Calculation and power collaboration system and method
By building a prediction and intelligent matching of electricity and computing power time series diagrams, we can achieve active scheduling of future electricity supply and computing power demand, solve the problem of system instability in existing technologies, and improve the utilization rate of new energy electricity and the stability of computing power services.
Patent Information
- Application Number
- CN202511187120.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-25
- Publication Date
- 2025-09-23
- Estimated Expiration
- 2045-08-25
AI Technical Summary
Existing computing-electricity collaborative technologies lack the ability to predict future trends in electricity supply and computing power demand, resulting in system instability and the inability to fully utilize the computing power system's adjustable load role in the power system. In addition, task classification and scheduling strategies are extensive and lack flexible adjustment potential.
The data acquisition module is used to construct a time series diagram of power supply and computing power demand, and meteorological data and historical analysis are combined to generate forecast values. The intelligent matching module conducts supply and demand balance analysis, and the task scheduling module performs differentiated scheduling. The feedback optimization module dynamically adjusts the forecast parameters and matching thresholds to achieve proactive scheduling decisions.
It has significantly improved the utilization rate of new energy electricity and the supply stability of computing power services, improved the accuracy of new energy power generation forecasts, and can actively adapt to changes in the operating status of the power system, realizing accurate identification and differentiated processing of supply and demand imbalances.
Smart Images

Figure CN120691373A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer-electronics collaborative technology, and in particular to a computer-electronics collaborative system and method. Background Art
[0002] With the rapid development of the digital economy and the deepening implementation of the "dual carbon" goals, the proportion of energy consumption in large-scale computing centers in total electricity consumption continues to rise. The integration of diverse entities such as renewable energy power generation, energy storage systems, and computing infrastructure into the power system has brought new challenges to the balance of power supply and demand. The coordinated optimization of computing infrastructure and the power system has become an important way to reduce carbon emissions and improve energy efficiency.
[0003] Existing computing-power coordination technologies primarily make scheduling decisions based on real-time power supply and demand, lacking the ability to predict future trends in power supply and computing power demand. This static matching model, faced with the volatility of renewable energy generation and the spatiotemporal variations in computing power demand, can easily lead to frequent supply and demand adjustments, causing system instability. Furthermore, existing computing power task classification and scheduling strategies are relatively crude, insufficiently exploring the flexible adjustment potential of different task types, and unable to fully utilize the computing power system's role as an adjustable load within the power system.
[0004] The invention patent with the Chinese publication number CN119518737A discloses a distribution network planning method for the coordination of power grid and energy storage. The method forms a net load curve through source-load prediction, determines the energy storage configuration capacity based on the power capacity shortage calculation, and establishes a cost optimization model for the coordination of grid and storage to achieve the comprehensive optimization of power grid construction cost and energy storage configuration cost. This method mainly targets the problem of insufficient power supply capacity of the power grid or insufficient new energy carrying capacity, alleviates the pressure on the power grid through the peak-shaving function of the energy storage system, and considers the cost comparison of various energy storage configuration methods such as leasing energy storage, response energy storage and new energy storage. However, this method mainly focuses on the energy storage configuration planning on the grid side, and does not involve the active scheduling of computing load and the optimization of computing-electricity coordination, and lacks consideration and utilization of the spatiotemporal migration characteristics of computing tasks. Summary of the Invention
[0005] In view of this, the present invention proposes a computing-electricity collaboration system and method, which solves the defect of independent operation of the computing power system and the power system in the existing technology, realizes computing-electricity collaboration, and significantly improves the absorption and utilization rate of new energy electricity and the supply stability of computing power services.
[0006] The technical solution of the present invention is implemented as follows: The present invention provides a computer-electronics collaborative system, comprising: The data acquisition module is used to collect power data, computing power data, and price data, and to construct a power supply timing diagram and a computing power demand timing diagram based on the power data and computing power data respectively; The prediction module is used to generate power supply forecast values and computing power demand forecast values based on the power supply timing diagram and computing power demand timing diagram, combined with historical data statistical analysis and real-time meteorological data correction; The intelligent matching module is used to generate a matching decision plan between computing tasks and power resources based on power supply forecasts and computing power demand forecasts, combined with time-sharing and regional power price data, through supply and demand balance analysis, cost-benefit analysis, and matching threshold judgment; A task scheduling module is used to obtain computing power tasks and classify them, perform differentiated scheduling on different types of computing power tasks based on the matching decision scheme, computing power data and price data, and generate task scheduling results; The feedback optimization module is used to evaluate the task scheduling results to generate evaluation results, feed the evaluation results back to the prediction module and the intelligent matching module, and dynamically adjust the prediction parameters of the prediction module and the matching threshold of the intelligent matching module.
[0007] Based on the above technical solution, preferably, the power data includes adjustable power generation capacity data and non-adjustable power generation capacity data, wherein the adjustable power generation capacity data includes coal power data, gas power data, hydropower data and nuclear power data, and the non-adjustable power generation capacity data includes wind power data and photovoltaic data; Computing power data includes computing power node power consumption data, task execution time data, network transmission delay data, and task resource requirement data; Price data includes time-based and regional electricity price data and computing power service price data.
[0008] On the basis of the above technical solution, preferably, the prediction module includes a power supply prediction unit and a computing power demand prediction unit, wherein: a power supply forecasting unit, configured to calculate, based on the power supply timing diagram and current power supply data, an adjustable capacity correction coefficient based on adjustable power generation capacity data, a meteorological correction coefficient based on meteorological data, and a trend correction coefficient based on historical power generation deviations, to generate a power supply forecast value; The computing power demand prediction unit is used to calculate the mode correction coefficient through load mode identification based on the computing power demand timing diagram and the computing power demand data at the current moment, combined with the computing power node power consumption data and the task execution duration data, and calculate the resource correction coefficient based on the task resource demand data and the network transmission delay data, and calculate the business correction coefficient based on the business volume change to generate a computing power demand prediction value.
[0009] On the basis of the above technical solution, preferably, the specific steps of the power supply prediction unit include: Obtain historical power supply data based on the power supply time series diagram, obtain current meteorological data, including wind speed data, wind direction data, solar radiation data, and temperature data, and obtain adjustable power generation capacity data, including the current operating status, available capacity, and rated capacity information of various adjustable generator sets; The theoretical value of wind power is calculated using the wind turbine power curve based on wind speed and direction data, and the theoretical value of photovoltaic power is calculated using the photovoltaic power generation physical model based on solar radiation and temperature data. The adjustment capacity correction coefficient is calculated based on the ratio of the available capacity to the rated capacity of the adjustable generator set. Calculating a historical average power generation based on historical power supply data, and calculating a meteorological correction factor based on the historical average power generation and theoretical power generation; Obtain actual power generation data and calculate the trend correction coefficient based on the deviation between the actual power generation data and historical power generation; The power supply forecast value is calculated based on historical power supply data, regulation capacity correction factor, meteorological correction factor and trend correction factor.
[0010] Based on the above technical solution, preferably, the specific steps of the computing power demand prediction unit include: Obtain historical computing power demand data based on the computing power demand time series diagram, and build a comprehensive load pattern sample library based on computing power node power consumption data and task execution duration data. The load pattern sample library includes daily load change patterns classified based on power consumption characteristics, weekly load change patterns classified based on execution duration characteristics, and monthly load change patterns classified based on comprehensive resource demand characteristics. Obtain computing power demand data within the set time window before the current moment, calculate the computing power demand change curve based on the current computing power node power consumption data and task execution duration data, and generate the current computing power demand change pattern; Calculate and sort the similarity between the current computing power demand change pattern and each load change pattern in the comprehensive load pattern sample library, determine the historical computing power demand value corresponding to the highest similarity, and calculate the pattern correction coefficient based on the historical computing power demand value and computing power node power consumption data; Analyze the resource demand distribution characteristics of the current task based on the task resource demand data, evaluate the resource consumption impact of cross-node task execution in combination with network transmission delay data, and calculate the resource correction coefficient; Obtain the current number of computing task submissions and calculate the business correction coefficient based on the deviation between the current number of tasks and the historical average number of tasks; Generate a computing power demand forecast value based on historical computing power demand data, model correction factor, resource correction factor, and business correction factor.
[0011] On the basis of the above technical solution, preferably, the calculation formula of the power supply forecast value is: ; ; ; in, express The power supply forecast value at the moment, Indicates the current moment, represents the prediction time step, express The historical power supply value at time represents the meteorological correction factor, represents the trend correction coefficient, express Theoretical power generation at the time, express The historical average power generation at the time, represents the trend sensitivity coefficient, represents the i-th day before the current moment, express Actual power generation data at the moment, express The historical power generation at the time, represents the long-term average value of historical power generation, express The adjustment capability correction factor of the moment, express The available capacity of the adjustable generator set at any time, express The rated capacity of the adjustable generator set at any time, represents the adjustment capacity weight coefficient, Indicates the length of the historical time window for trend analysis.
[0012] Based on the above technical solution, preferably, the calculation formula of the computing power demand prediction value is: ; ; ; in, express The predicted value of computing power demand at the moment, express The historical computing power demand value at the moment, represents the mode correction coefficient, represents the business correction coefficient, express The historical computing power demand value corresponding to the most similar mode at the moment, express The historical average computing power demand value at the moment, represents the business sensitivity coefficient, express The actual number of tasks submitted at the moment, Shows the historical average number of task submissions, Indicates the resource modification factor.
[0013] On the basis of the above technical solution, preferably, the intelligent matching module includes three matching rules, namely a first matching rule, a second matching rule and a third matching rule. The specific steps of the intelligent matching module include: Setting a supply and demand safety threshold and a supply and demand gap threshold, and obtaining the current supply and demand difference based on the power supply forecast and the computing power demand forecast, while calculating the current power cost coefficient based on the time-sharing and regional power price data; wherein the supply and demand safety threshold is less than the supply and demand gap threshold; When the power supply forecast value is greater than the computing power demand forecast value, and the current supply and demand difference is greater than the supply and demand safety threshold, the adjustable capacity range of the generator set at the current moment is evaluated in combination with the adjustable power generation capacity data, and the supply surplus is calculated by accumulating the supply and demand difference and the power generation regulation capacity within the forecast time step using the first matching rule. A cost-benefit analysis is performed in combination with the computing power node power consumption data and the current power cost coefficient, and an active scheduling decision plan is generated based on the supply surplus and the cost-benefit analysis results; the active scheduling decision plan instructs the task scheduling module to prioritize the execution of high-power consumption tasks; When the power supply forecast value is less than the computing power demand forecast value, and the absolute value of the supply and demand difference at the current moment is greater than the supply and demand gap threshold, the second matching rule is used to determine the moment when the power supply meets the computing power demand by forward calculation moment by moment, determine the duration of the power supply gap, and calculate the time cost of task migration based on the task execution time data and network transmission delay data. A conservative scheduling decision plan is generated based on the duration of the power supply gap and the task migration cost; the conservative scheduling decision plan instructs the task scheduling module to perform task delay scheduling or spatial migration scheduling; When the absolute value of the supply and demand difference at the current moment is between the supply and demand safety threshold and the supply and demand gap threshold, the third matching rule is used to calculate the historical supply and demand forecast accuracy, and a comprehensive cost analysis is performed based on the task resource demand data and the computing power service price data. A balanced scheduling decision plan is generated based on the historical supply and demand forecast accuracy and the comprehensive cost analysis results; the balanced scheduling decision plan instructs the task scheduling module to adjust the scheduling decision plan according to the historical supply and demand forecast accuracy.
[0014] More preferably, the specific steps of the task scheduling module include: Obtain computing power tasks and divide them into non-migratable tasks, time-migratable tasks, and space-migratable tasks; When the matching decision scheme is the active scheduling decision scheme, the power consumption information of each task in the current period is obtained, and the non-migratable tasks are sorted from high to low according to power consumption, and high-power consumption tasks are scheduled first. At the same time, the execution time of the time-migratable tasks is advanced to the current power-sufficient period; When the matching decision plan is a conservative scheduling decision plan, the normal execution of non-migratable tasks is maintained, the execution of time-migratable tasks is delayed until the power supply gap ends, and the execution of spatially migratable tasks is transferred to other computing power nodes with sufficient power supply; When the matching decision scheme is a balanced scheduling decision scheme, the historical supply and demand forecast accuracy is obtained. When the forecast accuracy is higher than the set threshold, the active scheduling decision scheme is executed. When the forecast accuracy is lower than the set threshold, the scheduling ratio of high-power consumption tasks is reduced and the time interval for task execution is increased.
[0015] The present invention provides a computer-electronic collaboration algorithm, which is applied to the above-mentioned collaboration model and includes the following steps: S1. Collect power data, computing power data, and price data, and construct a power supply timing diagram and a computing power demand timing diagram based on the power data and computing power data respectively; S2. Based on the power supply timing diagram and the computing power demand timing diagram, combined with historical data statistical analysis and real-time meteorological data correction, generate power supply forecast values and computing power demand forecast values; S3. Based on the power supply forecast and computing power demand forecast, combined with time-based and regional power price data, a matching decision plan for computing power tasks and power resources is generated through supply and demand balance analysis, cost-benefit analysis, and matching threshold judgment. S4. Obtain computing power tasks and classify them, perform differentiated scheduling on different types of computing power tasks based on the matching decision plan, computing power data, and price data, and generate task scheduling results; S5. Evaluate the task scheduling results and generate evaluation results. Dynamically adjust the power supply forecast value, computing power demand forecast value, and matching threshold of the intelligent matching module based on the evaluation results.
[0016] The computer-electronics collaborative system and method of the present invention have the following beneficial effects compared to the prior art: (1) The prediction module predicts power supply and computing power demand, the intelligent matching module generates proactive scheduling decisions based on the prediction results, and the task scheduling module performs flexible scheduling of tasks in time and space. Compared with the traditional passive response scheduling method, it significantly improves the utilization rate of new energy power and the supply stability of computing power services; (2) Through the dual correction mechanism of meteorological correction coefficient and trend correction coefficient, the accuracy and practicality of renewable energy power generation forecast are significantly improved, providing a more reliable data basis for computer-electricity collaborative decision-making and effectively reducing the adverse impact of prediction errors on scheduling decisions; (3) By setting supply and demand safety thresholds and supply and demand gap thresholds for multi-level risk identification, accurate identification and differentiated treatment of different degrees of supply and demand imbalance are achieved. When the supply and demand difference is in different threshold ranges, the corresponding matching strategy is automatically selected, so that the computing power task scheduling can actively adapt to the changes in the operating status of the power system, effectively improving the computing and power coordination capabilities. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0018] Figure 1 This is a system block diagram of a computer-electronic collaborative system of the present invention; Figure 2 This is a flow chart of a computer-electronic collaboration method of the present invention. DETAILED DESCRIPTION
[0019] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0020] like Figure 1 As shown, the present invention provides a computer-electronics collaborative system, comprising: The data acquisition module is used to collect power data, computing power data, and price data, and to construct a power supply timing diagram and a computing power demand timing diagram based on the power data and computing power data respectively; The prediction module is used to generate power supply forecast values and computing power demand forecast values based on the power supply timing diagram and computing power demand timing diagram, combined with historical data statistical analysis and real-time meteorological data correction; The intelligent matching module is used to generate a matching decision plan between computing tasks and power resources based on power supply forecasts and computing power demand forecasts, combined with time-sharing and regional power price data, through supply and demand balance analysis, cost-benefit analysis, and matching threshold judgment; A task scheduling module is used to obtain computing power tasks and classify them, perform differentiated scheduling on different types of computing power tasks based on the matching decision scheme, computing power data and price data, and generate task scheduling results; The feedback optimization module is used to evaluate the task scheduling results to generate evaluation results, feed the evaluation results back to the prediction module and the intelligent matching module, and dynamically adjust the prediction parameters of the prediction module and the matching threshold of the intelligent matching module.
[0021] The present invention predicts power supply and computing power demand through a prediction module, the intelligent matching module generates proactive scheduling decisions based on the prediction results, and the task scheduling module performs flexible spatiotemporal scheduling of tasks. Compared with traditional passive response scheduling methods, it significantly improves the utilization rate of new energy power and the supply stability of computing power services.
[0022] In one embodiment of the present invention, the power data includes adjustable power generation capacity data and non-adjustable power generation capacity data, wherein the adjustable power generation capacity data includes coal power data, gas power data, hydropower data and nuclear power data, and the non-adjustable power generation capacity data includes wind power data and photovoltaic data; Computing power data includes computing power node power consumption data, task execution time data, network transmission delay data, and task resource requirement data; Price data includes time-based and regional electricity price data and computing power service price data.
[0023] It can be understood that adjustable power generation capacity data refers to power generation resources in the power system that can accept dispatching instructions and change the output level within a certain period of time to maintain the supply and demand balance of the power system. The characteristics of this type of power source are controllable output, timely response, wide adjustment range, and the ability to actively adjust the power generation power according to the dispatching needs of the power grid; unadjustable power generation capacity data refers to power generation resources whose output is mainly restricted by external natural conditions and is difficult to actively adjust according to dispatching needs. The characteristics of this type of power source are large output fluctuations, strong intermittentness, and relatively poor predictability. Its power generation power mainly depends on natural factors such as meteorological conditions.
[0024] Among them, coal-fired power data is obtained through the power plant monitoring system to obtain key parameters such as the power plant's rated capacity, current load rate, and ramp rate, and is quantified as a power range that can be adjusted every 15 minutes; gas-fired power data is obtained through the operation monitoring system of the gas turbine unit to obtain parameters such as start-up and shutdown status, fuel supply conditions, and regulation rate, and is quantified as rapid adjustment capability; hydropower data is obtained through the hydropower plant automation system to obtain parameters such as reservoir water level, downstream flow, and unit operating status, and is quantified as an adjustable output range; nuclear power data is obtained through the nuclear power plant control system to obtain parameters such as reactor power level and regulation mode, and is quantified as limited adjustment capability.
[0025] Wind power data is obtained through wind farm monitoring systems and meteorological observation equipment, including wind speed, wind direction, unit availability, and other parameters, which are quantified in combination with wind turbine power curves to obtain real-time power generation. Photovoltaic data is obtained through photovoltaic power station monitoring systems and meteorological stations, including solar radiation intensity, component temperature, inverter efficiency, and other parameters, which are quantified in combination with photovoltaic power generation physical models to obtain real-time power generation.
[0026] Electricity price data is obtained through the electricity trading platform, and electricity price information for different periods and regions is obtained and quantified into cost parameters. Computing power service price data is obtained through the cloud service platform, and the price system of different computing power services is obtained, providing a reference for cost optimization of task scheduling.
[0027] In one embodiment of the present invention, the prediction module includes a power supply prediction unit and a computing power demand prediction unit, wherein: a power supply forecasting unit, configured to calculate, based on the power supply timing diagram and current power supply data, an adjustable capacity correction coefficient based on adjustable power generation capacity data, a meteorological correction coefficient based on meteorological data, and a trend correction coefficient based on historical power generation deviations, to generate a power supply forecast value; The computing power demand prediction unit is used to calculate the mode correction coefficient through load mode identification based on the computing power demand timing diagram and the computing power demand data at the current moment, combined with the computing power node power consumption data and the task execution duration data, and calculate the resource correction coefficient based on the task resource demand data and the network transmission delay data, and calculate the business correction coefficient based on the business volume change to generate a computing power demand prediction value.
[0028] It can be understood that the collection of meteorological data adopts multi-source fusion technology, which comprehensively utilizes ground meteorological station observation data, high-altitude detection data, satellite remote sensing data and numerical weather forecast model output to form a meteorological element field with high temporal and spatial resolution.
[0029] The prediction module of the present invention significantly improves the accuracy and practicality of renewable energy power generation prediction through the dual correction mechanism of meteorological correction coefficient and trend correction coefficient, provides a more reliable data basis for computer-electricity collaborative decision-making, and effectively reduces the adverse impact of prediction errors on scheduling decisions.
[0030] In one embodiment of the present invention, the specific steps of the power supply prediction unit include: Obtain historical power supply data based on the power supply time series diagram, obtain current meteorological data, including wind speed data, wind direction data, solar radiation data, and temperature data, and obtain adjustable power generation capacity data, including the current operating status, available capacity, and rated capacity information of various adjustable generator sets; The theoretical value of wind power is calculated using the wind turbine power curve based on wind speed and direction data, and the theoretical value of photovoltaic power is calculated using the photovoltaic power generation physical model based on solar radiation and temperature data. The adjustment capacity correction coefficient is calculated based on the ratio of the available capacity to the rated capacity of the adjustable generator set. Calculating a historical average power generation based on historical power supply data, and calculating a meteorological correction factor based on the historical average power generation and theoretical power generation; Obtain actual power generation data and calculate the trend correction coefficient based on the deviation between the actual power generation data and historical power generation; The power supply forecast value is calculated based on historical power supply data, regulation capacity correction factor, meteorological correction factor and trend correction factor. The calculation formula is: ; ; ; in, express The power supply forecast value at the moment, Indicates the current moment, represents the prediction time step, express The historical power supply value at time represents the meteorological correction factor, represents the trend correction coefficient, express Theoretical power generation at the time, express The historical average power generation at the time, represents the trend sensitivity coefficient, represents the i-th day before the current moment, express Actual power generation data at the moment, express The historical power generation at the time, represents the long-term average value of historical power generation, express The adjustment capability correction factor of the moment, express The available capacity of the adjustable generator set at any time, express The rated capacity of the adjustable generator set at any time, represents the adjustment capacity weight coefficient, Indicates the length of the historical time window for trend analysis.
[0031] Understandably, is the average historical power generation up to time t. Wind power forecasting uses the wind turbine power curve method, calculating the theoretical wind power value based on wind speed and direction data. Wind turbine power generation is closely related to wind speed. When the wind speed is below the cut-in wind speed (the minimum wind speed at which the wind turbine can start and generate electricity), the wind turbine cannot start, and the power generation is zero. When the wind speed exceeds the cut-out wind speed (the maximum wind speed at which the wind turbine must be shut down for safety), the wind turbine shuts down for protection, and the power generation drops to zero. Normal power generation is only possible between the cut-in and cut-out wind speeds, maintaining the power generation at the rated power level. Photovoltaic power generation is primarily affected by solar radiation intensity, exhibiting a distinct daily variation pattern. It is also affected by factors such as cloud cover and temperature fluctuations, resulting in significant volatility.
[0032] The meteorological correction coefficient reflects the degree of deviation of the current meteorological conditions from the historical average meteorological conditions. When the meteorological conditions are better than the historical average, the correction coefficient is greater than 1, indicating that the power generation capacity will be higher than the historical average; when the meteorological conditions are worse than the historical average, the correction coefficient is less than 1, indicating that the power generation capacity will be lower than the historical average.
[0033] The trend correction coefficient is used to capture systematic deviations and short-term trend changes. That is, by analyzing the deviation between the recent actual power generation and the historical average, the trend correction coefficient can be used to identify the changing trend of the system operation status. When the recent actual power generation continues to be higher than the historical average, the trend correction coefficient is greater than 1, indicating that the forecast value should be appropriately adjusted upward; when the recent actual power generation continues to be lower than the historical average, the trend correction coefficient is less than 1, indicating that the forecast value should be appropriately adjusted downward. Trend sensitivity coefficient The least squares regression method is used to determine the parameter value by minimizing the prediction error. The optimal , and is regularly updated based on operational data.
[0034] In one embodiment of the present invention, the historical average power generation power is calculated by a sliding window statistical method to sum and average the power generation power, which can capture the seasonal variation pattern of the power generation power while maintaining sensitivity to recent variation trends through the sliding window.
[0035] In one embodiment of the present invention, a hierarchical evaluation method is used to assess adjustable power generation capacity data in power supply forecasts. This method dynamically calculates the adjustable capacity correction coefficient by monitoring the operating status of various adjustable generating units in real time. For coal-fired power units, operating parameters such as the unit's current load factor, boiler temperature, and turbine status are obtained, and the adjustable capacity within the forecast period is calculated based on the unit's ramp rate and adjustment range. For gas-fired power units, the focus is on gas supply conditions, combustion chamber temperature, and turbine blade status. For hydropower units, key parameters such as the reservoir's current water level, inflow and outflow, and the unit's turbine operating status are obtained. For nuclear power units, core operating parameters such as reactor power level, steam generator status, and control rod position are monitored.
[0036] The calculation of the regulation capacity correction coefficient adopts a weighted comprehensive evaluation method. The system calculates the coal-fired power regulation coefficient, gas-fired power regulation coefficient, hydropower regulation coefficient and nuclear power regulation coefficient respectively according to the real-time available capacity, regulation rate characteristics and operating constraints of various adjustable units, and then performs weighted synthesis based on the installed capacity ratio of each unit in the system and the current operating status.
[0037] In one embodiment of the present invention, the specific steps of the computing power demand prediction unit include: Obtain historical computing power demand data based on the computing power demand time series diagram, and build a comprehensive load pattern sample library based on computing power node power consumption data and task execution duration data. The load pattern sample library includes daily load change patterns classified based on power consumption characteristics, weekly load change patterns classified based on execution duration characteristics, and monthly load change patterns classified based on comprehensive resource demand characteristics. Obtain computing power demand data within the set time window before the current moment, calculate the computing power demand change curve based on the current computing power node power consumption data and task execution duration data, and generate the current computing power demand change pattern; Calculate and sort the similarity between the current computing power demand change pattern and each load change pattern in the comprehensive load pattern sample library, determine the historical computing power demand value corresponding to the highest similarity, and calculate the pattern correction coefficient based on the historical computing power demand value and computing power node power consumption data; Analyze the resource demand distribution characteristics of the current task based on the task resource demand data, evaluate the resource consumption impact of cross-node task execution in combination with network transmission delay data, and calculate the resource correction coefficient; Obtain the current number of computing task submissions and calculate the business correction coefficient based on the deviation between the current number of tasks and the historical average number of tasks; The computing power demand forecast value is generated based on historical computing power demand data, model correction factor, resource correction factor, and business correction factor. The calculation formula is: ; ; ; ; ;in, express The predicted value of computing power demand at the moment, express The historical computing power demand value at the moment, represents the mode correction coefficient, represents the business correction coefficient, express The historical computing power demand value corresponding to the most similar mode at the moment, express The historical average computing power demand value at the moment, Indicates the business sensitivity coefficient, the value range is [0, 1], express The actual number of tasks submitted at the moment, Indicates the historical average number of task submissions, represents the resource correction factor, express The average resource demand intensity of the current task at the moment, represents the historical average task resource demand intensity, Indicates the standard deviation of the number of historical task submissions.
[0038] It can be understood that the average resource demand intensity is obtained by taking a weighted average of the number of CPU cores, memory requirements, and storage requirements of the computing task.
[0039] The pattern correction coefficient reflects the degree of deviation of the current demand pattern from the historical average pattern. When the current pattern is similar to the historical high demand pattern, the correction coefficient is greater than 1; when the current pattern is similar to the historical low demand pattern, the correction coefficient is less than 1.
[0040] In one embodiment of the present invention, a multi-dimensional feature fusion method is used to combine computing power node power consumption data and task execution duration data in computing power demand forecasting. This method analyzes the power consumption and time characteristics of different types of tasks to construct a more accurate load pattern sample library. Computing power tasks are classified according to their power consumption level into high-power tasks (power consumption exceeding 1000W), medium-power tasks (power consumption between 500-1000W), and low-power tasks (power consumption less than 500W). Furthermore, they are classified according to their execution duration into short-duration tasks (execution time less than 1 hour), medium-duration tasks (execution time 1-8 hours), and long-duration tasks (execution time exceeding 8 hours). This two-dimensional classification method establishes nine basic task type patterns.
[0041] In one embodiment of the present invention, the current computing power demand change pattern is generated using a sliding window technique. This technique obtains computing power demand data within a set time window before the current moment and calculates the computing power demand change curve. The time window is selected based on the forecast time scale: for hourly forecasts, data from the past 24 hours is used; for daily forecasts, data from the past 7 days is used; and for weekly forecasts, data from the past 4 weeks is used.
[0042] In one embodiment of the present invention, the similarity calculation adopts the dynamic time warping algorithm, and the calculation formula is: in, Indicates the current computing power demand change pattern and historical load change patterns The similarity between Indicates the current computing power demand change pattern, Represents the historical load change pattern, Indicates the current computing power demand change pattern and historical load change patterns The dynamic time warping distance between Indicates the current computing power demand change pattern and historical load change patterns The maximum absolute value of all elements in the sequence between.
[0043] In one embodiment of the present invention, the intelligent matching module includes three matching rules, namely a first matching rule, a second matching rule and a third matching rule. The specific steps of the intelligent matching module include: Setting a supply and demand safety threshold and a supply and demand gap threshold, and obtaining the current supply and demand difference based on the power supply forecast and the computing power demand forecast, while calculating the current power cost coefficient based on the time-sharing and regional power price data; wherein the supply and demand safety threshold is less than the supply and demand gap threshold; When the power supply forecast value is greater than the computing power demand forecast value, and the current supply and demand difference is greater than the supply and demand safety threshold, the adjustable capacity range of the generator set at the current moment is evaluated in combination with the adjustable power generation capacity data, and the supply surplus is calculated by accumulating the supply and demand difference and the power generation regulation capacity within the forecast time step using the first matching rule. A cost-benefit analysis is performed in combination with the computing power node power consumption data and the current power cost coefficient, and an active scheduling decision plan is generated based on the supply surplus and the cost-benefit analysis results; the active scheduling decision plan instructs the task scheduling module to prioritize the execution of high-power consumption tasks; When the power supply forecast value is less than the computing power demand forecast value, and the absolute value of the supply and demand difference at the current moment is greater than the supply and demand gap threshold, the second matching rule is used to determine the moment when the power supply meets the computing power demand by forward calculation moment by moment, determine the duration of the power supply gap, and calculate the time cost of task migration based on the task execution time data and network transmission delay data. A conservative scheduling decision plan is generated based on the duration of the power supply gap and the task migration cost; the conservative scheduling decision plan instructs the task scheduling module to perform task delay scheduling or spatial migration scheduling; When the absolute value of the supply and demand difference at the current moment is between the supply and demand safety threshold and the supply and demand gap threshold, the third matching rule is used to calculate the historical supply and demand forecast accuracy, and a comprehensive cost analysis is performed based on the task resource demand data and the computing power service price data. A balanced scheduling decision plan is generated based on the historical supply and demand forecast accuracy and the comprehensive cost analysis results; the balanced scheduling decision plan instructs the task scheduling module to adjust the scheduling decision plan according to the historical supply and demand forecast accuracy.
[0044] It can be understood that the priority scheduling of high-power consumption tasks is implemented based on the power consumption sorting mechanism. The task scheduling module obtains the power consumption information of each task in the current period, sorts them from high to low according to power consumption, and gives priority to executing computing tasks with higher power consumption, thereby maximizing the utilization efficiency of power resources.
[0045] The supply and demand security threshold is determined based on the statistical analysis of historical electricity supply and demand data, and the supply and demand gap threshold is based on the maximum bearing capacity and risk control requirements of the power system, which can timely trigger the threshold value of the conservative scheduling strategy.
[0046] In one embodiment of the present invention, the specific steps of the feedback optimization module include evaluation index calculation, deviation analysis and parameter adjustment, specifically including: By regularly collecting statistics on key performance indicators such as task scheduling success rate, average task completion time, power resource utilization, and system load balance within a certain time window, and comparing and analyzing them with preset target values, the achievement and deviation of each indicator are calculated; By analyzing the direction and magnitude of deviations in various evaluation indicators, we can identify the main problems and areas for improvement in the current scheduling strategy. For example, when the task scheduling success rate is lower than the target value, further analysis is needed to determine whether this is due to insufficient prediction accuracy, improper matching threshold settings, or overly strict task classification standards. By analyzing the distribution characteristics and changing trends of prediction errors, an adaptive adjustment strategy is adopted to gradually optimize the parameter values, making the prediction results closer to the actual situation.
[0047] The present invention performs multi-level risk identification by setting supply and demand safety thresholds and supply and demand gap thresholds, thereby realizing accurate identification and differentiated treatment of different degrees of supply and demand imbalance. When the supply and demand difference is in different threshold intervals, the corresponding matching strategy is automatically selected, so that computing power task scheduling can actively adapt to changes in the operating status of the power system, effectively improving the computing and power coordination capabilities.
[0048] In one embodiment of the present invention, when meteorological data indicates that the wind speed will be higher or the solar radiation intensity will be higher in the future period, the wind power and photovoltaic power generation power are expected to increase, the supply surplus will increase accordingly, and the intensity of active scheduling can be appropriately strengthened.
[0049] In one embodiment of the present invention, the conservative scheduling decision scheme includes task delay scheduling and space migration scheduling. Task delay scheduling is for time-migratable tasks, that is, the execution time of time-migratable tasks is postponed until the power supply gap ends; space migration scheduling is for space-migratable tasks, that is, the execution time of space-migratable tasks is transferred to other computing power nodes with sufficient power supply.
[0050] In one embodiment of the present invention, the historical supply and demand forecast accuracy is calculated using a sliding window method, and the calculation formula is: in, represents the accuracy of historical supply and demand forecasts, Indicates the length of the statistical window, usually set to 168, which is the number of hours in a week. Indicates the time sequence number, represents the power supply forecast value at the jth moment, represents the actual power supply value at the jth moment.
[0051] In one embodiment of the present invention, the specific steps of the task scheduling module include: Obtain computing power tasks and divide them into non-migratable tasks, time-migratable tasks, and space-migratable tasks; When the matching decision scheme is the active scheduling decision scheme, the power consumption information of each task in the current period is obtained, and the non-migratable tasks are sorted from high to low according to power consumption, and high-power consumption tasks are scheduled first. At the same time, the execution time of the time-migratable tasks is advanced to the current power-sufficient period; When the matching decision plan is a conservative scheduling decision plan, the normal execution of non-migratable tasks is maintained, the execution of time-migratable tasks is delayed until the power supply gap ends, and the execution of spatially migratable tasks is transferred to other computing power nodes with sufficient power supply; When the matching decision scheme is a balanced scheduling decision scheme, the historical supply and demand forecast accuracy is obtained. When the forecast accuracy is higher than the set threshold, the active scheduling decision scheme is executed. When the forecast accuracy is lower than the set threshold, the scheduling ratio of high-power consumption tasks is reduced and the time interval for task execution is increased.
[0052] It can be understood that a non-migratable task refers to a task that has strict requirements on execution time and execution location, and must be executed at a specified time and node; a time-migratable task refers to a task with a relatively fixed execution location but a certain degree of flexibility in execution time; a space-migratable task refers to a task with a relatively fixed execution time but flexible execution location.
[0053] The generation of balanced scheduling decision plans is based on the prediction accuracy threshold judgment mechanism. When the historical supply and demand prediction accuracy is greater than the set threshold, it indicates that the prediction module has high reliability, and the balanced scheduling decision plan tends to be active scheduling. It is executed according to the active scheduling decision plan of the first matching rule, that is, high-power consumption tasks are scheduled first and time-migratable tasks are executed in advance. When the historical supply and demand prediction accuracy is not greater than the set threshold, it indicates that the reliability of the prediction module is limited, and the balanced scheduling decision plan tends to be conservative scheduling, which is specifically manifested in reducing the scheduling ratio of high-power consumption tasks and increasing the time interval for task execution.
[0054] The present invention divides computing tasks and adopts differentiated scheduling decision-making schemes for different categories of tasks, thereby overcoming the technical limitations of traditional scheduling methods in accurately distinguishing task types. Through the coordinated optimization of task execution time and space, it improves the utilization rate of power resources and reduces the cost of computing task execution.
[0055] In one embodiment of the present invention, the advance execution strategy of time-migratable tasks is implemented using a time window optimization algorithm, which determines the optimal task execution time by analyzing the power supply sufficient period and task time constraints.
[0056] In one embodiment of the present invention, the evaluation of adjustable power generation capacity data utilizes a unit adjustability analysis method. This method analyzes the current operating status of various adjustable power generation units to determine the system's power generation adjustment space. For coal-fired power units, the current load factor, maximum output, and minimum output data are obtained to calculate the range of power generation that can be increased or decreased within the forecast period. For gas-fired power units, the gas turbine's start / stop status and fuel supply status are obtained. For hydropower units, the hydropower's adjustability is assessed by combining reservoir water levels, downstream flow constraints, and unit operating status. For nuclear power units, the reactor power level, control rod position, and steam generator operating parameters are monitored to assess the unit's load tracking capability.
[0057] When the intelligent matching module detects that the power supply forecast value is greater than the computing power demand forecast value and the difference exceeds the safety threshold, it analyzes whether the adjustable generator set has room to reduce its output. If the adjustable generator set is currently operating in a high-load state and has sufficient adjustment margin, the active scheduling decision plan will instruct to appropriately reduce the output of some adjustable units to reserve power space for the increase in computing power load, thereby achieving dynamic coordination between power supply and computing power demand.
[0058] In one embodiment of the present invention, a dynamic cost optimization strategy is used for time-sharing and regional electricity price data to guide the scheduling decisions of computing tasks based on the periodic changes and regional differences in electricity prices. The calculation of the electricity cost coefficient is based on the ratio of the current electricity price to the historical average electricity price. When the electricity cost coefficient is greater than 1, it means that the current electricity price is higher than the average level. When the electricity cost coefficient is less than 1, it means that the current electricity price is lower than the average level. When generating an active scheduling decision plan, the intelligent matching module gives priority to increasing the execution ratio of high-power consumption tasks during periods when electricity prices are relatively low, and executes low-power consumption tasks or delays some time-migratable tasks during periods when electricity prices are relatively high.
[0059] The cost-benefit analysis includes task execution cost calculation and cross-regional cost comparison. The task execution cost calculation is based on the computing power node power consumption data and the current electricity price data. The electricity cost of the task is obtained by multiplying the expected power consumption of the task by the corresponding electricity price, and then the total execution cost of the task is calculated in combination with the computing power service price data. The cross-regional cost includes the electricity cost of the target node, the computing power service cost, and the delay cost caused by network transmission.
[0060] In one embodiment of the present invention, the time cost of task migration is calculated using a comprehensive evaluation method. The execution time characteristics of different types of tasks are analyzed based on the task execution time data, and the tasks are divided into short-term tasks (execution time less than 1 hour), medium-term tasks (execution time 1-8 hours) and long-term tasks (execution time more than 8 hours) according to their execution time. Different types of tasks have different sensitivities to delays. Short-term tasks are most sensitive to delays, and long-term tasks have a relatively high tolerance for delays.
[0061] By real-time monitoring of network delays between nodes with different computing power, a dynamic network delay matrix is established. When the intelligent matching module needs to generate a spatial migration scheduling plan, it comprehensively considers the power supply status of the target node, network transmission delay, and the task's sensitivity to delay. For delay-sensitive tasks, neighboring nodes with lower network delays will be prioritized for migration. For delay-tolerant tasks, the power cost advantage of the target node will be considered.
[0062] In one embodiment of the present invention, the comprehensive cost analysis adopts a multi-dimensional cost modeling method, incorporating electricity costs, computing power service costs, network transmission costs, and time delay costs into a unified cost assessment framework. When generating a balanced scheduling decision plan, the intelligent matching module first calculates the weight coefficients of various costs. The weight coefficient of electricity cost is determined based on the degree of deviation between the current electricity price and the historical electricity price. The greater the deviation, the higher the weight coefficient. The weight coefficient of computing power service cost is determined based on the resource supply and demand ratio. The scarcer the resource, the higher the weight coefficient. The weight coefficient of network transmission cost is determined based on delay sensitivity. The higher the delay sensitivity, the higher the weight coefficient. The weight coefficient of time delay cost is determined based on the urgency of the task. The higher the urgency, the higher the weight coefficient.
[0063] The intelligent matching module divides the forecast period into multiple time windows. Within each time window, it analyzes the executable task combinations and their corresponding comprehensive costs. Through recursive calculations, it finds the task scheduling sequence that minimizes the comprehensive cost for the entire forecast period. When the historical supply and demand forecast accuracy is high, task scheduling is based on cost differences across time periods. When the historical supply and demand forecast accuracy is low, conservative decisions are made based on the current cost situation.
[0064] like Figure 2 As shown, the present invention also provides a computer-electronics collaborative method, which is applied to the computer-electronics collaborative system as described above, and includes the following steps: S1. Collect power data, computing power data, and price data, and construct a power supply timing diagram and a computing power demand timing diagram based on the power data and computing power data respectively; S2. Based on the power supply timing diagram and the computing power demand timing diagram, combined with historical data statistical analysis and real-time meteorological data correction, generate power supply forecast values and computing power demand forecast values; S3. Based on the power supply forecast and computing power demand forecast, combined with time-based and regional power price data, a matching decision plan for computing power tasks and power resources is generated through supply and demand balance analysis, cost-benefit analysis, and matching threshold judgment. S4. Obtain computing power tasks and classify them, perform differentiated scheduling on different types of computing power tasks based on the matching decision plan, computing power data, and price data, and generate task scheduling results; S5. Evaluate the task scheduling results and generate evaluation results. Dynamically adjust the power supply forecast value, computing power demand forecast value, and matching threshold of the intelligent matching module based on the evaluation results.
[0065] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A computer-electronics collaborative system, characterized by: include: The data acquisition module is used to collect power data, computing power data, and price data, and to construct a power supply timing diagram and a computing power demand timing diagram based on the power data and computing power data respectively; The prediction module is used to generate power supply forecast values and computing power demand forecast values based on the power supply timing diagram and computing power demand timing diagram, combined with historical data statistical analysis and real-time meteorological data correction; The intelligent matching module is used to generate a matching decision plan between computing tasks and power resources based on power supply forecasts and computing power demand forecasts, combined with time-sharing and regional power price data, through supply and demand balance analysis, cost-benefit analysis, and matching threshold judgment; A task scheduling module is used to obtain computing power tasks and classify them, perform differentiated scheduling on different types of computing power tasks based on the matching decision scheme, computing power data and price data, and generate task scheduling results; The feedback optimization module is used to evaluate the task scheduling results to generate evaluation results, feed the evaluation results back to the prediction module and the intelligent matching module, and dynamically adjust the prediction parameters of the prediction module and the matching threshold of the intelligent matching module.
2. The computer-electronics collaborative system according to claim 1, wherein: The power data includes adjustable power generation capacity data and non-adjustable power generation capacity data, wherein the adjustable power generation capacity data includes coal power data, gas power data, hydropower data and nuclear power data, and the non-adjustable power generation capacity data includes wind power data and photovoltaic data; Computing power data includes computing power node power consumption data, task execution time data, network transmission delay data, and task resource requirement data; Price data includes time-based and regional electricity price data and computing power service price data.
3. The computer-electronics collaborative system according to claim 1, wherein: The prediction module includes a power supply prediction unit and a computing power demand prediction unit, wherein: a power supply forecasting unit, configured to calculate, based on the power supply timing diagram and current power supply data, an adjustable capacity correction coefficient based on adjustable power generation capacity data, a meteorological correction coefficient based on meteorological data, and a trend correction coefficient based on historical power generation deviations, to generate a power supply forecast value; The computing power demand prediction unit is used to calculate the mode correction coefficient through load mode identification based on the computing power demand timing diagram and the computing power demand data at the current moment, combined with the computing power node power consumption data and the task execution duration data, and calculate the resource correction coefficient based on the task resource demand data and the network transmission delay data, and calculate the business correction coefficient based on the business volume change to generate a computing power demand prediction value.
4. The computer-electronics collaborative system according to claim 3, wherein: The specific steps of the power supply prediction unit include: Obtain historical power supply data based on the power supply time series diagram, obtain current meteorological data, including wind speed data, wind direction data, solar radiation data, and temperature data, and obtain adjustable power generation capacity data, including the current operating status, available capacity, and rated capacity information of various adjustable generator sets; The theoretical value of wind power is calculated using the wind turbine power curve based on wind speed and direction data, and the theoretical value of photovoltaic power is calculated using the photovoltaic power generation physical model based on solar radiation and temperature data. The adjustment capacity correction coefficient is calculated based on the ratio of the available capacity to the rated capacity of the adjustable generator set. Calculating a historical average power generation based on historical power supply data, and calculating a meteorological correction factor based on the historical average power generation and theoretical power generation; Obtain actual power generation data and calculate the trend correction coefficient based on the deviation between the actual power generation data and historical power generation; The power supply forecast value is calculated based on historical power supply data, regulation capacity correction factor, meteorological correction factor and trend correction factor.
5. The computer-electronics collaborative system according to claim 3, wherein: The specific steps of the computing power demand prediction unit include: Obtain historical computing power demand data based on the computing power demand time series diagram, and build a comprehensive load pattern sample library based on computing power node power consumption data and task execution duration data. The load pattern sample library includes daily load change patterns classified based on power consumption characteristics, weekly load change patterns classified based on execution duration characteristics, and monthly load change patterns classified based on comprehensive resource demand characteristics. Obtain computing power demand data within the set time window before the current moment, calculate the computing power demand change curve based on the current computing power node power consumption data and task execution duration data, and generate the current computing power demand change pattern; Calculate and sort the similarity between the current computing power demand change pattern and each load change pattern in the comprehensive load pattern sample library, determine the historical computing power demand value corresponding to the highest similarity, and calculate the pattern correction coefficient based on the historical computing power demand value and computing power node power consumption data; Analyze the resource demand distribution characteristics of the current task based on the task resource demand data, evaluate the resource consumption impact of cross-node task execution in combination with network transmission delay data, and calculate the resource correction coefficient; Obtain the current number of computing task submissions and calculate the business correction coefficient based on the deviation between the current number of tasks and the historical average number of tasks; Generate a computing power demand forecast value based on historical computing power demand data, model correction factor, resource correction factor, and business correction factor.
6. A computer-electronics collaborative system according to claim 5, characterized in that: The calculation formula of the power supply forecast value is: ; ; ; in, express The power supply forecast value at the moment, Indicates the current moment, represents the prediction time step, express The historical power supply value at time represents the meteorological correction factor, represents the trend correction coefficient, express Theoretical power generation at the time, express The historical average power generation at the time, represents the trend sensitivity coefficient, represents the i-th day before the current moment, express Actual power generation data at the moment, express The historical power generation at the time, represents the long-term average value of historical power generation, express The adjustment capability correction factor of the moment, express The available capacity of the adjustable generator set at any time, express The rated capacity of the adjustable generator set at any time, represents the adjustment capacity weight coefficient, Indicates the length of the historical time window for trend analysis.
7. A computer-electronics collaborative system according to claim 6, characterized in that: The calculation formula for the predicted computing power demand value is: ; ; ; in, express The predicted value of computing power demand at the moment, express The historical computing power demand value at the moment, represents the mode correction coefficient, represents the business correction coefficient, express The historical computing power demand value corresponding to the most similar mode at the moment, express The historical average computing power demand value at the moment, represents the business sensitivity coefficient, express The actual number of tasks submitted at the moment, Shows the historical average number of task submissions, Indicates the resource modification factor.
8. A computer-electronics collaborative system according to claim 7, characterized in that: The intelligent matching module includes three matching rules, namely a first matching rule, a second matching rule and a third matching rule. The specific steps of the intelligent matching module include: Setting a supply and demand safety threshold and a supply and demand gap threshold, and obtaining the current supply and demand difference based on the power supply forecast and the computing power demand forecast, while calculating the current power cost coefficient based on the time-sharing and regional power price data; wherein the supply and demand safety threshold is less than the supply and demand gap threshold; When the power supply forecast value is greater than the computing power demand forecast value, and the current supply and demand difference is greater than the supply and demand safety threshold, the adjustable capacity range of the generator set at the current moment is evaluated in combination with the adjustable power generation capacity data, and the supply surplus is calculated by accumulating the supply and demand difference and the power generation regulation capacity within the forecast time step using the first matching rule. A cost-benefit analysis is performed in combination with the computing power node power consumption data and the current power cost coefficient, and an active scheduling decision plan is generated based on the supply surplus and the cost-benefit analysis results; the active scheduling decision plan instructs the task scheduling module to prioritize the execution of high-power consumption tasks; When the power supply forecast value is less than the computing power demand forecast value, and the absolute value of the supply and demand difference at the current moment is greater than the supply and demand gap threshold, the second matching rule is used to determine the moment when the power supply meets the computing power demand by forward calculation moment by moment, determine the duration of the power supply gap, and calculate the time cost of task migration based on the task execution time data and network transmission delay data. A conservative scheduling decision plan is generated based on the duration of the power supply gap and the task migration cost; the conservative scheduling decision plan instructs the task scheduling module to perform task delay scheduling or spatial migration scheduling; When the absolute value of the supply and demand difference at the current moment is between the supply and demand safety threshold and the supply and demand gap threshold, the third matching rule is used to calculate the historical supply and demand forecast accuracy, and a comprehensive cost analysis is performed based on the task resource demand data and the computing power service price data. A balanced scheduling decision plan is generated based on the historical supply and demand forecast accuracy and the comprehensive cost analysis results; the balanced scheduling decision plan instructs the task scheduling module to adjust the scheduling decision plan according to the historical supply and demand forecast accuracy.
9. A computer-electronics collaborative system according to claim 8, characterized in that: The specific steps of the task scheduling module include: Obtain computing power tasks and divide them into non-migratable tasks, time-migratable tasks, and space-migratable tasks; When the matching decision scheme is the active scheduling decision scheme, the power consumption information of each task in the current period is obtained, and the non-migratable tasks are sorted from high to low according to power consumption, and high-power consumption tasks are scheduled first. At the same time, the execution time of the time-migratable tasks is advanced to the current power-sufficient period; When the matching decision plan is a conservative scheduling decision plan, the normal execution of non-migratable tasks is maintained, the execution of time-migratable tasks is delayed until the power supply gap ends, and the execution of spatially migratable tasks is transferred to other computing power nodes with sufficient power supply; When the matching decision scheme is a balanced scheduling decision scheme, the historical supply and demand forecast accuracy is obtained. When the forecast accuracy is higher than the set threshold, the active scheduling decision scheme is executed. When the forecast accuracy is lower than the set threshold, the scheduling ratio of high-power consumption tasks is reduced and the time interval for task execution is increased.
10. A computer-electronics collaborative method, characterized in that: A computer-electronics collaborative system applied to any of the above items includes the following steps: S1. Collect power data, computing power data, and price data, and construct a power supply timing diagram and a computing power demand timing diagram based on the power data and computing power data respectively; S2. Based on the power supply timing diagram and the computing power demand timing diagram, combined with historical data statistical analysis and real-time meteorological data correction, generate power supply forecast values and computing power demand forecast values; S3. Based on the power supply forecast and computing power demand forecast, combined with time-based and regional power price data, a matching decision plan for computing power tasks and power resources is generated through supply and demand balance analysis, cost-benefit analysis, and matching threshold judgment. S4. Obtain computing power tasks and classify them, perform differentiated scheduling on different types of computing power tasks based on the matching decision plan, computing power data, and price data, and generate task scheduling results; S5. Evaluate the task scheduling results and generate evaluation results. Dynamically adjust the power supply forecast value, computing power demand forecast value, and matching threshold of the intelligent matching module based on the evaluation results.
Citation Information
Patent Citations
Power grid and energy storage coordinated power distribution network planning method
CN119518737A
Computing power network task scheduling method, task processing method, device and system
CN118827668A
Method and device for adjusting power supply capability of computing power infrastructure
CN118863414A
Scheduling system for power load side resources
CN119051015A
Optical-data-storage integrated energy management optimization system and method based on computing task allocation
CN119742768A
Cited By
Computation fusion data application method based on full life cycle quality assurance
CN120910814A
Intelligent computing center resource management and control method, system and equipment based on computing and power collaboration
CN121364938A
A cross-domain computer collaboration low-latency prediction scheduling method and system
CN122526842A
