Power grid load scheduling method and device and storage medium

Abstract

The invention discloses a power grid load scheduling method and device and a storage medium, and the method comprises the steps: collecting first target load data for electrical equipment under the jurisdiction of an edge computing node in a power grid in the edge computing node, and collecting second target load data for electrical equipment under the jurisdiction of the edge computing node in the edge computing node; and inputting the load data into a preset short-term load prediction model, predicting future second target load data, generating a preliminary load scheduling scheme in the edge computing node according to the first target load data and the second target load data, transmitting the preliminary load scheduling scheme to a cloud computing center, and transmitting the preliminary load scheduling scheme to the cloud computing center. The initial load scheduling scheme is corrected into a global load scheduling scheme according to the operation state of the power grid, and the global load scheduling scheme is transmitted to each edge computing node to be executed in the cloud computing center so as to schedule the load of the power grid, so that the load scheduling efficiency and stability of the power grid are improved; and balanced distribution of the whole network load is realized through dynamic adjustment.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of power system planning, and in particular to a power grid load scheduling method, device and storage medium. Background Art

[0002] With the increasing complexity of power systems and energy demand, traditional centralized power dispatching methods are becoming increasingly difficult to cope with the growing load fluctuations and dynamic adjustment needs. The increasing uncertainty and volatility of power loads have brought new challenges to the stable operation of power systems.

[0003] Traditional power dispatching relies on centralized control, with the central dispatching center making dispatching decisions based on the load data of the entire network. Due to the large amount of load data in the entire network, the traditional power dispatching method has high computational complexity, slow response speed and low efficiency. Summary of the invention

[0004] The present invention provides a power grid load dispatching method, equipment and storage medium, which are used to improve the power grid load dispatching efficiency.

[0005] In a first aspect, an embodiment of the present invention provides a power grid load scheduling method, wherein the power Internet of Things has multiple edge computing nodes and a cloud computing center, including:

[0006] In the edge computing node, first target load data is collected from electrical equipment under the edge computing node in the power grid;

[0007] In the edge computing node, the load data is input into a preset short-term load forecasting model to predict a second target load data in the future;

[0008] In the edge computing node, a preliminary load scheduling plan is generated according to the first target load data and the second target load data, and the preliminary load scheduling plan is transmitted to the cloud computing center;

[0009] In the cloud computing center, the preliminary load scheduling plan is revised into a global load scheduling plan according to the operating status of the power grid;

[0010] In the cloud computing center, the global load scheduling scheme is transmitted to each of the edge computing nodes for execution, so as to schedule the load of the power grid.

[0011] In a second aspect, an embodiment of the present invention further provides a power grid load dispatching device, wherein the power Internet of Things has multiple edge computing nodes and a cloud computing center, including:

[0012] A first target load data collection module is used to collect first target load data from electrical equipment under the edge computing node in the power grid in the edge computing node;

[0013] A second target load data prediction module is used to input the load data into a preset short-term load prediction model in the edge computing node to predict future second target load data;

[0014] a preliminary load scheduling scheme generating module, configured to generate a preliminary load scheduling scheme in the edge computing node according to the first target load data and the second target load data, and transmit the preliminary load scheduling scheme to the cloud computing center;

[0015] A global load scheduling scheme correction module is used to correct the preliminary load scheduling scheme into a global load scheduling scheme according to the operating status of the power grid in the cloud computing center;

[0016] A global load scheduling scheme transmission module is used to transmit the global load scheduling scheme to each edge computing node in the cloud computing center for execution, so as to schedule the load of the power grid.

[0017] In a third aspect, an embodiment of the present invention further provides a computer device, including:

[0018] one or more processors;

[0019] A storage device for storing one or more programs;

[0020] When the one or more programs are executed by the one or more processors, the one or more processors implement the power grid load scheduling method provided in the first aspect of the present invention.

[0021] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the power grid load scheduling method provided in the first aspect of the present invention.

[0022] In a fifth aspect, an embodiment of the present invention further provides a computer program product, wherein the computer program product includes a computer program, and when the computer program is executed by a processor, the power grid load scheduling method provided in the first aspect of the present invention is implemented.

[0023] In this embodiment, the electric power Internet of Things has multiple edge computing nodes and a cloud computing center. In the edge computing node, first target load data is collected for electrical equipment under the edge computing node in the power grid. In the edge computing node, the load data is input into a preset short-term load forecasting model to predict the future second target load data. In the edge computing node, a preliminary load scheduling plan is generated based on the first target load data and the second target load data, and the preliminary load scheduling plan is transmitted to the cloud computing center. In the cloud computing center, the preliminary load scheduling plan is revised into a global load scheduling plan based on the operating status of the power grid. In the cloud computing center, the global load scheduling plan is transmitted to each edge computing node for execution to schedule the load of the power grid. In this embodiment, multiple edge computing nodes cooperate with the cloud computing center. The edge computing nodes are responsible for local load scheduling, and the cloud computing center is responsible for global optimization, thereby realizing distributed load scheduling optimization. The edge computing nodes undertake a large amount of computing work, which greatly reduces the computing pressure of the cloud computing center, effectively improves the response speed, and improves the efficiency of load scheduling. Multiple edge computing nodes and the cloud computing center continuously monitor and dynamically schedule loads on electrical equipment in the power grid, respond to load fluctuations in a timely manner, avoid local overloads, improve the load scheduling efficiency and stability of the power grid, and dynamically adjust to achieve balanced distribution of loads across the entire network. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flow chart of a power grid load scheduling method provided in Embodiment 1 of the present invention;

[0025] Figure 2 A structural block diagram of a power grid load dispatching device provided in Embodiment 2 of the present invention;

[0026] Figure 3 A schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments 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 creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can cover sequential implementations other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] Embodiment 1

[0030] See also Figure 1 , shows a flow chart of a power grid load dispatching method provided by the first embodiment of the present invention, the method can be executed by a power grid load dispatching detection device, the power grid load dispatching detection device can be implemented in the form of hardware and / or software, and the power grid load dispatching detection device can be configured in a computer device. Figure 1 As shown, the method includes:

[0031] Step 101: In an edge computing node, first target load data is collected from electrical equipment under the edge computing node in a power grid.

[0032] The power grid is the infrastructure for power transmission and distribution, and the power Internet of Things uses advanced information technology and smart devices to transform the power grid into an intelligent one. The power Internet of Things collects real-time data from all aspects of the power grid and uses this data to optimize the operation of the power grid, improve reliability and flexibility, enhance the power grid's ability to respond to emergencies, and support two-way interaction between users and the power grid, thereby achieving more efficient and reliable power system management. The power Internet of Things has multiple edge computing nodes and cloud computing centers.

[0033] Edge computing nodes refer to computing units that are distributed at important points in the power system and close to data sources. These nodes are usually installed in substations, distribution networks, power plants, etc. Edge computing nodes play a bridge role in the power Internet of Things, combining local data processing with global data analysis, improving the efficiency and responsiveness of the system.

[0034] The cloud computing center refers to a core platform for centralized storage and processing of large-scale data. It supports the optimized management and intelligent decision-making of the power system by centralized processing and analysis of large amounts of data.

[0035] Jurisdiction refers to the scope or equipment under the jurisdiction of a management or control node. Specifically, the edge computing node is responsible for data processing and analysis at the edge of the power grid.

[0036] In this embodiment, in the edge computing node, the sensor installed on the edge computing node receives the first target load data collected by the electrical equipment deployed near the location of the edge computing node of the power grid. The first target load data includes environmental data (temperature, humidity, etc.) and timestamp (day of the week, holidays, etc.).

[0037] Step 102: In the edge computing node, the load data is input into a preset short-term load forecasting model to predict the second target load data in the future.

[0038] In this embodiment, a trained short-term load forecasting model is preset in the edge computing node, and the first target load data collected by the electrical equipment under the edge computing node is input into the trained short-term load forecasting model, and the second target load data in the future is predicted by the trained short-term load forecasting model. The second target load data is a future load data relative to the first target load data.

[0039] Short-term load forecasting models are based on neural networks and are designed to process time series data, capturing complex patterns and dependencies in time series through deep learning techniques. Such models usually include multiple layers, such as convolutional layers, recursive layers, or long short-term memory (LSTM) layers, which can effectively extract and learn features and patterns in historical data. Neural networks are trained on past load data to identify time dependencies and periodicity in the data, thereby predicting future loads. This approach can not only handle linear relationships, but also capture nonlinear relationships to improve forecasting accuracy. By continuously adjusting the network weights and structure, the model can continuously optimize the forecast results to adapt to changing load demands and environmental conditions.

[0040] In one embodiment of the present invention, the short-term load forecasting model includes an autoregressive integrated moving average (ARIMA) model and a long short-term memory network (LSTM), and step 102 may include the following steps:

[0041] Step 1011: pre-process the first target load data.

[0042] In this embodiment, first target load data collected from electrical equipment under the edge computing node is preprocessed, and the first target load data is load data collected in real time.

[0043] Preprocessing refers to standardizing or normalizing the first target load data to ensure consistency of data scale.

[0044] Generate time series data blocks through sliding window technology to provide continuous time series input for the trained short-term load forecasting model, and extract key features from the first target load data, such as the rate of change of voltage, current, power, and temperature change trend. Create derived features, such as the moving average of load, the interactive features of load and temperature, etc. Ensure that the range and distribution of all first target load data features are similar to improve the application effect of the short-term load forecasting model.

[0045] Step 1012: If the preprocessing is completed, the first target load data is input into the autoregressive integral moving average model to generate smoothed target load data.

[0046] In this embodiment, if the first target load data has completed preprocessing, the first target load data is input into a trained autoregressive integral moving average model. The trained autoregressive integral moving average model smoothes the first target load data to eliminate short-term fluctuations, so that the first target load data presents a clearer trend and periodic pattern, and generates smoothed target load data.

[0047] The Autoregressive Integrated Moving Average (ARIMA) model can handle non-stationary time series data, making the data stationary through differencing and capturing patterns in the data using autoregressive and moving average components. The Autoregressive Integrated Moving Average (ARIMA) model can provide accurate forecasts in many practical applications and can be used when the data has a clear trend and no seasonal component.

[0048] Step 1013: Input the smoothed target load data into the long short-term memory network to predict the future second target load data.

[0049] In this embodiment, the generated smooth target load data is input into a trained long short-term memory network. The trained long short-term memory network captures the long-term dependencies in the time series data, and predicts the future second target load data by analyzing and learning the smooth target load data.

[0050] Long short-term memory network (LSTM) is good at processing and predicting time series data, and can effectively capture long-term dependencies. By introducing a gating mechanism, LSTM can control the retention and forgetting of information, overcome the gradient vanishing and explosion problems in traditional recurrent neural networks, and make it more stable and accurate when processing complex sequence data. Long short-term memory network (LSTM) includes input layer, LSTM layer, fully connected layer and output layer. The input layer receives the smooth target load data time series and processes it in combination with environmental characteristics and time characteristics; the LSTM layer is used to process the long-term dependencies in the smooth target load data time series and capture the complex nonlinear relationships in the smooth target load data; the fully connected layer further processes the output of the LSTM layer to generate short-term load forecast results; the output layer generates the second target load data for a period of time in the future.

[0051] Exemplarily, the short-term load forecasting model operation structure is as follows:

[0052] The input layer includes power load time series data: length is 24 hours, one data point per hour, dimension is (24,1), environmental characteristics: including temperature, humidity, etc., one data point per hour, dimension is (24,2), time characteristics: hour (0-23), day (1-7), week (1-4), dimension is (24,3) and total input dimension: (24,6) (load data 1 dimension + environmental characteristics 2 dimensions + time characteristics 3 dimensions).

[0053] The ARIMA model layers include the order: p = 3 (autoregressive term), d = 1 (difference term), q = 2 (moving average term) and output: smoothed load time series with dimension (24, 1).

[0054] The LSTM model layers include:

[0055] (1) First layer LSTM: Number of units: 128 LSTM units; Activation function: tanh (state update), sigmoid (gating mechanism); Dropout (dropout rate): the dropout rate is set to 0.3.

[0056] (2) Second layer LSTM: Number of units: 64 LSTM units; Activation function: tanh (state update), sigmoid (gating mechanism); Dropout: the dropout rate is set to 0.3.

[0057] The LSTM layer output includes a shape of (24, 64) (the output of the second LSTM layer).

[0058] The fully connected layer includes the number of layers: 1 fully connected layer, the number of units: 24 (outputting the load forecast value for the next 24 hours), and the activation function: linear activation (linear).

[0059] The output layer includes the output of the power load forecast for the next 24 hours, with a dimension of (24,1).

[0060] In one embodiment of the present invention, training a short-term load forecasting model may include the following steps:

[0061] Step 1021: Collect first sample load data.

[0062] In this embodiment, first sample load data is collected from electrical equipment under the edge computing node in the power grid, and the first sample load data is labeled with label load data. The label load data is historical load data, which is future load data relative to the first sample load data, that is, the load data actually measured, and is used as the standard answer for training the short-term load forecasting model to help the short-term load forecasting model learn how to make accurate predictions.

[0063] Step 1022: pre-process the first sample load data.

[0064] In this embodiment, preprocessing refers to standardizing or normalizing the first sample load data to ensure consistency of data scale.

[0065] Generate time series data blocks through sliding window technology to provide continuous time series input for training short-term load forecasting models, and extract key features from the first sample load data, such as the rate of change of voltage, current, power, and temperature change trend. Create derived features, such as the moving average of load, the interactive features of load and temperature, etc. Ensure that the range and distribution of all first sample load data features are similar to improve the training effect of short-term load forecasting models.

[0066] Step 1023: If the preprocessing is completed, the first sample load data is input into the autoregressive integrated moving average model to generate smoothed sample load data.

[0067] In this embodiment, if the first sample load data is preprocessed, the first sample load data is input into an autoregressive integrated moving average (ARIMA) model. The autoregressive integrated moving average (ARIMA) model can capture the long-term trend and seasonal fluctuations of the first sample load data and generate smoothed sample load data as input of the long short-term memory network.

[0068] Step 1024: Input the smoothed sample load data into the long short-term memory network to predict the future second sample load data.

[0069] In this embodiment, the long short-term memory network (LSTM) includes an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer receives a time series of smoothed sample load data and processes it in combination with environmental characteristics and time characteristics; the LSTM layer is used to process the long-term dependencies in the time series of smoothed sample load data and capture the complex nonlinear relationships in the smoothed sample load data; the fully connected layer further processes the output of the LSTM layer to generate a short-term load forecast result; and the output layer generates a second sample load data for a period of time in the future.

[0070] Step 1025: Calculate the total loss value according to the second sample load data and the label load data.

[0071] In this embodiment, the total loss value is calculated by the following formula:

[0072]

[0073] Total Loss=α·MSE+β·MAPE+γ·WAE+δ·Bias Penalty;

[0074] Among them, Total Loss is the total loss value, MAPE is the mean absolute percentage error, WAE is the weighted absolute error, MSE is the mean square error, Bias Penalty is the deviation penalty term, and y i is the label load data, is the second sample load data, N is the number of the first sample load data, w i represents the weight of the i-th first sample load data, λ is the penalty coefficient, and α, β, γ, and δ are all hyperparameters.

[0075] In this embodiment, MSE can smooth large errors, optimize the overall accuracy of the short-term load forecasting model, and ensure that the short-term load forecasting model fits the actual data better. The short-term load forecasting model will give a higher penalty weight to large errors and may be sensitive to outliers; MAPE measures the relative size of the prediction error, making the short-term load forecasting model more sensitive to data of different orders of magnitude, which helps to reduce the impact of the model's improper processing of load data of different magnitudes. WAE allows different weights to be given to different data points by introducing weights, which can give higher attention to key data in special time periods or regions in power load forecasting, thereby improving the local prediction ability of the short-term load forecasting model. The deviation penalty term effectively reduces the situation where the predicted value systematically deviates from the actual value by limiting the prediction deviation of the short-term load forecasting model, and improves the fairness and stability of the prediction. The penalty term is controlled by the penalty coefficient λ to ensure that the deviation will not be too large to affect the overall performance of the short-term load forecasting model. By adjusting the weights of the hyperparameters α, β, γ, and δ, the short-term load forecasting model can find a balance between different loss indicators, ensuring that while improving the accuracy of the model, the relative error, weighted error, and deviation control in practical applications are taken into account. This loss function that combines multiple indicators to calculate the total loss value can better adapt to complex load forecasting scenarios and improve the robustness and generalization ability of the short-term load forecasting model. The use of the combined loss function ensures that the short-term load forecasting model can not only make accurate predictions globally, but also perform more refined optimization in key areas, relative errors, deviations, etc., to adapt to the diverse needs of power load forecasting.

[0076] Step 1026: Update the autoregressive integrated moving average model and the long short-term memory network according to the total loss value.

[0077] In this embodiment, the total loss value is back-propagated, the gradient of the loss function for the ARIMA model and the LSTM parameters is calculated, the parameters of the LSTM and ARIMA models are updated using the calculated gradient, and steps 1021 to 1026 are repeated, and the gradient of the loss function for the ARIMA model and the LSTM parameters is calculated, and the parameters of the LSTM and ARIMA models are updated. The change of the total loss value is monitored, and if the total loss value no longer changes, it is determined that the LSTM and ARIMA models have converged, and the training of the autoregressive integrated moving average model and the long short-term memory network is completed.

[0078] Step 103: In the edge computing node, a preliminary load scheduling plan is generated according to the first target load data and the second target load data, and the preliminary load scheduling plan is transmitted to the cloud computing center.

[0079] In this embodiment, in the edge computing node, the first target load data provides the real state of the current system to ensure the immediacy and accuracy of the scheduling plan, and the second target load data helps to foresee future load demand to optimize the foresight and effectiveness of the scheduling plan. A preliminary load scheduling plan is generated based on the first target load data and the second target load data, which can reasonably plan future load demand while ensuring the existing load management, thereby achieving more efficient load scheduling, and the generated preliminary load scheduling plan is transmitted to the cloud computing center to formulate a load scheduling plan suitable for the entire power Internet of Things.

[0080] In one embodiment of the present invention, step 103 may include the following steps:

[0081] Step 1031, initialize the population.

[0082] In this embodiment, a genetic algorithm (GA) is used to initialize a population of load scheduling schemes for electrical equipment under the jurisdiction of the same edge computing node, and individuals in the population represent start and stop schemes for the electrical equipment.

[0083] Genetic algorithm (GA) is an optimization algorithm based on natural selection and genetic mechanism. Its principle imitates the biological evolution process to solve complex problems. First, an initial population of a set of solutions is randomly generated, each of which is called an individual. Each individual represents a possible solution, such as the start-stop solution of electrical equipment. Then, the algorithm evaluates the fitness of these individuals, usually measuring the quality of its problem-solving through an objective function. After the evaluation, individuals with higher fitness are selected as "parents" and new individuals ("offspring") are generated through crossover (recombination) and mutation operations. These offspring individuals will compete with individuals in the original population and gradually optimize the solution through multiple generations of evolution. Genetic algorithms can mainly handle complex, multi-peak optimization problems, have global search capabilities, can jump out of local optimal solutions, and are suitable for situations where the solution space is vast and difficult to solve with traditional methods. In the load scheduling of electrical equipment, genetic algorithms can explore different start-stop solutions, and find the best solution that can effectively balance load demand and equipment operating costs through iterative improvements, thereby improving the overall efficiency and reliability of the system.

[0084] Step 1032: Use the first target load data and the second target load data to set a first objective function and a first constraint condition for the population.

[0085] In this embodiment, the first target load data provides the current actual load condition, so that the first objective function can optimize the load dispatching scheme based on the actual situation, such as minimizing energy consumption or improving efficiency. At the same time, the second target load data predicts future load fluctuations, which helps to set the forward-looking first constraint condition to ensure that the dispatching scheme can also operate effectively under future load changes; the first objective function is used to measure the pros and cons of the load dispatching scheme, and the first constraint condition is used to ensure that the load dispatching scheme is feasible in practical applications.

[0086] Among them, the first objective function is:

[0087] min f(u)=α1·C total (u)+α2·R total (u)+α3·V total (u);

[0088] C total (u)=∑ t ∑ i [u i (t)·S i (t)·P electric (t)];

[0089] R total (u)=∑ t ∑ i [u i (t)·R i (t)];

[0090]

[0091] Where f(u) is the first objective function, min means that the goal of optimizing the first objective function is to minimize the output value of the first objective function, u is the start-stop strategy of the electrical equipment, C total (u) is the total cost, R total (u) is the total response time, V total (u) is the total load fluctuation, α1, α2, α3, u i (t) is the start / stop state of the i-th electrical equipment at time t in the first target load data, S i (t) is the operating power of the i-th electrical equipment at time t in the first target load data, P electric (t) is the electricity price at time t, R i (t) is the response time of the i-th electrical equipment at time t in the first target load data, is the load value at time t in the second target load data.

[0092] The first constraint condition includes at least one of the following:

[0093] Power limit constraints, load balance constraints, equipment start and stop constraints, response time constraints, and electricity price constraints.

[0094] The power limit constraint condition indicates that the operating power of each electrical device is less than or equal to the rated power;

[0095] The load balance constraint indicates that the total load of the edge computing nodes satisfies the supply and demand balance of the power grid;

[0096] Equipment start and stop constraint conditions indicate that the number of starts and stops of electrical equipment meets the number range;

[0097] The response time constraint indicates that the response time of the critical electrical equipment must meet the time range;

[0098] The electricity price constraint represents the upper limit of the electricity price.

[0099] Step 1033: Calculate the fitness of the individual under the constraint of the first constraint condition with the goal of optimizing the first objective function.

[0100] In this embodiment, the goal is to optimize the first objective function, and the individual calculates the fitness under given constraints to evaluate how each solution or individual approaches the optimization goal. If the first constraint is violated, a corresponding penalty will be imposed, that is, the fitness value will be reduced.

[0101] The specific fitness is:

[0102] F(u)=-(α1·C total (u)+α2·R total (u)+α3·V total (u))-λ·Penalty(u);

[0103] In the formula, F(u) is the fitness, u is the start-stop strategy of the electrical equipment, C total (u) is the total cost, R total (u) is the total response time, V total (u) is the total load fluctuation, Penalty(u) is the penalty term, and α1, α2, α3, and λ all represent weight parameters.

[0104] Step 1034: Perform selection operation, crossover operation and mutation operation on the individuals in sequence to obtain new individuals.

[0105] Selection: Individuals with high fitness are selected to enter the next generation through the roulette wheel selection method. The next generation (or "new generation") refers to the new population generated from the current population after completing an iteration or evolution.

[0106] Crossover: Generate new individuals through crossover operation, mix the start and stop plans of the two parent individuals, form a new scheduling plan, and through crossover, integrate the characteristics of the parent individuals to explore possible better solutions, thereby improving the overall fitness of the population.

[0107] Mutation: Randomly select some individuals for mutation operations, change the start and stop status of some devices in the individuals, increase population diversity, and avoid falling into local optimality.

[0108] In this embodiment, the use of selection, crossover and mutation is intended to mimic the natural selection process, gradually improving the quality of the solution by continuously updating the population. Selection ensures the retention of excellent solutions, crossover promotes information recombination, and mutation increases diversity to generate new individuals.

[0109] Step 1035, determine whether the preset genetic conditions are met; if so, execute step 1036; if not, return to execute step 1032.

[0110] In this embodiment, the genetic condition generally includes the stopping condition of the algorithm, such as reaching the maximum number of iterations, the fitness reaching a certain threshold, or the fitness improvement becoming insignificant. If these conditions are met, it means that the current genetic operation (selection, crossover, mutation) is basically completed, a suitable solution is found, and the next step, i.e., step 1036, can be entered. If these conditions are not met, it is necessary to return to step 1032, recalculate the fitness and perform the genetic operation to continue to improve the solution until the stopping condition is met.

[0111] Step 1036: Select the individual with the highest fitness and obtain a preliminary load scheduling plan.

[0112] In this embodiment, the individual with the highest fitness is the one that performs best among all optimized individuals, that is, the load scheduling plan that best meets the requirements under the first objective function. The individual with the highest fitness is selected to obtain a preliminary load scheduling plan.

[0113] Step 104: In the cloud computing center, the preliminary load scheduling plan is revised into a global load scheduling plan according to the operation status of the power grid.

[0114] In this embodiment, during the process of revising the preliminary load scheduling plan, the cloud computing center receives the preliminary load scheduling plan generated by each edge computing node and obtains the real-time operation status data of the power grid, including load distribution, power generation capacity, and equipment status, etc. The cloud computing center analyzes these data in detail, evaluates the current status of the power grid, identifies aspects that need to be adjusted, and adjusts and optimizes the preliminary plan based on the analysis results, generates a global load scheduling plan, and realizes the coordination and optimization of the power grid.

[0115] In this embodiment, a second objective function and a second constraint condition are set for the preliminary load dispatching plan according to the operating status of the power grid.

[0116] Among them, the second objective function is:

[0117]

[0118] Where F(x) is the second objective function, Minimize means that the goal of optimizing the second objective function is to minimize the output value of the second objective function, N is the number of edge computing nodes, and AdjustmegtCost i The cost of scheduling load for edge computing node i, LoadVariation i is the load fluctuation of edge computing node i, α i , β i are all weight parameters.

[0119] The second constraint condition includes at least one of the following:

[0120] Power constraints and network-wide load balancing constraints.

[0121] Among them, the power constraint condition means that the load of each edge computing node is less than or equal to the power upper limit;

[0122] The network-wide load balance constraint means that the sum of the loads of all edge settlement nodes is consistent with the supply and demand balance of the power grid.

[0123] In this embodiment, the second objective function is optimized as the goal, and under the constraint of the second constraint condition, the preliminary load scheduling plan is corrected to the global load scheduling plan. The second objective function and the second constraint condition are substituted into the Lagrangian function to calculate the local adjustment amount, and the preliminary load scheduling plan is adjusted according to the local adjustment amount to obtain the global load scheduling plan.

[0124] Among them, the Lagrange multiplication function is:

[0125]

[0126] in, is the Lagrangian function, F(x) is the second objective function, λ j is the Lagrange multiplier vector, g j (x) is the second constraint.

[0127] The Lagrange multiplication function is used to solve optimization problems with constraints. By introducing the Lagrange multiplier, the original constrained optimization problem is transformed into an unconstrained problem, so that the extreme value of the objective function can be found under the constraints. This method constructs the Lagrange function, combines the objective function with the constraints, and thus finds the optimal solution while satisfying the constraints.

[0128] Step 105: In the cloud computing center, the global load scheduling plan is transmitted to each edge computing node for execution to schedule the load of the power grid.

[0129] In this embodiment, in the cloud computing center, the global load scheduling plan is transmitted to each edge computing node for execution. The edge devices and edge nodes continuously monitor the current power load, equipment status and scheduling execution status, and feed back the monitored data to the cloud computing center for dynamic adjustment and optimization to achieve grid load scheduling.

[0130] In an embodiment of the present invention, the electric power Internet of Things has multiple edge computing nodes and a cloud computing center. In the edge computing node, first target load data is collected for electrical equipment under the edge computing node in the power grid. In the edge computing node, the load data is input into a preset short-term load forecasting model to predict future second target load data. In the edge computing node, a preliminary load scheduling plan is generated based on the first target load data and the second target load data, and the preliminary load scheduling plan is transmitted to the cloud computing center. In the cloud computing center, the preliminary load scheduling plan is revised into a global load scheduling plan based on the operating status of the power grid. In the cloud computing center, the global load scheduling plan is transmitted to each edge computing node for execution to schedule the load of the power grid. In this embodiment, multiple edge computing nodes cooperate with the cloud computing center. The edge computing nodes are responsible for local load scheduling, and the cloud computing center is responsible for global optimization, thereby realizing distributed load scheduling optimization. The edge computing nodes undertake a large amount of computing work, which greatly reduces the computing pressure of the cloud computing center, effectively improves the response speed, and improves the efficiency of load scheduling. Multiple edge computing nodes and the cloud computing center continuously monitor and dynamically schedule loads on electrical equipment in the power grid, respond to load fluctuations in a timely manner, avoid local overloads, improve the load scheduling efficiency and stability of the power grid, and dynamically adjust to achieve balanced distribution of loads across the entire network.

[0131] Embodiment 2

[0132] Figure 2 A schematic diagram of a power grid load dispatching device provided in the second embodiment of the present invention is shown in FIG. 1 , wherein the power Internet of Things has multiple edge computing nodes and a cloud computing center, such as Figure 2 As shown, the device comprises:

[0133] A first target load data collection module 201 is used to collect first target load data from electrical equipment under the edge computing node in the power grid in the edge computing node;

[0134] A second target load data prediction module 202 is used to input the load data into a preset short-term load prediction model in the edge computing node to predict future second target load data;

[0135] A preliminary load scheduling scheme generating module 203 is used to generate a preliminary load scheduling scheme in the edge computing node according to the first target load data and the second target load data, and transmit the preliminary load scheduling scheme to the cloud computing center;

[0136] A global load scheduling scheme correction module 204 is used to correct the preliminary load scheduling scheme into a global load scheduling scheme according to the operation status of the power grid in the cloud computing center;

[0137] The global load scheduling scheme transmission module 205 is used to transmit the global load scheduling scheme to each edge computing node in the cloud computing center for execution, so as to schedule the load of the power grid.

[0138] In one embodiment of the present invention, the short-term load forecasting model includes an autoregressive integrated moving average model and a long short-term memory network; the second target load data forecasting module 202 includes:

[0139] A first target load data preprocessing module, used for preprocessing the first target load data;

[0140] a smoothed target load data generating module, configured to input the first target load data into the autoregressive integral moving average model to generate smoothed target load data if the preprocessing is completed;

[0141] The second target load data acquisition module is used to input the smoothed target load data into the long short-term memory network to predict future second target load data.

[0142] In one embodiment of the present invention, the second target load data prediction module 202 further includes:

[0143] A first sample load data collection module is used to collect first sample load data; the first sample load data is labeled with label load data;

[0144] A first sample load data preprocessing module, used for preprocessing the first sample load data;

[0145] a smoothed sample load data generating module, configured to input the first sample load data into the autoregressive integrated moving average model to generate smoothed sample load data if the preprocessing is completed;

[0146] A second sample load data acquisition module, used for inputting the smoothed sample load data into the long short-term memory network to predict future second sample load data;

[0147] a total loss value calculation module, configured to calculate a total loss value according to the second sample load data and the label load data;

[0148] A model updating module, used for updating the autoregressive integrated moving average model and the long short-term memory network according to the total loss value;

[0149] Wherein, the total loss value is expressed as:

[0150]

[0151] TotalLoss=α·MSE+β·MAPE+γ·WAE+δ·Bias Penalty;

[0152] Among them, Total Loss is the total loss value, MAPE is the mean absolute percentage error, WAE is the weighted absolute error, MSE is the mean square error, Bias Penalty is the deviation penalty term, and y i is the tag load data, is the second sample load data, N is the number of the first sample load data, w i represents the weight of the i-th sample load data, λ is the penalty coefficient, and α, β, γ, and δ are all hyperparameters.

[0153] In one embodiment of the present invention, the preliminary load scheduling scheme generating module 203 includes:

[0154] A population initialization module, used to initialize the population; individuals in the population represent the start and stop scheme of the electrical equipment;

[0155] A first objective function and first constraint condition setting module, used to set a first objective function and a first constraint condition for the population using the first target load data and the second target load data;

[0156] A fitness calculation module, configured to calculate the fitness of the individual under the constraint of the first constraint condition with the goal of optimizing the first objective function;

[0157] A new individual acquisition module, used to sequentially perform selection operation, crossover operation and mutation operation on the individuals to obtain new individuals;

[0158] The genetic condition judgment module is used to judge whether the preset genetic condition is met; if so, the individual with the highest fitness is selected to obtain a preliminary load scheduling plan; if not, the execution is returned to the step of optimizing the first objective function and calculating the fitness of the individual under the constraint of the first constraint condition.

[0159] In one embodiment of the present invention, the first objective function is:

[0160] min f(u)=α1·C total (+α2·R total (u)+α3·V total (u);

[0161] C total (u)=∑ t ∑ i [u i (t)·S i (t)·P electric (t)];

[0162] R total (u)=∑ t ∑ i [u i (t)·R i (t)];

[0163]

[0164] The fitness is expressed as:

[0165] F(u)=-(α1·C total (u)+α2·R total (u)+α3·V total (u))-λ·Penalty(u)

[0166] Wherein, f(u) is the first objective function, F(u) is the fitness, min indicates that the goal of optimizing the first objective function is to minimize the output value of the first objective function, and u is the start-stop strategy of the electrical equipment; C total (u) is the total cost, R total (u) is the total response time, V total (u) is the total load fluctuation, Penalty(u) is the penalty term, and α1, α2, α3, and λ all represent weight parameters;

[0167] u i (t) is the start / stop state of the i-th electrical equipment at time t in the first target load data, S i(t) is the operating power of the i-th electrical equipment at time t in the first target load data, P electric (t) is the electricity price at time t, R i (t) is the response time of the i-th electrical device at time t in the first target load data, is the load value at time t in the second target load data;

[0168] The first constraint condition includes at least one of the following:

[0169] Power limit constraints, load balance constraints, equipment start and stop constraints, response time constraints, and electricity price constraints;

[0170] Wherein, the power limit constraint condition indicates that the operating power of each of the electrical devices is less than or equal to the rated power;

[0171] The load balancing constraint condition indicates that the total load of the edge computing node satisfies the supply and demand balance of the power grid;

[0172] The equipment start and stop constraint condition indicates that the number of starts and stops of the electrical equipment meets a number range;

[0173] The response time constraint condition indicates that the response time of the critical electrical device meets a time range;

[0174] The power price constraint condition indicates an upper limit of the power price.

[0175] In one embodiment of the present invention, the global load scheduling scheme modification module 204 includes:

[0176] A second objective function and a second constraint condition setting module, used to set a second objective function and a second constraint condition for the preliminary load dispatching scheme according to the operating state of the power grid;

[0177] A global load scheduling scheme acquisition module is used to optimize the second objective function and, under the constraints of the second constraint condition, correct the preliminary load scheduling scheme into a global load scheduling scheme.

[0178] In one embodiment of the present invention, the second objective function is:

[0179]

[0180] Where F(x) is the second objective function, Minimize indicates that the goal of optimizing the second objective function is to minimize the output value of the second objective function, N is the number of edge computing nodes, and AdjustmengtCost iis the cost of scheduling load for the edge computing node i, LoadVariation i is the load fluctuation of the edge computing node i, α i , β i All are weight parameters;

[0181] The second constraint condition includes at least one of the following:

[0182] Power constraints and network-wide load balancing constraints;

[0183] The power constraint condition indicates that the load of each edge computing node is less than or equal to the power upper limit;

[0184] The network-wide load balancing constraint condition indicates that the sum of the loads of all the edge settlement nodes is consistent with the supply and demand balance of the power grid.

[0185] In one embodiment of the present invention, the global load scheduling solution acquisition module includes:

[0186] A local adjustment amount calculation module, used for substituting the second objective function and the second constraint condition into a Lagrangian function to calculate a local adjustment amount;

[0187] A preliminary load scheduling scheme adjustment module, used to adjust the preliminary load scheduling scheme according to the local adjustment amount to obtain a global load scheduling scheme;

[0188] Wherein, the Lagrange multiplication function is:

[0189]

[0190] in, is the Lagrangian function, F(x) is the second objective function, λ j is the Lagrange multiplier vector, g j (x) is the second constraint condition.

[0191] The power grid load dispatching device provided in the embodiment of the present invention can execute the power grid load dispatching method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the power grid load dispatching method.

[0192] Embodiment 3

[0193] See also Figure 3, showing a schematic diagram of the structure of a computer device provided by an embodiment of the present invention. The computer device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, blade servers, mainframe computers, and other suitable computers. The computer device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0194] like Figure 3 As shown, the computer device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the computer device 10 can also be stored. The processor 11, ROM 12 and RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0195] A number of components in the computer device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the computer device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0196] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a power grid load dispatching method.

[0197] In some embodiments, the power grid load dispatching method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the computer device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the power grid load dispatching method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the power grid load dispatching method in any other appropriate manner (e.g., by means of firmware).

[0198] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0199] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0200] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0201] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0202] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0203] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0204] Embodiment 4

[0205] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the power grid load scheduling method provided in any embodiment of the present invention is implemented.

[0206] In the process of implementation, the computer program product can be written in one or more programming languages ​​or a combination thereof to perform the computer program code of the present invention, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0207] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0208] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A power grid load dispatching method, characterized in that: The power Internet of Things has multiple edge computing nodes and a cloud computing center, and the method includes: In the edge computing node, first target load data is collected from electrical equipment under the edge computing node in the power grid; In the edge computing node, the load data is input into a preset short-term load forecasting model to predict a second target load data in the future; In the edge computing node, a preliminary load scheduling plan is generated according to the first target load data and the second target load data, and the preliminary load scheduling plan is transmitted to the cloud computing center; In the cloud computing center, the preliminary load scheduling plan is revised into a global load scheduling plan according to the operating status of the power grid; In the cloud computing center, the global load scheduling scheme is transmitted to each of the edge computing nodes for execution, so as to schedule the load of the power grid.

2. The method according to claim 1, characterized in that The short-term load forecasting model includes an autoregressive integrated moving average model and a long short-term memory network; the step of inputting the load data into a preset short-term load forecasting model to forecast future second target load data includes: Preprocessing the first target load data; If the preprocessing is completed, the first target load data is input into the autoregressive integrated moving average model to generate smoothed target load data; The smoothed target load data is input into the long short-term memory network to predict the second target load data in the future.

3. The method according to claim 2, characterized in that The step of inputting the load data into a preset short-term load forecasting model in the edge computing node to forecast future second target load data further includes: Collecting first sample load data; the first sample load data is labeled with label load data; Preprocessing the first sample load data; If the preprocessing is completed, the first sample load data is input into the autoregressive integrated moving average model to generate smoothed sample load data; Inputting the smoothed sample load data into the long short-term memory network to predict future second sample load data; Calculating a total loss value according to the second sample load data and the label load data; Updating the autoregressive integrated moving average model and the long short-term memory network according to the total loss value; Wherein, the total loss value is expressed as: Total Loss=α·MSE+β·MAPE+γ·WAE+δ·Bias Penalty; Among them, Total Loss is the total loss value, MAPE is the mean absolute percentage error, WAE is the weighted absolute error, MSE is the mean square error, Bias Penalty is the deviation penalty term, and y i is the tag load data, is the second sample load data, N is the number of the first sample load data, w i represents the weight of the i-th sample load data, λ is the penalty coefficient, and α, β, γ, and δ are all hyperparameters.

4. The method according to claim 1, characterized in that: The generating a preliminary load dispatching scheme according to the first target load data and the second target load data comprises: Initializing a population; individuals in the population represent a start-stop plan for the electrical equipment; Using the first target load data and the second target load data, a first target function and a first constraint condition are set for the population; Taking optimization of the first objective function as a goal, and subject to the first constraint condition, calculating the fitness of the individual; Performing selection operation, crossover operation and mutation operation on the individuals in sequence to obtain new individuals; Determine whether the preset genetic conditions are met; if so, select the individual with the highest fitness to obtain a preliminary load scheduling plan; if not, return to execute the step of optimizing the first objective function and calculating the fitness of the individual under the constraints of the first constraint.

5. The method according to claim 4, characterized in that The first objective function is: min f(u)=α1·C total (u)+α2·R total (u)+α3·V total (u); C total (u)=Σ t Σ i [u i (t)·S i (t)·P electric (t)]; R total (u)=∑ t ∑ i [u i (t)·R i (t)]; The fitness is expressed as: F(u)=-(α1·C total (u)+α2·R total (u)+α3·V total (u))-λ·Penalty(u); Wherein, f(u) is the first objective function, F(u) is the fitness, min means that the goal of optimizing the first objective function is to minimize the output value of the first objective function, u is the start-stop strategy of the electrical equipment, C total (u) is the total cost, R total (u) is the total response time, V total (u) is the total load fluctuation, Penalty(u) is the penalty term, and α1, α2, α3, and λ all represent weight parameters; u i (t) is the start / stop state of the i-th electrical equipment at time t in the first target load data, S i (t) is the operating power of the i-th electrical equipment at time t in the first target load data, P electric (t) is the electricity price at time t, R i (t) is the response time of the i-th electrical device at time t in the first target load data, is the load value at time t in the second target load data; The first constraint condition includes at least one of the following: Power limit constraints, load balance constraints, equipment start and stop constraints, response time constraints, and electricity price constraints; Wherein, the power limit constraint condition indicates that the operating power of each of the electrical devices is less than or equal to the rated power; The load balancing constraint condition indicates that the total load of the edge computing node satisfies the supply and demand balance of the power grid; The equipment start and stop constraint condition indicates that the number of starts and stops of the electrical equipment meets a number range; The response time constraint condition indicates that the response time of the critical electrical device meets a time range; The power price constraint condition indicates an upper limit of the power price.

6. The method according to any one of claims 1 to 5, characterized in that The step of amending the preliminary load dispatching scheme into a global load dispatching scheme according to the operating state of the power grid includes: Setting a second objective function and a second constraint condition for the preliminary load dispatching scheme according to the operating state of the power grid; With the goal of optimizing the second objective function, under the constraint of the second constraint condition, the preliminary load scheduling plan is revised into a global load scheduling plan.

7. The method according to claim 6, characterized in that The second objective function is: Where F(x) is the second objective function, Minimize indicates that the goal of optimizing the second objective function is to minimize the output value of the second objective function, N is the number of edge computing nodes, and AdjustmengtCost i is the cost of scheduling load for the edge computing node i, LoadVariation i is the load fluctuation of the edge computing node i, α i , β i They are all weight parameters; The second constraint condition includes at least one of the following: Power constraints and network-wide load balancing constraints; The power constraint condition indicates that the load of each edge computing node is less than or equal to the power upper limit; The network-wide load balancing constraint condition indicates that the sum of the loads of all the edge settlement nodes is consistent with the supply and demand balance of the power grid.

8. The method according to claim 6, characterized in that The step of optimizing the second objective function as a goal and modifying the preliminary load dispatching scheme into a global load dispatching scheme under the constraint of the second constraint condition includes: Substituting the second objective function and the second constraint condition into the Lagrangian function to calculate the local adjustment amount; Adjust the preliminary load dispatching plan according to the local adjustment amount to obtain a global load dispatching plan; Wherein, the Lagrange multiplication function is: in, is the Lagrangian function, F(x) is the second objective function, λ j is the Lagrange multiplier vector, g j (x) is the second constraint condition.

9. A computer device, characterized in that: include: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the power grid load scheduling method as described in any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the power grid load scheduling method as described in any one of claims 1 to 8 is implemented.

Images