with the highest ranking are regarded as discoverers, the rest as followers, and the sentinels are randomly selected.

[0044] Furthermore, in step (3), the location update rule is specifically:

[0045] In each iteration, the sparrow fitness function is calculated, the population is divided in descending order, and then the sparrow position information is calculated according to the encoding rules. Finally, the updated position information is calculated according to different update rules; the discoverer is a sparrow in a better position, with a better fitness function value, exploring potential high-quality computing resources or configurations globally. Its position update calculation formula is:

[0046]

[0047] In the formula, represents the location information of sparrow i1 after iteration; r is the alarm value, ranging from [0,1], safe is the safety threshold, ranging from [0.5,1]; α is a random number, ranging from [0,1], K is a random number that obeys the standard normal distribution; genmax is the maximum number of iterations;

[0048] Followers are the largest group. They use the information provided by the discoverer to conduct a more detailed search in the local area to find a better allocation of computing power. The position update calculation formula in the first half is:

[0049]

[0050] In the formula, Indicates the best position in this iteration; A takes the value of -1 or 1. The processor needs to be reallocated in the second half due to the poor position;

[0051] In the process of computing power allocation, by introducing the mechanism of updating the position of the sentinel, we can prevent the potential better computing power allocation plan from being missed due to falling into the local optimum too early; the calculation formula for updating the position of the sentinel is:

[0052]

[0053] In the formula, represents the global optimal position in the current iteration, represents the worst position of the current iteration, fit(best), fit(worse), and fit(i1) represent the global worst, best, and current individual fitness values, respectively; β is the step size control parameter that obeys the standard normal distribution; B is a random number between [-1,1]; ε is a small number to ensure that the denominator is not 0, and its value must be of the same order of magnitude as the difference in the fitness function value;

[0054] In order to balance the global search and local development capabilities, a decay function related to the number of iterations is introduced to reduce the value of β, so that the algorithm performs extensive search in the early iteration and fine local development in the later iteration. The formula is as follows:

[0055]

[0056] In the formula, β init is the initial value of β, t is the current number of iterations, genmax is the maximum number of iterations, and θ is a constant less than 1;

[0057] When the maximum number of iterations is reached, the optimal computing power allocation strategy and corresponding performance indicators are output.

[0058] The beneficial effects of the present invention are: 1. By combining photovoltaic power generation and flexible computing power, the computing power needs of key industries can be fully explored, the existing computing power resources can be activated, the guidance of the demand for new computing power construction can be strengthened, the overall utilization rate of computing resources can be effectively improved, the promotion and application of green and low-carbon technologies can be strengthened, the proportion of green electricity used in data centers can be increased, the level of green computing power supply can be enhanced, and the level of independent protection of computing power infrastructure can be further strengthened to ensure the safety and reliability of computing power infrastructure; 2. Due to the addition of photovoltaic power generation technology, the present invention directly provides power to the computing power center through the photovoltaic power generation system, which can reduce the loss in the power transmission process and improve energy utilization efficiency; 3. The present invention also uses a new mathematical model to sort the task priorities more accurately and avoid singleness; 4. The present invention also uses the sparrow optimization algorithm to find the best solution to improve the utilization rate of computing power resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a structural schematic diagram of a technology for absorbing intelligent green energy according to the present invention;

[0060] Figure 2 It is an algorithm flow chart of the sparrow optimization algorithm in the present invention. DETAILED DESCRIPTION

[0061] The specific technical scheme of the present invention is further described in detail below with reference to specific examples.

[0062] As shown in the figure, the present invention describes an optimization method for utilizing intelligent green energy consumption technology and its operating status, which combines a new mathematical model with photovoltaic power generation and flexible computing power to ensure stable operation of computing equipment and rational use and allocation of resources. This technology can effectively improve the overall utilization rate of computing resources and strengthen the promotion and application of green and low-carbon technologies.

[0063] The technical solution of the present invention is: a cloud platform-enterprise carbon emission and carbon footprint assessment device described in the present invention includes photovoltaic power generation equipment, light sensors, power distribution modules, power supply AC energy storage systems, maintenance systems, monitoring systems, and computing equipment composed of an algorithm optimization system and a computing power allocation module;

[0064] Electricity is generated through photovoltaic power generation facilities. According to the computing power requirements of computing equipment, the operating status of photovoltaic power generation and power supply AC energy storage system is dynamically adjusted to meet the power demand in different periods of time. Light sensors can effectively monitor the operating status of photovoltaic power generation. The power supply AC energy storage system releases stored electric energy during peak power demand periods to reduce grid load pressure. Electric energy is stored during low power demand periods in case of emergency.

[0065] Furthermore, the power supply AC energy storage system is configured on the AC side of the photovoltaic power supply, and uses an independent charge and discharge controller and inverter to charge or invert the battery; it can be added to the existing photovoltaic power generation system as an independent energy storage device to form an in-station energy storage system, or it can be built into a completely independent energy storage power station according to the different needs of the equipment; when photovoltaic power generation cannot meet the minimum power required for the operation of the computing equipment, the discharge of the power supply AC energy storage system is used to ensure the normal power supply of the computing equipment, thereby ensuring the stable operation of the computing equipment.

[0066] Furthermore, the computing device mainly allocates computing power according to the priority of the task, and can program and execute specific logic on the data center; first, the tasks need to be classified and the priority of each task needs to be set; the priority can be set according to factors such as the urgency, importance, customer requirements, and business impact of the task; the tasks can be divided into three priorities: high, medium, and low;

[0067] Evaluate the currently available computing resources, including processors, memory, storage, electricity, etc.; develop a computing power allocation plan based on task priority and resource requirements; high-priority tasks should receive more resources to ensure that they can be completed as quickly as possible.

[0068] Furthermore, the power distribution module obtains the operating status of the photovoltaic power generation equipment and the supply and demand status of the power of the computing equipment through the monitoring system and the light sensor;

[0069] The power distribution module can be used to balance the supply and demand of computing equipment. When photovoltaic power generation generates excess electricity, it can be stored in the power supply AC energy storage system to provide power support when the computing equipment needs peak, reducing the risk of equipment stopping operation.

[0070] Furthermore, the light sensor is widely used in the monitoring of solar photovoltaic power generation systems; in photovoltaic power stations, it is mainly installed on the surface of solar panels to monitor light intensity in real time, understand the lighting conditions in the current environment, and thus predict and adjust the output of the photovoltaic power generation system.

[0071] Furthermore, the monitoring system monitors the operating temperature of the computing equipment and the real-time operating status of the computing equipment in real time, issues a real-time alarm when any of the above is abnormal, and cooperates with the maintenance system to repair it; when the power supply condition is abnormal, the power distribution module and the power supply AC side energy storage system will promptly dispatch and adjust the power distribution.

[0072] Furthermore, when the temperature of the computing equipment becomes abnormal, the maintenance system adjusts the operation of the water circulation refrigeration system to lower the temperature of the computing equipment and maintain it at an appropriate operating temperature; when various equipment fails, it promptly repairs the fault and regularly performs hardware maintenance on various equipment; when new technologies emerge, it promptly updates the equipment and upgrades various systems.

[0073] Furthermore, an optimization method using intelligent green energy consumption technology and its operating status, the algorithm optimization system and the computing power allocation module work together, the improved sparrow algorithm optimization system is optimized, and the computing power allocation module is adjusted to improve the computing power resources and energy utilization rate; the specific optimization steps are:

[0074] (1) Task priority sorting:

[0075] After Sparrow assigns a processor to each task, the order of task scheduling is more important. Due to different priority rankings, the scheduling results are quite different. When the priority ranking strategy in the current task scheduling algorithm is single, different types of tasks use the same ranking method, and the differences between different types of tasks are not analyzed, which will lead to inaccurate ranking results. To address this problem, a task priority diversion ranking strategy is adopted. According to the communication-computation ratio of the task graph, the task graph is divided into computing-intensive and communication-intensive tasks to ensure that high-priority tasks get computing resources first, thereby improving the overall operation efficiency and response speed of the system.

[0076] The priority ranking formula for communication-intensive tasks is:

[0077]

[0078] Among them, ranku(i) represents the priority value of communication-intensive task i; The average cost of the task is calculated on three processors; son(v i ) represents the subtask set of task i; C ij is the inter-task communication overhead;

[0079] The priority ranking formula for computationally intensive tasks is:

[0080]

[0081] In the formula, rankU(i) represents the priority value of computationally intensive task i, σ i Calculate the cost standard deviation for task i on three processors. The cost standard deviation can be used to obtain a method that meets the requirements of computationally intensive tasks, and the result of sorting the task values ​​in descending order is used as the final scheduling order.

[0082] (2) Fitness function:

[0083] In order to ensure that different computing power allocation schemes are effectively compared on a unified metric basis, so as to accurately select the optimal solution, the optimization target is set as the processor load balance index for the problem of processor load imbalance, and its calculation formula is:

[0084]

[0085] Where nump(m,i) represents the cumulative number of tasks of processor i in the sparrow m allocation scheme; The average number of tasks assigned to each processor; p is the number of processors; the smaller the pld value, the better;

[0086] The ultimate goal of the task scheduling algorithm is to reduce the overall completion time. Therefore, when the processor load tends to be balanced, the scheduling length is as short as possible. The scheduling length makespan calculation formula is:

[0087] makespan(m)=EFT(m,v exit )

[0088] In the formula, EFT(m,v exit ) represents the completion time of the export task in the sparrow m allocation scheme, and the smaller the scheduling length value, the better;

[0089] The combination of scheduling length and processor load balancing index is used as the final fitness function, but the values ​​of the two are quite different, so they cannot be added directly; in order to make the two at the same measurement level, they are first normalized;

[0090] The normalized processing formula of processor load balancing index (pld) is:

[0091]

[0092] Where max(pld) is the maximum value of the load balancing index in this iteration; normal_pld(m) represents the load balancing index after normalization of sparrow m;

[0093] The normalization formula for scheduling length makespan is:

[0094]

[0095] Where max(makespan) is the maximum scheduling length of this iteration, and normal_makespan(m) represents the normalized scheduling length of sparrow m;

[0096] After normalization, the fitness function calculation formula is:

[0097] fitness(m)=normal_pld(m)+normal_makespan(m)

[0098] After obtaining the fitness function value, the sparrows are sorted in descending order as the basis for the division of labor among the sparrow population. The sparrows with the highest ranking are regarded as discoverers, the rest as followers, and the vigilants are randomly selected;

[0099] (3) Location update rules:

[0100] In each iteration, the sparrow fitness function is calculated, the population is divided in descending order, and then the sparrow position information is calculated according to the encoding rules. Finally, the updated position information is calculated according to different update rules; the discoverer is a sparrow in a better position, with a better fitness function value, exploring potential high-quality computing resources or configurations globally. Its position update calculation formula is:

[0101]

[0102] In the formula, represents the location information of sparrow i1 after iteration; r is the alarm value, ranging from [0,1], safe is the safety threshold, ranging from [0.5,1]; α is a random number, ranging from [0,1], K is a random number that obeys the standard normal distribution; genmax is the maximum number of iterations;

[0103] Followers are the largest group. They use the information provided by the discoverer to conduct a more detailed search in the local area to find a better allocation of computing power. The position update calculation formula in the first half is:

[0104]

[0105] In the formula, Indicates the best position in this iteration; A takes the value of -1 or 1. The processor needs to be reallocated in the second half due to the poor position;

[0106] In the process of computing power allocation, by introducing the mechanism of updating the position of the sentinel, we can prevent the potential better computing power allocation plan from being missed due to falling into the local optimum too early; the calculation formula for updating the position of the sentinel is:

[0107]

[0108] In the formula, represents the global optimal position in the current iteration, represents the worst position of the current iteration, fit(best), fit(worse), and fit(i1) represent the global worst, best, and current individual fitness values, respectively; β is the step size control parameter that obeys the standard normal distribution; B is a random number between [-1,1]; ε is a small number to ensure that the denominator is not 0, and its value must be of the same order of magnitude as the difference in the fitness function value;

[0109] In order to balance the global search and local development capabilities, a decay function related to the number of iterations is introduced to reduce the value of β, so that the algorithm performs extensive search in the early iteration and fine local development in the later iteration. The formula is as follows:

[0110]

[0111] In the formula, β init is the initial value of β, t is the current number of iterations, genmax is the maximum number of iterations, and θ is a constant less than 1;

[0112] When the maximum number of iterations is reached, the optimal computing power allocation strategy and corresponding performance indicators are output.

[0113] Implementation scenarios: The present invention provides an optimization method using an intelligent green energy consumption system and its operating status, which is mainly used in data centers and green computing environments. In these scenarios, the technology integrates photovoltaic power generation, energy storage systems, intelligent computing power allocation and optimization algorithms to achieve efficient use of green electricity and refined management of computing resources. In the actual operating environment of the data center, the technology can dynamically adjust the operating status of photovoltaic power generation and energy storage systems according to the power demand and task priority of the computing equipment, thereby ensuring efficient use of computing resources and maximum consumption of green electricity.

[0114] Possible implementation cases:

[0115] The present invention is applied to green computing power base stations in remote areas. Through the intelligent green energy consumption system, these base stations can realize independent power supply, use photovoltaic power generation equipment to provide stable electricity, and configure energy storage systems to cope with insufficient light or peak power consumption periods; the system can intelligently allocate computing power and schedule according to the business needs and task priorities of the base stations to ensure that key tasks are processed first. The efficient operation and maintenance system monitors the operating status and power supply and demand in real time through the monitoring system and maintenance system, promptly discovers and solves problems, and improves operation and maintenance efficiency; in addition, this technology can also be applied to the comprehensive green transformation of data centers, by deploying photovoltaic power generation and energy storage systems, integrating algorithm optimization and computing power allocation, and intelligently scheduling power supply and demand to ensure the maximum consumption of green electricity.

Claims

1. An intelligent green energy consumption system, characterized in that: It includes photovoltaic power generation equipment, light sensors, power distribution modules, power supply AC energy storage systems, maintenance systems, monitoring systems, and computing equipment composed of algorithm optimization systems and computing power distribution modules; Electricity is generated through photovoltaic power generation equipment. According to the computing power requirements of the computing equipment, the operating status of photovoltaic power generation and power supply AC energy storage system is dynamically adjusted to meet the power demand at different times. Light sensors can monitor the operating status of photovoltaic power generation equipment. The power supply AC energy storage system releases the stored electric energy during the peak period of electricity demand, thereby reducing the load pressure on the power grid and storing electric energy during the period of low electricity demand.

2. The intelligent green energy consumption system according to claim 1 is characterized in that: The power supply AC energy storage system is configured on the AC side of the photovoltaic power supply, and uses an independent charge and discharge controller and inverter to charge or invert the battery. It is added to the existing photovoltaic power generation system as an independent energy storage device to form an in-station energy storage system. It can also be built into a completely independent energy storage power station according to the different needs of the equipment. When photovoltaic power generation cannot meet the minimum power for the operation of the computing equipment, the discharge of the power supply AC energy storage system is used to ensure the normal power supply of the computing equipment, thereby ensuring the stable operation of the computing equipment.

3. The intelligent green energy consumption system according to claim 1 is characterized in that: The computing device allocates computing power according to the priority of the task and can be programmed and execute specific logic in the data center; First, you need to categorize tasks and set the priority of each task. The priority can be set based on the urgency, importance, customer requirements, and business impact of the task. Categorize tasks into three priorities: high, medium, and low; Evaluate the currently available computing resources, including processors, memory, storage, and electricity; and develop a computing allocation plan based on task priorities and resource requirements.

4. The intelligent green energy consumption system according to claim 1, characterized in that: The power distribution module obtains the operating status of the photovoltaic power generation equipment and the supply and demand status of the computing equipment power through the monitoring system and the light sensor; through the power distribution module, the supply and demand balance of the computing equipment power can be met. When the photovoltaic power generation generates excess power, it is stored in the power supply AC energy storage system to provide power support when the computing equipment demand peaks; The light sensor is used in monitoring solar photovoltaic power generation systems. In a photovoltaic power station, it is installed on the surface of a solar panel to monitor light intensity in real time, thereby predicting and adjusting the output of the photovoltaic power generation system.

5. The intelligent green energy consumption system according to claim 1, characterized in that: The monitoring system monitors the operating temperature and real-time operating status of the computing equipment in real time, issues a real-time alarm when any of the above is abnormal, and cooperates with the maintenance system to repair it; when the power supply condition is abnormal, the power distribution module and the power supply AC side energy storage system will promptly dispatch and adjust the power distribution.

6. The intelligent green energy consumption system according to claim 1 is characterized in that: In the maintenance system, when the temperature of the computing equipment is abnormal, the operation of the water circulation refrigeration system is adjusted to lower the temperature of the computing equipment and maintain the operating temperature; when various equipment fails, the fault is repaired in time, and hardware maintenance of various equipment is performed regularly.

7. The method for optimizing the operating status of an intelligent green energy consumption system according to claim 1, characterized in that: The algorithm optimization system and the computing power allocation module work together to optimize the improved sparrow algorithm optimization system and adjust the computing power allocation module to improve computing power resources and energy utilization. The optimization steps are as follows: Step (1), task priority sorting; Step (2), fitness function; Step (3), location update rules.

8. The method for optimizing the operating status of an intelligent green energy consumption system according to claim 7, characterized in that: In step (1), the task priority ranking is specifically: After Sparrow assigns a processor to each task, the scheduling results are different due to different priority rankings; when the priority ranking strategy in the current task scheduling algorithm is single, different types of tasks use the same ranking method, and the differences between different types of tasks are not analyzed, resulting in incorrect ranking results; to address this problem, a task priority diversion ranking strategy is adopted; according to the communication-computation ratio of the task graph, the task graph is divided into computing-intensive and communication-intensive tasks to ensure that high-priority tasks get computing resources first, thereby improving the overall operating efficiency and response speed of the system; The priority ranking formula for communication-intensive tasks is: Where, ranku(i) represents the priority value of communication-intensive task i; The average cost of the task is calculated on three processors; son(v i ) represents the subtask set of task i; C ij is the inter-task communication overhead; The priority ranking formula for computationally intensive tasks is: In the formula, rankU(i) represents the priority order of computationally intensive task i, σ i The cost standard deviation is calculated for task i on three processors. The method for sorting computationally intensive tasks can be obtained by calculating the cost standard deviation, and the result of sorting the task sequence values ​​in descending order is used as the final scheduling order.

9. The method for optimizing the operating status of an intelligent green energy consumption system according to claim 7, characterized in that: In step (2), the fitness function is specifically: In order to ensure that different computing power allocation schemes are effectively compared on a unified metric basis, so as to select the optimal solution, the optimization target is set as the processor load balance index for the problem of processor load imbalance, and its calculation formula is: Where nump(m,i) represents the cumulative number of tasks of processor i in the sparrow m allocation scheme; The average number of tasks assigned to each processor; p is the number of processors; The smaller the pld value, the better; The ultimate goal of the task scheduling algorithm is to reduce the overall completion time. Therefore, when the processor load tends to be balanced, the scheduling length is as short as possible. The scheduling length makespan calculation formula is: makespan(m)=EFT(m,v exit ) In the formula, EFT(m,v exit ) represents the completion time of the export task in the sparrow m allocation scheme, and the smaller the scheduling length value, the better; The combination of scheduling length and processor load balancing index is used as the final fitness function, but the values ​​of the two are very different, so they cannot be directly added together; in order to make the two at the same measurement level, they are first normalized; The normalization formula for the processor load balancing index is: Where max(pld) is the maximum value of the load balancing index in this iteration; normal_pld(m) represents the load balancing index after normalization of sparrow m; The normalization formula for scheduling length makespan is: Where max(makespan) is the maximum scheduling length of this iteration, and normal_makespan(m) represents the normalized scheduling length of sparrow m; After normalization, the fitness function calculation formula is: fitness(m)=normal_pld(m)+normal_makespan(m) After obtaining the fitness function value, it is sorted in descending order as the basis for the division of labor among the sparrow population. The sparrows with the highest ranking are regarded as discoverers, the rest as followers, and the sentinels are randomly selected.

10. The method for optimizing the operating status of an intelligent green energy consumption system according to claim 7, characterized in that: In step (3), the location update rule is specifically: In each iteration, the sparrow fitness function is calculated, the population is divided into descending order, and then the sparrow position information is calculated according to the encoding rules, and finally the updated position information is calculated according to different update rules; the discoverer is a sparrow in a good position, with a fitness function value, exploring potential high-quality computing resources or configurations globally, and its position update calculation formula is: In the formula, represents the location information of sparrow i1 after iteration; r is the alarm value, ranging from [0,1], safe is the safety threshold, ranging from [0.5,1]; α is a random number, ranging from [0,1], K is a random number that obeys the standard normal distribution; genmax is the maximum number of iterations; Followers are the largest group. They use the information provided by the discoverer to conduct a more detailed search in the local area to find a better allocation of computing power. The position update calculation formula in the first half is: In the formula, Indicates the best position in this iteration; A takes the value of -1 or 1. The processor needs to be reallocated in the second half due to the poor position; In the process of computing power allocation, by introducing the mechanism of updating the position of the sentinel, we can prevent the potential better computing power allocation plan from being missed due to falling into the local optimum too early; the calculation formula for updating the position of the sentinel is: In the formula, represents the global optimal position in the current iteration, represents the worst position of the current iteration, fit(best), fit(worse), and fit(i1) represent the global worst, best, and current individual fitness values, respectively; β is the step size control parameter that obeys the standard normal distribution; B is a random number between [-1,1]; ε is a small number to ensure that the denominator is not 0, and its value must be of the same order of magnitude as the difference in the fitness function value; In order to balance the global search and local development capabilities, a decay function related to the number of iterations is introduced to reduce the value of β, so that the algorithm performs extensive search in the early iteration and fine local development in the later iteration. The formula is as follows: In the formula, β init is the initial value of β, t is the current number of iterations, genmax is the maximum number of iterations, and θ is a constant less than 1; When the maximum number of iterations is reached, the optimal computing power allocation strategy and corresponding performance indicators are output.

Images