Method and device for determining resource bearing capacity of power grid and electronic equipment

Abstract

The invention discloses a method and device for determining the resource bearing capacity of a power grid and electronic equipment. The method comprises the steps that current power grid operation data are collected through a plurality of sensors installed in a power grid, and the current power grid operation data comprise voltage, current, active power, reactive power, distributed energy output, load requirements and power grid topological structure information; dynamically evaluating the current power grid operation data by using the bearing capacity evaluation model to obtain a dynamic evaluation result of the distributed resource bearing capacity of the power grid; performing time-phased evaluation on the distributed resource bearing capacity of the power grid in the future time period to obtain a time-phased evaluation result; and based on the dynamic evaluation result and the time-phased evaluation result, target decision information is generated, and the target decision information comprises a distributed resource access time period, a potential overload risk time period and scheduling optimization information. According to the method and the device, the technical problem of relatively low evaluation accuracy of the distributed resource bearing capacity of the power grid in related technologies is solved.

Description

Technical Field

[0001] The present invention relates to the field of power grids, and in particular to a method, device and electronic equipment for determining the resource carrying capacity of a power grid. Background Art

[0002] With the large-scale access of distributed energy and renewable energy, the structure and operation mode of the power system are undergoing profound changes. The evaluation of the distributed resource carrying capacity of the power grid is of great significance to ensure the safe and stable operation of the power grid and improve the renewable energy absorption capacity.

[0003] In related technologies, static assessment models are usually used to assess the carrying capacity of distributed resources. However, due to the dynamic characteristics of the carrying capacity of distributed resources in power grids, static assessment models have too many limitations. In addition, as the scale of the power grid increases, the calculation complexity of the model increases, resulting in high limitations in the method of determining the carrying capacity of power grid resources. This in turn leads to the technical problem of low accuracy in assessing the carrying capacity of distributed resources in power grids in related technologies.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present invention provide a method, device and electronic device for determining the resource carrying capacity of a power grid, so as to at least solve the technical problem of low accuracy in evaluating the distributed resource carrying capacity of a power grid in the related art.

[0006] According to one aspect of an embodiment of the present invention, a method for determining the resource carrying capacity of a power grid is provided, comprising: collecting current power grid operation data through multiple sensors installed in the power grid, wherein the current power grid operation data includes: voltage, current, active power, reactive power, distributed energy output, load demand and power grid topology information; dynamically evaluating the current power grid operation data using a carrying capacity assessment model to obtain a dynamic assessment result of the power grid's distributed resource carrying capacity; performing time period assessment on the power grid's distributed resource carrying capacity in a future time period to obtain a time period assessment result; generating target decision information based on the dynamic assessment result and the time period assessment result, wherein the target decision information includes a distributed resource access time period, a potential overload risk time period and scheduling optimization information.

[0007] Furthermore, the current power grid operation data is dynamically evaluated using the carrying capacity assessment model to obtain a dynamic assessment result of the power grid distributed resource carrying capacity, including: preprocessing the current power grid operation data to obtain preprocessed power grid operation data; inputting the preprocessed power grid operation data into the carrying capacity assessment model, and obtaining a dynamic assessment result through forward propagation calculation.

[0008] Furthermore, the current power grid operation data is preprocessed to obtain preprocessed power grid operation data, including: detecting the current power grid operation data to determine abnormal values ​​in the current power grid operation data; interpolating the abnormal values ​​to obtain first power grid operation data; standardizing the first power grid operation data to obtain second power grid operation data; constructing time features based on the second power grid operation data, calculating sliding average and differential features, and extracting frequency domain features through fast Fourier transform; summarizing the time features, sliding average, differential features and frequency domain features to obtain preprocessed power grid operation data.

[0009] Furthermore, the carrying capacity of the distributed resources of the power grid in the future time period is evaluated by time period to obtain the evaluation results by time period, including: using the carrying capacity prediction model to perform a rolling time domain prediction on the carrying capacity of the distributed resources of the power grid in the future time period to obtain the rolling prediction results; based on the rolling prediction results, the carrying capacity of the distributed resources of the power grid in the future time period is divided into multiple time periods; for each time period, the average carrying capacity, maximum carrying capacity, minimum carrying capacity and volatility index are calculated to generate the evaluation results by time period.

[0010] Furthermore, based on the dynamic evaluation results and the time period evaluation results, target decision information is generated, including: based on the dynamic evaluation results, combined with the power grid security constraints and economic indicators, an access strategy for distributed resources is generated through multi-objective optimization, wherein the power grid security constraints are used to ensure the constraints for safe operation of the power grid, and the economic indicators are used to characterize the costs and benefits of distributed resources; a scheduling plan for distributed resources is generated based on a model predictive control framework; and target decision information is generated based on the dynamic evaluation results, the access strategy, the scheduling plan and the time period evaluation results.

[0011] Furthermore, the method also includes: visually presenting the dynamic evaluation results, access strategies, scheduling schemes and time period evaluation results.

[0012] Furthermore, the method also includes: collecting power grid operation data through multiple sensors, generating a training data set, and constructing a historical database, wherein the historical database is used to store the training data set and the corresponding historical assessment results of the power grid distributed resource carrying capacity; based on the training data set, a deep neural network structure is used to construct a carrying capacity assessment model.

[0013] Furthermore, based on the training data set, a deep neural network structure is used to construct a bearing capacity assessment model, including: based on the training data set, a multi-branch deep neural network model is constructed, wherein the multi-branch deep neural network model includes: a convolutional neural network branch, a long short-term memory network branch and a fully connected layer, the convolutional neural network branch is used to extract the topological characteristics of the power grid, the long short-term memory network branch is used to capture the timing characteristics, and the fully connected layer is used to fuse the multi-dimensional characteristics output by the convolutional neural network branch and the long short-term memory network branch; determine a pre-trained model constructed by power grid data, and migrate the parameters of the pre-trained model to the multi-branch deep neural network model to obtain a target deep neural network model; based on the target deep neural network model, construct multiple deep neural network sub-models with different structures, and use different data subsets to train each sub-model, integrate each sub-model to obtain an integrated model; dynamically adjust the learning rate, dynamically adjust the regularization parameter, perform incremental learning and concept drift detection, neural architecture search and dynamic network adjustment on the integrated model to obtain a bearing capacity assessment model.

[0014] Furthermore, the learning rate of the integrated model is dynamically adjusted, including: calculating the validation set performance based on the integrated model; applying the cosine annealing strategy to update the learning rate when the validation set performance has not been improved for multiple consecutive training cycles; updating the parameters of the integrated model; dynamically adjusting the regularization parameters of the integrated model, including: determining the complexity of the integrated model; dynamically adjusting the regularization strength based on the complexity; adaptively adjusting the discard rate of the integrated model based on the network layer depth of the integrated model; performing incremental learning and concept drift detection on the integrated model, including: obtaining sample data in the data buffer; assigning weights to the sample data; incrementally updating the parameters of the integrated model based on the sample data and the corresponding weights; retraining the integrated model when the data distribution change rate of the updated model is greater than the preset change rate; performing neural architecture search and dynamic network adjustment on the integrated model, including: determining the architecture search space of the integrated model; optimizing the network structure of the integrated model based on the policy gradient controller; dynamically pruning the integrated model based on the weight amplitude; and implementing network growth to gradually increase the network width and depth of the integrated model when the performance of the integrated model reaches the preset performance.

[0015] Furthermore, the method also includes: determining multiple power grid evaluation indicators based on current power grid operation data, wherein the multiple power grid evaluation indicators include: voltage deviation indicator, line load rate indicator, distributed resource penetration rate and system loss rate; determining target optimization measures from power grid topology reconstruction measures, demand-side response measures and energy storage system scheduling measures based on multiple power grid evaluation indicators; establishing a dynamic real-time electricity price mechanism, and determining the current electricity price based on the power grid system state; constructing a dynamic equation of energy storage charge state based on charging and discharging efficiency and capacity, and determining the energy storage charge state using the dynamic equation of energy storage charge state; using a control strategy prediction model to process multiple power grid evaluation indicators, energy storage charge state, current electricity price and target optimization measures to generate a power grid control strategy; and using the power grid control strategy for hierarchical control.

[0016] Furthermore, based on multiple power grid evaluation indicators, target optimization measures are determined from power grid topology reconstruction measures, demand-side response measures and energy storage system scheduling measures, including: comparing multiple power grid evaluation indicators with corresponding indicator thresholds to obtain multiple comparison results, wherein the indicator threshold is adaptively adjusted based on historical data and seasonal changes; determining a target trigger level based on multiple comparison results and a multi-level trigger mechanism, wherein the multi-level trigger mechanism is used to characterize the mapping relationship between different comparisons and trigger levels; based on the target trigger level, determining target optimization measures from power grid topology reconstruction measures, demand-side response measures and energy storage system scheduling measures.

[0017] Furthermore, the method also includes: generating initial reconstruction measures; based on the initial reconstruction measures, combining the equipment switch capacity and operation number constraints, constructing a multi-objective optimization function, wherein the multi-objective optimization function includes: voltage deviation, line load rate, distributed resource penetration rate, system loss rate and switch operation number; solving the multi-objective optimization function to obtain grid topology reconstruction measures; dividing the load into interruptible load, adjustable load and uncontrollable load, and establishing a price-sensitive load response model; based on the load response model, obtaining demand-side response measures; constructing a multi-time scale coordinated scheduling strategy, wherein the multi-time scale coordinated scheduling strategy includes: day-ahead scheduling strategy, hourly scheduling strategy and minute-level scheduling strategy; constructing the optimization objective function of the multi-time scale coordinated scheduling strategy, and solving the optimization objective function to obtain energy storage system scheduling measures, wherein the optimization objective function includes: electricity purchase cost, energy storage depreciation cost and ancillary service income.

[0018] Furthermore, the method also includes: incrementally updating the load response model and adaptively adjusting the control parameters using the Bayesian optimization method; determining the technical indicators and economic indicators after optimized scheduling, wherein the technical indicators after optimized scheduling include: voltage qualification rate, line utilization rate, distributed energy consumption rate and system loss rate, and the economic indicators after optimized scheduling include: operating cost, distributed energy income and user benefit; based on the technical indicators and economic indicators after optimized scheduling, performing system performance evaluation to obtain performance evaluation results; and dynamically adjusting the power grid control strategy according to the performance evaluation results.

[0019] Furthermore, an optimization objective function of a multi-time-scale coordinated scheduling strategy is constructed, and the optimization objective function is solved to obtain scheduling measures for the energy storage system, including: using a deterministic mixed integer linear programming model to construct the first objective function and the first constraint of the day-ahead scheduling strategy, and using a branch and bound algorithm to condition the first objective function based on the first constraint to obtain the day-ahead scheduling strategy; constructing a state space model and a second objective function of the hourly scheduling strategy, generating multiple load and energy processing scenarios, using an alternating direction multiplier method to solve the control strategy corresponding to each load and energy processing scenario, and determining the hourly scheduling strategy from the control strategies corresponding to multiple load and energy processing scenarios; combining a rule-based fast control strategy with fuzzy logic control to determine the control frequency in the minute-level scheduling strategy, and using a fuzzy logic controller to determine the reactive power regulation amount in the minute-level scheduling strategy; performing multi-time-scale coordination based on the day-ahead scheduling strategy, the hourly scheduling strategy, and the minute-level scheduling strategy to obtain scheduling measures for the energy storage system.

[0020] According to another aspect of an embodiment of the present invention, there is also provided a device for determining the resource carrying capacity of a power grid, comprising: a collection module, for collecting current power grid operation data through multiple sensors installed in the power grid, wherein the current power grid operation data includes: voltage, current, active power, reactive power, distributed energy output, load demand and power grid topology information; a first evaluation module, for dynamically evaluating the current power grid operation data using a carrying capacity evaluation model to obtain a dynamic evaluation result of the distributed resource carrying capacity of the power grid; a second evaluation module, for performing a time period evaluation on the distributed resource carrying capacity of the power grid in a future time period to obtain a time period evaluation result; a generation module, for generating target decision information based on the dynamic evaluation result and the time period evaluation result, wherein the target decision information includes a distributed resource access time period, a potential overload risk time period and scheduling optimization information.

[0021] According to another aspect of an embodiment of the present invention, there is further provided an electronic device, comprising: a memory storing an executable program; and a processor for running the program, wherein the method in each embodiment of the present invention is executed when the program is running.

[0022] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium includes a stored executable program, wherein when the executable program is running, the device where the computer-readable storage medium is located is controlled to execute the methods in various embodiments of the present invention.

[0023] According to another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the method in each embodiment of the present invention is implemented.

[0024] According to an embodiment of the present invention, the current power grid operation data is collected by multiple sensors installed in the power grid, and then the current power grid operation data is dynamically evaluated using the carrying capacity evaluation model to obtain the dynamic evaluation result of the power grid distributed resource carrying capacity, and then the power grid distributed resource carrying capacity in the future time period is evaluated by time period to obtain the time period evaluation result, and at the same time, based on the dynamic evaluation result and the time period evaluation result, the target decision information is generated, so as to achieve the purpose of improving the accuracy of determining the power grid resource carrying capacity. It is easy to notice that the current power grid operation data obtained include: voltage, current, active power, reactive power, distributed energy output, load demand and power grid topology information, and the generated target decision information includes distributed resource access period, potential overload risk period and scheduling optimization information. By obtaining a variety of current power grid operation data and evaluating the power grid resource carrying capacity of the current time period and the future time period respectively, the target decision information is finally obtained, so as to achieve the purpose of improving the accuracy of the power grid resource carrying capacity evaluation, and thus solve the technical problem of low evaluation accuracy of the distributed resource carrying capacity of the power grid in the related technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0026] Figure 1 is a flow chart of a method for determining resource carrying capacity of a power grid according to the present invention;

[0027] Figure 2 It is a schematic diagram of a device for determining resource carrying capacity of a power grid according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] 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.

[0029] 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 interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order 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.

[0030] According to an embodiment of the present invention, an embodiment of a method for determining the resource carrying capacity of a power grid is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0031] Figure 1 is a flow chart of a method for determining resource carrying capacity of a power grid according to the present invention, such as Figure 1 As shown, the method comprises the following steps:

[0032] Step S102, collecting current grid operation data through multiple sensors installed in the grid, wherein the current grid operation data includes: voltage, current, active power, reactive power, distributed energy output, load demand and grid topology information.

[0033] The current grid operation data in the above steps is used to describe the various parameters and information generated during the operation of the grid. The current grid operation data is used to reflect the real-time status and performance of the grid. The types of current grid operation data may include but are not limited to voltage, current, active power, reactive power, distributed energy output, load demand and grid topology information. The specific current grid operation data needs to be determined according to actual needs and is not limited here.

[0034] The collection methods in the above steps may include but are not limited to the following:

[0035] The first method: using multiple sensors installed in the power grid for data collection;

[0036] The second method: using telemetry technology to collect current power grid data and send it to the power grid control center through remote transmission equipment;

[0037] The third method: Use smart meters and fiber optic communications to collect current grid operation data and transmit it to the grid control center.

[0038] The above collection method is only an example. The specific collection method needs to be determined based on environmental factors and the actual grid architecture, and is not limited here.

[0039] In an optional embodiment, multiple sensors deployed in the power grid are used to collect current power grid operation data. Based on the collected current power grid operation data, the operating status of the power grid can be fully understood, providing key information for the stable operation, fault diagnosis, energy management and optimized scheduling of the power grid. Real-time monitoring and analysis of these data help to improve the reliability and efficiency of the power grid, ensure the quality and safety of power supply, and also provide data support for the intelligent and automated management of the power grid.

[0040] Step S104, dynamically evaluating the current power grid operation data using the carrying capacity evaluation model to obtain a dynamic evaluation result of the power grid distributed resource carrying capacity.

[0041] The carrying capacity assessment model in the above steps is used to describe a mathematical model for assessing the maximum load or resource capacity that a power grid can carry under specific conditions. The carrying capacity assessment model can be used to predict the resource carrying capacity of the power grid, as well as to assess the stability and reliability of the power grid under different conditions, helping power grid operators make reasonable dispatching decisions and ensuring safe and economical operation of the power grid.

[0042] The dynamic evaluation in the above steps is used to describe the continuous and real-time evaluation process of power grid performance and status under real-time changing conditions. The types of dynamic evaluation may include but are not limited to real-time monitoring of current data, power grid carrying capacity evaluation and predictive maintenance. The specific dynamic evaluation needs to be determined according to actual needs and is not limited here. The role of dynamic evaluation may include but is not limited to real-time monitoring of power grid status, rapid response to power grid changes, to maintain stable operation of the power grid, etc.

[0043] The distributed resource carrying capacity of the power grid in the above steps is used to describe the maximum capacity of the power grid to operate safely and stably when distributed resource power generation is connected. The distributed resources include but are not limited to solar power generation, wind power generation, photovoltaic power generation, etc. The types of distributed resource carrying capacity of the power grid may include but are not limited to the solar power generation carrying capacity of the power grid, the photovoltaic power generation carrying capacity of the power grid, the wind power generation carrying capacity of the power grid, etc. The specific distributed resource carrying capacity of the power grid needs to be determined according to the resource type and the power grid structure, and is not limited here.

[0044] The dynamic evaluation results in the above steps are used to describe the real-time data and analysis results about the power grid status and performance obtained during the dynamic evaluation process. The dynamic evaluation results may include but are not limited to power grid load forecasting, fault risk assessment, power grid stability analysis and other information. The specific dynamic evaluation results need to be determined according to actual needs and are not limited here. The dynamic evaluation results can be used to provide real-time decision support for power grid operators, helping them to adjust power grid operation strategies in a timely manner and respond to various emergencies.

[0045] In an optional embodiment, the real-time operation data of the power grid is deeply analyzed through the carrying capacity assessment model, so as to dynamically assess the carrying capacity of the power grid for distributed resources and obtain dynamic assessment results. Through this dynamic assessment, the power grid operator can better predict and respond to the fluctuation of distributed resources, and achieve more efficient and environmentally friendly power supply. In addition, it can also help the power grid operator make more realistic decisions in the face of changing energy demand and supply conditions, thereby improving the adaptability and efficiency of the entire power system.

[0046] Step S106, performing time-based evaluation on the distributed resource carrying capacity of the power grid in the future time period to obtain time-based evaluation results.

[0047] The future time period in the above steps is used to describe the time period from the current time point to a certain time point in the future. The future time period may include but is not limited to a few minutes, hours, days, weeks and months. The specific future time period needs to be determined according to the evaluation purpose and specific needs. No limitation is made here. The role of the future time period may include but is not limited to determining the time range of the evaluation, providing a time frame for power grid planning and operation, etc.

[0048] The time period assessment in the above steps is used to describe the further subdivision of future time periods into smaller time units, and to assess the distributed resource carrying capacity of the power grid in each small time period. The types of time period assessment may include but are not limited to minute-level time period assessment, hour-level time period assessment, day-ahead time period assessment, etc. The specific time period assessment needs to be determined according to the time scale and assessment purpose, and is not limited here. The time period assessment can be used to more accurately predict and analyze the operating status of the power grid in different time periods, and provide more detailed information for the dispatch and control of the power grid.

[0049] The time period assessment results in the above steps are used to describe the quantitative results of the distributed resource carrying capacity of the power grid obtained through the time period assessment. The time period assessment results may include but are not limited to the distributed resource carrying capacity of the power grid in the time period, load forecast, power generation forecast and power grid stability indicators, etc. The time period assessment results can be used to provide decision support for power grid operators, conduct reasonable resource allocation, formulate scheduling plans, prevent and respond to power grid failures, etc.

[0050] In an optional embodiment, by conducting a detailed time-period evaluation of the distributed resource carrying capacity of the power grid in the future time period, an accurate evaluation result of the power grid's ability to cope with distributed resource access in different time periods can be obtained. The evaluation result can help power grid operators predict and optimize power supply and ensure the stability and reliability of the power grid. Through the above segmented evaluation process, the differences in the distributed resource carrying capacity of the power grid in different time periods can be identified, thereby providing a basis for power grid upgrades, the formulation of load management strategies, and the effective integration of renewable energy.

[0051] Step S108, generating target decision information based on the dynamic evaluation results and the time period evaluation results, wherein the target decision information includes the distributed resource access period, the potential overload risk period and the scheduling optimization information.

[0052] The target decision information in the above steps is used to describe the decision support information formulated based on the evaluation results to achieve specific goals, where the specific goals may include but are not limited to the rational use of distributed resources connected to the power grid, ensuring power supply security, improving power grid performance, etc. The types of target decision information may include but are not limited to resource allocation decisions, load forecasting, demand response management, fault response decisions and market operation decisions, etc. The specific target decision information needs to be determined according to actual needs and is not limited here. The role of the target decision information may include but is not limited to providing system status analysis for the required time period, helping decision makers identify problems, assess risks, and formulate corresponding response strategies, etc.

[0053] The distributed resource access period in the above steps is used to describe the specific time period for distributed resources to access the power grid. The types of distributed resource access periods may include but are not limited to dynamic access periods based on predictions, access during peak hours, and access during off-peak hours. The specific distributed resource access period needs to be determined based on decision-making information and is not limited here. The role of the distributed resource access period may include but is not limited to clarifying the access time of distributed resources, avoiding resource waste, and improving the resource carrying capacity of the power grid.

[0054] The potential overload risk period in the above steps is used to describe the time period when some parts of the power grid may face overload risks due to excessive load or insufficient resources. The potential overload risk period needs to be determined based on decision-making information and is not limited here. The role of the potential overload risk period may include but is not limited to taking measures in advance, such as adjusting load distribution, increasing backup resources or limiting certain non-critical loads to ensure the stable operation of the power grid.

[0055] The dispatch optimization information in the above steps is used to describe the optimization strategies and suggestions for guiding the dispatch of the power system. The types of dispatch optimization information may include but are not limited to demand-side management strategies, integrated dispatch strategies for distributed resources, economic dispatch strategies, etc. The specific dispatch optimization information needs to be determined based on actual needs and target decision information, and is not limited here. The role of dispatch optimization information may include but is not limited to guiding the power system to meet power demand, optimize resource allocation, and reduce operating costs.

[0056] In an optional embodiment, target decision information is generated based on dynamic evaluation results and time period evaluation results to optimize the access and scheduling of distributed resources. Specifically, the system will adjust the power grid state according to the distributed resource access period, potential overload risk period and scheduling optimization information in the target decision information, including but not limited to: identifying the best time for distributed resource access to ensure the stability and efficiency of the power grid; predicting potential overload risk periods and taking measures in advance to avoid power supply interruptions or equipment damage; guiding power grid operators to make reasonable resource allocation decisions in different time periods based on the provided scheduling optimization information, thereby improving the operating efficiency and reliability of the entire power grid. Through the above process, the intelligence level of the power grid is jointly improved, providing strong support for achieving safer and more economical power supply.

[0057] According to an embodiment of the present invention, the current power grid operation data is collected by multiple sensors installed in the power grid, and then the current power grid operation data is dynamically evaluated using a carrying capacity assessment model to obtain a dynamic assessment result of the power grid distributed resource carrying capacity, and then the power grid distributed resource carrying capacity in the future time period is evaluated by time period to obtain the time period evaluation result, and at the same time, based on the dynamic evaluation result and the time period evaluation result, target decision information is generated, thereby achieving the purpose of improving the accuracy of determining the power grid resource carrying capacity.

[0058] It is easy to notice that the current grid operation data obtained include: voltage, current, active power, reactive power, distributed energy output, load demand and grid topology information. The generated target decision information includes distributed resource access time period, potential overload risk time period and scheduling optimization information. By obtaining a variety of current grid operation data and evaluating the grid resource carrying capacity in the current time period and future time period respectively, the target decision information is finally obtained, thereby achieving the purpose of improving the accuracy of grid resource carrying capacity assessment, and thus solving the technical problem of low accuracy in assessing the distributed resource carrying capacity of the grid in related technologies.

[0059] Optionally, the current power grid operation data is dynamically evaluated using a carrying capacity assessment model to obtain a dynamic assessment result of the power grid distributed resource carrying capacity, including: preprocessing the current power grid operation data to obtain preprocessed power grid operation data; inputting the preprocessed power grid operation data into the carrying capacity assessment model, and obtaining a dynamic assessment result through forward propagation calculation.

[0060] The above-mentioned preprocessing is used to describe the process of preprocessing the current power grid operation data to improve data quality before data analysis or model training. The types of preprocessing may include but are not limited to outlier detection, feature construction, standardization processing, etc. The specific preprocessing needs to be determined according to actual needs and is not limited here. The role of preprocessing may include but is not limited to improving data quality, reducing the impact of outliers, data screening, and improving model accuracy.

[0061] The above-mentioned forward propagation calculation is used to describe the process in which input data is passed through the network layer by layer until the output in the machine learning model. The role of the forward propagation calculation may include but is not limited to processing the input data, calculating dynamic evaluation results, etc.

[0062] In an optional embodiment, the current power grid operation data is first preprocessed to ensure the accuracy and availability of the data and provide high-quality input for subsequent evaluation; then the preprocessed data is input into the carrying capacity evaluation model, and the forward propagation mechanism of the model is used to calculate the dynamic evaluation results of the carrying capacity of the distributed resources of the power grid. The current power grid operation data is screened and standardized through preprocessing to obtain data that meets the standards, thereby improving the accuracy of the model. At the same time, the preprocessed data is predicted through the forward propagation mechanism in the carrying capacity evaluation model, which helps the power grid operator make more accurate operation decisions, thereby improving the stability and reliability of the power grid.

[0063] Optionally, the current power grid operation data is preprocessed to obtain preprocessed power grid operation data, including: detecting the current power grid operation data to determine abnormal values ​​in the current power grid operation data; interpolating the abnormal values ​​to obtain first power grid operation data; standardizing the first power grid operation data to obtain second power grid operation data; constructing time features based on the second power grid operation data, calculating sliding average and differential features, and extracting frequency domain features through fast Fourier transform; summarizing the time features, sliding average, differential features and frequency domain features to obtain preprocessed power grid operation data.

[0064] The above-mentioned outliers are used to describe data points in the current power grid operation data that deviate significantly from other observed values.

[0065] The above interpolation is used to describe the process of estimating and replacing outliers in the current power grid operation data.

[0066] The first power grid operation data mentioned above is used to describe the power grid operation data after the abnormal value interpolation processing, and the first operation data is used to ensure the integrity and consistency of the power grid operation data.

[0067] The above-mentioned standardization process is used to describe scaling the data so that it falls into a small specific interval, such as [0,1]. The role of the standardization process may include but is not limited to eliminating the influence of different dimensions and magnitudes.

[0068] The second power grid operation data mentioned above is used to describe the power grid operation data after standardization. The second power grid operation data can be used to provide a standardized data basis for subsequent feature extraction and model training.

[0069] The above-mentioned time features are used to describe the features extracted from time series data. The types of time features may include but are not limited to periodic features, trend features, etc. The specific time features need to be determined according to actual needs and are not limited here. The role of time features may include but is not limited to helping the model understand the patterns and laws of data changes over time.

[0070] The above-mentioned sliding average is used to describe a method of smoothing time series data. The sliding average reduces random fluctuations by calculating the average value of a series of consecutive data points.

[0071] The above-mentioned differential features are used to describe the features extracted by differentiating time series data.

[0072] The above-mentioned frequency domain features are used to describe the features extracted by converting time series data from the time domain to the frequency domain. The role of frequency domain features may include but is not limited to helping identify periodic components and frequency features in the data.

[0073] In an optional embodiment, the original power grid operation data is first detected to identify and determine the abnormal values ​​in the data. When an abnormal value appears, interpolation processing is performed to correct or replace the abnormal value, thereby obtaining the first power grid operation data; then the first power grid operation data is standardized to eliminate the influence of different dimensions and magnitudes, thereby obtaining the second power grid operation data; on this basis, by constructing time features, calculating sliding average and differential features, and using fast Fourier transform to extract frequency domain features, the dimension and depth of the data are further enriched; finally, these features are summarized to obtain the preprocessed power grid operation data. By preprocessing the current power grid operation data, the quality of the power grid operation data and the accuracy of the analysis can be improved, thereby providing strong data support for the stable operation and safety management of the power grid.

[0074] Optionally, the carrying capacity of the distributed resources of the power grid in the future time period is evaluated by time period to obtain the evaluation results by time period, including: using the carrying capacity prediction model to perform a rolling time domain prediction on the carrying capacity of the distributed resources of the power grid in the future time period to obtain the rolling prediction results; based on the rolling prediction results, the carrying capacity of the distributed resources of the power grid in the future time period is divided into multiple time periods; for each time period, the average carrying capacity, maximum carrying capacity, minimum carrying capacity and volatility index are calculated to generate the evaluation results by time period.

[0075] The above-mentioned rolling time domain prediction is used to describe a prediction method that reflects the latest data and trend dynamics by updating the prediction model at each time point. The types of rolling time domain predictions may include but are not limited to random forests, neural networks, etc. The specific rolling time domain prediction needs to be determined according to the actual model structure and is not limited here. The role of the rolling time domain prediction may include but is not limited to dynamically predicting the changes in carrying capacity in the future time period to better plan and schedule resources.

[0076] The above-mentioned rolling prediction results are used to describe the predicted values ​​output by the rolling time domain prediction model at each time point. The rolling prediction results reflect the expectations of the distributed resource carrying capacity of the power grid in the future time period.

[0077] The above average carrying capacity is used to describe the average value of the distributed resource carrying capacity of the power grid in each time period.

[0078] The above maximum carrying capacity is used to describe the maximum value of the distributed resource carrying capacity of the power grid in each time period.

[0079] The above minimum carrying capacity is used to describe the minimum value of the distributed resource carrying capacity of the power grid in each time period.

[0080] The above volatility index is a statistical quantity used to describe the degree of change in the carrying capacity of distributed resources in the power grid in each time period. The types of volatility indicators may include but are not limited to standard deviation, variance, peak-to-valley difference, etc. The specific volatility index needs to be determined according to actual needs and is not limited here.

[0081] In an optional embodiment, the carrying capacity prediction model is first used to perform a rolling time domain prediction of the carrying capacity of the power grid in the future time period to obtain continuous prediction results; then, based on these rolling prediction results, the prediction time period is divided into multiple hourly segments, and the carrying capacity in each time period is analyzed in detail to calculate the average carrying capacity, maximum carrying capacity, minimum carrying capacity and volatility index. The calculation of the above indicators helps to gain a deeper understanding of the carrying capacity of the power grid in different time periods, and provide more accurate data support for the operation and planning of the power grid, thereby rationalizing the allocation of power grid resources and improving the stability and reliability of the power grid.

[0082] Optionally, based on the dynamic evaluation results and the time period evaluation results, target decision information is generated, including: based on the dynamic evaluation results, combined with the power grid security constraints and economic indicators, an access strategy for distributed resources is generated through multi-objective optimization, wherein the power grid security constraints are used to ensure the constraints for safe operation of the power grid, and the economic indicators are used to characterize the costs and benefits of distributed resources; a scheduling plan for distributed resources is generated based on a model predictive control framework; and target decision information is generated based on the dynamic evaluation results, the access strategy, the scheduling plan and the time period evaluation results.

[0083] The above-mentioned multi-objective optimization is used to describe the simultaneous optimization of multiple conflicting objective functions. The role of multi-objective optimization may include but is not limited to balancing different operating objectives to achieve better power grid performance, where the multi-objective may include but is not limited to dynamic evaluation results, network safety constraints and economic indicators, etc. The specific multi-objective and multi-objective optimization need to be adjusted according to actual conditions and are not limited here.

[0084] The above access strategy is used to describe the rules and methods for accessing distributed resources such as solar photovoltaic and wind power to the power grid. The role of the access strategy may include but is not limited to ensuring that the access of distributed resources will not have a negative impact on the stability and security of the power grid, while improving the utilization efficiency of resources.

[0085] The above-mentioned model predictive control framework is used to describe the use of future prediction information to optimize current control decisions. The role of the model predictive control framework can include but is not limited to predicting future grid states and outputs of distributed resources, providing real-time scheduling solutions to cope with the uncertainty and volatility of renewable energy, etc.

[0086] The above-mentioned dispatching scheme is used to describe the resource allocation and operation plan formulated according to the needs of the power grid and the availability of distributed resources. The types of dispatching schemes may include but are not limited to dispatching schemes based on different dispatching cycles, dispatching schemes based on different optimization objectives, etc. The specific dispatching scheme needs to be determined according to actual needs and is not limited here. The role of the dispatching scheme may include but is not limited to ensuring the balance of supply and demand of the power grid, optimizing resource utilization, and reducing costs.

[0087] In an optional embodiment, firstly, the access strategy of distributed resources is formulated by using the dynamic evaluation results, combined with the safe operation constraints and economic indicators of the power grid, and using multi-objective optimization technology; then, the model predictive control framework is used to generate the scheduling plan of distributed resources to achieve efficient use of resources; finally, the dynamic evaluation results, access strategy, scheduling plan and time period evaluation results are integrated to generate the final target decision information to guide the management of distributed resources, thereby improving the operation efficiency and economic benefits of the power grid. The above process helps to realize the intelligent management of the power grid and the reasonable allocation of resources, thereby improving the stability and economy of the power grid.

[0088] Optionally, the method further includes: visually presenting the dynamic evaluation results, access strategies, scheduling schemes and time period evaluation results.

[0089] The above-mentioned visual presentation is used to describe the process of displaying dynamic evaluation results, access strategies, scheduling plans and time-segment evaluation results through a visual interface. The role of visual presentation may include but is not limited to converting complex data or information into graphics, charts or other visual formats, and displaying them through a visual interface for users to review, understand and communicate.

[0090] Visual presentation methods include but are not limited to the following:

[0091] Method 1: Display the above data results to the user through a visual interface;

[0092] Method 2: Convert the above data results into line graphs, interactive charts, etc., and then display them to users through a visual interface.

[0093] The above visualization methods are only examples. The specific visualization methods need to be determined according to the actual situation and are not limited here.

[0094] In an optional embodiment, complex evaluation results and scheduling decisions such as dynamic evaluation results, access strategies, scheduling schemes, and time-segment evaluation results are presented in a visual form, so that non-professional users can easily understand and monitor network performance. The above visualization function not only improves transparency, but also enhances user trust and satisfaction with network management decisions, thereby improving network service quality while also enhancing user participation and satisfaction.

[0095] Optionally, the method also includes: collecting power grid operation data through multiple sensors, generating a training data set, and constructing a historical database, wherein the historical database is used to store the training data set and the corresponding historical assessment results of the power grid distributed resource carrying capacity; based on the training data set, a deep neural network structure is used to construct a carrying capacity assessment model.

[0096] The above-mentioned training data set is used to describe the data set used to train the model. The types of training data sets may include but are not limited to power grid operation time data, power grid operation equipment data, power grid operation weather data, etc. The specific training data set needs to be determined according to the actual situation and is not limited here. The role of the training data set may include but is not limited to training the model so that it can identify the relationship between power grid operation data and carrying capacity.

[0097] The above historical evaluation results are used to describe the result records of past evaluations of the distributed resource carrying capacity of the power grid. The historical evaluation results can be used to train and verify the carrying capacity evaluation model.

[0098] In an optional embodiment, the power grid operation data is first collected through multiple sensors, and a training data set is generated based on the power grid operation data, and a historical database is further constructed; then a deep neural network structure is used to construct a carrying capacity assessment model. The historical database in the above process not only stores the training data set, but also records the historical assessment results of the power grid distributed resource carrying capacity, which can provide a reference for the analysis and prediction of the future power grid distributed resource carrying capacity. At the same time, the constructed carrying capacity assessment model can improve the accuracy and efficiency of the power grid carrying capacity assessment, thereby improving the power grid resource allocation and operation management.

[0099] Optionally, based on the training data set, a deep neural network structure is used to construct a bearing capacity assessment model, including: based on the training data set, constructing a multi-branch deep neural network model, wherein the multi-branch deep neural network model includes: a convolutional neural network branch, a long short-term memory network branch and a fully connected layer, the convolutional neural network branch is used to extract power grid topology features, the long short-term memory network branch is used to capture timing features, and the fully connected layer is used to fuse the multi-dimensional features output by the convolutional neural network branch and the long short-term memory network branch; determining a pre-trained model constructed using power grid data, and migrating the parameters of the pre-trained model to the multi-branch deep neural network model to obtain a target deep neural network model; based on the target deep neural network model, constructing multiple deep neural network sub-models with different structures, and using different data subsets to train each sub-model, integrating each sub-model to obtain an integrated model; dynamically adjusting the learning rate, dynamically adjusting the regularization parameter, performing incremental learning and concept drift detection, neural architecture search and dynamic network adjustment on the integrated model to obtain a bearing capacity assessment model.

[0100] The above-mentioned multi-branch deep neural network model is used to describe a deep neural network model including multiple parallel branches, where each branch can learn different features or representations of input data. The multi-branch deep neural network model structure includes a convolutional neural network branch, a long short-term memory network branch, and a fully connected layer. The role of the multi-branch deep neural network model can include but is not limited to capturing complex data patterns and serving as a benchmark model for the carrying capacity assessment model.

[0101] The above-mentioned pre-trained model is used to describe a model that is pre-trained based on power grid data. The role of the pre-trained model may include but is not limited to serving as the basis for transfer learning, migrating the learned parameters to a new model, thereby reducing training time and improving model performance.

[0102] The above-mentioned target deep neural network model is used to describe a multi-branch deep neural network model combined with pre-trained model parameters. The structure of the target deep neural network model is the same as that of the multi-branch deep neural network model.

[0103] The above-mentioned deep neural network sub-model is used to describe the various parts that constitute a complex deep neural network model. The types of deep neural network sub-models may include but are not limited to different layers of the network or different network structures, such as convolutional neural network layers, long short-term memory network layers, and fully connected layers. The specific deep neural network sub-model needs to be determined according to the target deep neural network structure, which is not limited here. The deep neural network sub-model can be used to improve the generalization ability and robustness of the model.

[0104] The above-mentioned dynamic adjustment of learning rate is used to describe the process of automatically adjusting the learning rate according to the training progress.

[0105] The above-mentioned dynamic adjustment of the regularization parameters is used to describe the process of automatically adjusting the regularization parameters according to the training effect.

[0106] The above incremental learning is used to describe that the model can gradually learn new data during the training process without the need to start training from scratch.

[0107] The above-mentioned concept drift detection is used to describe an important step to monitor model performance in real time and update the model in a timely manner. Concept drift detection can be used to timely discover concept drift and adjust the model to avoid model obsolescence and maintain the accuracy and effectiveness of the model.

[0108] The above-mentioned neural architecture search is used to describe the process of automatically designing the neural network structure, where the design includes the number of layers, layer type, connection method, etc. The specific design needs to be determined according to the neural network structure and is not limited here.

[0109] The above-mentioned dynamic network adjustment is used to describe the dynamic adjustment of network structure or parameters according to performance feedback during model training or deployment. The types of dynamic network adjustment may include but are not limited to network pruning, network growth, parameter fine-tuning, etc. The specific dynamic network adjustment needs to be determined according to the actual situation and is not limited here. Dynamic network adjustment can be used to help the model maintain good performance when facing new data or new tasks.

[0110] In an optional embodiment, a multi-branch deep neural network model is first constructed using a training data set, the model includes a convolutional neural network branch to extract power grid topology features, a long short-term memory network branch to capture timing features, and a fully connected layer to fuse these features; then the target deep neural network model is initialized by migrating the parameters of the pre-trained model; then multiple deep neural network sub-models with different structures are constructed and trained using different data subsets, and finally the generalization ability of the model is improved through ensemble learning; in addition, the method also includes dynamic adjustment of the learning rate and regularization parameters of the model, incremental learning, concept drift detection, neural architecture search, and dynamic network adjustment to optimize model performance and adapt to the changing data environment. The above method utilizes multiple features of power grid data, improves the accuracy and robustness of the model through pre-training and ensemble learning, and adapts to data changes through dynamic adjustment and search mechanisms, thereby achieving more accurate carrying capacity assessment.

[0111] Optionally, the learning rate of the integrated model is dynamically adjusted, including: calculating the validation set performance based on the integrated model; applying the cosine annealing strategy to update the learning rate when the validation set performance has not improved for multiple consecutive training cycles; updating the parameters of the integrated model; dynamically adjusting the regularization parameters of the integrated model, including: determining the complexity of the integrated model; dynamically adjusting the regularization strength based on the complexity; adaptively adjusting the discard rate of the integrated model based on the network layer depth of the integrated model; performing incremental learning and concept drift detection on the integrated model, including: obtaining sample data in the data buffer; assigning weights to the sample data; incrementally updating the parameters of the integrated model based on the sample data and the corresponding weights; retraining the integrated model when it is detected that the data distribution change rate of the updated model is greater than a preset change rate; performing neural architecture search and dynamic network adjustment on the integrated model, including: determining the architecture search space of the integrated model; optimizing the network structure of the integrated model based on a policy gradient controller; dynamically pruning the integrated model based on the weight amplitude; and when the performance of the integrated model reaches the preset performance, implementing network growth to gradually increase the network width and depth of the integrated model.

[0112] The above cosine annealing strategy is used to describe a learning rate adjustment method. The cosine annealing strategy dynamically adjusts the learning rate according to the periodic changes of the cosine function.

[0113] The above-mentioned data distribution change rate is used to describe the degree of change in the distribution of model input data relative to the previous period within a certain period of time.

[0114] The above-mentioned preset change rate is used to describe a preset threshold for determining when the model needs to respond to changes in data distribution. The preset change rate may include but is not limited to 10%, 15%, 20%, 25%, etc. The specific preset change rate needs to be set based on prior knowledge and is not limited here. The role of the preset change rate may include but is not limited to being a standard for determining whether the model needs to be retrained. When the actual data distribution change rate exceeds this threshold, the retraining or adjustment of the model is triggered.

[0115] The above retraining is used to describe the process of performing a new round of training on the model when the performance of the model no longer meets the requirements or the data distribution changes significantly.

[0116] The above architecture search space is used to describe the set of all possible network structures in neural architecture search.

[0117] The above policy gradient controller is used to describe an algorithm that adjusts the policy by gradient ascent or descent to maximize or minimize a certain objective function.

[0118] The above weight amplitude is used to describe the absolute value of the weights in the neural network. In dynamic pruning, the weight amplitude can be used to evaluate which weights contribute less to the model and can be pruned.

[0119] The above dynamic pruning is used to describe the technique of removing unimportant weights or neurons in a neural network based on the weight magnitude.

[0120] The preset performance mentioned above is used to describe a performance goal that is set in advance during the model training or adjustment process. When the model reaches this goal, further training or adjustment can be stopped. The preset performance can be used as the termination condition for model training and adjustment to ensure that the model is not over-trained after reaching the expected performance.

[0121] In an optional embodiment, the performance and adaptability of the model are improved through multiple methods such as dynamic adjustment of learning rate, updating of model parameters and regularization parameters, incremental learning, and concept drift detection. These strategies work together to ensure that the model remains efficient and accurate in a changing data environment, while being able to adapt to new data distributions and concept drift, ultimately achieving continuous updating and improvement of model performance.

[0122] Optionally, the method also includes: determining multiple power grid evaluation indicators based on current power grid operation data, wherein the multiple power grid evaluation indicators include: voltage deviation indicator, line load rate indicator, distributed resource penetration rate and system loss rate; determining target optimization measures from power grid topology reconstruction measures, demand-side response measures and energy storage system scheduling measures based on multiple power grid evaluation indicators; establishing a dynamic real-time electricity price mechanism, and determining the current electricity price based on the power grid system status; constructing a dynamic equation of energy storage charge state based on charging and discharging efficiency and capacity, and determining the energy storage charge state using the dynamic equation of energy storage charge state; using a control strategy prediction model to process multiple power grid evaluation indicators, energy storage charge state, current electricity price and target optimization measures to generate a power grid control strategy; and using the power grid control strategy for hierarchical control.

[0123] The above-mentioned grid topology reconstruction measures are used to describe measures to improve the operating status of the grid by changing the physical connection method of the grid. The types of grid topology reconstruction measures may include but are not limited to network reconstruction, distribution network reconstruction, transmission network reconstruction, etc. The specific grid topology reconstruction measures need to be determined according to the actual situation and are not limited here.

[0124] The above-mentioned demand-side response measures are used to describe the adjustment of the consumption behavior of electricity users through incentive or compensation mechanisms. Demand-side response measures can be used to reduce the load of the power grid during peak hours and improve the operating efficiency of the power grid.

[0125] The above-mentioned energy storage system dispatching measures are used to describe the reasonable dispatching of energy storage and release in the energy storage system according to the operation requirements of the power grid.

[0126] The above-mentioned target optimization measures are used to describe the selection of the most appropriate grid management measures according to grid evaluation indicators. The role of the target optimization measures may include but is not limited to ensuring that the grid operates in a good state, improving economic and social benefits, etc.

[0127] The above-mentioned dynamic real-time electricity price mechanism is used to describe a mechanism for dynamically adjusting electricity prices according to the real-time supply and demand conditions and operating costs of the power grid.

[0128] The above energy storage charge state dynamic equation is a mathematical model used to describe the change of charge state of the energy storage system during the charging and discharging process.

[0129] The above energy storage charge state is used to describe the percentage of electrical energy stored in the energy storage system at a certain moment in its total capacity.

[0130] The above-mentioned power grid control strategy is used to describe a series of decisions and actions to dispatch and control various equipment and resources in the power grid in order to achieve stable, safe and economical operation of the power grid.

[0131] The above-mentioned hierarchical control is used to describe a control strategy that decomposes complex power grid control strategies into multiple levels, with each level being responsible for different control tasks. Hierarchical control helps improve the flexibility and response speed of the power grid, enabling the power grid to more effectively respond to various changes and challenges.

[0132] In an optional embodiment, by comprehensively analyzing the operating data of the current power grid, key power grid evaluation indicators such as voltage deviation, line load rate, distributed resource penetration rate and system loss rate are determined. Then, based on the above indicators, the system can select better target optimization measures from a variety of measures such as power grid topology reconstruction, demand-side response and energy storage system scheduling. In addition, a dynamic real-time electricity price mechanism is established by adjusting the electricity price in real time according to the state of the power grid system. By constructing a dynamic equation for the energy storage charge state, the system can accurately predict the charge state of the energy storage system. Finally, the control strategy prediction model is used to combine the power grid evaluation indicators, the energy storage charge state, the current electricity price and the target optimization measures to generate an accurate power grid control strategy and realize hierarchical control of the power grid.

[0133] Optionally, based on multiple power grid evaluation indicators, target optimization measures are determined from power grid topology reconstruction measures, demand-side response measures and energy storage system scheduling measures, including: comparing multiple power grid evaluation indicators with corresponding indicator thresholds to obtain multiple comparison results, wherein the indicator threshold is adaptively adjusted based on historical data and seasonal changes; based on multiple comparison results and a multi-level trigger mechanism, a target trigger level is determined, wherein the multi-level trigger mechanism is used to characterize the mapping relationship between different comparisons and trigger levels; based on the target trigger level, target optimization measures are determined from power grid topology reconstruction measures, demand-side response measures and energy storage system scheduling measures.

[0134] The above comparison results are used to describe the output obtained after comparing the power grid evaluation index with the corresponding index threshold. The comparison results show the difference between the current state of the power grid and the preset threshold.

[0135] The above-mentioned target trigger level is used to describe the target level determined based on multiple comparison results and a multi-level trigger mechanism. The target trigger level can be used to characterize the severity of the power grid state and the urgency of the response measures to be taken.

[0136] In an optional embodiment, multiple grid evaluation indicators are first compared with corresponding indicator thresholds to generate multiple comparison results; then a multi-level trigger mechanism is used to determine the target trigger level based on the comparison results, and the target trigger level can map the relationship between different comparison results and trigger levels; finally, based on the determined target trigger level, the most appropriate optimization measure is selected from measures such as grid topology reconstruction, demand side response, and energy storage system scheduling. The above process can realize real-time monitoring and dynamic adjustment of the grid operation status, thereby improving the stability and efficiency of the grid.

[0137] Optionally, the method also includes: generating initial reconstruction measures; based on the initial reconstruction measures, combining the equipment switch capacity and operation number constraints, constructing a multi-objective optimization function, wherein the multi-objective optimization function includes: voltage deviation, line load rate, distributed resource penetration rate, system loss rate and switch operation number; solving the multi-objective optimization function to obtain grid topology reconstruction measures; dividing the load into interruptible load, adjustable load and uncontrollable load, and establishing a price-sensitive load response model; based on the load response model, obtaining demand-side response measures; constructing a multi-time scale coordinated scheduling strategy, wherein the multi-time scale coordinated scheduling strategy includes: day-ahead scheduling strategy, hourly scheduling strategy and minute-level scheduling strategy; constructing the optimization objective function of the multi-time scale coordinated scheduling strategy, and solving the optimization objective function to obtain energy storage system scheduling measures, wherein the optimization objective function includes: electricity purchase cost, energy storage depreciation cost and ancillary service income.

[0138] The above initial reconstruction measures are used to describe the preliminary adjustment plan formulated according to the current state and expected goals of the power grid during the power grid topology reconstruction process.

[0139] The above-mentioned equipment switch capacity and operation number constraints are used to describe the maximum number of closures and operation times of the equipment switch. The equipment switch capacity and operation number constraints may include but are not limited to the maximum number of closures of the switch, the maximum load capacity of the transformer, etc. The specific equipment switch capacity and operation number constraints need to be determined according to the equipment type and are not limited here. The role of the equipment switch capacity and operation number constraints may include but is not limited to ensuring that the power grid equipment operates within a safe and prescribed operating range, avoiding equipment overload or damage, etc.

[0140] The above-mentioned load response model is a model used to describe the changes in user load behavior according to changes in electricity prices or other incentive mechanisms.

[0141] The above-mentioned multi-time scale coordinated dispatching strategy is used to describe the power grid dispatching strategy that comprehensively considers different time scales. The types of multi-time scale coordinated dispatching strategies may include but are not limited to day-ahead dispatching strategies, hourly dispatching strategies, minute-level dispatching strategies, etc. The specific multi-time scale coordinated dispatching strategy needs to be determined according to actual needs and is not limited here. The role of the multi-time scale coordinated dispatching strategy includes but is not limited to coordination at different time scales to achieve dynamic balance of the power grid and improve the flexibility and responsiveness of the power grid.

[0142] The above-mentioned day-ahead dispatch strategy is used to describe the energy storage system dispatch plan formulated before the beginning of the day based on the predicted load and energy prices. The day-ahead dispatch strategy can be used to provide preliminary guidance for the daily operation of the energy storage system and reduce the pressure of real-time dispatch.

[0143] The above-mentioned hourly scheduling strategy is used to describe the energy storage system scheduling plan formulated at each hour level based on the latest load and energy price information. The hourly scheduling strategy can be used to provide more refined scheduling control to adapt to real-time changes.

[0144] The minute-level dispatch strategy described above is used to describe the fine-tuning of the operation of the energy storage system at each minute level to respond to the instantaneous demand of the power grid. The minute-level dispatch strategy can be used to provide the fastest response to maintain the stability of the power grid and the efficient operation of the energy storage system.

[0145] In an optional embodiment, the initial reconstruction measures of the power grid are first generated and combined with the switch capacity and operation number constraints of the equipment to construct a multi-objective optimization function. By solving this multi-objective optimization function, the optimized power grid topology reconstruction measures can be obtained; then the load is divided into interruptible load, adjustable load and uncontrollable load, and a price-sensitive load response model is established to obtain demand-side response measures according to the model, and a multi-time scale coordinated scheduling strategy is constructed, including day-ahead, hourly and minute-level scheduling strategies, and a corresponding optimization objective function is constructed. By solving the optimization objective function, the scheduling measures of the energy storage system are obtained. The above method can improve the operating efficiency of the power grid, reduce costs, and improve demand-side management, thereby realizing intelligent scheduling and economic operation of the power grid.

[0146] Optionally, the method also includes: incrementally updating the load response model and adaptively adjusting the control parameters using the Bayesian optimization method; determining the technical indicators and economic indicators after optimized scheduling, wherein the technical indicators after optimized scheduling include: voltage qualification rate, line utilization rate, distributed energy consumption rate and system loss rate, and the economic indicators after optimized scheduling include: operating cost, distributed energy income and user benefit; based on the technical indicators and economic indicators after optimized scheduling, performing system performance evaluation to obtain performance evaluation results; dynamically adjusting the power grid control strategy according to the performance evaluation results.

[0147] The above-mentioned incremental update is used to describe the small-step iteration of the load response model to adapt to changes in the operating state of the power grid. The types of incremental updates may include but are not limited to incremental adjustments of model parameters, or incremental optimization of the model structure. Incremental updates can be used to improve the adaptability and response speed of the model.

[0148] The above performance evaluation results are used to describe the comprehensive evaluation of the system performance after the grid is optimized and dispatched, including the evaluation of technical indicators and economic indicators. The performance evaluation results can be used to guide the dynamic adjustment of the grid control strategy to ensure that the grid operates in a good state.

[0149] The above technical indicators are used to describe a series of quantitative indicators to measure the operating performance of the power grid. The types of technical indicators may include but are not limited to voltage qualification rate, line utilization rate, distributed energy consumption rate and system loss rate. The specific technical indicators need to be determined according to the actual situation and are not limited here. The technical indicators can be used to evaluate the operating efficiency and stability of the power grid and are an important reference for the optimization and scheduling of the power grid.

[0150] The above economic indicators are a series of quantitative indicators used to describe the economic benefits of grid operation. The types of economic indicators may include but are not limited to operating costs, distributed energy benefits and user benefits. Specific economic indicators need to be determined based on actual conditions and are not limited here. Economic indicators can be used to evaluate the economic benefits of grid optimization and scheduling and are an important basis for grid operation decision-making.

[0151] In an optional embodiment, the load response model is incrementally updated, and the control parameters are adaptively adjusted using Bayesian optimization technology to optimize the grid dispatch. During the optimization process, the system will comprehensively consider multiple technical and economic indicators. By optimizing the dispatch of these indicators, the system can determine the best dispatch plan; then, based on these optimized technical and economic indicators, the system performance is evaluated to obtain performance evaluation results. These evaluation results will be used to dynamically adjust the grid control strategy to ensure the efficiency and economy of the grid operation, while meeting the dual requirements of technical specifications and economic benefits. The above process can improve the operating efficiency and economic benefits of the grid while realizing the automation and intelligence of grid dispatch, and ensure the stability and reliability of the grid by considering both technical and economic indicators.

[0152] Optionally, an optimization objective function of a multi-time-scale coordinated scheduling strategy is constructed, and the optimization objective function is solved to obtain scheduling measures for the energy storage system, including: using a deterministic mixed integer linear programming model to construct a first objective function and a first constraint condition of the day-ahead scheduling strategy, and using a branch and bound algorithm to constrain the first objective function based on the first constraint condition to obtain the day-ahead scheduling strategy; constructing a state space model and a second objective function of an hourly scheduling strategy, generating multiple load and energy processing scenarios, using an alternating direction multiplier method to solve the control strategy corresponding to each load and energy processing scenario, and determining the hourly scheduling strategy from the control strategies corresponding to the multiple load and energy processing scenarios; combining a rule-based fast control strategy with fuzzy logic control to determine the control frequency in the minute-level scheduling strategy, and using a fuzzy logic controller to determine the reactive power regulation amount in the minute-level scheduling strategy; performing multi-time-scale coordination based on the day-ahead scheduling strategy, the hourly scheduling strategy, and the minute-level scheduling strategy to obtain scheduling measures for the energy storage system.

[0153] The above-mentioned deterministic mixed integer linear programming model is used to describe a mathematical optimization model, in which the objective function and the constraints are linear, but the decision variables include continuous variables and integer variables.

[0154] The above-mentioned first objective function is used to describe the day-ahead scheduling strategy, and achieves specific goals through the first objective function. The above-mentioned specific goals may include but are not limited to minimizing costs or maximizing benefits. The role of the first objective function may include but is not limited to guiding the goal achievement process to ensure that the scheduling plan with the lowest cost or the highest benefit is found while meeting all constraints.

[0155] The first constraint mentioned above is used to describe the conditions that must be met in the day-ahead dispatch strategy. The first constraint may include but is not limited to the physical limitations of the energy storage system, grid operation rules, market demand, etc. The first constraint can be used to ensure the feasibility and safety of the dispatch strategy.

[0156] The above-mentioned second objective function is used to describe the realization of relevant goals through the second objective function in the hourly scheduling strategy. The above-mentioned relevant goals may include but are not limited to real-time response, adjustment speed, etc. The second objective function can be used to guide the adjustment of the hourly scheduling strategy to adapt to real-time changes.

[0157] In an optional embodiment, the day-ahead scheduling strategy is first constructed through a deterministic mixed integer linear programming model, and the first objective function and the first constraint are solved using a branch-and-bound algorithm; then, the hourly scheduling strategy is constructed through a state space model and a second objective function, and the control strategy is solved for different load and energy processing scenarios using an alternating direction multiplier method; finally, the control frequency and reactive power regulation in the minute-level scheduling strategy are determined by combining a rule-based fast control strategy and fuzzy logic control. In the above process, a comprehensive scheduling measure for the energy storage system is obtained through three levels of coordination, which realizes efficient management and optimized scheduling of energy, is conducive to improving the flexibility and response speed of the energy storage system, and at the same time ensures the stability and economy of energy supply.

[0158] According to another aspect of an embodiment of the present invention, a device for determining the resource carrying capacity of a power grid is also provided. The device can execute the method for determining the resource carrying capacity of the power grid provided in the above embodiment. The specific implementation method and preferred application scenario are the same as the above embodiment and will not be repeated here.

[0159] Figure 2 is a schematic diagram of a device for determining resource carrying capacity of a power grid according to an embodiment of the present invention, such as Figure 2As shown, the device includes: a collection module 202, which is used to collect current power grid operation data through multiple sensors installed in the power grid, wherein the current power grid operation data includes: voltage, current, active power, reactive power, distributed energy output, load demand and power grid topology information; a first evaluation module 204, which is used to dynamically evaluate the current power grid operation data using a carrying capacity evaluation model to obtain a dynamic evaluation result of the power grid distributed resource carrying capacity; a second evaluation module 206, which is used to perform a time period evaluation on the power grid distributed resource carrying capacity in a future time period to obtain a time period evaluation result; a generation module 208, which is used to generate target decision information based on the dynamic evaluation result and the time period evaluation result, wherein the target decision information includes a distributed resource access time period, a potential overload risk time period and scheduling optimization information.

[0160] An embodiment of the present application further provides an electronic device, comprising: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of the present invention when running.

[0161] An embodiment of the present application further provides a computer-readable storage medium, which includes a stored executable program, wherein when the executable program is running, the device where the computer-readable storage medium is located is controlled to execute the methods in various embodiments of the present invention.

[0162] An embodiment of the present application further provides a computer program product, including a computer program, which implements the methods in various embodiments of the present invention when executed by a processor.

[0163] An embodiment of the present application further provides a computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium is used to store a computer program, and when the computer program is executed by a processor, the method in each embodiment of the present invention is implemented.

[0164] The embodiments of the present application further provide a computer program, which implements the methods in the above-mentioned embodiments of the present invention when executed by a processor.

[0165] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0166] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0167] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0168] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0169] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0170] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for determining resource carrying capacity of a power grid, characterized in that: include: Collecting current grid operation data through multiple sensors installed in the grid, wherein the current grid operation data includes: voltage, current, active power, reactive power, distributed energy output, load demand and grid topology information; Dynamically evaluating the current power grid operation data using a carrying capacity evaluation model to obtain a dynamic evaluation result of the power grid distributed resource carrying capacity; Conduct time-based evaluation of the distributed resource carrying capacity of the power grid in the future time period and obtain time-based evaluation results; Based on the dynamic evaluation results and the time period evaluation results, target decision information is generated, wherein the target decision information includes a distributed resource access time period, a potential overload risk time period and scheduling optimization information.

2. The method according to claim 1, characterized in that: The current power grid operation data is dynamically evaluated using the carrying capacity evaluation model to obtain a dynamic evaluation result of the power grid distributed resource carrying capacity, including: Preprocessing the current power grid operation data to obtain preprocessed power grid operation data; The preprocessed power grid operation data is input into the load-bearing capacity assessment model, and the dynamic assessment result is obtained through forward propagation calculation.

3. The method according to claim 2, characterized in that Preprocessing the current power grid operation data to obtain preprocessed power grid operation data includes: Detecting the current power grid operation data to determine abnormal values ​​in the current power grid operation data; Interpolating the abnormal value to obtain first power grid operation data; performing standardization processing on the first power grid operation data to obtain second power grid operation data; constructing time features based on the second power grid operation data, calculating sliding average and differential features, and extracting frequency domain features through fast Fourier transform; The time feature, the sliding average, the differential feature and the frequency domain feature are summarized to obtain the preprocessed power grid operation data.

4. The method according to claim 1, characterized in that The distributed resource carrying capacity of the power grid in the future time period is evaluated by time period, and the evaluation results by time period are obtained, including: Use the carrying capacity prediction model to perform rolling time domain prediction on the distributed resource carrying capacity of the power grid in the future time period to obtain the rolling prediction result; Based on the rolling prediction result, dividing the distributed resource carrying capacity of the power grid in the future time period into multiple time periods; The average carrying capacity, maximum carrying capacity, minimum carrying capacity and volatility index are calculated for each time period to generate the time period evaluation results.

5. The method according to claim 1, characterized in that Based on the dynamic evaluation result and the time-division evaluation result, target decision information is generated, including: Based on the dynamic evaluation results, combined with power grid security constraints and economic indicators, an access strategy for distributed resources is generated through multi-objective optimization, wherein the power grid security constraints are used to ensure the constraints for safe operation of the power grid, and the economic indicators are used to characterize the costs and benefits of distributed resources; Generate scheduling solutions for distributed resources based on a model predictive control framework; The target decision information is generated based on the dynamic evaluation result, the access strategy, the scheduling scheme and the time period evaluation result.

6. The method according to claim 5, characterized in that The method further comprises: The dynamic evaluation results, the access strategy, the scheduling scheme and the time period evaluation results are visualized.

7. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: Collecting power grid operation data through the multiple sensors, generating a training data set, and constructing a historical database, wherein the historical database is used to store the training data set and the corresponding historical evaluation results of the power grid distributed resource carrying capacity; Based on the training data set, a deep neural network structure is used to construct the bearing capacity assessment model.

8. The method according to claim 7, characterized in that Based on the training data set, the bearing capacity assessment model is constructed using a deep neural network structure, including: Based on the training data set, a multi-branch deep neural network model is constructed, wherein the multi-branch deep neural network model includes: a convolutional neural network branch, a long short-term memory network branch and a fully connected layer, the convolutional neural network branch is used to extract power grid topology features, the long short-term memory network branch is used to capture timing features, and the fully connected layer is used to fuse the multi-dimensional features output by the convolutional neural network branch and the long short-term memory network branch; Determine a pre-trained model constructed by using power grid data, and migrate parameters of the pre-trained model to the multi-branch deep neural network model to obtain a target deep neural network model; Based on the target deep neural network model, construct multiple deep neural network sub-models with different structures, use different data subsets to train each sub-model, and integrate each sub-model to obtain an integrated model; The integrated model is subjected to dynamic adjustment of learning rate, dynamic adjustment of regularization parameters, incremental learning and concept drift detection, neural architecture search and dynamic network adjustment to obtain the carrying capacity assessment model.

9. The method according to claim 8, characterized in that The learning rate of the integrated model is dynamically adjusted, including: Calculating the validation set performance based on the integrated model; applying the cosine annealing strategy to update the learning rate when the validation set performance has not been improved in a plurality of consecutive training cycles; and updating the parameters of the integrated model; Dynamically adjusting the regularization parameters of the integrated model includes: determining the complexity of the integrated model; dynamically adjusting the regularization strength based on the complexity; adaptively adjusting the discard rate of the integrated model based on the network layer depth of the integrated model; The integrated model is subjected to incremental learning and concept drift detection, including: acquiring sample data in a data buffer; assigning weights to the sample data; incrementally updating parameters of the integrated model based on the sample data and the corresponding weights; and retraining the integrated model when a data distribution change rate of the updated model is detected to be greater than a preset change rate; The integrated model is subjected to neural architecture search and dynamic network adjustment, including: determining the architecture search space of the integrated model; optimizing the network structure of the integrated model based on a policy gradient controller; dynamically pruning the integrated model based on weight amplitude; and when the performance of the integrated model reaches a preset performance, implementing network growth to gradually increase the network width and depth of the integrated model.

10. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: Based on the current power grid operation data, a plurality of power grid evaluation indicators are determined, wherein the plurality of power grid evaluation indicators include: a voltage deviation indicator, a line load rate indicator, a distributed resource penetration rate, and a system loss rate; Based on the multiple power grid evaluation indicators, determining target optimization measures from power grid topology reconstruction measures, demand-side response measures, and energy storage system scheduling measures; Establish a dynamic real-time electricity price mechanism and determine the current electricity price based on the status of the power grid system; Based on the charge and discharge efficiency and capacity, construct an energy storage state of charge dynamic equation, and determine the energy storage state of charge using the energy storage state of charge dynamic equation; Processing the plurality of power grid evaluation indicators, the energy storage charge state, the current electricity price and the target optimization measures using a control strategy prediction model to generate a power grid control strategy; The power grid control strategy is used to perform hierarchical control.

11. The method according to claim 10, characterized in that Based on the multiple grid evaluation indicators, target optimization measures are determined from grid topology reconstruction measures, demand-side response measures, and energy storage system dispatch measures, including: Comparing the plurality of power grid assessment indicators with corresponding indicator thresholds to obtain a plurality of comparison results, wherein the indicator thresholds are adaptively adjusted based on historical data and seasonal changes; Determining a target trigger level based on the multiple comparison results and a multi-level trigger mechanism, wherein the multi-level trigger mechanism is used to characterize a mapping relationship between different comparisons and trigger levels; Based on the target trigger level, the target optimization measure is determined from the power grid topology reconstruction measure, the demand side response measure and the energy storage system scheduling measure.

12. The method according to claim 10, characterized in that The method further comprises: Generate initial reconstruction measures; based on the initial reconstruction measures, construct a multi-objective optimization function in combination with equipment switch capacity and operation number constraints, wherein the multi-objective optimization function includes: voltage deviation, line load rate, distributed resource penetration rate, system loss rate and switch operation number; solve the multi-objective optimization function to obtain the grid topology reconstruction measures; The load is divided into interruptible load, adjustable load and uncontrollable load, and a price-sensitive load response model is established; based on the load response model, the demand-side response measures are obtained; A multi-time-scale coordinated scheduling strategy is constructed, wherein the multi-time-scale coordinated scheduling strategy includes: a day-ahead scheduling strategy, an hourly scheduling strategy, and a minute-level scheduling strategy; an optimization objective function of the multi-time-scale coordinated scheduling strategy is constructed, and the optimization objective function is solved to obtain the energy storage system scheduling measures, wherein the optimization objective function includes: electricity purchase cost, energy storage depreciation cost, and ancillary service income.

13. The method according to claim 12, characterized in that The method further comprises: Incrementally updating the load response model and adaptively adjusting control parameters using a Bayesian optimization method; Determine the technical indicators and economic indicators after optimized scheduling, wherein the technical indicators after optimized scheduling include: voltage qualification rate, line utilization rate, distributed energy consumption rate and system loss rate, and the economic indicators after optimized scheduling include: operating cost, distributed energy income and user benefit; Based on the technical indicators and economic indicators after the optimized scheduling, a system performance evaluation is performed to obtain a performance evaluation result; The power grid control strategy is dynamically adjusted according to the performance evaluation result.

14. The method according to claim 12, characterized in that Constructing an optimization objective function of the multi-time scale coordinated scheduling strategy, and solving the optimization objective function to obtain the energy storage system scheduling measures, including: A deterministic mixed integer linear programming model is used to construct a first objective function and a first constraint condition of the day-ahead scheduling strategy, and a branch and bound algorithm is used to condition the first objective function based on the first constraint condition to obtain the day-ahead scheduling strategy; Constructing a state space model and a second objective function of the hourly scheduling strategy, generating multiple load and energy processing scenarios, solving the control strategy corresponding to each load and energy processing scenario using an alternating direction multiplier method, and determining the hourly scheduling strategy from the control strategies corresponding to the multiple load and energy processing scenarios; A rule-based fast control strategy combined with fuzzy logic control is used to determine the control frequency in the minute-level scheduling strategy, and a fuzzy logic controller is used to determine the reactive power adjustment amount in the minute-level scheduling strategy; Multi-time-scale coordination is performed based on the day-ahead scheduling strategy, the hourly scheduling strategy, and the minute-level scheduling strategy to obtain the energy storage system scheduling measures.

15. A device for determining resource carrying capacity of a power grid, characterized in that: include: A collection module, used to collect current grid operation data through multiple sensors installed in the grid, wherein the current grid operation data includes: voltage, current, active power, reactive power, distributed energy output, load demand and grid topology information; A first evaluation module is used to dynamically evaluate the current power grid operation data using a carrying capacity evaluation model to obtain a dynamic evaluation result of the power grid distributed resource carrying capacity; The second evaluation module is used to evaluate the distributed resource carrying capacity of the power grid in the future time period by time period and obtain the evaluation results by time period; A generation module is used to generate target decision information based on the dynamic evaluation results and the time period evaluation results, wherein the target decision information includes a distributed resource access period, a potential overload risk period and scheduling optimization information.

16. An electronic device, characterized in that: include: A memory storing an executable program; A processor, configured to run the program, wherein the program executes the method according to any one of claims 1 to 14 when running.

Images