Method and device for controlling cluster power parameters

Abstract

The invention discloses a method and device for controlling cluster power parameters, and the method comprises the steps: obtaining the operation data of a first cluster in a plurality of historical periods when a first period starts, and enabling the historical periods to comprise a period before the first period; predicting the predicted carbon emission of the first cluster in the first period according to the operation data of the plurality of historical periods; based on the floating quota of the first cluster, the predicted carbon emission in the first period and the actual carbon emission in the second period, power parameters of the first cluster in the first period are controlled, and the floating quota comprises the carbon emission difference of the first cluster and the carbon emission difference of the second cluster in the historical period; the carbon emission difference of the cluster in any period is determined according to the actual emission amount and the carbon emission quota of the cluster in the corresponding period.

Description

Technical Field

[0001] The present application relates to the field of control technology, and in particular to a method and device for controlling cluster power parameters. Background Art

[0002] In order to meet the needs of energy conservation and emission reduction, some computer clusters can control the power parameters of the cluster based on a specified carbon emission quota. For example, the cluster can periodically detect the actual carbon emissions generated by the cluster operation. If the emissions exceed the quota, the power parameters are lowered to reduce carbon emissions.

[0003] The problem with the above control method is that when there are multiple clusters, different clusters have different loads and therefore require different power parameters. In this case, adjusting the power parameters according to the above control method will result in the power parameters of the cluster being unable to meet its load requirements. Summary of the invention

[0004] To this end, this application discloses the following technical solutions:

[0005] A first aspect of the present application provides a method for controlling cluster power parameters, comprising:

[0006] At the beginning of the first cycle, the operation data of the first cluster in a plurality of historical cycles are obtained, wherein the historical cycles include the cycles before the first cycle;

[0007] Predicting the predicted carbon emissions of the first cluster in the first period based on the operation data of the multiple historical periods;

[0008] Based on the floating quota of the first cluster, the predicted carbon emissions in the first period and the actual carbon emissions in the second period, the power parameters of the first cluster in the first period are controlled, the floating quota includes the carbon emission difference of the first cluster and the carbon emission difference of the second cluster in the historical period, and the carbon emission difference of the cluster in any period is determined according to the actual emissions and carbon emission quota of the cluster in the corresponding period.

[0009] Optionally, the operation data includes power consumption data;

[0010] The predicting, based on the operation data of the plurality of historical periods, the predicted carbon emissions of the first cluster in the first period includes:

[0011] Processing the power consumption data of the multiple historical periods according to the power consumption prediction model to obtain predicted power consumption data of the first cluster in the first period;

[0012] Based on a first coefficient and the predicted electricity consumption data, the predicted carbon emissions of the first cluster in the first period are determined, wherein the first coefficient represents a proportional relationship between electricity consumption and carbon emissions of the first cluster.

[0013] Optionally, the operation data includes device status data related to power consumption;

[0014] The predicting, based on the operation data of the plurality of historical periods, the predicted carbon emissions of the first cluster in the first period includes:

[0015] Processing the device status data of the first cluster in the multiple historical periods according to the status prediction model to obtain predicted device status data of the first cluster in the first period;

[0016] Determine predicted power consumption data of the first cluster in the first period based on the predicted device status data and a predetermined first association relationship, where the first association relationship is an association relationship between the device status data of the first cluster and power consumption;

[0017] Based on a first coefficient and the predicted electricity consumption data, the predicted carbon emissions of the first cluster in the first period are determined, wherein the first coefficient represents a proportional relationship between electricity consumption and carbon emissions of the first cluster.

[0018] Optionally, the method for determining the first association relationship includes:

[0019] Obtaining device status data and power consumption data of the first cluster in the multiple historical periods;

[0020] The first association relationship is determined according to the device status data and power consumption data of the multiple historical periods.

[0021] Optionally, controlling the power parameter of the first cluster in the first period based on the floating quota of the first cluster, the predicted carbon emissions in the first period, and the actual carbon emissions in the second period includes:

[0022] If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both less than or equal to the carbon emission quota of the corresponding period, control the power parameters of the first cluster in the first period to be consistent with the power parameters in the second period, and add the carbon emission difference of the first cluster in the second period to the floating quota;

[0023] If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both greater than the carbon emission quota of the corresponding period, and the carbon emission difference of the first cluster in the second period is greater than the floating quota, control the power parameter of the first cluster in the first period to be less than the power parameter in the second period;

[0024] If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both greater than the carbon emission quota of the corresponding period, and the carbon emission difference of the first cluster in the second period is less than the floating quota, control the power parameters of the first cluster in the first period to be consistent with the power parameters in the second period, and subtract the carbon emission difference of the first cluster in the second period from the floating quota.

[0025] Optionally, the subtracting the carbon emission difference of the first cluster in the second period from the floating quota includes:

[0026] The carbon emission differences in the floating quota are deducted in order from the largest to the smallest duration until the total of the deducted carbon emission differences matches the carbon emission difference of the first cluster in the second period. The duration is the time elapsed since the carbon emission difference was added to the floating quota.

[0027] Optionally, also include:

[0028] Determine a duration threshold corresponding to the first cluster and the second cluster;

[0029] The carbon emission difference whose duration is greater than the duration threshold is deleted from the floating quota.

[0030] Optionally, also include:

[0031] Determining a unit quota of the first cluster in each unit time period of the first cycle according to the carbon emission quota of the first cluster in the first cycle, where the first cycle includes a plurality of unit time periods;

[0032] At the end of each unit time period in the first cycle, the unit quota of the next unit time period and the power parameters of the first cluster in the next unit time period are updated according to the unit carbon emission difference of the previous unit time period, and the unit carbon emission difference is the difference between the emissions of the first cluster in the corresponding unit time period and the unit quota.

[0033] Optionally, also include:

[0034] At the end of the first period, determining the difference between the actual emissions of the first cluster in the first period and the carbon emission quota;

[0035] The carbon emission difference of the first cluster in the first period is determined according to the difference and the unit carbon emission difference of the first cluster in the last unit time period in the first period.

[0036] A second aspect of the present application provides a device for controlling cluster power parameters, including:

[0037] an obtaining unit, configured to obtain, at the beginning of a first cycle, operation data of the first cluster in a plurality of historical cycles, wherein the historical cycles include a cycle before the first cycle;

[0038] a prediction unit, configured to predict the predicted carbon emissions of the first cluster in the first period based on the operation data of the plurality of historical periods;

[0039] A control unit is used to control the power parameters of the first cluster in the first period based on the floating quota of the first cluster, the predicted carbon emissions in the first period and the actual carbon emissions in the second period, the floating quota includes the carbon emission difference of the first cluster and the carbon emission difference of the second cluster in the historical period, and the carbon emission difference of the cluster in any period is determined according to the actual emissions and carbon emission quota of the cluster in the corresponding period. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0041] Figure 1 is a flow chart of a method for controlling cluster power parameters provided by an embodiment of the present application;

[0042] Figure 2 is a schematic diagram of multiple clusters provided in an embodiment of the present application;

[0043] Figure 3 It is a flow chart of a method for controlling power parameters within a unit time period provided by an embodiment of the present application;

[0044] Figure 4 It is a structural schematic diagram of a device for controlling cluster power parameters provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0046] This embodiment provides a method for controlling cluster power parameters. Figure 1 , is a flow chart of the method, and the method may include the following steps.

[0047] The method execution subject of this embodiment may be an independent server device capable of controlling the power parameters of multiple clusters, or may be a computer program deployed on multiple clusters, such as a control service for controlling the power parameters of each cluster.

[0048] S101, when a first cycle starts, obtaining operation data of a first cluster in a plurality of historical cycles, where the historical cycles include cycles before the first cycle.

[0049] The first cluster may be any one of a plurality of computer clusters controlled by the method of this embodiment.

[0050] The length of the cycle can be set as needed without limitation. As some examples, one day (24 hours) can be set as a cycle, one week can be set as a cycle, 30 hours can be set as a cycle, etc.

[0051] Step S101 can be executed at the end of any cycle and the beginning of the next cycle. The next cycle that just begins is the first cycle mentioned above, and each cycle before the first cycle can be regarded as a historical cycle. In combination with the above example, if one day is a cycle, it can end on October 20 and start on October 21, that is, step S101 is executed at 00:00 on October 21. In this case, the whole day of October 21 can be regarded as the first cycle mentioned above, and each day before October 21 can be regarded as a historical cycle.

[0052] In this embodiment, during the operation of one cluster, the operation data of the first cluster can be detected in real time and the detected operation data can be recorded in a related storage medium. When S101 is executed, the operation data of multiple historical periods can be read from the storage medium.

[0053] The operation data of a cluster in a historical period may include any one or more types of data related to the actual carbon emissions of the cluster in the historical period.

[0054] The actual carbon emissions of a cluster in a cycle refers to the total mass of greenhouse gases (mainly carbon dioxide) emitted to generate the electricity consumed by the cluster in that cycle, and the unit can be kilograms.

[0055] In this embodiment, the operating data of N consecutive historical cycles before the first cycle can be obtained. The value of N can be set as needed without limitation. For example, N can be set equal to 5, or equal to 7. Correspondingly, the operating data of the most recent 5 historical cycles can be obtained, or the operating data of the most recent 7 historical cycles can be obtained.

[0056] S102: predicting the carbon emissions of the first cluster in the first period based on the operation data of multiple historical periods.

[0057] In step S102, a pre-built data processing model with prediction capability may be used to process the operation data of multiple historical periods, so as to obtain the predicted carbon emissions of the first cluster in the first period according to the processing results output by the data processing model.

[0058] The predicted carbon emissions of the first cluster in the first period refer to the total mass of greenhouse gases that are expected to be emitted in order to generate the electrical energy consumed by the first cluster in the first period.

[0059] There are many ways to process the operation data of the historical period to obtain the predicted carbon emissions, and this embodiment does not limit the prediction method.

[0060] S103, based on the floating quota of the first cluster, the predicted carbon emissions in the first period and the actual carbon emissions in the second period, control the power parameters of the first cluster in the first period, the floating quota includes the carbon emission difference of the first cluster and the carbon emission difference of the second cluster in the historical period, and the carbon emission difference of the cluster in any period is determined according to the actual emissions and carbon emission quota of the cluster in the corresponding period.

[0061] The second cluster may be a cluster other than the first cluster among the multiple computer clusters controlled by the method of this embodiment. There may be one or more second clusters.

[0062] Each cluster has a preset carbon emission quota in each cycle, which is equivalent to the limit of the actual carbon emission of the corresponding cluster in the corresponding cycle. In the relevant technology, if the actual carbon emission of a cluster in a cycle exceeds the carbon emission quota, it means that the cluster exceeds the standard in the cycle, and the power of the cluster can be reduced to reduce the actual carbon emission. If the actual carbon emission of a cluster in a cycle is less than or equal to the carbon emission quota, it means that the cluster meets the standard in the cycle, and the cluster can continue to operate at the current power.

[0063] The carbon emission quotas of different clusters can be the same or different. The carbon emission quotas of the same cluster in different periods can be the same or different. The carbon emission quotas of the same cluster in different periods can be fixed values ​​or floating values ​​adjusted according to certain strategies.

[0064] In step S103, the predicted carbon emissions of the first cluster in the first period and the carbon emission quota in the first period can be compared to predict whether the first cluster is in an over-standard state or in a standard state in the first period, and the actual carbon emissions of the first cluster in the second period and the carbon emission quota in the second period can be compared to determine whether the first cluster is in an over-standard state or in a standard state in the second period. Then, based on whether the first cluster meets the standard in the first period and the second period, as well as the floating quota of the first cluster, the power parameters of the first cluster in the first period are controlled.

[0065] The carbon emission difference of a cluster in a cycle can be defined as the difference between the carbon emission quota of the cluster in the cycle and the actual carbon emissions. When the difference is negative, it means that the cluster exceeds the standard in the cycle. When the difference is positive or 0, it means that the cluster meets the standard in the cycle.

[0066] The initial value of the floating quota may be 0. During the control of multiple clusters according to the method of this embodiment, the floating quota may be updated based on the carbon emission difference of each cluster.

[0067] See also Figure 2 In some embodiments, the floating quota can be equal to the positive carbon emission difference of all clusters in each historical period. At the end of each period, the carbon emission difference of each cluster in this period can be calculated. If the carbon emission difference of a cluster in this period is positive, the carbon emission difference is added to the floating quota.

[0068] In some embodiments, the floating quota may also include a positive carbon emission difference that meets certain conditions. For example, for any cluster, at the end of a cycle, the predicted carbon emissions of the cluster in the next cycle can be predicted according to the method of steps S101 and S102, and the predicted carbon emissions and the emission quota of the cluster in the next cycle can be compared to predict whether the cluster meets the standard in the next cycle. If the cluster meets the standard in both the cycle that just ended and the next cycle, the positive carbon emission difference of the cluster in the cycle that just ended is added to the floating quota. If the cluster meets the standard in the cycle that just ended, but it is predicted that it will exceed the standard in the next cycle, the positive carbon emission difference of the cluster in the cycle that just ended is not added to the floating quota.

[0069] The beneficial effects of this embodiment are:

[0070] When controlling the power parameters of multiple clusters, the method of this embodiment can share the carbon emission difference of the second cluster with the first cluster based on the predicted carbon emissions, actual carbon emissions and floating quota of the first cluster, so that the power parameters of the first cluster can meet its actual load requirements.

[0071] In some embodiments, the operation data obtained in S101 may include power consumption data of the cluster. For example, the power consumption data of the first cluster in a historical period may be 5000 kWh, indicating that the first cluster consumed a total of 5000 kWh (ie, 5000 degrees) of electricity in the historical period.

[0072] Based on the above operation data, the method for obtaining the predicted carbon emissions of the first cluster in the first period may be:

[0073] Processing the electricity consumption data of multiple historical periods according to the electricity consumption prediction model to obtain the predicted electricity consumption data of the first cluster in the first period;

[0074] Based on the first coefficient and the predicted electricity consumption data, the predicted carbon emissions of the first cluster in the first period are determined, and the first coefficient represents the proportional relationship between the electricity consumption and the carbon emissions of the first cluster.

[0075] The power consumption of a cluster in multiple consecutive cycles may conform to a certain variation pattern. In this embodiment, the power consumption data of multiple clusters in multiple historical cycles can be collected, and a power consumption prediction model can be trained using these data so that the power consumption prediction model can learn the variation pattern of power consumption.

[0076] Therefore, the electricity consumption data of the first cluster in multiple historical periods can be input into the trained electricity consumption prediction model. The electricity consumption prediction model is used in combination with the electricity consumption data of the historical periods and the law of electricity consumption changes to predict the amount of electricity that the first cluster may consume in the first period, which is the above-mentioned predicted electricity consumption data.

[0077] The first coefficient may be any coefficient that can characterize the proportional relationship between the electricity consumption and carbon emissions of the first cluster. As an example, the first coefficient may be the national grid emission factor of the region where the first cluster is located. The national grid emission factor of a region may indicate how much mass of greenhouse gas a power plant will emit for every unit of electricity (e.g., 1 kWh of electricity) generated in the region. The value of the first coefficient may be pre-configured by the relevant user before applying the method of this embodiment, or obtained from a query on the Internet.

[0078] In this embodiment, the predicted electricity consumption data of the first cluster in the first period may be multiplied by the first coefficient, and the obtained product may be used as the predicted carbon emissions of the first cluster in the first period.

[0079] In some optional embodiments, the operation data may include device status data related to power consumption. The device status data may include any one or more of parameters such as the central processing unit (CPU) usage rate of the computer devices contained in the first cluster, the CPU temperature, the water inlet temperature of the cooling module of the device, the water outlet temperature of the cooling module of the device, and the power (wattage) of the device when it is working.

[0080] Based on the above operation data, the method of predicting the predicted carbon emissions of the first cluster in the first period may be:

[0081] Processing the device status data of the first cluster in multiple historical periods according to the status prediction model to obtain predicted device status data of the first cluster in the first period;

[0082] Determine predicted power consumption data of the first cluster in the first period based on the predicted device status data and a predetermined first association relationship, where the first association relationship is an association relationship between the device status data of the first cluster and power consumption;

[0083] Based on the first coefficient and the predicted electricity consumption data, the predicted carbon emissions of the first cluster in the first period are determined, and the first coefficient represents the proportional relationship between the electricity consumption and the carbon emissions of the first cluster.

[0084] The device status data of a cluster in multiple consecutive cycles may also conform to certain changing rules. In this embodiment, the device status data of multiple clusters in multiple historical cycles can be collected, and a state prediction model can be trained using these data so that the state prediction model can learn the changing rules of the device status data.

[0085] In this way, the equipment status data of the first cluster in multiple historical periods can be input into the trained state prediction model, so that the state prediction model combines the equipment status data of the historical period and the change law of the equipment status data to predict the possible equipment status data of the first cluster in the first period, that is, predict the equipment status data.

[0086] The first association relationship may have various forms. As some examples, the first association relationship may be one or a group of tables, in which a plurality of device status data with different values ​​and the power consumption data of the first cluster corresponding to each device status data are recorded, or the first association relationship may be a specific function or model, in which case the first association relationship may accept the device status data of the first cluster within a cycle as input, calculate the input and output the power consumption data of the first cluster within the cycle.

[0087] The first association relationship may be determined before predicting the predicted carbon emissions of the first period. The method for determining the first association relationship may be:

[0088] Obtaining device status data and power consumption data of the first cluster in multiple historical periods;

[0089] A first association relationship is determined based on the device status data and the power consumption data of multiple historical periods.

[0090] For the first association relationship in the form of a table, in the above embodiment, the device status data and the power consumption data of multiple historical periods can be recorded correspondingly in a blank table to obtain the first association relationship.

[0091] For the first association relationship in function form, the device status data and power consumption data of multiple historical periods can be fitted based on various data fitting methods in related technologies to obtain a function that can reflect the law of change of power consumption data with device status data, and the function is used as the first association relationship.

[0092] For the first association relationship in model form, an initial model can be determined by a random initialization method, and then the initial model is trained using device status data and power consumption data from multiple historical periods. After the training is completed, the trained model is used as the first association relationship.

[0093] The prediction models (power consumption prediction model and state prediction model) used in the above embodiments may have a model structure based on any machine learning or deep learning algorithm in the relevant technology, including but not limited to regression analysis algorithm, time series algorithm, artificial neural network algorithm, support vector machine algorithm, convolutional neural network algorithm, recurrent neural network algorithm, etc. As an example, the prediction model may include a convolutional neural network and a recurrent neural network.

[0094] In some embodiments, the power parameters of the first cluster in the first period may be controlled as follows:

[0095] 1. If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both less than or equal to the carbon emission quota of the corresponding period, the power parameters of the first cluster in the first period are controlled to be consistent with the power parameters in the second period, and the carbon emission difference of the first cluster in the second period is added to the floating quota;

[0096] Second, if the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both greater than the carbon emission quota of the corresponding period, and the carbon emission difference of the first cluster in the second period is greater than the floating quota, the power parameter of the first cluster in the first period is controlled to be less than the power parameter in the second period;

[0097] 3. If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both greater than the carbon emission quotas of the corresponding periods, and the carbon emission difference of the first cluster in the second period is less than the floating quota, control the power parameters of the first cluster in the first period to be consistent with the power parameters in the second period, and subtract the carbon emission difference of the first cluster in the second period from the floating quota.

[0098] The first situation is equivalent to the first cluster meeting the target in the second cycle and is expected to meet the target in the first cycle. This means that the load of the first cluster has been relatively light in recent period and its carbon emission quota may be surplus. Therefore, the surplus in the second cycle, that is, the carbon emission difference of the first cluster in the second cycle, can be added to the floating quota, so that the carbon emission difference of the first cluster in the second cycle can be used by other clusters.

[0099] The second and third cases are equivalent to the first cluster exceeding the standard in the second cycle and being expected to exceed the standard in the first cycle as well, which indicates that the first cluster has been heavily loaded in recent times and its carbon emission quota may not be able to meet the load requirements of the first cluster.

[0100] If it is the second case, it means that the carbon emission quota of the first cluster is seriously insufficient compared to its load, and the current floating quota is not enough to compensate for the missing part of the carbon emission quota. Therefore, it is necessary to reduce the power parameters of the first cluster to reduce the load of the first cluster.

[0101] If it is the third case, it means that the carbon emission quota of the first cluster is slightly insufficient compared to its load, and the current floating quota can compensate for the missing part of the carbon emission quota. At this time, the missing part of the carbon emission quota of the first cluster (that is, the emission difference in the second cycle) can be offset by the floating quota to avoid reducing the power parameters of the first cluster, so that the first cluster can still process tasks at a higher power.

[0102] Among them, the carbon emission difference of the first cluster in the second period is greater than the floating quota, which means that the absolute value of the carbon emission difference in the second period is greater than the floating quota. For example, if the emission difference in the second period is -15 kilograms and the floating quota is 10 kilograms, then the carbon emission difference of the first cluster in the second period is greater than the floating quota; if the emission difference in the second period is -10 kilograms and the floating quota is 20 kilograms, then the carbon emission difference of the first cluster in the second period is less than the floating quota.

[0103] As an example, assuming that it is determined in S103 that the first cluster meets the first situation, the floating quota is 20 kg, and the carbon emission difference of the first cluster in the second cycle is 5 kg, the 5 kg can be added to the floating quota, that is, the floating quota is updated to 25 kg, and the power parameter of the first cluster remains unchanged;

[0104] Assuming that it is determined in S103 that the first cluster meets the second situation, the floating quota is 20 kg, and the carbon emission difference of the first cluster in the second cycle is -25 kg, then the power parameter of the first cluster is reduced in the first cycle, and the floating quota remains unchanged;

[0105] Assuming that it is determined in S103 that the first cluster meets the third situation, the floating quota is 20 kg, and the carbon emission difference of the first cluster in the second cycle is -15 kg, then in the first cycle, the floating quota is subtracted from 15 kg of the carbon emission difference in the second cycle, and the floating quota is updated to 5 kg, and the power parameters of the first cluster remain unchanged.

[0106] In some embodiments, if when executing S103, it is found that the first cluster meets the standard in the second cycle and is expected to exceed the standard in the first cycle, or it is found that the first cluster exceeds the standard in the second cycle and is expected to meet the standard in the first cycle, it means that the current carbon emission quota of the first cluster can just meet the load demand of the first cluster, and no other clusters need to compensate, and there is no excess difference to provide other clusters with use. Therefore, in this case, the power parameters of the first cluster in the first cycle can be controlled to be consistent with the power parameters in the second cycle, and the floating quota is not updated based on the emission difference of the first cluster in the second cycle. That is, the positive emission difference is neither added to the floating quota nor the negative emission difference is deducted from the floating quota.

[0107] In some embodiments, the floating quota can be shared among multiple clusters. That is, for each cluster controlled by the method of this embodiment, the corresponding carbon emission difference can be subtracted from the floating quota according to the control method for the first cluster, and the carbon emission difference of the cluster can be added to the floating quota according to the method for updating the floating quota.

[0108] Optionally, when deducting the carbon emission difference of the first cluster in the second period from the floating quota, the deduction can be made as follows:

[0109] The carbon emission differences in the floating quota are deducted in order from the largest to the smallest duration until the total of the deducted carbon emission differences matches the carbon emission difference of the first cluster in the second cycle. The duration is the time that has passed since the carbon emission difference was added to the floating quota.

[0110] In this embodiment, the floating quota may be recorded in a table, which includes each carbon emission difference added to the floating quota and the corresponding adding time.

[0111] Exemplarily, the floating quota can be recorded in descending order of duration: 10 kg of carbon emission difference is added at T1, 7 kg of carbon emission difference is added at T2, 8 kg of carbon emission difference is added at T3, and so on.

[0112] Based on the above recording method, when it is necessary to deduct the emission difference of the first cluster in the second cycle from the floating quota, the longest carbon emission difference in the floating quota can be found first. If the carbon emission difference is greater than the absolute value of the emission difference of the first cluster in the second cycle, the excess can be retained and wait for the next deduction;

[0113] If the carbon emission difference with the longest duration is less than the absolute value of the emission difference of the first cluster in the second period, the carbon emission difference will be deleted from the floating quota, and then the carbon emission difference with the longest duration in the floating quota will be found, and part or all of it will be deleted. The process will be repeated until the amount deleted from the floating quota is equal to the absolute value of the emission difference of the first cluster in the second period.

[0114] Combined with the above example, assuming that the emission difference of the first cluster in the second period is -5 kg, then only the carbon emission difference of 5 kg added at time T1 can be deducted from the floating quota. After deduction, the floating quota can include: 5 kg of carbon emission difference added at time T1, 7 kg of carbon emission difference added at time T2, and 8 kg of carbon emission difference added at time T3;

[0115] Assuming that the emission difference of the first cluster in the second cycle is 19 kilograms, the carbon emission difference of 10 kilograms added at T1, the carbon emission difference of 7 kilograms added at T2, and the carbon emission difference of 2 kilograms added at T3 can be deducted from the floating quota record. After deduction, the floating quota can include: the carbon emission difference of 6 kilograms added at T3.

[0116] Optionally, also include:

[0117] Determine a duration threshold corresponding to the first cluster and the second cluster;

[0118] Remove carbon emission differences from floating quotas whose duration is greater than the duration threshold.

[0119] In this embodiment, the duration threshold can be determined according to the region where the cluster is located, and the duration threshold can be different for clusters in different regions. In view of this situation, when the power parameters of multiple clusters are controlled according to the method of this embodiment, the multiple clusters can be divided by region, multiple clusters in region A share one floating quota, and multiple clusters in region B share another floating quota. The floating quota in region A is not related to the floating quota in region B. Accordingly, the first cluster and the second cluster in the aforementioned embodiment can be multiple clusters in the same region.

[0120] The regions here can be divided by administrative regions, geographical regions, or other dimensions, without limitation.

[0121] After determining the duration threshold, it is possible to regularly determine whether the duration of each carbon emission difference added to the floating quota is greater than the duration threshold. Whenever a carbon emission difference is found to have a duration greater than the duration threshold, the carbon emission difference will be deleted from the floating quota.

[0122] For example, assuming that the current floating quota is 30 kilograms, which includes a carbon emission difference of 6 kilograms, if it is found that the duration of the carbon emission difference exceeds the duration threshold, the carbon emission difference will be deleted and the floating quota will be updated to 24 kilograms.

[0123] In some embodiments, the effective carbon footprint of each cluster may also be recorded. The recording method of this parameter is as follows: when a cluster is initially controlled according to the method of this embodiment, the effective carbon footprint of the cluster is set equal to 0 kg. Thereafter, at the end of each cycle, the carbon emission difference of the cluster in the past cycle is calculated, and the effective carbon footprint of the cluster is updated with the carbon emission difference.

[0124] For example, at the end of the first cycle, the carbon emission difference of cluster X is -10 kilograms, that is, cluster X exceeded the standard in the first cycle, and the actual carbon emissions were 10 kilograms more than the carbon emission quota, so the effective carbon footprint of cluster X is updated to 0-10 kilograms, that is, -10 kilograms. At the end of the second cycle, the carbon emission difference is +6 kilograms, that is, cluster X met the standard in the second cycle, and the actual carbon emissions were 6 kilograms less than the carbon emission quota, so the effective carbon footprint of cluster X is updated to -10+6 kilograms, that is, -4 kilograms. At the end of the third cycle, the carbon emission difference is +7 kilograms, so the effective carbon footprint of cluster X is updated to -4+7 kilograms, that is, +3 kilograms, and so on.

[0125] The purpose of recording the effective carbon footprint of a cluster is that it can be used to understand whether the cumulative carbon emissions of a cluster to date exceed or meet the standard, as well as the degree of exceeding or meeting the standard. If the effective carbon footprint of a cluster is a large negative value, it means that the cluster has frequently exceeded the standard in multiple cycles to date. At this time, the cluster load can be appropriately reduced to further reduce the actual carbon emissions of the cluster. If the effective carbon footprint of a cluster is a large positive value, it means that the cluster has met the standard many times in multiple cycles to date. At this time, the cluster load can be appropriately increased to make full use of the processing capacity of the cluster.

[0126] Optional, see Figure 3 Each cycle can be further divided into a plurality of unit time periods. In the first cycle, the power parameters of the first cluster in each unit time period can be controlled as follows:

[0127] S301 : Determine a unit quota of the first cluster in each unit time period of the first cycle according to the carbon emission quota of the first cluster in the first cycle, where the first cycle includes a plurality of unit time periods.

[0128] S302, at the end of each unit time period in the first cycle, the unit quota of the next unit time period and the power parameters of the first cluster in the next unit time period are updated according to the unit carbon emission difference of the previous unit time period, and the unit carbon emission difference is the difference between the emission of the first cluster in the corresponding unit time period and the unit quota.

[0129] In step S301, the duration of the unit time period can be set as needed, and the condition that the duration of a cycle is an integer multiple of the duration of the unit time period is satisfied. For example, when a cycle is one day (24 hours) or 30 hours, the unit time period can be one hour; when a cycle is one week, the unit time period can be one day.

[0130] In a cycle, the unit quota for each unit time period can be the same or different.

[0131] In the former case, the carbon emission quota of the first cluster in the first cycle can be evenly distributed to each unit time period to obtain the unit quota of each unit time period. For example, if a cycle includes 24 unit time periods, the carbon emission quota can be divided by 24 to obtain the unit quota.

[0132] For the latter case, the actual carbon emissions of the first cluster in different unit time periods in the past multiple historical cycles can be collected, and then the carbon emission quota of the first cycle can be allocated to each unit time period in the first cycle based on the changing pattern of the actual carbon emissions in different unit time periods in the historical cycles.

[0133] In step S302, the power parameters of the first cluster can be detected in real time within the first cycle. At the end of each unit time period in the first cycle, the power consumption data of the first cluster in the unit time period is determined based on the power parameters detected in the unit time period, and then the actual carbon emissions of the first cluster in the unit time period are calculated according to the first coefficient and the power consumption data in the unit time period (that is, the emissions of the first cluster in the corresponding unit time period in S302), and the actual carbon emissions are subtracted from the unit quota of the first cluster in the unit time period, and the difference obtained is used as the unit carbon emissions difference of the first cluster in the unit time period.

[0134] At the end of a unit time period, if the unit carbon emission difference of the first cluster in the past unit time period is greater than 0, it means that the first cluster has a surplus of unit quota in the past unit time period. In this case, the power parameter of the first cluster can be increased in the next unit time period, so that the first cluster can make full use of the unit quota in the next unit time period;

[0135] At the end of a unit time period, if the unit carbon emission difference of the first cluster in the past unit time period is less than 0, it means that the unit quota of the first cluster in the past unit time period is insufficient. In this case, the power parameter of the first cluster can be reduced in the next unit time period, so that the first cluster reduces carbon emissions in the next unit time period;

[0136] At the end of a unit time period, if the unit carbon emission difference of the first cluster in the past unit time period is equal to 0, the power parameter of the first cluster can be kept consistent with that of the past unit time period in the next unit time period.

[0137] In some embodiments, at the end of each unit time period, the unit quota of the next unit time period can be updated according to the unit carbon emission difference of the ended unit time period. For example, the unit carbon emission difference of the ended unit time period and the unit quota of the next unit time period can be directly added together, and the result can be used as the updated unit quota of the next unit time period. If the unit quota of the unit time period is updated, the updated unit quota shall be used as the basis for calculating the unit carbon emission difference of the unit time period.

[0138] In the above embodiment, if the unit quota of the previous unit time period is insufficient, the insufficient part (i.e., the unit carbon emission difference less than 0) will be deducted from the unit quota of the next unit time period accordingly; if there is a surplus of the unit quota of the previous unit time period, the surplus part (i.e., the unit carbon emission difference greater than 0) will be added to the unit quota of the next unit time period.

[0139] As an example, assuming that the unit quota for each unit time period is 10 kilograms, at the end of the first unit time period, the unit carbon emission difference for the first unit time period is calculated to be 2 kilograms, so these 2 kilograms are added to the second unit time period, and the updated unit quota for the second unit time period is 12 kilograms, and the power parameters of the first cluster are increased during the second unit time period. At the end of the second unit time period, the unit carbon emission difference for the second unit time period is calculated to be -4 kilograms based on the updated unit quota, so these -4 kilograms are added to the third unit time period, the updated unit quota for the third unit time period is 10-4=6 kilograms, and the power parameters of the first cluster are reduced during the third unit time period, and so on.

[0140] In this embodiment, the power parameter of the first cluster in the first unit time period of the first cycle may be determined in the manner of determining the power parameter of the first cycle in the aforementioned embodiment, which is not described in detail.

[0141] The beneficial effect of this embodiment is that it can share carbon emission quotas among multiple clusters on a periodic basis, thereby solving the problem of a single cluster having no quota due to carbon emissions exceeding the quota of the cluster, and can also solve the problem of jitter caused by too long a period span when multiple clusters share.

[0142] Optionally, in the case where the power parameters in each cycle are controlled according to the above-mentioned unit time period, at the end of a cycle, taking the first cycle as an example, the carbon emission difference of the first cluster in the cycle may be determined in the following manner:

[0143] At the end of the first period, determining the difference between the actual emissions of the first cluster in the first period and the carbon emission quota;

[0144] The carbon emission difference of the first cluster in the first period is determined according to the difference and the unit carbon emission difference of the first cluster in the last unit time period in the first period.

[0145] In this embodiment, at the end of the first cycle, which is also the end of the last unit time period of the first cycle, the difference between the actual emissions of the first cluster and the carbon emission quota in the first cycle can be obtained, recorded as the cycle difference, and the unit carbon emission difference of the first cluster in the last unit time period in the first cycle can be obtained;

[0146] Then, the period difference and the unit carbon emission difference of the last unit time period can be calculated according to the preset second coefficient, and the result obtained can be used as the carbon emission difference of the first cluster in the first period. The calculation method can be: carbon emission difference of the first period = period difference + (unit carbon emission difference of the last unit time period * second coefficient), that is, the product of the second coefficient and the unit carbon emission difference of the last unit time period, plus the period difference, and the result obtained is used as the carbon emission difference of the first cluster in the first period.

[0147] The second coefficients corresponding to different clusters may be the same or different.

[0148] In any embodiment of the present application, the power parameter of the control cluster may be any one or more of the parameters such as the rated power, the number of operating devices, and the operating frequency of the control cluster.

[0149] For example, increasing the power parameter of the first cluster may be to increase the rated power of the devices in operation in the first cluster, or to increase the number of devices in operation in the first cluster, or to increase the operating frequency of the devices in operation in the first cluster;

[0150] Reducing the power parameter of the first cluster may be to reduce the rated power of the devices in operation in the first cluster, or to reduce the number of the devices in operation in the first cluster, or to reduce the operating frequency of the devices in operation in the first cluster.

[0151] This embodiment also provides a device for controlling cluster power parameters, see Figure 4 , is a schematic diagram of the structure of the device, and the device may include the following units.

[0152] The obtaining unit 401 is used to obtain the operation data of the first cluster in multiple historical periods when the first period starts, and the historical periods include the period before the first period;

[0153] A prediction unit 402, configured to predict the predicted carbon emissions of the first cluster in the first period according to the operation data of the plurality of historical periods;

[0154] The control unit 403 is used to control the power parameters of the first cluster in the first period based on the floating quota of the first cluster, the predicted carbon emissions in the first period and the actual carbon emissions in the second period. The floating quota includes the carbon emission difference of the first cluster and the carbon emission difference of the second cluster in the historical period. The carbon emission difference of the cluster in any period is determined according to the actual emissions and carbon emission quota of the cluster in the corresponding period.

[0155] Optionally, the operating data includes power consumption data;

[0156] When predicting the predicted carbon emissions of the first cluster in the first period based on the operation data of multiple historical periods, the prediction unit 402 is used to:

[0157] Processing the electricity consumption data of multiple historical periods according to the electricity consumption prediction model to obtain the predicted electricity consumption data of the first cluster in the first period;

[0158] Based on the first coefficient and the predicted electricity consumption data, the predicted carbon emissions of the first cluster in the first period are determined, and the first coefficient represents the proportional relationship between the electricity consumption and the carbon emissions of the first cluster.

[0159] Optionally, the operating data includes device status data related to power consumption;

[0160] When predicting the predicted carbon emissions of the first cluster in the first period based on the operation data of multiple historical periods, the prediction unit 402 is used to:

[0161] Processing the device status data of the first cluster in multiple historical periods according to the status prediction model to obtain predicted device status data of the first cluster in the first period;

[0162] Determine predicted power consumption data of the first cluster in the first period based on the predicted device status data and a predetermined first association relationship, where the first association relationship is an association relationship between the device status data of the first cluster and power consumption;

[0163] Based on the first coefficient and the predicted electricity consumption data, the predicted carbon emissions of the first cluster in the first period are determined, and the first coefficient represents the proportional relationship between the electricity consumption and the carbon emissions of the first cluster.

[0164] Optionally, the method by which the prediction unit 402 determines the first association relationship includes:

[0165] Obtaining device status data and power consumption data of the first cluster in multiple historical periods;

[0166] A first association relationship is determined based on the device status data and the power consumption data of multiple historical periods.

[0167] Optionally, when the control unit 403 controls the power parameter of the first cluster in the first period based on the floating quota of the first cluster, the predicted carbon emissions in the first period, and the actual carbon emissions in the second period, it is configured to:

[0168] If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both less than or equal to the carbon emission quota of the corresponding period, the power parameters of the first cluster in the first period are controlled to be consistent with the power parameters in the second period, and the carbon emission difference of the first cluster in the second period is added to the floating quota;

[0169] If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both greater than the carbon emission quota of the corresponding period, and the carbon emission difference of the first cluster in the second period is greater than the floating quota, the power parameter of the first cluster in the first period is controlled to be less than the power parameter in the second period;

[0170] If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both greater than the carbon emission quota of the corresponding period, and the carbon emission difference of the first cluster in the second period is less than the floating quota, the power parameters of the first cluster in the first period are controlled to be consistent with the power parameters in the second period, and the carbon emission difference of the first cluster in the second period is subtracted from the floating quota.

[0171] Optionally, when the control unit 403 subtracts the carbon emission difference of the first cluster in the second period from the floating quota, it is configured to:

[0172] The carbon emission differences in the floating quota are deducted in order from the largest to the smallest duration until the total of the deducted carbon emission differences matches the carbon emission difference of the first cluster in the second cycle. The duration is the time that has passed since the carbon emission difference was added to the floating quota.

[0173] Optionally, the control unit 403 is further configured to:

[0174] Determine a duration threshold corresponding to the first cluster and the second cluster;

[0175] Remove carbon emission differences from floating quotas whose duration is greater than the duration threshold.

[0176] Optionally, the control unit 403 is further configured to:

[0177] Determine a unit quota of the first cluster for each unit time period in the first period according to the carbon emission quota of the first cluster in the first period, where the first period includes a plurality of unit time periods;

[0178] At the end of each unit time period in the first cycle, the unit quota of the next unit time period and the power parameter of the first cluster in the next unit time period are updated according to the unit carbon emission difference of the previous unit time period. The unit carbon emission difference is the difference between the emissions of the first cluster in the corresponding unit time period and the unit quota.

[0179] Optionally, the control unit 403 is further configured to:

[0180] At the end of the first period, determining the difference between the actual emissions of the first cluster in the first period and the carbon emission quota;

[0181] The carbon emission difference of the first cluster in the first period is determined according to the difference and the unit carbon emission difference of the first cluster in the last unit time period in the first period.

[0182] The working principle of the device for controlling cluster power parameters provided in this embodiment can refer to the relevant steps in the method for controlling cluster power parameters provided in any embodiment, and will not be described in detail.

[0183] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0184] For the convenience of description, the above system or device is described by dividing it into various modules or units according to its functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0185] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0186] Finally, it should be noted that, in this article, relational terms such as first, second, third and fourth are only used 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 "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0187] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for controlling cluster power parameters, comprising: At the beginning of the first cycle, the operation data of the first cluster in a plurality of historical cycles are obtained, wherein the historical cycles include the cycles before the first cycle; Predicting the predicted carbon emissions of the first cluster in the first period based on the operation data of the multiple historical periods; Based on the floating quota of the first cluster, the predicted carbon emissions in the first period and the actual carbon emissions in the second period, the power parameters of the first cluster in the first period are controlled, the floating quota includes the carbon emission difference of the first cluster and the carbon emission difference of the second cluster in the historical period, and the carbon emission difference of the cluster in any period is determined according to the actual emissions and carbon emission quota of the cluster in the corresponding period.

2. The method according to claim 1, wherein the operation data comprises power consumption data; The predicting, based on the operation data of the plurality of historical periods, the predicted carbon emissions of the first cluster in the first period includes: Processing the power consumption data of the multiple historical periods according to the power consumption prediction model to obtain predicted power consumption data of the first cluster in the first period; Based on a first coefficient and the predicted electricity consumption data, the predicted carbon emissions of the first cluster in the first period are determined, wherein the first coefficient represents a proportional relationship between electricity consumption and carbon emissions of the first cluster.

3. The method according to claim 1, wherein the operating data includes device status data related to power consumption; The predicting, based on the operation data of the plurality of historical periods, the predicted carbon emissions of the first cluster in the first period includes: Processing the device status data of the first cluster in the multiple historical periods according to the status prediction model to obtain predicted device status data of the first cluster in the first period; Determine predicted power consumption data of the first cluster in the first period based on the predicted device status data and a predetermined first association relationship, where the first association relationship is an association relationship between the device status data of the first cluster and power consumption; Based on a first coefficient and the predicted electricity consumption data, the predicted carbon emissions of the first cluster in the first period are determined, wherein the first coefficient represents a proportional relationship between electricity consumption and carbon emissions of the first cluster.

4. According to the method of claim 3, the method of determining the first association relationship comprises: Obtaining device status data and power consumption data of the first cluster in the multiple historical periods; The first association relationship is determined according to the device status data and power consumption data of the multiple historical periods.

5. The method according to claim 1, wherein the controlling the power parameter of the first cluster in the first period based on the floating quota of the first cluster, the predicted carbon emissions in the first period, and the actual carbon emissions in the second period comprises: If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both less than or equal to the carbon emission quota of the corresponding period, control the power parameters of the first cluster in the first period to be consistent with the power parameters in the second period, and add the carbon emission difference of the first cluster in the second period to the floating quota; If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both greater than the carbon emission quota of the corresponding period, and the carbon emission difference of the first cluster in the second period is greater than the floating quota, control the power parameter of the first cluster in the first period to be less than the power parameter in the second period; If the predicted carbon emissions of the first cluster in the first period and the actual carbon emissions in the second period are both greater than the carbon emission quota of the corresponding period, and the carbon emission difference of the first cluster in the second period is less than the floating quota, control the power parameters of the first cluster in the first period to be consistent with the power parameters in the second period, and subtract the carbon emission difference of the first cluster in the second period from the floating quota.

6. The method according to claim 5, wherein the step of subtracting the carbon emission difference of the first cluster in the second period from the floating quota comprises: The carbon emission differences in the floating quota are deducted in order from the largest to the smallest duration until the total of the deducted carbon emission differences matches the carbon emission difference of the first cluster in the second period. The duration is the time elapsed since the carbon emission difference was added to the floating quota.

7. The method according to claim 6, further comprising: Determine a duration threshold corresponding to the first cluster and the second cluster; The carbon emission difference whose duration is greater than the duration threshold is deleted from the floating quota.

8. The method according to claim 1, further comprising: Determining a unit quota of the first cluster in each unit time period of the first cycle according to the carbon emission quota of the first cluster in the first cycle, where the first cycle includes a plurality of unit time periods; At the end of each unit time period in the first cycle, the unit quota of the next unit time period and the power parameters of the first cluster in the next unit time period are updated according to the unit carbon emission difference of the previous unit time period, and the unit carbon emission difference is the difference between the emissions of the first cluster in the corresponding unit time period and the unit quota.

9. The method according to claim 8, further comprising: At the end of the first period, determining the difference between the actual emissions of the first cluster in the first period and the carbon emission quota; The carbon emission difference of the first cluster in the first period is determined according to the difference and the unit carbon emission difference of the first cluster in the last unit time period in the first period.

10. A device for controlling cluster power parameters, comprising: an obtaining unit, configured to obtain, at the beginning of a first cycle, operation data of the first cluster in a plurality of historical cycles, wherein the historical cycles include a cycle before the first cycle; a prediction unit, configured to predict the predicted carbon emissions of the first cluster in the first period based on the operation data of the plurality of historical periods; A control unit is used to control the power parameters of the first cluster in the first period based on the floating quota of the first cluster, the predicted carbon emissions in the first period and the actual carbon emissions in the second period, the floating quota includes the carbon emission difference of the first cluster and the carbon emission difference of the second cluster in the historical period, and the carbon emission difference of the cluster in any period is determined according to the actual emissions and carbon emission quota of the cluster in the corresponding period.

Images