Data analysis method and system applied to intelligent computing power center energy consumption system

Abstract

The application provides a data analysis method and system applied to an intelligent computing center energy consumption system, aiming to obtain and analyze energy consumption use hotspot data and energy consumption flow data, map the energy consumption flow path to the energy consumption distribution network, and then generate a region set corresponding to the energy consumption flow path. At the same time, by extracting key energy consumption region data related to each energy consumption use hotspot, the energy consumption use hotspot is mapped to the energy consumption distribution network, thereby generating the association information between the regions and the energy consumption use hotspots in the energy consumption distribution network. Based on the above association information, the energy consumption use hotspot set associated with the energy consumption flow path can be determined, and then the energy consumption monitoring data analysis of the intelligent computing center energy consumption system is carried out for these energy consumption use hotspot sets, which not only can more accurately grasp the energy consumption situation of the intelligent computing center, but also can provide strong data support for subsequent energy consumption optimization, thereby realizing more efficient and more environmentally friendly energy consumption management.

Description

Technical Field

[0001] This application relates to the field of computer technology, and more specifically, to a data analysis method and system applied to the energy consumption system of an intelligent computing center. Background Technology

[0002] With the rapid development of information technology, intelligent computing centers are playing an increasingly important role in data processing, storage, and transmission. However, the energy consumption of intelligent computing centers is also becoming increasingly prominent, becoming a key factor restricting their sustainable development. Therefore, in order to achieve energy conservation, emission reduction, and lower operating costs, refined management and optimization of the energy consumption of intelligent computing centers are particularly important.

[0003] Traditional energy management methods often focus only on overall energy consumption data, neglecting the flow and usage of energy consumption between different areas and devices. This makes it difficult to accurately identify energy hotspots and critical paths during energy management and optimization, thus hindering targeted improvements. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of this application, embodiments of this application provide a data analysis method applied to an energy consumption system in an intelligent computing center, the method comprising:

[0005] Obtain energy consumption hotspot data and energy flow data of the intelligent computing center energy consumption system;

[0006] Based on the energy flow data, the energy flow path is mapped to an energy distribution network that includes multiple regions, generating a set of regions corresponding to the energy flow path;

[0007] Extract key energy consumption area data that are respectively related to each energy consumption hotspot from the energy consumption hotspot data;

[0008] Based on the key energy consumption area data, each energy consumption hotspot is mapped to the energy consumption distribution network, generating association information between the areas and energy consumption hotspots in the energy consumption distribution network;

[0009] Based on the set of regions corresponding to the energy flow path and the associated information, a set of energy consumption hotspots associated with the energy flow path is determined, and energy consumption monitoring data analysis of the intelligent computing center energy consumption system is performed based on the set of energy consumption hotspots associated with the energy flow path.

[0010] In one possible implementation of the first aspect, the energy flow data includes multiple energy flow nodes; the step of mapping the energy flow path to an energy distribution network comprising multiple regions based on the energy flow data, and generating a region set corresponding to the energy flow path, includes:

[0011] Identify the energy flow nodes in the energy flow data that correspond to the energy flow path, and generate an energy flow node diagram of the energy flow path.

[0012] The energy flow node graph of the energy flow path is mapped to an energy distribution network that includes multiple regions, thereby generating a set of regions corresponding to the energy flow path.

[0013] In one possible implementation of the first aspect, mapping the energy flow node graph of the energy flow path to an energy distribution network comprising multiple regions to generate a set of regions corresponding to the energy flow path includes:

[0014] Walk through each energy flow node in the energy flow node graph along the energy flow path, and determine the node attribute information of the energy flow node that is walked to.

[0015] Determine the region attribute information associated with the node attribute information;

[0016] If the regional attribute information does not exist in the regional sequence of the energy consumption flow path, then the regional attribute information is loaded into the regional sequence;

[0017] Continue traversing the next energy flow node, return to the step of determining the node attribute information of the energy flow node reached, and continue execution until all energy flow nodes in the energy flow node graph have been traversed;

[0018] Based on the final generated region sequence, the set of regions corresponding to the energy consumption flow path is determined.

[0019] In one possible implementation of the first aspect, extracting key energy consumption area data related to each energy consumption hotspot from the energy consumption hotspot data includes:

[0020] Extract energy consumption area data for each energy consumption hotspot from the energy consumption hotspot data;

[0021] Based on the flow characteristic vectors between data from different energy consumption areas, the data from the energy consumption areas are selected to generate key energy consumption area data.

[0022] In one possible implementation of the first aspect, the step of selecting energy consumption region data based on the flow feature vectors between data from different energy consumption regions to generate key energy consumption region data includes:

[0023] Traverse any energy consumption region data and use the traversed energy consumption region data as the baseline energy consumption region data;

[0024] Other energy consumption region data within the target interval of the flow feature vector related to the benchmark energy consumption region data will be used as the linked energy consumption region data of the benchmark energy consumption region data.

[0025] The baseline energy consumption area data and the linked energy consumption area data are grouped into a data cluster;

[0026] Each data cluster is traversed, and key energy consumption area data is generated based on the flow rate corresponding to each energy consumption area data in the traversed data cluster. The flow rate corresponding to the energy consumption area data is determined by the energy consumption flow diagram composed of the energy consumption area data.

[0027] In one possible implementation of the first aspect, the step of traversing each data cluster and selecting based on the flow rate corresponding to each energy consumption region data in the traversed data cluster to generate key energy consumption region data includes:

[0028] Traverse any data cluster and determine the shared flow between every two energy consumption regions in the traversed data cluster;

[0029] If the shared flow meets the flow selection criteria, then the energy consumption area data with the larger flow is retained, and key energy consumption area data is generated.

[0030] In one possible implementation of the first aspect, the step of mapping each energy consumption hotspot to the energy consumption distribution network based on the key energy consumption area data, and generating association information between areas and energy consumption hotspots in the energy consumption distribution network, includes:

[0031] Extract energy consumption boundary information from the key energy consumption area data, and determine the minimum coverage block of the key energy consumption area data based on the energy consumption boundary information.

[0032] Extract each target energy consumption domain within the minimum coverage block from the energy consumption distribution network;

[0033] The system navigates through each target energy consumption domain. When the energy consumption focus position of the target energy consumption domain is within the energy consumption boundary information of the key energy consumption area data, the system obtains the regional attribute information of the target energy consumption domain.

[0034] The acquired regional attribute information is mapped to the energy consumption hotspots corresponding to the key energy consumption regional data, and the association information between regions and energy consumption hotspots in the energy consumption distribution network is generated.

[0035] In one possible implementation of the first aspect, the step of extracting energy consumption boundary information from the key energy consumption area data and determining the minimum coverage block of the key energy consumption area data based on the energy consumption boundary information includes:

[0036] Analyze the data of the key energy consumption areas to identify the boundary points or boundary lines of each energy consumption area, which serve as the energy consumption limit information;

[0037] Based on the extracted energy consumption limit information, the minimum coverage block that can completely contain the corresponding limit is calculated. The minimum coverage block provides the minimum coverage range of the energy consumption area in two-dimensional space.

[0038] In one possible implementation of the first aspect, determining the set of energy consumption hotspots associated with the energy consumption flow path based on the set of regions corresponding to the energy consumption flow path and the association information includes:

[0039] Based on the association information, determine the energy consumption hotspots associated with each region in the set of regions corresponding to the energy consumption flow path;

[0040] Based on time-series selection conditions or geographical distribution selection conditions, the energy consumption hotspots are selected to generate target energy consumption hotspots;

[0041] Based on the selected target energy consumption hotspots, a set of energy consumption hotspots associated with the energy flow path is generated.

[0042] In one possible implementation of the first aspect, the step of selecting the energy consumption hotspots based on temporal selection conditions or geographical distribution selection conditions to generate target energy consumption hotspots includes:

[0043] Obtain the continuous parameters corresponding to each energy consumption hotspot;

[0044] Remove energy consumption hotspots whose continuous parameters are less than the preset continuous parameters, and generate the first energy consumption hotspot;

[0045] Obtain the geographical distribution characteristics of the first energy consumption hotspot;

[0046] Based on the geographical distribution characteristics of the primary energy consumption hotspots, the data are cleaned to generate target energy consumption hotspots.

[0047] In one possible implementation of the first aspect, the step of analyzing the energy consumption monitoring data of the intelligent computing center energy consumption system based on the set of energy consumption hotspots associated with the energy consumption flow path includes:

[0048] The set of energy consumption hotspots associated with energy consumption flow paths is integrated with the original energy consumption monitoring data. For each energy consumption hotspot in the integrated data, a time-series prediction model is constructed based on the energy consumption monitoring data of the energy consumption hotspot. The time-series prediction model is used to predict future energy consumption trends.

[0049] Furthermore, clustering algorithms are used to perform clustering analysis on the energy consumption monitoring data of the energy consumption hotspots, different energy consumption patterns are identified, and the characteristic patterns of different energy consumption patterns are analyzed. Based on the characteristic patterns of different energy consumption patterns of the energy consumption hotspots, association rule mining algorithms are used to analyze the correlation between the energy consumption hotspots.

[0050] Abnormal energy consumption data is detected based on the correlation between the energy consumption hotspots, and corresponding energy consumption anomaly events are generated based on the abnormal energy consumption data.

[0051] In one possible implementation of the first aspect, the step of constructing a time-series prediction model based on energy consumption monitoring data of the energy consumption hotspot, the time-series prediction model being used to predict future energy consumption trends, includes:

[0052] For each energy consumption hotspot, capture the dynamic characteristics of time-series data from the energy consumption monitoring data;

[0053] The corresponding time series prediction model is selected based on the dynamic characteristics of the time series data, and a training set and a validation set are divided according to the dynamic characteristics of the time series data and the corresponding expert knowledge annotation data. The selected time series prediction model is trained using the training set, and the prediction performance of the time series prediction model is evaluated using the validation set.

[0054] The accuracy of the time series prediction model is evaluated by comparing the difference between the predicted value and the actual value. When the accuracy of the time series prediction model meets the expected conditions, the final time series prediction model is generated.

[0055] In one possible implementation of the first aspect, the step of detecting abnormal energy consumption data based on the correlation between the energy consumption hotspots and generating corresponding energy consumption anomaly events based on the abnormal energy consumption data includes:

[0056] Based on the correlation between energy consumption hotspots, a corresponding correlation network is constructed, where nodes in the correlation network represent energy consumption hotspots and edges represent the relationships between these hotspots.

[0057] The energy consumption monitoring data of the energy consumption hotspots are monitored in real time. The correlation network and predefined anomaly detection rules are used to detect anomalies in the energy consumption monitoring data. When abnormal energy consumption data is detected, an energy consumption anomaly event is generated.

[0058] In another aspect, embodiments of this application also provide a data analysis system applied to an energy consumption system of an intelligent computing center, including a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the machine-readable storage medium to implement the above-described method.

[0059] Based on the above, by comprehensively acquiring energy consumption hotspot data and energy flow data, and using these data for in-depth analysis and mapping, comprehensive and accurate energy consumption monitoring and analysis of the intelligent computing center's energy consumption system was achieved. Specifically, by mapping energy flow data to an energy distribution network encompassing multiple regions, a set of regions highly corresponding to energy flow paths was generated, improving the accuracy and efficiency of energy path analysis. Extracting key energy consumption region data closely related to each hotspot from the energy consumption hotspot data helps to quickly locate energy bottlenecks and optimization points. Mapping each energy consumption hotspot to the energy distribution network based on key energy consumption region data enhances the correlation information between regions and energy consumption hotspots. Through the set of regions corresponding to energy flow paths and the correlation information, a set of energy consumption hotspots closely related to the energy flow paths was identified, making energy consumption monitoring data analysis more focused and targeted. In-depth data analysis based on the energy consumption hotspot set obtained from the above steps helps to discover energy consumption anomalies, predict energy consumption trends, and provide decision support for energy management and energy-saving optimization in the intelligent computing center. Therefore, by comprehensively and accurately mapping and analyzing the energy consumption data and flow paths of the intelligent computing center's energy consumption system, the accuracy and efficiency of energy consumption monitoring have been effectively improved, providing solid technical support for the green and efficient operation of the intelligent computing center. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of the execution flow of the data analysis method applied to the energy consumption system of the intelligent computing center provided in the embodiments of this application.

[0061] Figure 2 This is a schematic diagram of the hardware architecture of a data analysis system applied to an energy consumption system in an intelligent computing center, provided in an embodiment of this application. Detailed Implementation

[0062] The present application will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a data analysis method for an energy consumption system in an intelligent computing center, provided in one embodiment of this application. The following is a detailed description of this data analysis method for an energy consumption system in an intelligent computing center.

[0063] Step S110: Obtain energy consumption hotspot data and energy flow data of the intelligent computing center energy consumption system.

[0064] In this embodiment, the server can obtain energy consumption hotspot data and energy flow data of the intelligent computing center energy consumption system from a database or other data storage system.

[0065] Specifically, the energy consumption hotspot data* can refer to the energy consumption characteristic data of devices or areas with high energy consumption in an intelligent computing center. This energy consumption characteristic data typically includes the device's power, operating time, and total energy consumption. For example, in a data center, some high-performance servers may become energy consumption hotspots due to their heavy computing tasks.

[0066] The energy flow data describes the flow of energy within a smart computing center from one device or area to another. For example, in a data center, electrical energy may first be received by a UPS (Uninterruptible Power Supply), then distributed to various server racks, and further distributed to individual servers. The power inflow and outflow at each stage of this process constitutes the energy flow data.

[0067] Step S120: Based on the energy flow data, map the energy flow path to an energy distribution network that includes multiple regions, and generate a set of regions corresponding to the energy flow path.

[0068] In this embodiment, the server receives energy flow data, which includes multiple energy flow nodes. Each energy flow node contains node attribute information, such as node ID, node name, and node type.

[0069] Furthermore, the server maps the energy flow path to an energy distribution network that includes multiple regions based on the node attribute information in the energy flow data. This means that the server needs to associate each energy flow node with the corresponding region in the energy distribution network.

[0070] Based on this, the server generates a set of regions corresponding to the energy flow path. This set of regions includes all regions involved in the energy flow path.

[0071] The energy flow path refers to the path of energy flow from the source to the destination, which may include multiple intermediate energy flow nodes. For example, the entire power supply path from the external power supply of a data center to a specific server is an energy flow path.

[0072] The energy consumption distribution network is a topology that represents the energy consumption distribution of all devices or areas within an intelligent computing center. It can include information such as physical location, device type, and energy intensity. For example, the floor layout, rack arrangement, and distribution of servers and network devices in a data center can all constitute an energy consumption distribution network.

[0073] The set of regions is a collection of all regions traversed by an energy flow path. For example, if an energy flow path passes through a server area, a network switching area, and a storage area, then these three areas constitute a set of regions.

[0074] Step S130: Extract key energy consumption area data related to each energy consumption hotspot from the energy consumption hotspot data.

[0075] In this embodiment, the server extracts key energy consumption area data related to each energy consumption hotspot from the energy consumption hotspot data. This key energy consumption area data may include information such as the location, area, and energy load of the energy consumption hotspot.

[0076] Step S140: Based on the key energy consumption area data, map each energy consumption hotspot to the energy consumption distribution network, and generate association information between the areas and energy consumption hotspots in the energy consumption distribution network.

[0077] The associated information describes the mapping relationship and attribute information between regions and energy consumption hotspots in the energy consumption distribution network.

[0078] In this embodiment, the server maps each energy consumption hotspot to the energy consumption distribution network based on key energy consumption area data. This means the server needs to associate each key energy consumption area with a corresponding area in the energy consumption distribution network. Then, the server generates association information between the areas in the energy consumption distribution network and the energy consumption hotspots. This association information may include the relationship between the energy consumption hotspots and the associated areas, their weights, etc.

[0079] Step S150: Based on the set of regions corresponding to the energy flow path and the associated information, determine the set of energy consumption hotspots associated with the energy flow path, and perform energy consumption monitoring data analysis of the intelligent computing center energy consumption system based on the set of energy consumption hotspots associated with the energy flow path.

[0080] In this embodiment, the server determines a set of energy consumption hotspots associated with the energy consumption flow path based on the set of regions corresponding to the energy consumption flow path and related information. This means that the server needs to filter out the energy consumption hotspots related to the energy consumption flow path from the related information.

[0081] Next, the server performs energy consumption monitoring data analysis on the intelligent computing center's energy consumption system based on the set of energy consumption hotspots associated with the energy flow path. This may include operations such as time-series prediction, cluster analysis, and association rule mining on the energy consumption monitoring data of the energy consumption hotspots.

[0082] Based on the above steps, by comprehensively acquiring energy consumption hotspot data and energy flow data, and using these data for in-depth analysis and mapping, comprehensive and accurate energy consumption monitoring and analysis of the intelligent computing center's energy consumption system is achieved. Specifically, by mapping energy flow data to an energy distribution network encompassing multiple regions, a set of regions highly corresponding to energy flow paths is generated, improving the accuracy and efficiency of energy path analysis. Extracting key energy consumption region data closely related to each hotspot from the energy consumption hotspot data helps to quickly locate energy bottlenecks and optimization points. Mapping each energy consumption hotspot to the energy distribution network based on key energy consumption region data enhances the correlation information between regions and energy consumption hotspots. Through the set of regions corresponding to energy flow paths and the correlation information, a set of energy consumption hotspots closely related to the energy flow paths is identified, making energy consumption monitoring data analysis more focused and targeted. In-depth data analysis based on the energy consumption hotspot set obtained in the above steps helps to discover energy consumption anomalies, predict energy consumption trends, and provide decision support for energy management and energy-saving optimization in the intelligent computing center. Therefore, by comprehensively and accurately mapping and analyzing the energy consumption data and flow paths of the intelligent computing center's energy consumption system, the accuracy and efficiency of energy consumption monitoring have been effectively improved, providing solid technical support for the green and efficient operation of the intelligent computing center.

[0083] In one possible implementation, the energy flow data includes multiple energy flow nodes. Step S120 may include:

[0084] Step S121: Determine the energy flow nodes in the energy flow data that correspond to the energy flow path, and generate an energy flow node diagram of the energy flow path.

[0085] Step S122: Map the energy flow node graph of the energy flow path to an energy distribution network that includes multiple regions, and generate a set of regions corresponding to the energy flow path.

[0086] To better understand and describe this process, a specific scenario will be used as an example below.

[0087] Suppose there is a data center containing multiple servers, network devices, and other IT equipment. The energy distribution network of this data center can be divided into several zones, such as a server zone, a network device zone, and a rack zone. The goal is to find energy flow paths based on real-time energy flow data and map these paths to the corresponding set of zones.

[0088] First, real-time energy flow data can be collected. This energy flow data can be obtained from various sensors, smart meters, energy monitoring systems, and other sources. Each device in the data center generates energy consumption data, which can be integrated to form a complete energy flow data package.

[0089] After collecting energy flow data, the energy flow nodes corresponding to the energy flow paths can be identified. These energy flow nodes can be servers, network devices, server racks, etc. By analyzing the energy consumption data, the starting and ending points of the energy flow, as well as the intermediate nodes between them, can be found.

[0090] Based on the identified energy flow nodes, an energy flow node graph can be constructed. This graph represents the energy transmission relationships between devices. In this graph, nodes represent devices, and edges represent energy transmission paths. Graph theory algorithms can be used to analyze this graph and find the energy flow paths.

[0091] After generating the energy flow node graph, this graph can be mapped to the actual energy distribution network. This means that the nodes and edges in the energy flow node graph need to be mapped to regions in the energy distribution network. This can be achieved by assigning nodes in the energy flow node graph to the corresponding regions. For example, nodes in the server region can be assigned to the server region, and nodes in the network device region can be assigned to the network device region.

[0092] Finally, based on the energy flow node graph mapped to the energy distribution network, a set of regions corresponding to the energy flow paths is generated. This set of regions helps to understand the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points.

[0093] By following the steps above, energy flow data can be collected, analyzed, and visualized, thereby helping data center managers better understand energy distribution and take corresponding measures to optimize energy consumption.

[0094] In one possible implementation, step S122 may include:

[0095] Step S1221: Walk through each energy flow node in the energy flow node graph of the energy flow path and determine the node attribute information of the energy flow node that is walked through.

[0096] Step S1222: Determine the region attribute information associated with the node attribute information.

[0097] Step S1223: If the regional attribute information does not exist in the regional sequence of the energy consumption flow path, then the regional attribute information is loaded into the regional sequence.

[0098] Step S1224: Continue walking to the next energy flow node, return to the step of determining the node attribute information of the energy flow node walked to, and continue to execute until all energy flow nodes in the energy flow node graph have been walked.

[0099] Step S1225: Based on the finally generated region sequence, determine the region set corresponding to the energy consumption flow path.

[0100] To better understand and describe this process, we will continue to use the previous data center scenario as an example.

[0101] After generating the energy flow node graph, you can traverse all nodes in the graph. You can start from one node and move along the edges to the next node until all nodes have been visited. During this process, you can record the visited nodes to avoid revisiting them.

[0102] During the process of traversing the network, attribute information for each visited node can be obtained. This attribute information may include device type, power consumption, location, etc. This information can be obtained by querying a device database or calling the device API.

[0103] Based on the node attribute information of the energy flow nodes that the network reaches, the associated regional attribute information can be determined. This can be achieved by matching the node attribute information with regions in the energy distribution network. For example, the node attribute information of the server region can be matched with the regional attribute information of the server region.

[0104] After determining the regional attribute information associated with the node attribute information, it can be checked whether this regional attribute information already exists in the regional sequence of the energy consumption flow path. If it does not exist, the regional attribute information can be loaded into the regional sequence.

[0105] After processing the current node, the process can continue to the next energy flow node. This process will be repeated until all energy flow nodes in the energy flow node graph have been visited.

[0106] After traversing all energy flow nodes, a region sequence containing region attribute information corresponding to the energy flow path will be obtained. This region sequence can be used to determine the region set corresponding to the energy flow path. For example, the region attribute information of server areas, network device areas, etc., can be combined into a region set.

[0107] By following the steps above, the energy flow node graph of the energy flow path can be mapped to an energy distribution network comprising multiple regions, generating a set of regions corresponding to the energy flow path. This helps to better understand the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points.

[0108] In one possible implementation, step S130 may include:

[0109] Step S131: Extract energy consumption area data for each energy consumption hotspot from the energy consumption hotspot data.

[0110] Step S132: Select the energy consumption area data based on the flow feature vector between different energy consumption area data to generate key energy consumption area data.

[0111] To better understand and describe this process, we will continue to use the previous data center scenario as an example.

[0112] After collecting energy consumption hotspot data, energy consumption area data related to each hotspot can be extracted from this data. This energy consumption area data can include device type, power consumption, location, etc. This information can be obtained by querying a device database or calling the device API.

[0113] After extracting energy consumption data from various energy-intensive areas, key energy consumption areas can be selected based on the flow feature vectors of this data. These flow feature vectors represent the energy transmission relationships between devices, such as the probability of device A transmitting energy to device B. Machine learning algorithms (such as clustering and classification) can be used to analyze these flow feature vectors to determine key energy consumption areas.

[0114] Here is a specific example scenario:

[0115] Assume a data center has three areas: a server area, a network equipment area, and a server rack area. Energy consumption hotspot data has been collected, including device power consumption information for these three areas. First, extract energy consumption data related to each energy consumption hotspot from this data; for example, server power consumption in the server area, switch power consumption in the network equipment area, and air conditioning power consumption in the server rack area.

[0116] Next, we can analyze the flow feature vectors of this data. We can calculate the energy transmission probability between the server area and other areas, as well as the energy transmission probability between the network device area and other areas. By clustering or classifying these probabilities, we can identify key energy-consuming areas. For example, we might find that the energy transmission probability is high between the server area and the network device area, therefore, the data from these two areas can be considered key energy-consuming areas.

[0117] By following the steps above, key energy consumption area data related to each energy consumption hotspot can be extracted from the energy consumption hotspot data. This helps to better understand the distribution of energy consumption within the data center, as well as potential energy consumption bottlenecks and optimization points.

[0118] In one possible implementation, step S132 includes:

[0119] Step S1321: Traverse any energy consumption area data and use the traversed energy consumption area data as the baseline energy consumption area data.

[0120] Step S1322: Other energy consumption region data within the target interval of the flow feature vector related to the benchmark energy consumption region data are used as the linked energy consumption region data of the benchmark energy consumption region data.

[0121] Step S1323: The baseline energy consumption area data and the linked energy consumption area data are grouped into a data cluster.

[0122] Step S1324: Traverse each data cluster and select key energy consumption region data based on the flow rate corresponding to each energy consumption region data in the traversed data cluster. The flow rate corresponding to the energy consumption region data is determined by the energy consumption flow diagram formed by the energy consumption region data.

[0123] In this process, key energy consumption area data can be selected based on the flow characteristic vectors between data from different energy consumption areas.

[0124] In detail, after collecting energy consumption area data, one of these energy consumption areas can be selected as the baseline energy consumption area data. This baseline energy consumption area data can be any one, such as data from the server area.

[0125] After determining the baseline energy consumption area data, related linked energy consumption area data can be found. This can be achieved by analyzing the flow characteristic vector of the baseline energy consumption area data. For example, data related to the network equipment area and server rack area can be found.

[0126] After identifying the baseline energy consumption region data and its associated energy consumption region data, they can be grouped into a data cluster. This data cluster can represent an energy consumption cluster, including the baseline energy consumption region data and the associated energy consumption region data.

[0127] After grouping the baseline energy consumption area data and its associated energy consumption area data into a single data cluster, each data cluster can be traversed, and key energy consumption area data can be selected based on the flow rate corresponding to each energy consumption area. This can be achieved by analyzing an energy flow diagram. An energy flow diagram can represent the energy transmission relationship between devices, such as the energy transmission relationship between a server area and a network device area. By analyzing these relationships, key energy consumption area data can be determined.

[0128] By following the steps above, key energy consumption area data can be selected based on the flow characteristic vectors between data from different energy consumption areas. This helps to better understand the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points.

[0129] In one possible implementation, step S1324 includes:

[0130] Step S1324-1: Traverse any data cluster and determine the shared flow between every two energy consumption regions in the traversed data cluster.

[0131] Step S1324-2: If the shared flow meets the flow selection criteria, then retain the energy consumption area data with larger flow and generate key energy consumption area data.

[0132] In this process, each data cluster can be traversed, and the flow corresponding to each energy consumption area data in the traversed data cluster can be selected to generate key energy consumption area data.

[0133] In detail, after collecting energy consumption region data, a data cluster can be selected from this data for traversal. During the traversal, the shared flow between every two energy consumption regions within the data cluster can be determined. This can be achieved by analyzing an energy flow graph. An energy flow graph can represent the energy transfer relationships between devices, such as the energy transfer relationship between a server area and a network device area. By analyzing these relationships, the shared flow between every two energy consumption regions within the data cluster can be determined.

[0134] After determining the shared flow between every two energy consumption regions in the data cluster, it can be judged whether these shared flows meet the flow selection criteria. The flow selection criteria can be defined according to actual needs, such as the shared flow exceeding a certain threshold. If the shared flow meets the criteria, the energy consumption region data with the larger flow can be retained and designated as the key energy consumption region data.

[0135] Here is a specific example scenario:

[0136] Suppose a data center has three areas: a server area, a network equipment area, and a server rack area. Energy consumption data has been collected, including power consumption information for the devices in these three areas. This data is first grouped into a single data cluster.

[0137] Next, traverse this data cluster to determine the shared flow between every two energy consumption areas within the cluster. For example, the shared flow between the server area and the network device area, as well as the shared flow between the server area and the rack area, can be calculated. By analyzing these shared flows, it can be found that the shared flow between the server area and the network device area is relatively large, therefore the data from these two areas can be considered as critical energy consumption areas.

[0138] By following the steps above, each data cluster can be traversed, and key energy consumption area data can be generated based on the flow rate corresponding to each energy consumption area data within the traversed data cluster. This helps to better understand the distribution of energy consumption within the data center, as well as potential energy consumption bottlenecks and optimization points.

[0139] In one possible implementation, step S140 may include:

[0140] Step S141: Extract energy consumption boundary information from the key energy consumption area data, and determine the minimum coverage block of the key energy consumption area data based on the energy consumption boundary information.

[0141] Step S142: Extract each target energy consumption domain within the minimum coverage block from the energy consumption distribution network.

[0142] Step S143: Walk through each of the target energy consumption domains. When the energy consumption focus position of the target energy consumption domain is within the energy consumption boundary information of the key energy consumption area data, obtain the regional attribute information of the target energy consumption domain.

[0143] Step S144: Establish a mapping relationship between the acquired regional attribute information and the energy consumption hotspots corresponding to the key energy consumption regional data, and generate the association information between regions and energy consumption hotspots in the energy consumption distribution network.

[0144] In this process, energy consumption hotspots can be mapped to an energy consumption distribution network based on key energy consumption area data, generating correlation information between areas in the energy consumption distribution network and energy consumption hotspots.

[0145] In detail, after collecting data on key energy consumption areas, energy consumption boundary information can be extracted from this data. Next, this energy consumption boundary information can be used to determine the minimum coverage area of ​​the key energy consumption area data. The minimum coverage area can represent the region in the energy consumption distribution network that is related to the key energy consumption area data.

[0146] After determining the minimum coverage area, target energy consumption domains can be extracted from these areas within the energy consumption distribution network. A target energy consumption domain can represent a specific area within the energy consumption distribution network that is related to key energy consumption area data.

[0147] When traversing target energy consumption domains, it is possible to check whether the energy consumption focus location of each target energy consumption domain is within the energy consumption boundary information of the key energy consumption area data. If the condition is met, the regional attribute information of the target energy consumption domain can be obtained. This attribute information may include device type, power consumption, location, etc.

[0148] After obtaining the regional attribute information of the target energy consumption domain, this information can be mapped to energy usage hotspots corresponding to key energy consumption areas. This can be achieved by associating regional attribute information with energy usage hotspots. The generated association information can help to better understand the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points.

[0149] Here is a specific example scenario:

[0150] Suppose a data center has three areas: a server area, a network equipment area, and a server rack area. Energy consumption data for each area has been collected, including power consumption information for the devices in these three areas. First, this data is grouped into a single data cluster, and then key energy consumption data is extracted.

[0151] Next, each energy consumption hotspot can be mapped to an energy consumption distribution network. The smallest coverage blocks related to key energy consumption area data, such as server and network device areas, can be extracted from the energy consumption distribution network. Then, these areas are traversed, checking whether the energy consumption focus location of each target energy consumption domain is within the energy consumption boundary information of the key energy consumption area data. If the condition is met, the regional attribute information of the target energy consumption domain can be obtained, such as device type, power consumption, and location of the server area.

[0152] Finally, a mapping relationship is established between the acquired regional attribute information and the energy consumption hotspots corresponding to the key energy consumption area data. For example, the regional attribute information of the server area can be associated with the server area data in the key energy consumption area data. Through the above steps, the association information between areas and energy consumption hotspots in the energy consumption distribution network can be generated. This helps to better understand the distribution of energy consumption within the data center, as well as potential energy consumption bottlenecks and optimization points.

[0153] In one possible implementation, step S141 includes:

[0154] Step S1411: Analyze the key energy consumption area data and identify the boundary points or boundary lines of each energy consumption area as the energy consumption limit information.

[0155] Step S1412: Based on the extracted energy consumption limit information, calculate the minimum coverage block that can completely contain the corresponding limit. The minimum coverage block provides the minimum coverage range of the energy consumption area in two-dimensional space.

[0156] In this process, energy consumption boundary information can be extracted from the data of key energy consumption areas, and the minimum coverage block of the key energy consumption area data can be determined based on this information.

[0157] In detail, after collecting data on key energy consumption areas, this data can be analyzed to identify the boundary points or boundary lines of each energy consumption area. These boundary points or boundary lines can represent the extent or limit of the energy consumption area. These boundary points or boundary lines can be used as energy consumption boundary information for use in subsequent steps.

[0158] After obtaining energy consumption limit information, the minimum coverage block that can completely contain the corresponding limit can be calculated based on this information. The minimum coverage block can represent the minimum coverage area of ​​the energy consumption region in two-dimensional space, which can help to better understand the distribution of energy consumption within the data center.

[0159] Suppose a data center has three areas: a server area, a network equipment area, and a server rack area. Energy consumption data for each area has been collected, including power consumption information for the devices in these three areas. First, this data is grouped into a single data cluster, and then key energy consumption data is extracted.

[0160] Next, energy consumption boundary information can be extracted from the data of key energy consumption areas. For example, boundary points or boundary lines of the server area and the network device area can be identified. By analyzing these boundary points or boundary lines, energy consumption boundary information in the data of key energy consumption areas can be obtained.

[0161] Then, the minimum coverage blocks can be calculated based on the extracted energy consumption boundary information. For example, the minimum coverage blocks for the server area and network device area can be calculated, representing the minimum coverage area of ​​the server area and network device area in two-dimensional space, respectively. Through the above steps, the minimum coverage blocks for key energy consumption area data can be determined. This helps to better understand the distribution of energy consumption within the data center, as well as potential energy consumption bottlenecks and optimization points.

[0162] In one possible implementation, step S150 may include:

[0163] Step S151: Based on the association information, determine the energy consumption hotspots associated with each region in the region set corresponding to the energy consumption flow path.

[0164] Step S152: Select the energy consumption hotspots based on the time sequence selection conditions or the geographical distribution selection conditions to generate target energy consumption hotspots.

[0165] Step S153: Generate a set of energy consumption hotspots associated with the energy consumption flow path based on the selected target energy consumption hotspots.

[0166] In this process, the set of energy consumption hotspots associated with the energy consumption flow path can be determined based on the set of regions corresponding to the energy consumption flow path and related information.

[0167] In detail, after collecting the relevant information, we can use this information to identify the energy consumption hotspots associated with each region in the set of regions corresponding to the energy consumption flow path.

[0168] After identifying energy consumption hotspots associated with each region in the set of regions corresponding to energy flow paths, these hotspots can be selected based on either temporal or geographical distribution criteria. Temporal selection criteria can be based on the temporal characteristics of the hotspots, such as selecting hotspots that occurred within a specific time period. Geographical selection criteria can be based on the geographical location of the hotspots, such as selecting hotspots located within a specific region. These selection criteria can then be used to generate target energy consumption hotspots.

[0169] After selecting target energy consumption hotspots, these hotspots can be associated with energy flow paths to generate a set of energy consumption hotspots linked to those paths. This set helps to better understand the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points.

[0170] Suppose a data center has three areas: a server area, a network equipment area, and a server rack area. Energy consumption data for each area has been collected, including power consumption information for the devices in these three areas. First, this data is grouped into a single data cluster, and then key energy consumption data is extracted.

[0171] Next, based on the set of regions corresponding to the energy flow path and the associated information, the set of energy consumption hotspots associated with the energy flow path can be determined. For example, energy consumption hotspots corresponding to the server area and the network device area can be identified.

[0172] These hotspots can then be selected based on either time-series or geographical distribution criteria. For example, hotspots in server and network device zones generated within a specific time period can be selected. These selection criteria allow for the generation of target energy consumption hotspots.

[0173] Finally, the selected target energy consumption hotspots are associated with energy flow paths, generating a set of energy consumption hotspots associated with each path. For example, target energy consumption hotspots in server areas and network device areas can be associated with energy flow paths. Through these steps, the set of energy consumption hotspots associated with energy flow paths can be determined. This helps to better understand the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points.

[0174] In one possible implementation, step S152 includes:

[0175] Step S1521: Obtain the continuous parameters corresponding to each energy consumption hotspot.

[0176] Step S1522: Eliminate energy consumption hotspots with continuous parameters less than the preset continuous parameters, and generate the first energy consumption hotspot.

[0177] Step S1523: Obtain the geographical distribution characteristics of the first energy consumption hotspot.

[0178] Step S1524: Clean the target energy consumption hotspots based on the geographical distribution characteristics of the first energy consumption hotspots.

[0179] In this process, energy consumption hotspots can be selected based on time-series selection conditions or geographical distribution selection conditions to generate target energy consumption hotspots.

[0180] In detail, after collecting data on energy consumption hotspots, the duration parameters for each hotspot can be obtained. These duration parameters represent information such as the duration of the hotspot and its continuous energy consumption. Hotspots can then be selected based on these duration parameters.

[0181] After obtaining the duration parameters for each hotspot, hotspots with duration parameters less than a preset duration parameter can be removed. The preset duration parameter can be set according to actual needs, such as 10 minutes or 100 watt-hours. By removing these hotspots, the first energy-consuming hotspot can be generated.

[0182] After generating the primary energy consumption hotspots, their geographical distribution characteristics can be obtained. These characteristics can include information such as the location and extent of the hotspots. The primary energy consumption hotspots can then be selected based on these characteristics.

[0183] After obtaining the geographical distribution characteristics of the primary energy consumption hotspots, these hotspots can be cleaned based on these characteristics. For example, hotspots can be filtered according to their location and extent, retaining only those that meet the criteria. These filtering conditions can then be used to generate target energy consumption hotspots.

[0184] For example, suppose a data center has three areas: a server area, a network equipment area, and a server rack area. Energy consumption data for each area has been collected, including power consumption information for the devices in these three areas. First, this data is grouped into a data cluster, and then key energy consumption data is extracted.

[0185] Next, energy consumption hotspots can be selected based on time-series or geographical distribution criteria. For example, energy consumption hotspots in server and network device areas can be obtained, along with their duration parameters. Then, hotspots with duration parameters lower than a preset duration parameter can be removed, generating the first energy consumption hotspot.

[0186] Next, the geographical distribution characteristics of the primary energy consumption hotspots can be obtained. For example, the location and range of the primary energy consumption hotspots in the server area and network device area can be obtained. Then, the hotspots can be filtered based on these characteristics, retaining only those that meet the criteria. Through these filtering conditions, target energy consumption hotspots can be generated.

[0187] Finally, the selected target energy consumption hotspots are associated with energy flow paths, generating a set of energy consumption hotspots associated with each path. For example, target energy consumption hotspots in server areas and network device areas can be associated with energy flow paths. Through these steps, the set of energy consumption hotspots associated with energy flow paths can be determined. This helps to better understand the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points.

[0188] In one possible implementation, step S150 may further include:

[0189] Step S154: Integrate the set of energy consumption hotspots associated with the energy consumption flow path with the original energy consumption monitoring data, and for each energy consumption hotspot in the integrated data, construct a time series prediction model based on the energy consumption monitoring data of the energy consumption hotspot. The time series prediction model is used to predict future energy consumption trends.

[0190] Step S155: Clustering algorithm is used to perform clustering analysis on the energy consumption monitoring data of the energy consumption hotspots to identify different energy consumption patterns. After analyzing the characteristic patterns of different energy consumption patterns, association rule mining algorithm is used to analyze the correlation between energy consumption hotspots based on the characteristic patterns of different energy consumption patterns of the energy consumption hotspots.

[0191] Step S156: Detect abnormal energy consumption data based on the correlation between the energy consumption hotspots, and generate corresponding abnormal energy consumption events based on the abnormal energy consumption data.

[0192] In this process, energy consumption monitoring data of the intelligent computing center's energy consumption system can be analyzed based on the set of energy consumption hotspots associated with energy consumption flow paths.

[0193] In detail, after collecting a set of energy usage hotspots associated with energy flow paths, these hotspots can be integrated with the raw energy monitoring data. The integrated data can help to better understand the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points.

[0194] After integrating the data, time-series predictive models can be built based on energy consumption monitoring data for each energy consumption hotspot. These models can help predict future energy consumption trends, allowing for proactive measures to address potential energy consumption issues.

[0195] After constructing the time-series prediction model, clustering algorithms can be used to perform cluster analysis on energy consumption monitoring data of energy consumption hotspots. Cluster analysis can help identify different energy consumption patterns and analyze the characteristic patterns of these patterns. Then, based on these characteristic patterns, association rule mining algorithms can be used to analyze the correlation between energy consumption hotspots.

[0196] After analyzing the correlations between energy consumption hotspots, abnormal energy consumption data can be detected based on these correlations. Abnormal energy consumption data can represent the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points. Then, corresponding energy consumption anomaly events can be generated based on this abnormal energy consumption data.

[0197] Here is a specific example scenario:

[0198] Suppose a data center has three areas: a server area, a network equipment area, and a server rack area. Energy consumption data for each area has been collected, including power consumption information for the devices in these three areas. First, this data is grouped into a single data cluster, and then key energy consumption data is extracted.

[0199] Next, energy consumption monitoring data from the intelligent computing center's energy consumption system can be analyzed based on sets of energy consumption hotspots associated with energy flow paths. For example, energy consumption hotspots in the server and network equipment areas can be integrated with the original energy consumption monitoring data, and time-series prediction models can be constructed based on the energy consumption monitoring data of these hotspots. Then, clustering algorithms can be used to perform cluster analysis on the energy consumption monitoring data of these hotspots, identifying different energy consumption patterns and analyzing the characteristic patterns of these patterns. Next, association rule mining algorithms can be used to analyze the correlations between energy consumption hotspots based on these characteristic patterns. Finally, abnormal energy consumption data can be detected based on these correlations, and corresponding energy consumption anomaly events can be generated based on these abnormal energy consumption data.

[0200] By following the steps above, energy consumption monitoring data from the intelligent computing center's energy consumption system can be analyzed to understand the distribution of energy consumption within the data center, as well as potential energy bottlenecks and optimization points. This helps to better understand the distribution of energy consumption within the data center, and identify potential energy bottlenecks and optimization points.

[0201] In one possible implementation, step S154 includes:

[0202] Step S1541: For each energy consumption hotspot, capture the dynamic characteristics of time series data from the energy consumption monitoring data.

[0203] Step S1542: Select the corresponding time series prediction model according to the dynamic characteristics of the time series data, and divide the training set and validation set according to the dynamic characteristics of the time series data and the corresponding expert knowledge annotation data. Use the training set to train the selected time series prediction model, and use the validation set to evaluate the prediction performance of the time series prediction model.

[0204] Step S1543: The accuracy of the time series prediction model is evaluated by comparing the difference between the predicted value and the actual value of the time series prediction model. When the accuracy of the time series prediction model meets the expected conditions, the final time series prediction model is generated.

[0205] In this process, time-series prediction models can be built based on energy consumption monitoring data of energy consumption hotspots to predict future energy consumption trends.

[0206] In detail, after collecting energy consumption monitoring data from energy consumption hotspots, the dynamic characteristics of time-series data can be captured from this data. These dynamic characteristics can include data trends, periodicity, and volatility. Statistical methods or machine learning methods can be used to capture these characteristics.

[0207] After capturing the dynamic characteristics of time series data, a suitable time series prediction model can be selected based on these characteristics. Common time series prediction models include ARIMA, LSTM, and GRU. Then, training and validation sets can be created based on the dynamic characteristics of the time series data and the corresponding expert knowledge-annotated data. The training set is used to train the selected time series prediction model, and the validation set is used to evaluate the model's predictive performance.

[0208] After training and validating the time series prediction model, the model's accuracy can be evaluated by comparing the differences between the model's predictions and the actual values. Metrics such as mean squared error (MSE) and mean absolute error (MAE) can be used to measure the model's predictive performance. If the model's accuracy meets the expected conditions (e.g., the prediction error is less than a certain threshold), the final time series prediction model can be generated.

[0209] Here is a specific example scenario:

[0210] Suppose a data center has an energy consumption hotspot, namely the server area. Energy consumption monitoring data for the server area has been collected, including energy consumption values ​​over a past period. The first step is to capture the dynamic characteristics of the time-series data, such as trends and periodicity.

[0211] Then, a suitable time-series prediction model, such as LSTM, is selected based on the captured dynamic features. Next, training and validation sets can be created based on the dynamic features of the time-series data and the corresponding expert knowledge annotation data. The selected LSTM model can be trained using the training set, and the model's predictive performance can be evaluated using the validation set.

[0212] After training and validating the model, its accuracy is evaluated by comparing the differences between the model's predictions and actual values. If the model's prediction error is less than a preset threshold, its accuracy can be considered to meet the expected conditions. At this point, the final time-series prediction model can be generated to predict future energy consumption trends in the server area.

[0213] By following the steps above, a time-series forecasting model can be built based on energy consumption monitoring data of energy usage hotspots to predict future energy consumption trends. This helps to take proactive measures to address potential energy consumption issues and optimize energy management in data centers.

[0214] In one possible implementation, step S156 may include:

[0215] Step S1561: Based on the correlation between energy consumption hotspots, construct a corresponding correlation network, where nodes in the correlation network represent energy consumption hotspots and edges represent the relationships between the energy consumption hotspots.

[0216] Step S1562: Real-time monitoring of energy consumption monitoring data of the energy consumption hotspots; using the correlation network and predefined anomaly detection rules to detect anomalies in the energy consumption monitoring data; and generating an energy consumption anomaly event when anomaly energy consumption data is detected.

[0217] In this embodiment, abnormal energy consumption data can be detected based on the correlation between energy consumption hotspots, and corresponding abnormal energy consumption events can be generated based on these abnormal energy consumption data.

[0218] In detail, after collecting the relationships between energy consumption hotspots, a relationship network can be constructed. Nodes in this network represent energy consumption hotspots, and edges represent the relationships between these hotspots. This relationship network can be constructed using graph theory or network analysis methods.

[0219] After constructing the correlation network, energy consumption monitoring data for energy usage hotspots can be monitored in real time. Real-time data stream processing technologies (such as Apache Kafka and Apache Flink) can be used to process and analyze this data in real time. During real-time monitoring, the correlation network and predefined anomaly detection rules can be used to detect anomalies in the energy consumption monitoring data. Anomaly detection rules can include threshold detection, clustering detection, and anomaly detection algorithms (such as IsolationForest and Local Outlier Factor). If abnormal energy consumption data is detected, corresponding energy consumption anomaly events can be generated based on this data.

[0220] Here is a specific example scenario:

[0221] Suppose a data center has three areas: a server area, a network equipment area, and a server rack area. Energy consumption monitoring data for these three areas, as well as the relationships between them, have been collected. First, a relationship network is constructed based on these relationships.

[0222] Next, energy consumption data for these areas can be monitored in real time. Real-time data stream processing technology can be used to process and analyze this data in real time. During real-time monitoring, correlation networks and predefined anomaly detection rules can be used to detect anomalies in the energy consumption monitoring data. For example, a threshold can be set; when the energy consumption of a certain area exceeds this threshold, it is considered that the area has abnormal energy consumption.

[0223] If abnormal energy consumption data is detected, corresponding energy consumption anomaly events can be generated based on this data. For example, if the energy consumption of a server area exceeds a preset threshold, an event related to abnormal energy consumption in that server area can be generated. This abnormal energy consumption event can include information such as the abnormal energy consumption data, the time of the anomaly, and the area affected. These events can be stored in a database for subsequent analysis and processing.

[0224] By following the steps above, abnormal energy consumption data can be detected based on the correlation between energy consumption hotspots, and corresponding energy consumption anomaly events can be generated based on this abnormal data. This helps to identify potential energy consumption problems in a timely manner and take appropriate measures for optimization.

[0225] Figure 2 This illustration shows the hardware structure of a data analysis system 100 for implementing the above-described data analysis method for an energy-consuming intelligent computing center, as provided in an embodiment of this application. Figure 2 As shown, the data analysis system 100 applied to the energy consumption system of the intelligent computing center may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

[0226] In one possible design, the data analysis system 100 applied to the energy consumption system of the intelligent computing center can be a single server or a server group. The server group can be centralized or distributed (e.g., the data analysis system 100 applied to the energy consumption system of the intelligent computing center can be a distributed system). In some embodiments, the data analysis system 100 applied to the energy consumption system of the intelligent computing center can be local or remote. For example, the data analysis system 100 applied to the energy consumption system of the intelligent computing center can access information and / or data stored in machine-readable storage medium 120 via a network. As another example, the data analysis system 100 applied to the energy consumption system of the intelligent computing center can be directly connected to the machine-readable storage medium 120 to access the stored information and / or data. In some embodiments, the data analysis system 100 applied to the energy consumption system of the intelligent computing center can be implemented on a data analysis system for the energy consumption system of the intelligent computing center. By way of example only, the data analysis system applied to the energy consumption system of the intelligent computing center can include private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, internal cloud, multi-layer cloud, etc., or any combination thereof.

[0227] Machine-readable storage medium 120 may store data and / or instructions. In some embodiments, machine-readable storage medium 120 may store data acquired from an external terminal. In some embodiments, machine-readable storage medium 120 may store data and / or instructions used by the data analysis system 100 applied to the energy consumption system of the intelligent computing center to perform or use in order to accomplish the exemplary methods described in this application.

[0228] In the specific implementation process, at least one processor 110 executes computer-executable instructions stored in the machine-readable storage medium 120, so that the processor 110 can execute the data analysis method applied to the energy consumption system of the intelligent computing center as described in the above method embodiment. The processor 110, the machine-readable storage medium 120 and the communication unit 140 are connected through the bus 130. The processor 110 can be used to control the sending and receiving actions of the communication unit 140.

[0229] The specific implementation process of processor 110 can be found in the various method embodiments executed by the data analysis system 100 applied to the energy consumption system of the intelligent computing center. The implementation principle and technical effect are similar, and will not be repeated here.

[0230] Furthermore, this application embodiment also provides a readable storage medium, wherein computer-executable instructions are preset in the readable storage medium, and when the processor executes the computer-executable instructions, the data analysis method applied to the energy consumption system of the intelligent computing center as described above is implemented.

[0231] It should be noted that, in order to simplify the description disclosed in this application and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of the embodiments of this application sometimes combines multiple features into a single embodiment, drawing, or description thereof. Similarly, it should be noted that, in order to simplify the description disclosed in this application and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of the embodiments of this application sometimes combines multiple features into a single embodiment, drawing, or description thereof.

Claims

1. A data analysis method applied to an energy consumption system in an intelligent computing center, characterized in that, The method includes: Obtain energy consumption hotspot data and energy flow data of the intelligent computing center energy consumption system; Based on the energy flow data, the energy flow path is mapped to an energy distribution network that includes multiple regions, generating a set of regions corresponding to the energy flow path; Extract key energy consumption area data that are respectively related to each energy consumption hotspot from the energy consumption hotspot data; Based on the key energy consumption area data, each energy consumption hotspot is mapped to the energy consumption distribution network, generating association information between the areas and energy consumption hotspots in the energy consumption distribution network; Based on the set of regions corresponding to the energy flow path and the associated information, a set of energy consumption hotspots associated with the energy flow path is determined, and energy consumption monitoring data analysis of the intelligent computing center energy consumption system is performed based on the set of energy consumption hotspots associated with the energy flow path. The extraction of key energy consumption area data related to each energy consumption hotspot from the energy consumption hotspot data includes: Extract energy consumption area data for each energy consumption hotspot from the energy consumption hotspot data; Based on the flow feature vectors between data from different energy consumption areas, the energy consumption area data is selected to generate key energy consumption area data; The step of selecting energy consumption region data based on the flow feature vectors between data from different energy consumption regions to generate key energy consumption region data includes: Traverse any energy consumption region data and use the traversed energy consumption region data as the baseline energy consumption region data; Other energy consumption region data within the target interval of the flow feature vector related to the benchmark energy consumption region data will be used as the linked energy consumption region data of the benchmark energy consumption region data. The baseline energy consumption area data and the linked energy consumption area data are grouped into a data cluster; Each data cluster is traversed, and key energy consumption area data is generated based on the flow rate corresponding to each energy consumption area data in the traversed data cluster. The flow rate corresponding to the energy consumption area data is determined by the energy consumption flow diagram composed of the energy consumption area data. Specifically, the process of traversing each data cluster and selecting key energy consumption area data based on the flow rate corresponding to each energy consumption area data in the traversed data cluster includes: Traverse any data cluster and determine the shared flow between every two energy consumption regions in the traversed data cluster; If the shared flow meets the flow selection criteria, then the energy consumption area data with the larger flow is retained to generate key energy consumption area data.

2. The data analysis method for an energy consumption system in an intelligent computing center according to claim 1, characterized in that, The energy flow data includes multiple energy flow nodes; the step of mapping the energy flow path to an energy distribution network comprising multiple regions based on the energy flow data, and generating a region set corresponding to the energy flow path, includes: Identify the energy flow nodes in the energy flow data that correspond to the energy flow path, and generate an energy flow node diagram of the energy flow path. The energy flow node graph of the energy flow path is mapped to an energy distribution network that includes multiple regions, thereby generating a set of regions corresponding to the energy flow path.

3. The data analysis method applied to the energy consumption system of an intelligent computing center according to claim 2, characterized in that, The step of mapping the energy flow node graph of the energy flow path to an energy distribution network comprising multiple regions, and generating a set of regions corresponding to the energy flow path, includes: Walk through each energy flow node in the energy flow node graph along the energy flow path, and determine the node attribute information of the energy flow node that is walked to. Determine the region attribute information associated with the node attribute information; If the regional attribute information does not exist in the regional sequence of the energy consumption flow path, then the regional attribute information is loaded into the regional sequence; Continue traversing the next energy flow node, return to the step of determining the node attribute information of the energy flow node reached, and continue execution until all energy flow nodes in the energy flow node graph have been traversed; Based on the final generated region sequence, the set of regions corresponding to the energy consumption flow path is determined.

4. The data analysis method for an energy consumption system in an intelligent computing center according to claim 1, characterized in that, The process of mapping each energy consumption hotspot to the energy consumption distribution network based on the key energy consumption area data, and generating association information between areas and energy consumption hotspots in the energy consumption distribution network, includes: Extract energy consumption boundary information from the key energy consumption area data, and determine the minimum coverage block of the key energy consumption area data based on the energy consumption boundary information; Extract each target energy consumption domain within the minimum coverage block from the energy consumption distribution network; The system navigates through each target energy consumption domain. When the energy consumption focus position of the target energy consumption domain is within the energy consumption boundary information of the key energy consumption area data, the system obtains the regional attribute information of the target energy consumption domain. The acquired regional attribute information is mapped to the energy consumption hotspots corresponding to the key energy consumption regional data, and the association information between regions and energy consumption hotspots in the energy consumption distribution network is generated. The step of extracting energy consumption boundary information from the key energy consumption area data and determining the minimum coverage block of the key energy consumption area data based on the energy consumption boundary information includes: Analyze the data of the key energy consumption areas to identify the boundary points or boundary lines of each energy consumption area, which serve as the energy consumption limit information. Based on the extracted energy consumption limit information, the minimum coverage block that can completely contain the corresponding limit is calculated. The minimum coverage block provides the minimum coverage range of the energy consumption area in two-dimensional space.

5. The data analysis method for an energy consumption system in an intelligent computing center according to claim 1, characterized in that, The step of determining the set of energy consumption hotspots associated with the energy consumption flow path based on the set of regions corresponding to the energy consumption flow path and the associated information includes: Based on the association information, determine the energy consumption hotspots associated with each region in the set of regions corresponding to the energy consumption flow path; Based on time-series selection conditions or geographical distribution selection conditions, the energy consumption hotspots are selected to generate target energy consumption hotspots; Based on the selected target energy consumption hotspots, a set of energy consumption hotspots associated with the energy flow path is generated.

6. The data analysis method for an energy consumption system in an intelligent computing center according to claim 5, characterized in that, The step of selecting energy consumption hotspots based on temporal or geographical selection criteria to generate target energy consumption hotspots includes: Obtain the continuous parameters corresponding to each energy consumption hotspot; Remove energy consumption hotspots whose continuous parameters are less than the preset continuous parameters, and generate the first energy consumption hotspot; Obtain the geographical distribution characteristics of the first energy consumption hotspot; Based on the geographical distribution characteristics of the primary energy consumption hotspots, the data are cleaned to generate target energy consumption hotspots.

7. The data analysis method for an energy consumption system in an intelligent computing center according to claim 1, characterized in that, The step of analyzing the energy consumption monitoring data of the intelligent computing center energy consumption system based on the set of energy consumption hotspots associated with the energy consumption flow path includes: The set of energy consumption hotspots associated with energy consumption flow paths is integrated with the original energy consumption monitoring data. For each energy consumption hotspot in the integrated data, a time-series prediction model is constructed based on the energy consumption monitoring data of the energy consumption hotspot. The time-series prediction model is used to predict future energy consumption trends. Furthermore, clustering algorithms are used to perform clustering analysis on the energy consumption monitoring data of the energy consumption hotspots, different energy consumption patterns are identified, and the characteristic patterns of different energy consumption patterns are analyzed. Based on the characteristic patterns of different energy consumption patterns of the energy consumption hotspots, association rule mining algorithms are used to analyze the correlation between the energy consumption hotspots. Abnormal energy consumption data is detected based on the correlation between the energy consumption hotspots, and corresponding energy consumption anomaly events are generated based on the abnormal energy consumption data. The step of constructing a time-series prediction model based on energy consumption monitoring data of the energy consumption hotspots, wherein the time-series prediction model is used to predict future energy consumption trends, includes: For each energy consumption hotspot, capture the dynamic characteristics of time-series data from the energy consumption monitoring data; The corresponding time series prediction model is selected based on the dynamic characteristics of the time series data, and a training set and a validation set are divided according to the dynamic characteristics of the time series data and the corresponding expert knowledge annotation data. The selected time series prediction model is trained using the training set, and the prediction performance of the time series prediction model is evaluated using the validation set. The accuracy of the time series prediction model is evaluated by comparing the difference between the predicted value and the actual value. When the accuracy of the time series prediction model meets the expected conditions, the final time series prediction model is generated. The step of detecting abnormal energy consumption data based on the correlation between energy consumption hotspots and generating corresponding energy consumption anomaly events based on the abnormal energy consumption data includes: Based on the correlation between energy consumption hotspots, a corresponding correlation network is constructed, where nodes in the correlation network represent energy consumption hotspots and edges represent the relationships between these hotspots. The energy consumption monitoring data of the energy consumption hotspots are monitored in real time. The correlation network and predefined anomaly detection rules are used to detect anomalies in the energy consumption monitoring data. When abnormal energy consumption data is detected, an energy consumption anomaly event is generated.

8. A data analysis system applied to the energy consumption system of an intelligent computing center, characterized in that, The data analysis system applied to the energy consumption system of the intelligent computing center includes a processor and a memory, the memory and the processor are connected, the memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to implement the data analysis method applied to the energy consumption system of the intelligent computing center as described in any one of claims 1-7.

Images