Docking station power consumption self-adaptive adjusting method and device based on edge calculation

Abstract

The invention discloses a docking station power consumption adaptive adjustment method and device based on edge computing, and relates to the technical field of edge computing, and the method comprises the steps: driving an edge node to monitor each port, and obtaining power consumption time sequence data of a connection device; performing complementary analysis on the time series data, and clustering and dividing complementary connection equipment groups; generating a resource compensation priority map by using the time sequence data; performing two-dimensional demand prediction on the equipment group, and outputting a prediction value; dynamically executing resource allocation adjustment according to a predicted value by taking the priority map as a constraint; and realizing closed-loop feedback optimization according to prediction time sequence updating. According to the invention, the technical problems of resource allocation conflict, poor stability and energy efficiency waste caused by imbalance of bandwidth and power consumption demand time between devices of the docking station in a multi-device high-load scene are solved, and complementary scheduling and dynamic priority compensation of the device group are realized through edge calculation; the technical effects of improving the multi-device cooperation stability, optimizing the bandwidth and power consumption resource cooperation distribution efficiency and reducing the overall energy consumption are achieved.

Description

Technical Field

[0001] This application relates to the field of edge computing technology, specifically to a method and apparatus for adaptive power adjustment of docking stations based on edge computing. Background Technology

[0002] With the rapid development of IoT technology and the widespread adoption of multi-device collaboration scenarios, docking stations, as core hubs connecting various terminals such as laptops, mobile phones, tablets, external monitors, and storage devices, have evolved from simple interface expansion to key platforms for multi-device collaboration. However, in high-load scenarios (such as simultaneously connecting multiple high-definition monitors, high-speed storage devices, and peripherals), docking stations often face the challenge of uneven resource demands among devices: some devices require high bandwidth (such as video transmission) or high power consumption (such as fast charging) at specific times, while other devices are under low load. Traditional docking stations, lacking intelligent adjustment mechanisms, often adopt fixed resource allocation strategies, leading to problems such as bandwidth congestion, power waste, or device overheating. System stability and energy efficiency urgently need optimization. In existing technologies, docking station resource management largely relies on hardware presets or simple threshold control, failing to dynamically perceive changes in actual device demands. For example, when multiple devices are transmitting data simultaneously, bandwidth resources may be monopolized by a few high-demand devices, causing other devices to lag due to insufficient resources; and if power allocation is not adjusted according to the real-time status of the devices, it may cause the overall power consumption of the docking station to exceed the limit, affecting heat dissipation or shortening device battery life. In addition, traditional solutions typically handle bandwidth and power consumption in isolation, lacking synergistic optimization between the two, making it difficult to reduce energy consumption while ensuring device performance. Summary of the Invention

[0003] This application provides a method and apparatus for adaptive power consumption adjustment of docking stations based on edge computing, which solves the technical problems of resource allocation conflicts, poor stability and energy waste caused by the unbalanced time of bandwidth and power consumption requirements between devices in high-load scenarios with multiple devices.

[0004] This application provides a power consumption adaptive adjustment method for an expansion dock based on edge computing. The method includes: driving an edge power control node to perform multi-dimensional data monitoring on multiple physical ports of the expansion dock to obtain multiple sets of power consumption-related time-series data of multiple connected devices connected to the multiple physical ports; performing resource scheduling complementarity analysis on the multiple sets of power consumption-related time-series data to cluster the multiple connected devices into N complementary connected device groups; using the multiple sets of power consumption-related time-series data to perform resource compensation priority analysis on the N complementary connected device groups to obtain N resource compensation priority maps; performing two-dimensional resource demand prediction on the N complementary connected device groups to output N predicted bandwidth demand groups and N predicted power consumption demand groups; using the N resource compensation priority maps as compensation priority constraints, performing intra-group resource allocation adjustment on the N complementary connected device groups according to the N predicted bandwidth demand groups and N predicted power consumption demand groups; and performing closed-loop feedback optimization on the N complementary connected device groups based on the time-series intermittent updates of the two-dimensional resource demand prediction.

[0005] This application also provides an edge computing-based docking station power consumption adaptive adjustment device, comprising: a data monitoring module: driving an edge power control node to perform multi-dimensional data monitoring on multiple physical ports of the docking station, obtaining multiple sets of power consumption-related time-series data of multiple connected devices connected to the multiple physical ports; a complementary analysis module: performing resource scheduling complementarity analysis on the multiple sets of power consumption-related time-series data, clustering the multiple connected devices into N complementary connected device groups; a compensation analysis module: using the multiple sets of power consumption-related time-series data to perform resource compensation priority analysis on the N complementary connected device groups, obtaining N resource compensation priority maps; a demand prediction module: performing two-dimensional resource demand prediction on the N complementary connected device groups, outputting N predicted bandwidth demand groups and N predicted power consumption demand groups; a resource adjustment module: using the N resource compensation priority maps as compensation priority constraints, performing intra-group resource allocation adjustment on the N complementary connected device groups according to the N predicted bandwidth demand groups and N predicted power consumption demand groups; and a closed-loop optimization module: performing closed-loop feedback optimization on the N complementary connected device groups based on the time-series intermittent updates of the two-dimensional resource demand prediction.

[0006] The proposed edge computing-based adaptive power consumption adjustment method and apparatus for expansion docks involves the following steps: First, an edge power control node is driven to perform multi-dimensional data monitoring on multiple physical ports of the expansion dock, obtaining multiple sets of power consumption-related time-series data for multiple connected devices connected to these physical ports. Then, resource scheduling complementarity analysis is performed on these multiple sets of power consumption-related time-series data, clustering the multiple connected devices into N complementary connected device groups. Next, resource compensation priority analysis is performed on the N complementary connected device groups using the multiple sets of power consumption-related time-series data, resulting in N resource compensation priority maps. Further, two-dimensional resource demand prediction is performed on the N complementary connected device groups, outputting N predicted bandwidth demand groups and N predicted power consumption demand groups. Then, the N resource compensation priority maps are used as compensation priority constraints, and intra-group resource allocation adjustment is performed on the N complementary connected device groups based on the N predicted bandwidth demand groups and N predicted power consumption demand groups. Finally, closed-loop feedback optimization is performed on the N complementary connected device groups based on the time-series intermittent updates of the two-dimensional resource demand predictions. It solves the technical problems of resource allocation conflicts, poor stability and energy waste caused by the uneven time of bandwidth and power consumption demand between devices in the docking station under high load scenarios with multiple devices. It achieves the technical effect of improving the stability of multi-device collaboration, optimizing the collaborative allocation efficiency of bandwidth and power consumption resources, and reducing overall energy consumption by realizing complementary scheduling and dynamic priority compensation of device groups through edge computing. Attached Figure Description

[0007] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the apparatus according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0008] Figure 1 This is a schematic diagram of the edge computing-based adaptive power consumption adjustment method for docking stations provided in an embodiment of this application.

[0009] Figure 2 A schematic diagram of the power consumption adaptive adjustment device for a docking station based on edge computing provided in this application embodiment.

[0010] Figure labeling: Data monitoring module 11, complementary analysis module 12, compensation analysis module 13, demand forecasting module 14, resource adjustment module 15, closed-loop optimization module 16. Detailed Implementation

[0011] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application.

[0012] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0013] In the following description, references to "some embodiments" describe a subset of all possible embodiments. However, it is understood that "some embodiments" can be the same or different subsets of all possible embodiments and can be combined with each other without conflict. The terms "first" and "second" are used merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or apparatuses. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.

[0014] This application provides an edge computing-based method for adaptive power consumption adjustment of docking stations, such as... Figure 1 As shown, the method includes: The driving edge power control node performs multi-dimensional data monitoring on multiple physical ports of the expansion dock to obtain multiple sets of power consumption correlation timing data of multiple connected devices connected to the multiple physical ports.

[0015] In this embodiment, the edge power control node monitors multiple physical ports of the expansion dock from multiple dimensions to obtain real-time power consumption and resource requirements of each connected device. These physical ports typically connect to various devices, such as laptops, external hard drives, monitors, and smart devices. When connected to the expansion dock, their respective power consumption, bandwidth requirements, and other resource usage fluctuate with changes in device usage status. To accurately understand the resource requirements of these devices and dynamically adjust them according to actual conditions, the edge power control node is driven to comprehensively monitor these devices, acquiring real-time power consumption, bandwidth utilization, temperature changes, and other data from multiple dimensions. This data is stored to form multiple sets of power consumption-related time-series data, each set corresponding to a physical port and a connected device. This data provides a reliable data source for subsequent complementarity analysis, resource compensation priority calculation, and dynamic resource allocation, ensuring intelligent decision-making based on actual device needs under high-load scenarios.

[0016] Table 1: Example Table of Power Consumption Correlation Timing Data ;

[0017] As shown above, Table 1 is an example table of power consumption-related timing data. This table displays the real-time power consumption, bandwidth utilization, and temperature data of device 1, device 2, and device 3 at multiple time points, which is used for subsequent power consumption-related timing data analysis to help with resource scheduling and optimization.

[0018] By performing resource scheduling complementarity analysis on the multiple sets of power consumption-related timing data, the multiple connected devices are clustered into N complementary connected device groups.

[0019] In one embodiment, after obtaining power consumption correlation timing data of multiple connected devices, the edge power control node further performs resource scheduling complementarity analysis on this data. This resource scheduling complementarity analysis refers to analyzing the changing trends of power consumption, bandwidth, and other resource requirements of each device at different time points to identify which devices have complementary relationships in resource requirements. That is, when the power consumption requirements of some devices are high, the power consumption requirements of other devices are low, and vice versa. For example, a laptop may consume a lot of power when working, while another device (such as an external monitor) consumes less power at the same time. Through this complementarity analysis, devices with different resource requirement characteristics can be grouped together, which allows for more reasonable resource allocation, avoids resource waste, and ensures the overall stable operation of the docking station when multiple devices are connected. After the analysis is completed, the edge power control node divides these devices into N complementary connected device groups according to their resource complementarity. The devices in each group exhibit complementary characteristics in terms of resource requirement time, enabling more effective resource scheduling and optimization. By clustering connected devices in the above manner, not only can the resource utilization efficiency of the docking station be improved, but also the stable operation of each device can be ensured under high load and high demand conditions.

[0020] Furthermore, this application provides a method for clustering the multiple connected devices into N complementary connected device groups by performing resource scheduling complementarity analysis on the multiple sets of power consumption-related timing data. The method includes: The multiple sets of power consumption-related timing data are decomposed to obtain multiple timing power consumption data, multiple timing temperature data, and multiple timing bandwidth utilization rates of the multiple connected devices. After parsing and obtaining multiple device descriptors of the multiple connected devices, the multiple device descriptors are used as retrieval features to load multiple typical power consumption pattern curves from the cloud database. The multiple timing power consumption data are calibrated according to the multiple typical power consumption pattern curves to obtain multiple timing correction data. Based on the multiple timing correction data, multiple timing temperature data, and multiple timing bandwidth utilization rates, resource scheduling complementarity analysis is performed to cluster the multiple connected devices into the N complementary connected device groups.

[0021] Preferably, after the edge power control node collects multiple sets of power-related timing data, it divides this data according to key names to obtain multiple timing power consumption data, multiple timing temperature data, and multiple timing bandwidth utilization rates for multiple connected devices. Each timing power consumption data, timing temperature data, and timing bandwidth utilization rate corresponds one-to-one with a connected device. After extracting the timing data for each connected device, the edge power control node parses the device descriptors of these devices. Device descriptors typically include the device ID, device type, device model, and device hardware configuration. Subsequently, the edge power control node uses this information as retrieval features to load corresponding typical power consumption pattern curves from the cloud database. These typical power consumption pattern curves are reference data extracted from historical data or technical documents provided by the manufacturer, based on the device type and configuration. Next, the edge power control node calibrates multiple timing power data based on these power mode curves. During calibration, a dynamic time warping algorithm is used to align the time axis of the measured timing power consumption with that of the reference curve. Then, the ratio of the timing power data and typical power data within the sliding window at the same moment is calculated. The calculated ratio is then averaged to obtain a correction coefficient. By dividing the timing power data at each moment by the correction coefficient, the corrected data for each moment is obtained. These corrected data are arranged in time sequence to form multiple timing corrected data sets. Then, the edge power control node uses the calibrated timing corrected data, timing temperature data, and timing bandwidth utilization data to perform resource scheduling complementarity analysis. By calculating a complementarity score, it identifies the complementary relationships between devices in terms of resource demand. Finally, based on the results of the complementarity analysis, the edge power control node divides multiple connected devices into N complementary connected device groups according to their complementarity. Devices in each group exhibit complementarity in resource demand; that is, when some devices have higher resource demands, other devices have lower demands, maximizing resource utilization and avoiding overconsumption, providing data support for subsequent resource scheduling and optimization.

[0022] Furthermore, this application provides a resource scheduling complementarity analysis based on the multiple time-series correction data, multiple time-series temperature data, and multiple time-series bandwidth utilization rates, clustering the multiple connected devices into the N complementary connected device groups. The method includes: The multiple connected devices are enumerated and arranged to obtain multiple sets of connected devices. Based on the first set of connected devices, a first set of time-series corrected data, a first set of time-series temperature data, and a first set of time-series bandwidth utilization are extracted and combined from the multiple time-series corrected data, multiple time-series temperature data, and multiple time-series bandwidth utilization data to output a first set of time-series corrected data, a first set of time-series temperature data, and a first set of time-series bandwidth utilization. The Pearson correlation coefficient is calculated on the first set of time-series corrected data, the first set of time-series temperature data, and the first set of time-series bandwidth utilization to obtain a first complementarity score. The multiple complementarity scores of the multiple sets of connected devices are calculated by analogy. Based on the multiple complementarity scores, the multiple connected devices are associated and aggregated to output the N sets of complementary connected devices.

[0023] Optionally, the edge power control node first enumerates the permutations and combinations of multiple connected devices, generating all possible device combinations. Each device combination represents a specific set of devices, containing multiple devices. The purpose of these permutations and combinations is to analyze the resource requirements of different devices under different combinations, such as power consumption, bandwidth, and temperature, thereby identifying the complementarity and mutual influence relationships between devices. For example, if the docking station connects 5 devices, the edge power control node will generate all possible combinations of the 5 devices. For instance, combination 1 includes device 1 and device 2, combination 2 includes device 2 and device 3, combination 3 includes device 1, device 2, and device 3, etc. These device combinations will be further analyzed for their resource requirement characteristics in subsequent steps. Subsequently, for the first group of connected devices, the edge power control node maps and extracts the corresponding time-series data from multiple timing correction data, multiple timing temperature data, and multiple timing bandwidth utilization extraction combinations, forming a first timing correction data group, a first timing temperature data group, and a first timing bandwidth utilization group. Each data group includes the timing data of all devices in that combination. Next, the edge power control node activates the Pearson correlation coefficient calculation function and uses it to perform correlation analysis on the first time-series corrected data group, the first time-series temperature data group, and the first time-series bandwidth utilization group. It calculates the Pearson correlation coefficient between each pair of data groups. The Pearson correlation coefficient is a statistic that measures the strength of the linear relationship between two variables, with a value range between -1 and 1, where 1 indicates a perfect positive correlation, -1 indicates a perfect negative correlation, and 0 indicates no linear relationship. The first complementarity score is obtained by averaging the absolute values ​​of the three calculated Pearson correlation coefficients. Similarly, the edge power control node also performs Pearson correlation coefficient calculations on other groups of connected devices to obtain a complementarity score for each combination. A higher complementarity score indicates stronger complementarity between the resource requirements of the devices within that combination, meaning that the power consumption and bandwidth requirements between devices can be better coordinated, reducing conflicts or excessive consumption. After the complementarity scores of all device combinations are calculated, the edge power control node will associate and aggregate the devices based on these complementarity scores. In this way, the edge power control node will eventually divide multiple connected devices into N complementary connected device groups. The devices in each device group show high complementarity in resource requirements, which can efficiently allocate resources in the subsequent resource scheduling process, ensuring that the power consumption, bandwidth and other resources of all devices in the expansion dock are used rationally.

[0024] Furthermore, this application provides a method for associating and aggregating the multiple connectivity devices based on the multiple complementarity scores, and outputting the N complementary connectivity device groups, the method comprising: The multiple connected devices are used as multiple topology nodes, and the multiple complementarity scores are used to quantize the topology connections to construct a device complementarity quantization topology; the topology connections that do not meet the preset quantization length in the device complementarity quantization topology are removed, and the device complementarity quantization topology is divided into N device complementarity sub-topologies; the device composition of the N device complementarity sub-topologies is extracted, and the N complementary connected device groups are output.

[0025] Optionally, multiple connected devices can be treated as nodes in the topology, and complementarity scores can be used as the topological connections between these devices to construct a device complementarity quantization topology. After generating the device complementarity quantization topology, the edge power control node filters the topology, removing connections with complementarity scores less than a preset quantization length. This eliminates device combinations with weak resource complementarity, preventing resource allocation to these combinations and improving resource utilization efficiency. After removing unnecessary connections, the device complementarity quantization topology is divided into several device complementarity sub-topologies. Each sub-topology represents a group of device nodes with strong resource complementarity. The devices in these sub-topologies have high complementarity, enabling more refined and efficient scheduling in subsequent resource allocation. Finally, the edge power control node extracts the internal device composition from each device complementarity sub-topology, i.e., the specific set of devices in that sub-topology, resulting in N complementary connected device groups. The devices in these groups not only have complementary resource requirements but can also be optimized for resource scheduling and management by the edge power control node, providing a basis for subsequent power and bandwidth scheduling.

[0026] The resource compensation priority analysis of the N complementary connected device groups is performed using the multiple sets of power consumption correlation timing data to obtain N resource compensation priority maps.

[0027] In one embodiment, after obtaining N complementary connected device groups, the edge power control node performs resource compensation priority analysis on the complementary sub-topology corresponding to each complementary connected device group based on the previously acquired multiple sets of power-related timing data. This determines which devices in each sub-topology should receive resource allocation first, thereby generating N resource compensation priority maps. Each resource compensation priority map corresponds to a complementary connected device group, providing a clear basis for subsequent resource scheduling and ensuring that the docking station's resources can be efficiently allocated to the most needed device groups under high load conditions.

[0028] Furthermore, this application provides a method for performing resource compensation priority analysis on the N complementary interconnected device groups using the aforementioned multiple sets of power consumption correlation timing data to obtain N resource compensation priority maps, the method comprising: Based on the multiple complementarity scores, calculate the sum of P complementarity scores for P topology nodes in the first device complementary sub-topology; perform directed priority connections for the P topology nodes based on the sum of the P complementarity scores, and output the first resource compensation priority graph; based on the multiple complementarity scores, perform directed priority connections for N-1 device complementary sub-topologies by analogy, and obtain N-1 resource compensation priority graphs.

[0029] Optionally, during resource compensation priority analysis, multiple complementarity scores calculated based on multiple sets of power consumption-related time-series data are obtained. Then, a device complementary sub-topology corresponding to one of the N complementary connected device groups is randomly selected as the first device complementary sub-topology. Subsequently, based on the P topology nodes in the first device complementary sub-topology, P complementarity scores for the P topology nodes are extracted from the multiple complementarity scores. The sum of the complementarity scores for each topology node is then calculated; that is, all complementarity scores for each topology node are added together. Afterward, the sums of the complementarity scores of adjacent topology nodes are compared. If the sum of the complementarity scores of topology node a is less than the sum of the complementarity scores of adjacent topology node b, the topology connection between topology node a and topology node b is modified to a directed topology connection from topology node b to topology node a. Through this process, the first device complementary sub-topology can be converted into a first resource compensation priority graph. This first resource compensation priority graph clearly shows the resource compensation priority between devices, aiding subsequent resource scheduling decisions. Similarly, for the other N-1 complementary sub-topologies, the same directed priority connections will be made, thereby converting the N-1 complementary sub-topologies into N-1 resource compensation priority maps. These resource compensation priority maps provide resource priority ranking for each device group, helping to rationally allocate resources such as power consumption and bandwidth during actual operation, so as to ensure the stable operation and efficient collaboration of multiple devices in the expansion dock.

[0030] Perform two-dimensional resource demand prediction on the N complementary connection device groups, and output N predicted bandwidth demand groups and N predicted power consumption demand groups.

[0031] In one embodiment, after identifying N complementary connected device groups, a pre-trained two-dimensional resource demand prediction model is used to analyze the timing correction data, timing temperature data, and timing bandwidth utilization of each complementary connected device group. The two-dimensional resource demand prediction model can predict N predicted bandwidth demand groups for the N complementary connected device groups through its internal bandwidth demand prediction sub-model, and N predicted power demand groups for the N complementary connected device groups through its internal power consumption demand prediction sub-model. Each predicted bandwidth demand group corresponds to a predicted bandwidth demand value for a complementary connected device group, representing the group's bandwidth demand over a future time period. Each predicted power consumption demand group corresponds to a predicted power consumption demand value for a complementary connected device group, representing the group's power consumption demand over a future time period. This two-dimensional resource demand prediction enables more intelligent management and allocation of resources in the expansion dock, ensuring efficient and stable resource utilization even under high load and simultaneous connection of multiple devices.

[0032] Furthermore, the method also includes: Separate the second timing correction data, second timing temperature data, and second timing bandwidth utilization rate of the second connected device from the multiple timing correction data, multiple timing temperature data, and multiple timing bandwidth utilization rates, wherein the second connected device belongs to the second complementary connected device group; based on the aligned timestamps of the second timing correction data, second timing temperature data, and second timing bandwidth utilization rate, the second timing power consumption record is locally retrieved; based on the aligned timestamp of the second timing bandwidth utilization rate, the second connected device is used as a search taboo to retrieve the second complementary timing power consumption group and the second cooperative timing feature vector group of the second complementary connected device group; the second timing temperature data, the second cooperative timing feature vector group, and the second timing power consumption record are used as training data to construct a power demand prediction sub-model; the second timing bandwidth utilization rate and the second cooperative timing feature vector group are used as training data to construct a bandwidth demand prediction sub-model; the power demand prediction sub-model and the bandwidth demand prediction sub-model are connected using a parallel architecture to obtain a second two-dimensional resource demand prediction model.

[0033] Optionally, firstly, from multiple time-series correction data, multiple time-series temperature data, and multiple time-series bandwidth utilization rates, the second time-series correction data, second time-series temperature data, and second time-series bandwidth utilization rate of the second connected device are separated in the same manner as described above. Then, these time-series data are timestamped to ensure that all datasets (power consumption, temperature, bandwidth utilization rate) have corresponding records at the same time point. After aligning the timestamps, the edge power control node locally retrieves the second time-series power consumption records based on the timestamps. These records include the power consumption change data of the device during that time period. Subsequently, the second connected device is used as a retrieval taboo, and based on the timestamp aligned with the second time-series bandwidth utilization rate, complementary device data is retrieved from the database to form a second complementary time-series power consumption group and a second collaborative time-series feature vector group. The second complementary time-series power consumption group reflects the power consumption time-series data of other devices within the device group, while the second collaborative time-series feature vector group contains feature vector data of other devices working collaboratively with this device, reflecting the collaborative relationship between devices, such as their collaborative patterns of bandwidth and power consumption requirements during a specific time period. Subsequently, using the retrieved data, the second time-series temperature data, the second collaborative time-series feature vector set, and the second time-series power consumption records were used as training data. A Long Short-Term Memory (LSTM) network was iteratively optimized through forward propagation, loss calculation (mean squared error), backpropagation, and parameter optimization (Adam optimizer) to construct a power consumption demand prediction sub-model. This model will learn how to predict the future power consumption demand of a device based on its historical power consumption, temperature, and data from other collaborative devices. Similarly, using the second time-series bandwidth utilization rate and the second collaborative time-series feature vector set as training data, a bandwidth demand prediction sub-model was constructed in the same manner. This model will learn how to predict the future bandwidth demand of a second connected device based on bandwidth utilization rate and the characteristics of collaborative devices. Finally, a parallel architecture was used to connect the power consumption demand prediction sub-model and the bandwidth demand prediction sub-model. This parallel architecture allows for simultaneous prediction of both power consumption and bandwidth resource demands, and the prediction results are integrated to form a two-dimensional resource demand prediction model. This parallel architecture allows edge power control nodes to simultaneously predict the power consumption and bandwidth requirements of a device at a single point in time, providing a comprehensive view of resource requirements. This enables the system to perform resource scheduling and optimization more efficiently, avoiding conflicts between bandwidth and power consumption requirements.

[0034] Furthermore, this application provides that both the power consumption demand prediction sub-model and the bandwidth demand prediction sub-model include long short-term memory networks.

[0035] Optionally, both the power consumption demand prediction sub-model and the bandwidth demand prediction sub-model are built on Long Short-Term Memory (LSTM) networks. LSTM is a variant of recurrent neural networks specifically designed for processing and predicting time-series data. Unlike traditional neural networks, LSTM can effectively capture long-term dependencies, meaning it can remember and use information from longer time intervals in the input data. LSTM is particularly suitable for power consumption and bandwidth demand prediction because these demands are usually time-series in nature, and the resource demand at the previous moment often affects the demand at the next moment.

[0036] Furthermore, this application provides a method for performing two-dimensional resource demand prediction on the N complementary interconnection device groups, outputting N predicted bandwidth demand groups and N predicted power consumption demand groups, the method comprising: The edge power control node is used to monitor the second physical port in real time to obtain short-term temperature data slices and short-term power data slices. Using the second complementary connection device group as a constraint, a short-term collaborative feature vector is extracted from the edge power control node. The short-term temperature data slices, short-term power data slices, and short-term collaborative feature vectors are input into the power demand prediction sub-model to obtain the second predicted power demand. Similarly, short-term data slices are collected, and the bandwidth demand prediction sub-model is driven to perform data prediction, outputting the second predicted bandwidth demand. In the second two-dimensional resource demand prediction model, the power demand prediction sub-model and the bandwidth demand prediction sub-model are executed in parallel for two-dimensional resource demand prediction. After obtaining multiple predicted power demands and multiple predicted bandwidth demands for the multiple connection devices, data assembly is performed based on the N complementary connection device groups to output the N predicted bandwidth demand groups and N predicted power demand groups.

[0037] Optionally, the edge power control node first performs real-time data monitoring on the connected devices at the second physical port, acquiring short-term temperature and power consumption data slices. The short-term temperature data slice reflects temperature changes within a short time window, indicating thermal variations under different operating conditions. The short-term power consumption data slice describes power consumption within the same time window, indicating short-term power consumption. Then, using the second complementary connected device group as a constraint, the edge power control node extracts short-term collaborative feature vectors from existing device data. These feature vectors describe the synergistic effect of resource demands across multiple devices in the short term, including power consumption and bandwidth requirements. This helps understand the interactions between devices, thereby optimizing resource scheduling. Next, the short-term temperature, power consumption, and collaborative feature vectors are input into the previously constructed power demand prediction sub-model. This model uses this input data to predict a second predicted power demand based on LSTM, reflecting the power consumption trend of the devices in the following period. Similar to power demand prediction, the edge power control node acquires real-time bandwidth utilization data through short-term data slices and drives the bandwidth demand prediction sub-model to predict bandwidth demand, outputting a second predicted bandwidth demand. Typically, the power consumption demand prediction sub-model and bandwidth demand prediction sub-model of the second two-dimensional resource demand prediction model are executed in parallel. Through parallel execution, the edge power control node can simultaneously obtain the power consumption and bandwidth demand prediction results of multiple devices, thereby improving the overall resource scheduling efficiency. After completing the power consumption and bandwidth demand prediction of multiple connected devices, the edge power control node will perform data assembly based on the complementarity and synergy of the devices. That is, based on the prediction results of N complementary connected device groups, the power consumption and bandwidth demands of each device group are summarized to obtain N predicted bandwidth demand groups and N predicted power consumption demand groups. This prediction data will help the docking station optimize resource allocation under high load conditions, ensuring that each device group can obtain the required bandwidth and power resources.

[0038] Using the N resource compensation priority maps as compensation priority constraints, the resource allocation within the N complementary connection device groups is adjusted based on the N predicted bandwidth demand groups and the N predicted power consumption demand groups.

[0039] In one embodiment, after completing the resource compensation priority analysis for each device group and generating N resource compensation priority maps, the edge power control node uses these maps as compensation priority constraints to guide the resource allocation process. The resource compensation priority maps clearly define the priority relationships of different device groups in terms of power consumption and bandwidth requirements. Subsequently, based on the N predicted bandwidth requirement groups and N predicted power consumption requirement groups, the corresponding N initial bandwidth quotas and N initial power consumption quotas are calculated. Resource allocation within each group is adjusted according to these quotas to ensure that high-priority device groups can obtain the required resources first, while low-priority device groups are allocated resources more efficiently when resources are scarce, thereby achieving efficient resource utilization and stable operation of each device group.

[0040] Furthermore, this application provides a method for adjusting the intra-group resource allocation of the N complementary connectivity device groups based on the N predicted bandwidth demand groups and the N predicted power consumption demand groups, using the N resource compensation priority maps as compensation priority constraints. The method includes: Based on the N predicted bandwidth demand combinations, calculate and output N initial bandwidth quotas; using the N resource compensation priority maps as compensation priority constraints, perform hierarchical resource scheduling and allocation of the N initial bandwidth quotas for the N complementary connection device groups; similarly calculate N initial power consumption quotas, and then perform hierarchical resource scheduling and allocation for the N complementary connection device groups.

[0041] Optionally, based on N predicted bandwidth demand groups, the edge power control node sums the predicted bandwidth demands of each connected device in each predicted bandwidth demand group to obtain the total expected bandwidth of each complementary connected device group. Then, it divides the total expected bandwidth of each complementary connected device group by the sum of the total expected bandwidths of all complementary connected device groups, and multiplies the quotient by the total bandwidth capacity of the docking station and the redundancy factor (e.g., 1.2) to obtain the initial bandwidth quota for each complementary connected device group. Subsequently, for each complementary connected device group, a device priority queue is generated based on the resource compensation priority map. Then, hierarchical resource scheduling and allocation of the N initial bandwidth quotas are performed according to priority from high to low. After satisfying the needs of high-priority devices, the remaining quota is allocated to low-priority devices. Similar to the calculation of the initial bandwidth quota, N initial power consumption quotas are also calculated based on the N predicted power consumption demand groups, and then hierarchical resource scheduling and allocation of the N complementary connected device groups are performed using the same scheduling method. After bandwidth and power consumption resources are allocated, the edge power control node will perform closed-loop optimization. Based on the real-time feedback from the device group and the actual resource requirements, the allocation of resources will be further optimized, thereby improving resource utilization efficiency and device stability.

[0042] Based on the time-series intermittent updates of the two-dimensional resource demand forecast, closed-loop feedback optimization is performed on the N complementary connected device groups.

[0043] In one embodiment, based on the time-series intermittent updates of dual-dimensional resource demand predictions, edge computing nodes perform closed-loop feedback optimization according to the real-time power consumption and bandwidth requirements of device groups. Whenever new time-series data is received, the edge power control node updates the existing bandwidth and power demand predictions, thereby adjusting the resource allocation strategy in real time. This feedback mechanism means that the resource allocation for each complementary connected device group is not static, but dynamically adjusted according to the actual needs and usage status of the devices. Through this real-time closed-loop feedback optimization, the accuracy and efficiency of resource scheduling can be continuously improved, enabling each device group to obtain the most suitable resource allocation at any given time based on the latest predictions and demands.

[0044] In the above text, refer to Figure 1 This paper describes in detail a docking station power consumption adaptive adjustment method based on edge computing according to an embodiment of the present invention. Next, reference will be made to... Figure 2 This invention describes an edge computing-based docking station power consumption adaptive adjustment device according to an embodiment of the present invention.

[0045] The edge computing-based adaptive power consumption adjustment device for docking stations according to embodiments of the present invention solves the technical problems of resource allocation conflicts, poor stability, and energy waste caused by the unbalanced time of bandwidth and power consumption demands among devices in high-load scenarios with multiple devices. It achieves the technical effect of improving the collaborative stability of multiple devices, optimizing the collaborative allocation efficiency of bandwidth and power consumption resources, and reducing overall energy consumption by realizing complementary scheduling and dynamic priority compensation of device groups through edge computing. The edge computing-based adaptive power consumption adjustment device for docking stations includes: a data monitoring module 11, a complementary analysis module 12, a compensation analysis module 13, a demand prediction module 14, a resource adjustment module 15, and a closed-loop optimization module 16.

[0046] Data monitoring module 11: Drives the edge power consumption control node to perform multi-dimensional data monitoring on multiple physical ports of the expansion dock, obtaining multiple sets of power consumption-related time-series data of multiple connected devices connected to the multiple physical ports; Complementary analysis module 12: Performs resource scheduling complementarity analysis on the multiple sets of power consumption-related time-series data, and clusters the multiple connected devices into N complementary connected device groups; Compensation analysis module 13: Uses the multiple sets of power consumption-related time-series data to perform resource compensation priority analysis on the N complementary connected device groups, obtaining N resource compensation priority maps; Demand prediction module 14: Performs two-dimensional resource demand prediction on the N complementary connected device groups, outputting N predicted bandwidth demand groups and N predicted power consumption demand groups; Resource adjustment module 15: Uses the N resource compensation priority maps as compensation priority constraints, and performs intra-group resource allocation adjustment on the N complementary connected device groups according to the N predicted bandwidth demand groups and N predicted power consumption demand groups; Closed-loop optimization module 16: Performs closed-loop feedback optimization on the N complementary connected device groups based on the time-series intermittent updates of the two-dimensional resource demand prediction.

[0047] Furthermore, the complementary analysis module 12 also includes: The multiple sets of power consumption-related timing data are decomposed to obtain multiple timing power consumption data, multiple timing temperature data, and multiple timing bandwidth utilization rates of the multiple connected devices. After parsing and obtaining multiple device descriptors of the multiple connected devices, the multiple device descriptors are used as retrieval features to load multiple typical power consumption pattern curves from the cloud database. The multiple timing power consumption data are calibrated according to the multiple typical power consumption pattern curves to obtain multiple timing correction data. Based on the multiple timing correction data, multiple timing temperature data, and multiple timing bandwidth utilization rates, resource scheduling complementarity analysis is performed to cluster the multiple connected devices into the N complementary connected device groups.

[0048] Furthermore, the complementary analysis module 12 also includes: The multiple connected devices are enumerated and arranged to obtain multiple sets of connected devices. Based on the first set of connected devices, a first set of time-series corrected data, a first set of time-series temperature data, and a first set of time-series bandwidth utilization are extracted and combined from the multiple time-series corrected data, multiple time-series temperature data, and multiple time-series bandwidth utilization data to output a first set of time-series corrected data, a first set of time-series temperature data, and a first set of time-series bandwidth utilization. The Pearson correlation coefficient is calculated on the first set of time-series corrected data, the first set of time-series temperature data, and the first set of time-series bandwidth utilization to obtain a first complementarity score. The multiple complementarity scores of the multiple sets of connected devices are calculated by analogy. Based on the multiple complementarity scores, the multiple connected devices are associated and aggregated to output the N sets of complementary connected devices.

[0049] Furthermore, the complementary analysis module 12 also includes: The multiple connected devices are used as multiple topology nodes, and the multiple complementarity scores are used to quantize the topology connections to construct a device complementarity quantization topology; the topology connections that do not meet the preset quantization length in the device complementarity quantization topology are removed, and the device complementarity quantization topology is divided into N device complementarity sub-topologies; the device composition of the N device complementarity sub-topologies is extracted, and the N complementary connected device groups are output.

[0050] Furthermore, the compensation analysis module 13 also includes: Based on the multiple complementarity scores, calculate the sum of P complementarity scores for P topology nodes in the first device complementary sub-topology; perform directed priority connections for the P topology nodes based on the sum of the P complementarity scores, and output the first resource compensation priority graph; based on the multiple complementarity scores, perform directed priority connections for N-1 device complementary sub-topologies by analogy, and obtain N-1 resource compensation priority graphs.

[0051] Furthermore, the demand forecasting module 14 also includes: Separate the second timing correction data, second timing temperature data, and second timing bandwidth utilization rate of the second connected device from the multiple timing correction data, multiple timing temperature data, and multiple timing bandwidth utilization rates, wherein the second connected device belongs to the second complementary connected device group; based on the aligned timestamps of the second timing correction data, second timing temperature data, and second timing bandwidth utilization rate, the second timing power consumption record is locally retrieved; based on the aligned timestamp of the second timing bandwidth utilization rate, the second connected device is used as a search taboo to retrieve the second complementary timing power consumption group and the second cooperative timing feature vector group of the second complementary connected device group; the second timing temperature data, the second cooperative timing feature vector group, and the second timing power consumption record are used as training data to construct a power demand prediction sub-model; the second timing bandwidth utilization rate and the second cooperative timing feature vector group are used as training data to construct a bandwidth demand prediction sub-model; the power demand prediction sub-model and the bandwidth demand prediction sub-model are connected using a parallel architecture to obtain a second two-dimensional resource demand prediction model.

[0052] Furthermore, the demand forecasting module 14 also includes: Both the power consumption demand prediction sub-model and the bandwidth demand prediction sub-model include long short-term memory networks.

[0053] Furthermore, the demand forecasting module 14 also includes: The edge power control node is used to monitor the second physical port in real time to obtain short-term temperature data slices and short-term power data slices. Using the second complementary connection device group as a constraint, a short-term collaborative feature vector is extracted from the edge power control node. The short-term temperature data slices, short-term power data slices, and short-term collaborative feature vectors are input into the power demand prediction sub-model to obtain the second predicted power demand. Similarly, short-term data slices are collected, and the bandwidth demand prediction sub-model is driven to perform data prediction, outputting the second predicted bandwidth demand. In the second two-dimensional resource demand prediction model, the power demand prediction sub-model and the bandwidth demand prediction sub-model are executed in parallel for two-dimensional resource demand prediction. After obtaining multiple predicted power demands and multiple predicted bandwidth demands for the multiple connection devices, data assembly is performed based on the N complementary connection device groups to output the N predicted bandwidth demand groups and N predicted power demand groups.

[0054] Furthermore, the resource adjustment module 15 also includes: Based on the N predicted bandwidth demand combinations, calculate and output N initial bandwidth quotas; using the N resource compensation priority maps as compensation priority constraints, perform hierarchical resource scheduling and allocation of the N initial bandwidth quotas for the N complementary connection device groups; similarly calculate N initial power consumption quotas, and then perform hierarchical resource scheduling and allocation for the N complementary connection device groups.

[0055] The edge computing-based docking station power consumption adaptive adjustment device provided in this embodiment of the invention can execute the edge computing-based docking station power consumption adaptive adjustment method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0056] Although this application makes various references to certain modules in the apparatus according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not intended to limit the scope of protection of this invention.

[0057] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for adaptive power consumption adjustment of a docking station based on edge computing, characterized in that, The method includes: The driving edge power control node performs multi-dimensional data monitoring on multiple physical ports of the expansion dock to obtain multiple sets of power consumption correlation timing data of multiple connected devices connected to the multiple physical ports; By performing resource scheduling complementarity analysis on the multiple sets of power consumption-related timing data, the multiple connected devices are clustered into N complementary connected device groups. The resource compensation priority analysis of the N complementary connected device groups is performed using the multiple sets of power consumption correlation timing data to obtain N resource compensation priority maps; Perform two-dimensional resource demand prediction on the N complementary connection device groups, and output N predicted bandwidth demand groups and N predicted power consumption demand groups; The N resource compensation priority maps are used as compensation priority constraints, and the resource allocation within the N complementary connection device groups is adjusted according to the N predicted bandwidth demand groups and the N predicted power consumption demand groups. Based on the time-series intermittent updates of the two-dimensional resource demand forecast, closed-loop feedback optimization is performed on the N complementary connected device groups.

2. The edge computing-based docking station power consumption adaptive adjustment method as described in claim 1, characterized in that, By performing resource scheduling complementarity analysis on the multiple sets of power consumption-related time-series data, the multiple connected devices are clustered into N complementary connected device groups. The method includes: Decompose the multiple sets of power consumption associated timing data to obtain multiple timing power consumption data, multiple timing temperature data and multiple timing bandwidth utilization of the multiple connected devices; After parsing and obtaining multiple device descriptors of the multiple connected devices, the multiple device descriptors are used as retrieval features to load multiple typical power consumption mode curves from the cloud database; The multiple timing power consumption data are calibrated based on the multiple typical power consumption mode curves to obtain multiple timing correction data; Based on the multiple time-series correction data, multiple time-series temperature data, and multiple time-series bandwidth utilization, a resource scheduling complementarity analysis is performed, and the multiple connected devices are clustered into the N complementary connected device groups.

3. The edge computing-based docking station power consumption adaptive adjustment method as described in claim 2, characterized in that, Based on the multiple time-series corrected data, multiple time-series temperature data, and multiple time-series bandwidth utilization, a resource scheduling complementarity analysis is performed, and the multiple connected devices are clustered into the N complementary connected device groups. The method includes: The multiple connection devices are arranged and combined enumerated to obtain multiple sets of connection devices; Based on the first group of connected devices, the mapping extracts and combines the multiple time-series correction data, multiple time-series temperature data, and multiple time-series bandwidth utilization data to output the first time-series correction data group, the first time-series temperature data group, and the first time-series bandwidth utilization group. Pearson correlation coefficients are calculated for the first time-series corrected data set, the first time-series temperature data set, and the first time-series bandwidth utilization set to obtain the first complementarity score. By analogy, multiple complementarity scores of the multiple sets of connected devices are calculated; Based on the multiple complementarity scores, the multiple connected devices are associated and aggregated to output the N complementary connected device groups.

4. The edge computing-based docking station power consumption adaptive adjustment method as described in claim 3, characterized in that, Based on the multiple complementarity scores, the multiple connected devices are associated and aggregated to output the N complementary connected device groups. The method includes: The multiple connected devices are used as multiple topology nodes, and the multiple complementarity scoring methods are used to quantize the topology connections to construct a device complementarity quantization topology; Remove the topology connections in the device complementary quantization topology that do not meet the preset quantization length, and divide the device complementary quantization topology into N device complementary sub-topologies; Extract the device composition of the N complementary sub-topologies and output the N complementary connected device groups.

5. The edge computing-based docking station power consumption adaptive adjustment method as described in claim 4, characterized in that, The resource compensation priority analysis is performed on the N complementary interconnected device groups using the multiple sets of power consumption correlation timing data to obtain N resource compensation priority maps. The method includes: Based on the multiple complementarity scores, calculate the sum of P complementarity scores for P topological nodes in the first device complementary sub-topology; Based on the P complementarity scores and the directed priority connections of the P topology nodes, a first resource compensation priority graph is output. Based on the multiple complementarity scores, directed priority connections are made to N-1 complementary sub-topologies of devices to obtain N-1 resource compensation priority maps.

6. The edge computing-based docking station power consumption adaptive adjustment method as described in claim 3, characterized in that, The method further includes: The second timing correction data, second timing temperature data, and second timing bandwidth utilization of the second connection device are separated from the multiple timing correction data, multiple timing temperature data, and multiple timing bandwidth utilization, wherein the second connection device belongs to the second complementary connection device group; Based on the aligned timestamps of the second timing correction data, the second timing temperature data, and the second timing bandwidth utilization, the second timing power consumption record is locally retrieved; Based on the aligned timestamp of the second timing bandwidth utilization, the second connected device is used as a search taboo to retrieve the second complementary timing power consumption group and the second cooperative timing feature vector group of the second complementary connected device group; The second time-series temperature data, the second collaborative time-series feature vector group, and the second time-series power consumption record are used as training data to construct a power consumption demand prediction sub-model. Using the second time-series bandwidth utilization rate and the second collaborative time-series feature vector group as training data, a bandwidth demand prediction sub-model is constructed. By using a parallel architecture to connect the power consumption demand prediction sub-model and the bandwidth demand prediction sub-model, a second two-dimensional resource demand prediction model is obtained.

7. The edge computing-based docking station power consumption adaptive adjustment method as described in claim 6, characterized in that, Both the power consumption demand prediction sub-model and the bandwidth demand prediction sub-model include long short-term memory networks.

8. The edge computing-based docking station power consumption adaptive adjustment method as described in claim 6, characterized in that, The method involves performing two-dimensional resource demand prediction on the N complementary interconnection device groups, outputting N predicted bandwidth demand groups and N predicted power consumption demand groups, wherein the method includes: The edge power control node is interacted with to perform real-time data monitoring of the second physical port, and short-term temperature data slices and short-term power data slices are obtained. Using the second complementary connectivity device group as a constraint, a short-term cooperative feature vector is extracted from the edge power control node; The short-term temperature data slice, the short-term power consumption data slice, and the short-term collaborative feature vector are input into the power consumption demand prediction sub-model to obtain the second predicted power consumption demand. By analogy, short-term data slices are collected, and the bandwidth demand prediction sub-model is driven to perform data prediction and output a second predicted bandwidth demand. In the second two-dimensional resource demand prediction model, the power consumption demand prediction sub-model and the bandwidth demand prediction sub-model are executed in parallel to predict the two-dimensional resource demand. After obtaining multiple predicted power consumption requirements and multiple predicted bandwidth requirements of the multiple connected devices by analogy, data assembly is performed based on the N complementary connected device groups to output the N predicted bandwidth requirement groups and N predicted power consumption requirement groups.

9. The edge computing-based docking station power consumption adaptive adjustment method as described in claim 1, characterized in that, Using the N resource compensation priority maps as compensation priority constraints, and based on the N predicted bandwidth demand groups and N predicted power consumption demand groups, the method performs intra-group resource allocation adjustment on the N complementary connection device groups, the method includes: Based on the N predicted bandwidth demand combinations, calculate and output N initial bandwidth quotas; Using the N resource compensation priority maps as compensation priority constraints, hierarchical resource scheduling and allocation of the N initial bandwidth quotas are performed on the N complementary connection device groups; After calculating N initial power consumption quotas by analogy, hierarchical resource scheduling and allocation are performed on the N complementary connected device groups.

10. A docking station power consumption adaptive adjustment device based on edge computing, characterized in that, The device is used to implement the edge computing-based docking station power consumption adaptive adjustment method according to any one of claims 1-9, and the device comprises: Data monitoring module: Drives the edge power consumption control node to perform multi-dimensional data monitoring on multiple physical ports of the expansion dock, and obtains multiple sets of power consumption correlation timing data of multiple connected devices connected to the multiple physical ports; Complementary analysis module: By performing resource scheduling complementarity analysis on the multiple sets of power consumption-related timing data, the multiple connected devices are clustered into N complementary connected device groups; Compensation Analysis Module: The module uses the multiple sets of power consumption-related timing data to perform resource compensation priority analysis on the N complementary connected device groups, and obtains N resource compensation priority maps. Demand forecasting module: performs two-dimensional resource demand forecasting on the N complementary connection device groups, and outputs N predicted bandwidth demand groups and N predicted power consumption demand groups; Resource adjustment module: Using the N resource compensation priority maps as compensation priority constraints, the module performs intra-group resource allocation adjustment on the N complementary connection device groups based on the N predicted bandwidth demand groups and the N predicted power consumption demand groups; Closed-loop optimization module: Based on the time-series intermittent updates of the two-dimensional resource demand prediction, it performs closed-loop feedback optimization of the N complementary connected device groups.

Images