Calculation and power collaborative optimization method and system based on digital twinning, electronic equipment and storage medium

Abstract

The invention provides an electricity calculation collaborative optimization method and system based on digital twinning, electronic equipment and a storage medium, and belongs to the technical field of electric power resource scheduling optimization, and the method comprises the steps: obtaining historical and real-time electric power system parameter information and calculation task information, and constructing a dynamic recursive state prediction model based on digital twinning; utilizing the prediction model to predict the power demand and the calculation task load at the next moment, and performing preliminary scheduling on the current power demand and the calculation task load according to a prediction result; and on the basis of the preliminary scheduling, performing collaborative optimization on the scheduling of the power demand and the calculation task load at the current moment according to a Lagrange optimization algorithm so as to generate a scheduling result which minimizes the scheduling deviation of the power demand and the calculation task load, and dynamically adjusting the scheduling result through a stability correction factor. According to the invention, accurate collaborative optimization between the power demand and the calculation task load is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power resource scheduling optimization, and in particular to a computer-electricity collaborative optimization method, system, electronic device and storage medium based on digital twins. Background Art

[0002] Computing power is a new type of productivity in the digital economy era and a key engine driving the construction of new power and energy systems. Data centers are the carriers of computing power. Equipped with a variety of servers, graphics cards, and other hardware devices, they deliver powerful computing capabilities and large-scale data processing capabilities to meet the needs of high-performance computing applications. With the continuous increase in global energy demand and the increasing complexity of computing tasks, the coordinated scheduling of power systems and data centers faces unprecedented challenges.

[0003] Traditional power scheduling methods often rely on historical data or static models, lacking the ability to adapt to dynamic changes and real-time loads. At the same time, the computing workloads in data centers continue to increase and fluctuate, making the proper allocation of computing resources even more crucial. To improve system energy efficiency and ensure the proper allocation of resources, achieving precise co-optimization between power demand and computing workloads has become a key issue in modern power and computing management. Summary of the Invention

[0004] In response to the above problems, the present invention provides a computer-electricity collaborative optimization method, system, electronic device and storage medium based on digital twins to achieve resource scheduling optimization of power systems and data centers.

[0005] The present invention provides a computer-electronic collaborative optimization method based on digital twins, comprising: Acquire historical power system parameter information and historical calculation task information within a first historical period, historical power system parameter information and historical calculation task information within a second historical period, and real-time power system parameter information and real-time calculation task information; Using the historical power system parameter information and historical computing task information in the first historical period, a dynamic recursive state prediction model based on digital twins is constructed; Using the constructed dynamic recursive state prediction model, based on the historical power system parameter information and historical computing task information in the second historical period, the power demand and computing task load at the next moment are predicted, and according to the prediction results, the power demand and computing task load at the current moment are preliminarily scheduled; Based on the preliminary scheduling, the scheduling of the current power demand and computing task load is collaboratively optimized according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjusts the scheduling result through a stability correction factor based on real-time power system parameter information and real-time computing task information.

[0006] Furthermore, the power system parameter information includes power load, current, voltage, power, and temperature, and the computing task information includes the execution progress, load distribution, response time, and energy consumption of the computing task.

[0007] Furthermore, the use of historical power system parameter information and historical computing task information within the first historical period to construct a dynamic recursive state prediction model based on digital twins includes constructing a dynamic recursive state prediction model for power demand and a dynamic recursive state prediction model for computing task load, wherein the constructed dynamic recursive state prediction model for power demand is as follows:

[0008] Where, is the forecast value of electricity demand; is the size of the historical data window, indicating the number of historical moments considered; Indicates the current moment, Indicates the next moment; It's a historic moment electricity demand; It's a historic moment The weighting coefficient of the electricity demand data indicates the influence of the past moments on the electricity demand forecast; It's a historic moment The dynamic adjustment coefficient of the power demand data is adjusted in real time as the load fluctuation rate changes; The dynamic recursive state prediction model of the computing task load is constructed as follows:

[0009] Where, is the predicted value of the computational task load; is the size of the historical data window, indicating the number of historical moments considered; Indicates the current moment, Indicates the next moment; It's a historic moment Computational task load; It's a historic moment The weighting coefficient of the computing task load data indicates the influence of the past time on the computing task load prediction; It's a historic moment The dynamic adjustment coefficient of the computing task load data is used to adapt to the rate of fluctuation of the computing task load.

[0010] Furthermore, the power demand and computing task load at the current moment are preliminarily scheduled according to the prediction result, including preliminarily scheduling the power demand and computing task load at the current moment by the following formula:

[0011]

[0012] Where, and It is at the moment Preliminary allocation of power requirements and computing task loads; and are the predicted values ​​of power demand and computing task load; is the real-time load adaptation factor, which is used to dynamically adjust power demand scheduling according to real-time load changes to cope with load fluctuations; and Separate moments The maximum power demand limit and the maximum computing task load limit are set to ensure that the system is not overloaded.

[0013] Furthermore, based on the preliminary scheduling, the scheduling of the power demand and the computing task load at the current moment is collaboratively optimized according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load. The objective function of determining the scheduling error between the power demand and the computing task load is included, and the allocation strategy is obtained by minimizing the objective function, wherein the objective function is:

[0014] Where, is the objective function of scheduling optimization, which is used to measure the power demand and computing workloads The scheduling error is calculated and penalized, with the goal of minimizing the scheduling error. is the total length of the scheduling period, indicating the number of moments in the entire scheduling cycle; and is the Lagrange coefficient, which is used to control the penalty for power demand scheduling error and computing task load scheduling error.

[0015] Furthermore, the dynamically adjusting the scheduling result by using a stability correction factor based on the real-time power system parameter information and the real-time computing task information includes determining the stability correction factor, wherein the formula of the stability correction factor is:

[0016] Where, is the stability correction factor, which means that at time The scheduling stability adjustment factor of is the stability correction factor, which is used to control the influence of the error on the stability correction factor. is a weight coefficient used to adjust the influence of the difference between the current actual load and the future predicted load on the stability correction. and It is the time obtained based on real-time power system parameter information and real-time calculation task information The power demand and computing task load.

[0017] Furthermore, the dynamically adjusting the scheduling result by the stability correction factor based on the real-time power system parameter information and the real-time computing task information includes determining the final power demand and computing task load scheduling result using the following formula:

[0018]

[0019] Where, and It is the minimization of the scheduling optimization objective function, and the optimized power demand scheduling and computing task load scheduling are obtained. is the stability correction factor.

[0020] Furthermore, the scheduling of the power demand and computing task load at the current moment is collaboratively optimized according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, including using constraints to constrain the optimization process, wherein the constraints include power demand constraints, computing task timeliness constraints, and system stability constraints.

[0021] Furthermore, the power demand constraint, computing task timeliness constraint, and system stability constraint are: Power demand constraints:

[0022] Computational task time constraints:

[0023] System stability constraints:

[0024]

[0025] Where, It's time electricity demand, For the moment Computational task load; It's time Maximum power demand limit; It's time The execution time of the corresponding computing task; is the maximum timeliness requirement of the computing task, that is, the maximum tolerable execution time of each task; is the tolerance of the power load error, indicating the time Power demand and predicted value The maximum error between is the tolerance of the prediction error of the computing task, indicating the time Computational workload and predicted value The maximum error between .

[0026] The present invention provides a computer-electricity collaborative optimization system based on digital twins, which includes a data acquisition module, a prediction model construction module, a preliminary scheduling module, a resource scheduling optimization module, and a scheduling solution output module; The data acquisition module is configured to acquire historical power system parameter information and historical calculation task information within a first historical period, historical power system parameter information and historical calculation task information within a second historical period, and real-time power system parameter information and real-time calculation task information; The prediction model construction module is used to use the historical power system parameter information and historical computing task information within the first historical period to construct a dynamic recursive state prediction model based on digital twins; The preliminary scheduling module is used to use the constructed dynamic recursive state prediction model to predict the power demand and computing task load at the next moment based on the historical power system parameter information and historical computing task information in the second historical period, and to perform preliminary scheduling of the power demand and computing task load at the current moment according to the prediction results; The resource scheduling optimization module is used to collaboratively optimize the scheduling of the current power demand and computing task load based on the preliminary scheduling according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjust the scheduling result through a stability correction factor based on real-time power system parameter information and real-time computing task information; The scheduling solution output module is used to output the adjusted scheduling result.

[0027] Furthermore, the data acquisition module includes a first historical data acquisition unit, a second historical data acquisition unit and a real-time data acquisition unit, wherein: The first historical data acquisition unit is used to acquire historical power system parameter information and historical calculation task information within a first historical period; The second historical data acquisition unit is used to acquire historical power system parameter information and historical calculation task information within a second historical period; The real-time data acquisition unit is used to acquire real-time power system parameter information and real-time computing task information.

[0028] Furthermore, the preliminary scheduling module includes a prediction unit and a scheduling unit, wherein: The prediction unit is configured to use the constructed dynamic recursive state prediction model to predict the power demand and computing task load at the next moment based on the historical power system parameter information and historical computing task information in the second historical period, and transmit the prediction result to the scheduling unit; The scheduling unit is used to perform preliminary scheduling of the power demand and computing task load at the current moment according to the prediction result of the prediction unit.

[0029] Furthermore, the resource scheduling optimization module includes an optimization unit and a constraint unit, wherein: The optimization unit is configured to collaboratively optimize the scheduling of the current power demand and computing task load based on the preliminary scheduling according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjust the scheduling result through a stability correction factor based on real-time power system parameter information and real-time computing task information; The constraint unit is used to constrain the scheduling result using constraint conditions.

[0030] The present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the following method: Acquire historical power system parameter information and historical calculation task information within a first historical period, historical power system parameter information and historical calculation task information within a second historical period, and real-time power system parameter information and real-time calculation task information; Using the historical power system parameter information and historical computing task information in the first historical period, a dynamic recursive state prediction model based on digital twins is constructed; Using the constructed dynamic recursive state prediction model, based on the historical power system parameter information and historical computing task information in the second historical period, the power demand and computing task load at the next moment are predicted, and according to the prediction results, the power demand and computing task load at the current moment are preliminarily scheduled; Based on the preliminary scheduling, the scheduling of the current power demand and computing task load is collaboratively optimized according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjusts the scheduling result through a stability correction factor based on real-time power system parameter information and real-time computing task information.

[0031] The present invention provides a storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the following method are implemented: Acquire historical power system parameter information and historical calculation task information within a first historical period, historical power system parameter information and historical calculation task information within a second historical period, and real-time power system parameter information and real-time calculation task information; Using the historical power system parameter information and historical computing task information in the first historical period, a dynamic recursive state prediction model based on digital twins is constructed; Using the constructed dynamic recursive state prediction model, based on the historical power system parameter information and historical computing task information in the second historical period, the power demand and computing task load at the next moment are predicted, and according to the prediction results, the power demand and computing task load at the current moment are preliminarily scheduled; Based on the preliminary scheduling, the scheduling of the current power demand and computing task load is collaboratively optimized according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjusts the scheduling result through a stability correction factor based on real-time power system parameter information and real-time computing task information.

[0032] The present invention provides a computer-electricity collaborative optimization method, system, electronic device and storage medium based on digital twins. By accurately predicting power demand and computing task load and performing intelligent scheduling based on the prediction results, it can more efficiently allocate the current power demand and computing task load, avoid energy waste and excessive use of computing resources, and thus significantly improve the energy efficiency of the system; by introducing a stability correction factor, the scheduling plan can be dynamically adjusted to cope with fluctuations in power load and computing task load; the real-time monitoring and feedback mechanism enables the system to maintain stable operation when the load changes, ensuring the effective distribution of power demand and computing task load at the current moment. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a flow chart of the computer-electrical collaborative optimization method based on digital twins according to the first embodiment of the present invention.

[0034] Figure 2 It is a structural diagram of the computer-electricity collaborative optimization system based on digital twins in the second embodiment of the present invention.

[0035] Explanation of the accompanying symbols: 10. Data acquisition module; 20. Prediction model construction module; 30. Preliminary scheduling module; 40. Resource scheduling optimization module; 50. Scheduling plan output module. DETAILED DESCRIPTION

[0036] The following is a combination of specific embodiments and appendix Figure 1-Figure 2 The invention is described in detail so that those skilled in the art can more fully understand the purpose, features and effects of the invention.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the art to which the invention belongs. In the event that the definition of a term in the present invention conflicts with the meaning commonly understood by those skilled in the art to which the invention belongs, the definition in the present invention shall prevail.

[0038] Existing power dispatching relies on historical data or static models, and often ignores the dynamic characteristics of load fluctuations. As a result, when the power load fluctuates significantly, the system cannot adjust the dispatching strategy in time, resulting in over-dispatching or under-dispatching.

[0039] Furthermore, the widespread adoption of renewable energy, particularly fluctuating energy sources like wind and solar, has led to increased uncertainty and challenges in power system scheduling. Traditional scheduling methods lack flexibility and cannot effectively integrate the fluctuating nature of renewable energy with the real-time scheduling needs of green energy. This leads to uneven resource utilization and compromises system stability and sustainability.

[0040] The rise of smart microgrids and green data centers requires that power systems be more efficiently coordinated with computing workload scheduling to achieve optimal system operation. This invention provides a digital twin-based computing-electricity collaborative optimization method, system, electronic device, and storage medium. Based on digital twin technology, this method improves the intelligence level of power demand and computing workload scheduling through real-time data monitoring and prediction models, thereby achieving more efficient and stable system operation.

[0041] Example 1 As a specific embodiment of the present invention, this embodiment provides a computer-electricity collaborative optimization method based on digital twins, referring to Figure 1 , the specific steps are as follows: S100, acquiring historical power system parameter information and historical computing task information within a first historical period, historical power system parameter information and historical computing task information within a second historical period, and real-time power system parameter information and real-time computing task information; S200, using historical power system parameter information and historical computing task information within a first historical period, constructing a dynamic recursive state prediction model based on digital twins; S300, using the constructed dynamic recursive state prediction model, based on historical power system parameter information and historical computing task information in a second historical period, predicting the power demand and computing task load at the next moment, and performing preliminary scheduling of the power demand and computing task load at the current moment according to the prediction results; S400. Based on the preliminary scheduling, the scheduling of the current power demand and the computing task load is collaboratively optimized according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjusts the scheduling result through a stability correction factor based on real-time power system parameter information and real-time computing task information.

[0042] The digital twin-based computing-electricity collaborative optimization solution of this embodiment can more efficiently allocate the current power demand and computing task load by accurately predicting power demand and computing task load, and performing intelligent scheduling based on the prediction results, thereby avoiding energy waste and excessive use of computing resources, and thus significantly improving the energy efficiency of the system; by introducing a stability correction factor, the scheduling plan can be dynamically adjusted to cope with fluctuations in power load and computing task load; the real-time monitoring and feedback mechanism enables the system to maintain stable operation when the load changes, ensuring the effective distribution of power demand and computing task load at the current moment.

[0043] Furthermore, in S100, the power system parameter information includes data such as power load, current, voltage, power, and temperature; and the computing task information includes data such as the execution progress, load distribution, response time, and energy consumption of the computing task.

[0044] Various sensors deployed in the power system and data center provide real-time access to power system parameter information such as power load, current, voltage, power, and temperature, as well as computing task information such as execution progress, load distribution, response time, and energy consumption. In one feasible embodiment, sensors deployed in the power system primarily collect operational data from the power grid, recording power load fluctuations, current and voltage changes, and changes in renewable energy generation during each time period. Sensors deployed in the data center primarily collect computing resource usage, recording computing task execution progress, load distribution, response time, and energy consumption.

[0045] In a feasible embodiment, the power system parameter information can be collected through smart electricity meters, current transformers, voltage sensors, frequency collectors, temperature sensors and energy storage system monitoring modules, including current, voltage, active power, reactive power, frequency, equipment temperature, energy storage capacity, load type identification and other data; the computing task information is collected through server management chips (such as BMC), operating system monitoring interfaces (such as SNMP), high-performance counters and task scheduling monitoring modules, including task arrival rate, task priority, computing node CPU / GPU usage, memory occupancy, node temperature, execution delay, resource waiting time and task completion rate and other data.

[0046] Historical power system parameter information and historical calculation task information record the past operation data of the power system and data center, which can be obtained from databases, log systems, and equipment collection records.

[0047] The method of constructing a dynamic recursive state prediction model based on digital twins by utilizing historical power system parameter information and historical computing task information within the first historical period includes constructing a dynamic recursive state prediction model of power demand and a dynamic recursive state prediction model of computing task load.

[0048] The digital twin model in this embodiment maps the physical world's power systems and data centers, constructing a high-fidelity dynamic equivalent model. Multiple sensors deployed on-site continuously and synchronously capture the current operating status. These sensors collect data on physical quantities such as power load, current, voltage, power, temperature, execution progress, load distribution, response time, and energy consumption, which is then fed into the virtual model in real time, ensuring that the digital twin model remains consistent with the actual system over time.

[0049] In this embodiment, power demand refers to the electrical load, which reflects the power required by the system at a certain moment; computing task load refers to the comprehensive reflection of the number, complexity and resource consumption intensity of computing tasks that the data center needs to process within a certain period of time, including the number of concurrent tasks, task complexity, resource usage, and task execution time.

[0050] This embodiment adopts a dynamic recursive state prediction modeling mechanism assisted by digital twins. Based on the original historical data, the digital twin model is used to normalize, dimensionally compress, and nonlinearly transform each variable to obtain a set of state variable sequences that can be used for modeling. A weighted influence function and disturbance enhancement function are constructed for them, thereby constructing a prediction expression with state recursive memory characteristics.

[0051] The dynamic recursive state prediction model of the power demand is constructed as follows:

[0052] Where, is the forecast value of electricity demand; is the size of the historical data window, indicating the number of historical moments considered; Indicates the current moment, Indicates the next moment; It's a historic moment The power demand reflects the power load at the past moment; It's a historic moment The weighting coefficient of the power demand data indicates the influence of the past moment on the power demand forecast, which is obtained by fitting historical data or using a machine learning algorithm; It's a historic moment The dynamic adjustment coefficient of the power demand data is adjusted in real time as the load fluctuation rate changes. By monitoring the power load fluctuation and the changes in renewable energy generation, the coefficient is dynamically adjusted to enable the model to respond to load fluctuations of different change speeds.

[0053] In this embodiment, the modeling of power load is not a simple superposition of historical load values, but the use of digital twin model to model the state at each moment. After encoding, weighted expression is performed, and then a dynamic disturbance term reflecting the changing trend is added .

[0054] By combining the changes in renewable energy power generation, it is possible to more reasonably allocate power resources, make full use of green energy, reduce dependence on traditional energy, and help promote the development of green electricity and smart microgrids.

[0055] The dynamic recursive state prediction model of the computing task load is constructed as follows:

[0056] Where, is the predicted value of the computational task load; is the size of the historical data window, indicating the number of historical moments considered; Indicates the current moment, Indicates the next moment; It's a historic moment The computing task load is obtained based on the actual operation records of the system; It's a historic moment The weighting coefficient of the computing task load data indicates the influence of the past time on the computing task load prediction; It's a historic moment The dynamic adjustment coefficient of the computing task load data is used to adapt to the rate of fluctuation of the computing task load.

[0057] In this embodiment, the state sequence is constructed by modeling the computing task load. , and introduce dynamic nonlinear adjustment coefficient , in order to simulate non-stationary factors such as sudden tasks and resource congestion in real systems.

[0058] By combining the historical computing task load and the fluctuation situation, the weighting coefficient and the dynamic adjustment coefficient in the dynamic recursive state prediction model of the computing task load are adjusted.

[0059] All input and It is not a directly sampled value, but a multi-parameter combination deduction result output by the digital twin model - that is, an equivalent "system state variable" formed on the basis of real-time simulation of the real system, combined with factors such as power fluctuation trends, task migration rules, and resource temperature coupling characteristics. These state variables become the core driver of the recursive model after embedded transformation and dynamic response modeling.

[0060] The model also provides a dynamic calibration mechanism. Whenever the prediction error deviates from the stability threshold set by the system, the digital twin will automatically call back the historical simulation state and readjust the weighting coefficients in the prediction parameters. , and dynamic adjustment coefficient , , ensuring that the model's responsiveness adapts over time. This is achieved through multi-period residual detection and sliding window regression iteration. The process does not rely on neural networks, but is based on state residual field modeling and parameter perturbation response analysis.

[0061] Although the dynamic recursive state prediction model is based on historical data for training, it needs to continuously introduce real-time data streams during operation to compare the error between the predicted value and the actual value and dynamically adjust the prediction coefficient. , timely capture non-periodic events (such as sudden computing tasks, local power grid jumps) to achieve model adaptation, otherwise the model will always be based on the past and will be difficult to cope with system mutations.

[0062] The dynamic recursive state prediction model driven by digital twins combines physical consistency, timing correlation and adaptive adjustment capabilities, and can achieve high-precision prediction of future power demand and computing task load. It has strong adaptability and good stability, and is particularly suitable for collaborative optimization scheduling scenarios with strong time-varying and complex coupling behaviors in green data centers and smart microgrids.

[0063] In this embodiment, a dynamic recursive state prediction model is constructed using digital twins to jointly forecast power demand and computing task load. To better predict future power demand and computing task load, this model considers the impact of historical data on the current moment and adapts to fluctuations in power and computing task loads through a dynamic adjustment coefficient. Power demand forecasting is achieved by weighting power demand data from the past period of time, and a dynamic adjustment coefficient is introduced to enable the forecast to adjust accordingly based on the rate of change in past load.

[0064] Furthermore, in S200, performing preliminary scheduling of the power demand and computing task load at the current moment according to the prediction result includes performing preliminary scheduling of the power demand and computing task load at the current moment according to the following formula:

[0065]

[0066] Where, and It is at the present moment Preliminary allocation of power requirements and computing task loads; and are the predicted values ​​of power demand and computing task load; It is a real-time load adaptation factor used to dynamically adjust power demand scheduling according to real-time load changes to cope with load fluctuations. It is obtained based on the current load change rate and the historical load fluctuation amplitude. It is measured by normalizing the increase in the current load compared to the previous moment, and taking into account the degree of fluctuation of the load from the average level (i.e., standard deviation) over a period of time. The adjustment coefficient is dynamically calculated according to the preset weight and is used to flexibly adjust the scheduling amplification ratio of the predicted load; and The current moment The maximum power demand limit and the maximum computing task load limit are set to ensure that the system is not overloaded. and Can be calculated based on historical data.

[0067] After obtaining the predicted results for future power demand and computing task load, respectively, using the dynamic recursive state prediction model for power demand and the dynamic recursive state prediction model for computing task load constructed in S100, resource scheduling optimization is performed. First, preliminary scheduling is performed. The goal of preliminary scheduling is to allocate preliminary power demand and computing task load for each moment based on the predicted results, perform preliminary resource allocation, and ensure that the power demand and computing task load at each moment are met to a certain extent in the preliminary scheduling. To ensure the rationality of preliminary scheduling, a real-time load adaptation factor is added to the preliminary scheduling formula.

[0068] The dynamic adjustment coefficient and real-time load adaptation factor can be flexibly adjusted according to the fluctuations of power load and computing task load, ensuring that the system can respond quickly to load changes and avoid over-scheduling or under-scheduling.

[0069] Furthermore, in S300, based on the initial scheduling, minimizing scheduling errors and the synergy between resources are considered to further optimize the current distribution of power demand and computing task load. A Lagrangian optimization algorithm is used to collaboratively optimize the scheduling of power demand and computing task load. The optimization goal is to minimize the deviation between power demand and computing task load scheduling while ensuring system stability through constraints. By introducing the Lagrangian multiplier method, the scheduling error between power demand and computing task load in the optimization problem is quantified as a penalty term, and the optimal resource allocation strategy is obtained by minimizing the objective function.

[0070] Specifically, the objective function of the scheduling error between the power demand and the computing task load is:

[0071] Where, is the objective function of scheduling optimization, which is used to measure the power demand and computing workloads The scheduling error is calculated and penalized, with the goal of minimizing the scheduling error to achieve more efficient resource scheduling. is the total length of the scheduling period, indicating the number of moments in the entire scheduling cycle; and is the Lagrange coefficient, which is used to control the penalty for power demand scheduling error and computing task load scheduling error.

[0072] The objective function calculates the scheduling error between power demand and computing task load, where the scheduling deviation at each time step is squared and weighted with the corresponding Lagrange coefficient. The goal is to balance the penalty for scheduling error by adjusting the coefficient so that the resource allocation of power and computing tasks is as close as possible to the initial scheduling value, thereby achieving optimal resource scheduling.

[0073] By minimizing the dispatch optimization objective function, the optimized power demand dispatch quantity is obtained and computing task load scheduling .

[0074] Preferably, after obtaining the optimized scheduling result and Finally, the scheduling results are adjusted by the stability correction factor to ensure stable system operation and adapt to load fluctuations. The stability correction factor is dynamically adjusted using the following formula:

[0075] Where, is the stability correction factor, which means that at time The scheduling stability adjustment factor of is the stability correction factor, which is used to control the influence of the error on the stability correction factor. is a weight coefficient used to adjust the influence of the difference between the current actual load and the future predicted load on the stability correction; and It is the time obtained based on real-time power system parameter information and real-time calculation task information The power demand and computing task load.

[0076] The dynamic adjustment of the scheduling result by the stability correction factor includes determining the final power demand and computing task load scheduling result using the following formula:

[0077]

[0078] In order to ensure the stability of the system, the optimization process is constrained. In this embodiment, the constraints include power demand constraints, computing task timeliness constraints, and system stability constraints.

[0079] Specifically, the constraints are as follows: Power demand constraints:

[0080] Computational task time constraints:

[0081] System stability constraints:

[0082]

[0083] Where, It's time electricity demand; It's time Maximum power demand limit; It's time The execution time of the corresponding computing task; is the maximum timeliness requirement of the computing task, that is, the maximum tolerable execution time of each task; is the tolerance of the power load error, indicating the time Power demand and predicted value The maximum error between is the tolerance of the prediction error of the computing task, indicating the time Computational workload and predicted value The maximum error between the current moment Scheduling and next moment The stability constraint between the predictions of The scheduled power demand and computing task load cannot deviate too far from future predictions to ensure that the scheduling has predictive consistency and forward stability of the system.

[0084] Constraints ensure that power dispatch does not exceed the system's maximum supply capacity , and the computing tasks can be completed on time, and the stability of the system is guaranteed. The constraints are combined with the objective function to form an overall optimization problem, which aims to minimize the error and satisfy the above constraints at the same time, and finally obtain the optimal scheduling result that meets all requirements and .

[0085] By using the Lagrangian optimization algorithm, which comprehensively considers the scheduling errors of power demand and computing task load, the allocation plan of power resources and computing task load can be accurately optimized, reducing resource waste and making the scheduling results closer to the optimal state. It not only optimizes the scheduling plan, but also ensures that constraints such as power demand and computing task timeliness are met, avoiding system instability caused by load overload or task delay, and improving the reliability and sustainability of the system.

[0086] Furthermore, according to the final power demand at the current moment and the computing task load scheduling result calculated in S300 and ,outputs the optimized scheduling scheme, which ensures the optimal distribution of power demand and computing task load, and satisfies the system stability and load constraints. In this embodiment, the final power demand and computing task load scheduling results are and It is the current moment The scheduling output of the next moment is used The purpose of the prediction result is to serve as an adjustment reference so that the current moment The scheduling output is more robust and forward-looking.

[0087] The digital twin-based computing-electricity collaborative optimization method provided in this implementation, based on preliminary scheduling, collaboratively optimizes the scheduling of power demand and computing task load through the Lagrangian optimization algorithm, minimizes scheduling errors and balances the synergy between resources, and dynamically adjusts the scheduling results by introducing a stability correction factor, taking into account the constraints of power demand, computing task timeliness and system stability, and ultimately obtaining the optimal scheduling solution that meets all constraints.

[0088] The digital twin-based computer-electricity collaborative optimization method provided in this embodiment solves the problem that the dynamic characteristics of load fluctuations are usually ignored in the existing technology, resulting in the system being unable to adjust the scheduling strategy in time when the power load fluctuates greatly, thus causing over-scheduling or under-scheduling; and the traditional scheduling method lacks flexibility and cannot effectively integrate the fluctuating characteristics of renewable energy and the real-time scheduling needs of green energy, resulting in uneven resource utilization and affecting the stability and sustainability of the system.

[0089] Example 2 As another specific embodiment of the present invention, this embodiment provides a computer-electricity collaborative optimization system based on digital twins, referring to Figure 2 , including a data acquisition module 10, a prediction model construction module 20, a preliminary scheduling module 30, a resource scheduling optimization module 40, and a scheduling plan output module 50.

[0090] The data acquisition module 10 is used to acquire historical power system parameter information and historical calculation task information within a first historical period, historical power system parameter information and historical calculation task information within a second historical period, and real-time power system parameter information and real-time calculation task information; The prediction model building module 20 is used to build a dynamic recursive state prediction model based on digital twins by using the historical power system parameter information and historical computing task information within the first historical period; The preliminary scheduling module 30 is configured to use the constructed dynamic recursive state prediction model to predict the power demand and computing task load at the next moment based on the historical power system parameter information and historical computing task information in the second historical period, and to perform preliminary scheduling of the power demand and computing task load at the current moment based on the prediction results; The resource scheduling optimization module 40 is configured to collaboratively optimize the scheduling of the current power demand and computing task load based on the preliminary scheduling using a Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjust the scheduling result using a stability correction factor based on real-time power system parameter information and real-time computing task information; The scheduling solution output module 50 is used to output the adjusted scheduling result.

[0091] Furthermore, the data acquisition module 10 includes a first historical data acquisition unit, a second historical data acquisition unit and a real-time data acquisition unit, wherein: The first historical data acquisition unit is used to acquire historical power system parameter information and historical calculation task information within a first historical period; The second historical data acquisition unit is used to acquire historical power system parameter information and historical calculation task information within a second historical period; The real-time data acquisition unit is used to acquire real-time power system parameter information and real-time computing task information. Furthermore, the preliminary scheduling module 30 includes a prediction unit and a scheduling unit, wherein: The prediction unit is configured to use the constructed dynamic recursive state prediction model to predict the power demand and computing task load at the next moment based on the historical power system parameter information and historical computing task information in the second historical period, and transmit the prediction result to the scheduling unit; The scheduling unit is used to perform preliminary scheduling of the power demand and computing task load at the current moment according to the prediction result of the prediction unit.

[0092] The goal of preliminary scheduling is to allocate preliminary power demand and computing task load for each moment based on the prediction results, make preliminary allocation of resources, and ensure that the power demand and computing task load at each moment are met to a certain extent in the preliminary scheduling.

[0093] Furthermore, the resource scheduling optimization module 40 includes an optimization unit and a constraint unit, wherein: The optimization unit is configured to collaboratively optimize the scheduling of the current power demand and computing task load based on the preliminary scheduling according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjust the scheduling result through a stability correction factor based on real-time power system parameter information and real-time computing task information; The constraint unit is used to constrain the optimization process using constraint conditions.

[0094] The resource scheduling optimization module 40 of this embodiment uses the Lagrangian optimization algorithm to optimize resource scheduling, minimize the scheduling errors of power demand and computing task load, and consider the synergy between loads; through optimization, the power demand and computing task load scheduling results at the current moment are obtained, and a stability correction factor is introduced to ensure the stability of the scheduling results.

[0095] In order to ensure that the scheduling results are optimized under the premise of meeting power demand, computing task timeliness and system stability, constraints are used for verification during the optimization process to make the scheduling results feasible and stable.

[0096] Example 3 As another specific embodiment of the present invention, this embodiment provides an electronic device, including a memory, a processor, and a computer program stored on the memory, wherein the processor executes the computer program to implement the steps of the computer-electronic collaborative optimization method based on digital twins described in Example 1.

[0097] Example 4 As another specific embodiment of the present invention, this embodiment provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the computer-electrical collaborative optimization method based on digital twins described in Example 1 are implemented.

[0098] The above description is merely a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent variation based on the technical essence of the present invention shall still fall within the scope of protection claimed by the present invention.

Claims

1. A computer-electronics collaborative optimization method based on digital twins, characterized in that: The method comprises: Acquire historical power system parameter information and historical calculation task information within a first historical period, historical power system parameter information and historical calculation task information within a second historical period, and real-time power system parameter information and real-time calculation task information; Using the historical power system parameter information and historical computing task information in the first historical period, a dynamic recursive state prediction model based on digital twins is constructed; Using the constructed dynamic recursive state prediction model, based on the historical power system parameter information and historical computing task information in the second historical period, the power demand and computing task load at the next moment are predicted, and according to the prediction results, the power demand and computing task load at the current moment are preliminarily scheduled; Based on the preliminary scheduling, the scheduling of the current power demand and computing task load is collaboratively optimized according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjusts the scheduling result through a stability correction factor based on real-time power system parameter information and real-time computing task information.

2. The computer-electronics collaborative optimization method based on digital twins according to claim 1 is characterized in that: The power system parameter information includes power load, current, voltage, power, and temperature, and the computing task information includes the execution progress, load distribution, response time, and energy consumption of the computing task.

3. The computer-electronics collaborative optimization method based on digital twins according to claim 1 is characterized in that: The construction of a dynamic recursive state prediction model based on digital twins using historical power system parameter information and historical computing task information within the first historical period includes constructing a dynamic recursive state prediction model for power demand and a dynamic recursive state prediction model for computing task load, wherein the constructed dynamic recursive state prediction model for power demand is as follows: Where, is the forecast value of electricity demand; is the size of the historical data window, indicating the number of historical moments considered; Indicates the current moment, Indicates the next moment; It's a historic moment electricity demand; It's a historic moment The weighting coefficient of the electricity demand data indicates the influence of the past moments on the electricity demand forecast; It's a historic moment The dynamic adjustment coefficient of the power demand data is adjusted in real time as the load fluctuation rate changes; The dynamic recursive state prediction model of the computing task load is constructed as follows: Where, is the predicted value of the computational task load; is the size of the historical data window, indicating the number of historical moments considered; Indicates the current moment, Indicates the next moment; It's a historic moment Computational task load; It's a historic moment The weighting coefficient of the computing task load data indicates the influence of the past time on the computing task load prediction; It's a historic moment The dynamic adjustment coefficient of the computing task load data is used to adapt to the rate of fluctuation of the computing task load.

4. The computer-electronics collaborative optimization method based on digital twins according to claim 1, characterized in that: The power demand and computing task load at the current moment are preliminarily scheduled according to the prediction results, including preliminarily scheduling the power demand and computing task load at the current moment by using the following formula: Where, and It is at the moment Preliminary allocation of power requirements and computing task loads; and are the predicted values ​​of power demand and computing task load; is the real-time load adaptation factor, which is used to dynamically adjust power demand scheduling according to real-time load changes to cope with load fluctuations; and Separate moments The maximum power demand limit and the maximum computing task load limit are set to ensure that the system is not overloaded.

5. The computer-electronics collaborative optimization method based on digital twins according to claim 4 is characterized in that: Based on the preliminary scheduling, the scheduling of the power demand and the computing task load at the current moment is collaboratively optimized according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load. The objective function of determining the scheduling error between the power demand and the computing task load is included, and the allocation strategy is obtained by minimizing the objective function, wherein the objective function is: Where, is the objective function of scheduling optimization, which is used to measure the power demand and computing workloads The scheduling error is calculated and penalized, with the goal of minimizing the scheduling error. is the total length of the scheduling period, indicating the number of moments in the entire scheduling cycle; and is the Lagrange coefficient, which is used to control the penalty for power demand scheduling error and computing task load scheduling error.

6. A computer-electricity collaborative optimization system based on digital twins, characterized in that: The system includes a data acquisition module (10), a prediction model construction module (20), a preliminary scheduling module (30), a resource scheduling optimization module (40), and a scheduling solution output module (50); The data acquisition module (10) is used to acquire historical power system parameter information and historical calculation task information within a first historical period, historical power system parameter information and historical calculation task information within a second historical period, and real-time power system parameter information and real-time calculation task information; The prediction model construction module (20) is used to construct a dynamic recursive state prediction model based on digital twins by using the historical power system parameter information and historical calculation task information within the first historical period; The preliminary scheduling module (30) is used to use the constructed dynamic recursive state prediction model to predict the power demand and computing task load at the next moment based on the historical power system parameter information and historical computing task information in the second historical period, and to perform preliminary scheduling of the power demand and computing task load at the current moment according to the prediction result; The resource scheduling optimization module (40) is used to collaboratively optimize the scheduling of the power demand and the computing task load at the current moment based on the preliminary scheduling according to the Lagrangian optimization algorithm to generate a scheduling result that minimizes the scheduling deviation between the power demand and the computing task load, and dynamically adjust the scheduling result through a stability correction factor based on real-time power system parameter information and real-time computing task information; The scheduling scheme output module (50) is used to output the adjusted scheduling result.

7. The computer-electronics collaborative optimization system based on digital twin according to claim 6, characterized in that: The data acquisition module (10) comprises a first historical data acquisition unit, a second historical data acquisition unit and a real-time data acquisition unit, wherein: The first historical data acquisition unit is used to acquire historical power system parameter information and historical calculation task information within a first historical period; The second historical data acquisition unit is used to acquire historical power system parameter information and historical calculation task information within a second historical period; The real-time data acquisition unit is used to acquire real-time power system parameter information and real-time computing task information.

8. The computer-electronics collaborative optimization system based on digital twin according to claim 6, characterized in that: The preliminary scheduling module (30) includes a prediction unit and a scheduling unit, wherein: The prediction unit is configured to use the constructed dynamic recursive state prediction model to predict the power demand and computing task load at the next moment based on the historical power system parameter information and historical computing task information in the second historical period, and transmit the prediction result to the scheduling unit; The scheduling unit is used to perform preliminary scheduling of the power demand and computing task load at the current moment according to the prediction result of the prediction unit.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory, wherein: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 5.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images