A method and system for assessing power generation losses based on digital twins

Abstract

This invention provides a method and system for assessing power generation losses based on digital twins, relating to the field of power system energy efficiency management technology. Based on the equipment loss score, grid topology loss score, environmental loss score, and management loss score of the power grid to be assessed in a first time period, a comprehensive power loss score for the power grid to be assessed in the first time period is determined. If the comprehensive power loss score of the power grid to be assessed in the first time period is less than or equal to a scoring threshold, the comprehensive power loss score of the power grid to be assessed in the second time period is predicted. If the difference between the comprehensive power loss score of the power grid to be assessed in the first time period and the comprehensive power loss score in the second time period is greater than a difference threshold, a first warning is issued. If the comprehensive power loss score of the power grid to be assessed in the second time period is greater than the scoring threshold, a second warning is issued. This invention achieves accurate assessment, dynamic early warning, and optimization of power losses through multi-dimensional data fusion, thereby improving the power grid's generation efficiency.

Description

Technical Field

[0001] This invention relates to the field of power system energy efficiency management technology, and in particular to a method and system for assessing power production losses based on digital twins. Background Technology

[0002] In power system operation, power loss assessment is a crucial step in ensuring the economical and efficient operation of the power grid. With the increasing intelligence of the power grid and the growing complexity of its operating conditions, accurately quantifying power loss levels has become a key area of ​​ongoing technological exploration for the industry.

[0003] Existing technologies primarily assess losses through static models based on physical formulas such as Ohm's law, or by threshold comparisons of single operating parameters such as current and voltage. For example, some solutions calculate technical losses by multiplying the conductor resistance by the square of the current, or estimate line loss rates based on differences in metering data, thus constructing a basic loss assessment framework.

[0004] However, such technologies have a fundamental flaw: their evaluation logic relies on static parameters or single-dimensional data (such as focusing only on equipment resistance loss), and cannot dynamically respond to the impact of real-time operating conditions such as load fluctuations, equipment aging, and changes in ambient temperature. This results in a large deviation between the quantitative results of power grid losses and the actual operating conditions, and the evaluation accuracy is difficult to meet the needs of modern power grid refined management. In particular, under complex operating conditions, it is easy to misjudge or miss loss sources, and cannot provide accurate data support for loss reduction optimization. Summary of the Invention

[0005] This invention provides a digital twin-based method and system for assessing power generation losses. It can solve the problem that existing technologies rely on static parameters or single-dimensional data for loss assessment, which cannot dynamically respond to real-time operating conditions such as load fluctuations, equipment aging, and changes in ambient temperature, resulting in low assessment accuracy. This invention enables high-precision dynamic assessment of power losses, providing accurate data support for power grid loss reduction and optimization.

[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0007] In a first aspect, this invention provides a method for assessing power generation losses based on digital twins. This method determines the comprehensive power loss score of the power grid to be assessed in the first time period based on equipment loss scores, grid topology loss scores, environmental loss scores, and management loss scores. The power grid to be assessed includes generation units, transmission units, and regulation units. The end time of the first time period is no later than the real-time time. The equipment loss score is determined based on the equipment attribute data and operating parameter data corresponding to the generation unit, transmission unit, and regulation unit, respectively. The grid topology loss score is determined based on the topology data of the power grid to be assessed. The environmental loss score is determined based on the environmental data corresponding to the environments of the generation unit, transmission unit, and regulation unit, respectively. The management loss score is determined based on the management data corresponding to the power grid to be assessed. The management data includes the accuracy, maintenance data, and reactive power compensation switching data of the power metering devices for the power grid to be assessed. If the comprehensive power loss score of the power grid to be assessed in the first time period is less than or equal to a scoring threshold, the comprehensive power loss score of the power grid to be assessed in the second time period is predicted. The start time of the second time period is no earlier than the real-time time. If the difference between the comprehensive power loss score of the power grid under evaluation in the first time period and the comprehensive power loss score in the second time period exceeds a threshold, a first warning message will be issued. If the comprehensive power loss score of the power grid under evaluation in the second time period exceeds a threshold, a second warning message will be issued.

[0008] The beneficial effects of this invention are:

[0009] By constructing a power production loss assessment system based on digital twins, multi-dimensional data such as equipment loss, power grid topology loss, environmental loss, and management loss are integrated to form a comprehensive power loss score, which can comprehensively and accurately reflect the actual power loss status of the power grid.

[0010] This invention utilizes a digital twin model to simulate and analyze historical data. It can not only predict the power loss trend in the second time period based on the comprehensive score of the first time period, but also trigger multiple types of early warning information in a timely manner by comparing the score differences between different time periods and comparing them with thresholds, thereby achieving dynamic monitoring and early warning of power loss.

[0011] Furthermore, in the process of determining (constructing) the digital twin model, this invention combines algorithms such as deep learning, graph neural networks, random forests, and reinforcement learning to construct and integrate twin models at the device level, topology level, environment level, and management level, respectively, quantifying the mutual influence between losses in different dimensions, making the prediction process more in line with the actual operation scenario of the power grid, and significantly improving the accuracy and reliability of power loss prediction.

[0012] Furthermore, when an early warning is triggered, this invention can generate optimization strategies such as equipment maintenance and adjustment, power grid topology reconfiguration, and reactive power compensation adjustment. It can also simulate and deduce the implementation cost, risk probability, and loss reduction of the strategies through digital twin models. After assessing the feasibility of the strategies, the optimal solution is executed, realizing closed-loop management from early warning to optimization. This can effectively reduce power production losses, improve power grid operating efficiency, and provide data support and technical guarantee for the refined management and scientific decision-making of the power system.

[0013] Based on the above technical solution, the present invention can be further improved as follows.

[0014] Furthermore, the power generation unit includes at least one power generation device, the transmission unit includes at least one transmission line, and the regulation unit includes at least one transformer. The equipment attribute data for the power generation unit includes the equipment type, rated output power, commissioning time, and service life of each power generation device. The operating parameter data for the power generation unit includes the real-time output power, operating efficiency, equipment temperature, vibration frequency, and number of start-stop cycles for each power generation device. The equipment attribute data for the transmission unit includes the resistivity, cross-sectional area, and length of each transmission line. The operating parameter data for the transmission unit includes the current value, voltage value, line loss rate, temperature distribution data, and corona discharge data for each transmission line. The equipment attribute data for the regulation unit includes the capacity, turns ratio, winding material, cooling method, and tap changer type for each transformer. The operating parameter data for the regulation unit includes the load rate, winding temperature, no-load loss data, load loss data, and partial discharge data for each transformer.

[0015] The beneficial effects of adopting the above-mentioned further solutions are: clarifying the equipment attribute data (such as equipment type, resistivity, capacity, etc.) and operating parameter data (such as real-time output power, line loss rate, load rate, etc.) of the power generation unit, transmission unit, and regulation unit, providing detailed data support for equipment loss scoring, and ensuring that the evaluation results are more consistent with the actual operating status of the equipment.

[0016] Furthermore, the topology data of the power grid to be evaluated includes the number of generating units, the number of transmission lines, the number of transformers, and the connection relationships between each generating unit, each transmission line, and each transformer.

[0017] The beneficial effects of adopting the above-mentioned further scheme are: defining the power grid topology data (number of devices and connection relationships), providing a quantitative basis for power grid topology loss scoring, and enabling the assessment to reflect the impact of topology on losses, such as line tortuosity and node connection efficiency.

[0018] Furthermore, the environmental data corresponding to the power generation unit includes the temperature, humidity, temperature baseline, humidity baseline, temperature threshold, humidity threshold, and meteorological disaster index of the environment where the power generation unit is located. The environmental data corresponding to the power transmission unit includes the temperature, humidity, temperature baseline, humidity baseline, temperature threshold, humidity threshold, and meteorological disaster index of the environment where the power transmission unit is located. The environmental data corresponding to the regulation unit includes the temperature, humidity, temperature baseline, humidity baseline, temperature threshold, humidity threshold, and meteorological disaster index of the environment where the regulation unit is located.

[0019] The beneficial effects of adopting the above-mentioned further scheme are: standardizing the environmental data of each unit (temperature, humidity, meteorological disaster index, etc.), incorporating environmental factors into the loss assessment system, and enabling the environmental loss score to dynamically respond to the impact of external conditions such as temperature changes and extreme weather on power grid losses.

[0020] Furthermore, the accuracy of the power metering device is determined based on its metering error rate. Maintenance data includes the failure rate of the power grid to be evaluated and the timeliness of maintenance for that grid. Reactive power compensation switching data includes the power factor, number of switching operations, average transmission distance, and optimal transmission distance of the reactive power compensation switching equipment corresponding to the power grid to be evaluated.

[0021] The beneficial effects of adopting the above-mentioned further solutions are: clarifying management data (metering error rate, maintenance timeliness rate, reactive power compensation switching data, etc.), incorporating management-level factors (such as metering accuracy and maintenance efficiency) into loss assessment, and improving the comprehensiveness and scientific nature of management loss scoring.

[0022] Furthermore, a digital twin model corresponding to the power grid to be evaluated is determined. This digital twin model includes equipment-level twin models, topology-level twin models, environmental-level twin models, and management-level twin models. Based on the equipment-level twin model, historical equipment attribute data and historical operating parameter data corresponding to the generation unit, transmission unit, and regulation unit are simulated to determine the equipment attribute data and operating parameter data corresponding to the generation unit, transmission unit, and regulation unit in the second time period. Based on the equipment attribute data and operating parameter data corresponding to the generation unit, transmission unit, and regulation unit in the second time period, the equipment loss score of the power grid to be evaluated in the second time period is determined. Based on the topology-level twin model, historical topology data corresponding to the power grid to be evaluated is simulated to determine the topology data corresponding to the power grid to be evaluated in the second time period. Based on the topology data corresponding to the power grid to be evaluated in the second time period, the power grid topology loss score of the power grid to be evaluated in the second time period is determined. Based on the environmental-level twin model, historical environmental data corresponding to the generation unit, transmission unit, and regulation unit are simulated to determine the environmental data corresponding to the generation unit, transmission unit, and regulation unit in the second time period. Based on the environmental data corresponding to the generation unit, transmission unit, and regulation unit in the second time period, the environmental loss score of the power grid to be evaluated in the second time period is determined. Simulations were performed using a management-level twin model based on historical management data of the power grid to be evaluated, determining the management data for the power grid in the second time period. Based on this management data, the management loss score for the power grid in the second time period was determined. Finally, based on the equipment loss score, grid topology loss score, environmental loss score, and management loss score for the power grid in the second time period, the overall power loss score for the power grid in the second time period was determined.

[0023] The beneficial effects of adopting the above-mentioned further solution are: using digital twin models (equipment level, topology level, environment level, and management level) to simulate historical data, predicting the comprehensive power loss score for the second time period, and improving the accuracy and reliability of the prediction through multi-dimensional model collaboration.

[0024] Furthermore, the digital twin model is constructed. The construction of the digital twin model includes:

[0025] Based on historical equipment attribute data and historical operating parameter data corresponding to the generation, transmission, and regulation units, respectively, a deep learning algorithm is used to construct an equipment-level twin model. The deep learning algorithm includes convolutional neural networks (CNNs) and long short-term memory (LSTM) networks. The CNNs output feature vectors representing the correlation features between elements in the equipment's multidimensional parameter matrix. Each element in the equipment's multidimensional parameter matrix is ​​determined based on the historical equipment attribute data and historical operating parameter data corresponding to the generation, transmission, and regulation units, respectively. The LTM network, based on the feature vectors representing the correlation features between elements in the equipment's multidimensional parameter matrix, outputs a prediction sequence representing the evolution of equipment losses over time. This prediction sequence is used to predict the equipment attribute data and operating parameter data corresponding to the generation, transmission, and regulation units at any given time. Based on the historical topology data of the power grid to be evaluated, a graph neural network is used to construct a topology-level twin model. The graph neural network outputs a topology feature vector representing the trend of topology changes in the power grid. This topology feature vector is used to predict the topology data of the power grid to be evaluated at any given time. Based on historical environmental data corresponding to the power generation unit, transmission unit, and regulation unit, an environmental-level twin model is constructed using the random forest algorithm. The random forest algorithm outputs environmental feature vectors representing the impact of environmental factors on equipment losses. These environmental feature vectors are used to predict the environmental data corresponding to the power generation unit, transmission unit, and regulation unit at any given time. Based on historical management data corresponding to the power grid to be evaluated, a management-level twin model is constructed using the reinforcement learning algorithm. The reinforcement learning algorithm outputs management feature vectors representing the impact of management strategies on power grid losses. These management feature vectors are used to predict the management data corresponding to the power grid to be evaluated at any given time. The equipment-level twin model, topology-level twin model, environmental-level twin model, and management-level twin model are integrated to obtain a digital twin model of the power grid to be evaluated. The integration includes constructing data interaction interfaces between the twin models and a loss transmission coefficient matrix, which is used to quantify the degree of mutual influence between losses in different dimensions.

[0026] The beneficial effects of adopting the above-mentioned further scheme are: by constructing twin models at each level through algorithms such as deep learning and graph neural networks, and integrating them to form a digital twin model, the mutual influence between losses in different dimensions (loss transmission coefficient matrix) is quantified, making the model more in line with the actual operation scenario of the power grid and enhancing the accuracy of prediction.

[0027] Furthermore, if the comprehensive power loss score of the power grid to be evaluated in the first time period is greater than the score threshold, a third early warning message will be issued.

[0028] The beneficial effects of adopting the above-mentioned further scheme are: supplementing the third early warning information, triggering an early warning when the comprehensive score in the first time period directly exceeds the threshold, improving the early warning mechanism, and realizing comprehensive monitoring of the power grid loss status.

[0029] Furthermore, upon receiving a target early warning message, a power loss optimization strategy is generated based on that message. The target early warning message can be any one of a first, second, or third early warning message. The power loss optimization strategy includes at least one of equipment maintenance and adjustment strategies, grid topology reconfiguration strategies, and reactive power compensation adjustment strategies. The power loss optimization strategy for the target early warning message is input into a digital twin model for simulation, predicting the implementation cost, risk probability, and the reduction in power loss of the grid to be evaluated. If the risk probability of implementing the power loss optimization strategy is less than a preset risk probability threshold, a strategy score is determined based on the implementation cost, risk probability, and the reduction in power loss of the grid to be evaluated. If the strategy score is greater than the preset strategy score, the power loss optimization strategy is executed.

[0030] The beneficial effects of adopting the above-mentioned further solutions are: based on the early warning information, optimization strategies (equipment maintenance, topology reconfiguration, reactive power compensation adjustment, etc.) are generated, and the implementation cost, risk and loss reduction effect of the strategies are simulated and deduced through digital twin models, thereby realizing closed-loop management from early warning to optimization and effectively reducing power production losses.

[0031] Secondly, the present invention provides a power generation loss assessment system based on digital twins, comprising:

[0032] The real-time scoring module is used to determine the comprehensive power loss score of the power grid under evaluation in the first time period, based on the equipment loss score, grid topology loss score, environmental loss score, and management loss score of the power grid under evaluation in the first time period. The power grid under evaluation includes generation units, transmission units, and regulation units. The end time of the first time period is no later than the real-time time. The equipment loss score is determined based on the equipment attribute data and operating parameter data corresponding to the generation unit, transmission unit, and regulation unit, respectively. The grid topology loss score is determined based on the topology data of the power grid under evaluation. The environmental loss score is determined based on the environmental data corresponding to the environment in which the generation unit, transmission unit, and regulation unit are located. The management loss score is determined based on the management data corresponding to the power grid under evaluation. The management data includes the accuracy of the power metering devices, maintenance data, and reactive power compensation switching data for the power grid under evaluation.

[0033] The prediction and scoring module is used to predict the comprehensive power loss score of the power grid to be evaluated in the second time period if the comprehensive power loss score of the power grid to be evaluated in the first time period is less than or equal to the scoring threshold; the start time of the second time period is no earlier than the real time.

[0034] The early warning module is used to issue a first early warning if the difference between the comprehensive power loss score of the power grid to be evaluated in the first time period and the comprehensive power loss score in the second time period is greater than a difference threshold; and to issue a second early warning if the comprehensive power loss score of the power grid to be evaluated in the second time period is greater than a score threshold.

[0035] The prediction and scoring module is also used to: determine the digital twin model corresponding to the power grid to be evaluated. The digital twin model includes equipment-level twin models, topology-level twin models, environmental-level twin models, and management-level twin models. Based on the equipment-level twin model, simulations are performed on the historical equipment attribute data and historical operating parameter data corresponding to the generation unit, transmission unit, and regulation unit, respectively, to determine the equipment attribute data and operating parameter data corresponding to the generation unit, transmission unit, and regulation unit in the second time period. Based on the equipment attribute data and operating parameter data corresponding to the generation unit, transmission unit, and regulation unit in the second time period, the equipment loss score of the power grid to be evaluated in the second time period is determined. Based on the topology-level twin model, simulations are performed on the historical topology data corresponding to the power grid to be evaluated, to determine the topology data corresponding to the power grid to be evaluated in the second time period. Based on the topology data corresponding to the power grid to be evaluated in the second time period, the power grid topology loss score of the power grid to be evaluated in the second time period is determined. Based on the environmental-level twin model, simulations are performed on the historical environmental data corresponding to the generation unit, transmission unit, and regulation unit, respectively, to determine the environmental data corresponding to the generation unit, transmission unit, and regulation unit in the second time period. Based on the environmental data corresponding to the generation unit, transmission unit, and regulation unit in the second time period, the environmental loss score of the power grid to be evaluated in the second time period is determined. Simulations are performed on the historical management data of the power grid to be evaluated using a management-level twin model to determine the management data of the power grid to be evaluated in the second time period. Based on the management data of the power grid to be evaluated in the second time period, the management loss score of the power grid to be evaluated in the second time period is determined. Based on the equipment loss score, power grid topology loss score, environmental loss score, and management loss score of the power grid to be evaluated in the second time period, the comprehensive power loss score of the power grid to be evaluated in the second time period is determined.

[0036] Thirdly, the present invention provides an electronic device, comprising: a memory, one or more processors; the memory and the processors being coupled; wherein the memory stores computer program code, the computer program code including computer instructions, which, when executed by the processor, cause the electronic device to perform the method described in any of the first aspects above.

[0037] Fourthly, a computer-readable storage medium is provided, including computer instructions that, when executed on an electronic device, cause the electronic device to perform the method described in any of the first aspects above.

[0038] Fifthly, a computer program product is provided that, when run on a computer, causes the computer to perform the method described in any of the first aspects above.

[0039] It is understood that the beneficial effects achieved by the system of the second aspect, the electronic device of the third aspect, the computer-readable storage medium of the fourth aspect, and the computer program product of the fifth aspect can be referred to the beneficial effects of the first aspect and any of its possible design embodiments, which will not be repeated here. Attached Figure Description

[0040] Figure 1 A flowchart illustrating a digital twin-based method for assessing power production losses, provided in an embodiment of the present invention.

[0041] Figure 2 A schematic diagram of the composition structure of a power grid to be evaluated, provided as an embodiment of the present invention;

[0042] Figure 3 A schematic diagram of the architecture of a digital twin model provided in an embodiment of the present invention;

[0043] Figure 4 A schematic diagram illustrating the generation and execution process of a power loss optimization strategy provided in an embodiment of the present invention;

[0044] Figure 5 This is a schematic diagram of a power production loss assessment system based on digital twins, provided as an embodiment of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. In the description of this application, unless otherwise stated, " / " indicates that the objects before and after are in an "or" relationship. For example, A / B can represent A or B. "And / or" in this application is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. Furthermore, in the description of this application, unless otherwise stated, "multiple" refers to two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple. Furthermore, to facilitate a clear description of the technical solutions in the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with substantially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that "first" and "second" are not necessarily different. Meanwhile, in the embodiments of this application, words such as "exemplary" or "for example" are used to indicate that something is being used as an example, illustration, or description. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes.

[0046] See Figure 1 This invention provides a method for assessing power generation losses based on digital twins, comprising the following steps:

[0047] S101: Based on the equipment loss score, grid topology loss score, environmental loss score and management loss score of the power grid to be evaluated in the first time period, determine the comprehensive power loss score of the power grid to be evaluated in the first time period.

[0048] Among them, the equipment loss score can be used to characterize the degree of power production loss caused by equipment factors corresponding to the equipment included in the power grid to be evaluated; the power grid topology loss score can be used to characterize the degree of power production loss caused by the power grid topology of the power grid to be evaluated; the environmental loss score can be used to characterize the degree of power production loss caused by the impact of the environment on the power grid to be evaluated; and the management loss score can be used to characterize the degree of power production loss caused by management factors of the power grid to be evaluated.

[0049] Furthermore, the end time of the first time period is no later than the real-time time. Those skilled in the art can set the duration between the start and end times of the first time period based on the actual scenario and needs. For example, the duration between the start and end times of the first time period can be set to 1 hour. This application embodiment does not impose any restrictions.

[0050] See Figure 2 The power grid to be evaluated includes a generation unit, a transmission unit, and a regulation unit. The generation unit includes at least one power generation device (e.g., thermal power generation device, hydropower generation device, wind power generation device, solar power generation device, nuclear power generation device, etc.), the transmission unit includes at least one transmission line, and the regulation unit includes at least one transformer device (e.g., power transformer, special transformer (such as rectifier transformer, electric furnace transformer, etc.), instrument transformer, etc.).

[0051] In some embodiments, the equipment loss score is determined based on the equipment attribute data and operating parameter data corresponding to the power generation unit, power transmission unit, and regulation unit, respectively.

[0052] The following provides a detailed explanation of the equipment attribute data and operating parameter data corresponding to the power generation unit, power transmission unit, and regulation unit, respectively.

[0053] The equipment attribute data corresponding to the power generation unit includes the equipment type, rated output power, commissioning time and service life of each power generation device.

[0054] The operating parameter data corresponding to the power generation unit includes the real-time output power, operating efficiency, equipment temperature, vibration frequency, and number of start-stop cycles for each power generation device.

[0055] The equipment attribute data corresponding to the power transmission unit includes the resistivity, cross-sectional area, and length of each power transmission line.

[0056] The operating parameter data corresponding to the power transmission unit includes the current value, voltage value, line loss rate, temperature distribution data, and corona discharge data for each transmission line.

[0057] The equipment attribute data corresponding to the regulating unit includes the capacity, turns ratio, winding material, cooling method, and tap changer type of each transformer.

[0058] The operating parameter data corresponding to the regulating unit includes the load rate, winding temperature, no-load loss data, load loss data, and partial discharge data for each transformer device. Among them, the partial discharge data refers to the relevant electrical and acoustic characteristic data generated when the internal insulation medium of the transformer device included in the regulating unit experiences partial non-penetrating discharge due to the electric field strength exceeding the tolerance threshold during operation. This data can be used to reflect the insulation status of the equipment and potential fault hazards.

[0059] The following is a detailed explanation of the process for determining equipment loss scores based on the equipment attribute data and operating parameter data corresponding to the power generation unit, transmission unit, and regulation unit, respectively.

[0060] ;

[0061] in, Indicates equipment wear and tear score; This indicates the loss score of the power generation unit equipment. The weights for the loss scores of power generation unit equipment are indicated (those skilled in the art can set specific values ​​for these weights based on actual scenarios and needs, and this application does not limit them). This indicates the loss score of the power transmission unit equipment. The weights for the loss scores of power transmission unit equipment are indicated (those skilled in the art can set specific values ​​for these weights based on actual scenarios and needs, and this application does not limit them). This indicates the equipment loss score of the regulating unit. The weight of the adjustment unit equipment loss score is indicated (those skilled in the art can set the specific value of the weight based on the actual scenario and needs, and the embodiments of this application are not limited thereto).

[0062] ;

[0063] In relation to In the calculation formula, i represents the power generation equipment included in the power generation unit. n represents the number of power generation equipment included in the power generation unit. This represents the equipment type influence coefficient corresponding to power generation equipment i. It is determined based on the equipment type of power generation equipment i, and different equipment types of power generation equipment have different equipment type influence coefficients. This represents the real-time output power of generator i. This indicates the rated output power of generator i. This represents the influence coefficient of the commissioning time of power generation equipment i. It is determined based on the commissioning time of power generation equipment i. The longer the commissioning time of power generation equipment i, the greater the influence coefficient. The larger. This represents the operational efficiency of power generation equipment i. This represents the lifespan correction factor for power generation equipment i, determined based on the commissioning time and service life of power generation equipment i. The larger the ratio of the commissioning time to the service life of power generation equipment i, the better. The larger. This indicates the temperature of power generation equipment i. and These represent the lower limit and upper limit of the temperature of power generation equipment i, respectively. This represents the vibration frequency of power generation equipment i. and These represent the lower limit and upper limit of the vibration frequency corresponding to power generation equipment i, respectively. This indicates the number of times generator i has been started and stopped. This represents the threshold number of start-stop cycles for generator i. - These represent the corresponding parameter weights (those skilled in the art can set the specific values ​​of each parameter weight based on actual scenarios and needs, but this application embodiment does not limit this).

[0064] ;

[0065] In relation to In the calculation formula, i represents the transmission lines included in the transmission unit. n represents the number of transmission lines included in the transmission unit. The material influence coefficient of transmission line i is represented by the resistivity of transmission line i. Determine the resistivity of transmission line i. The larger, The larger. This represents the influence coefficient of the line length of transmission line i. The longer the line length of transmission line i, the greater the influence coefficient. The larger. This represents the influence coefficient of the cross-sectional area of ​​transmission line i. The larger the cross-sectional area of ​​transmission line i, the better. The larger. This represents the line loss rate of transmission line i. This represents the temperature value of transmission line i at sampling time a, and m represents the number of sampling times. This indicates the preset normal operating temperature value of transmission line i. This represents the temperature threshold of transmission line i. This represents the corona discharge value of transmission line i. and These represent the upper limit and lower limit of the power loss range for transmission line i, respectively. This represents the current value flowing through transmission line i. This represents the resistance value of transmission line i. - These represent the corresponding parameter weights (those skilled in the art can set the specific values ​​of each parameter weight based on actual scenarios and needs, but this application embodiment does not limit this).

[0066] ;

[0067] In relation to In the calculation formula, i represents the transformer equipment included in the regulating unit. n represents the number of transformer equipment included in the regulating unit. This represents the capacity influence coefficient of transformer equipment i, which is determined based on the capacity of transformer equipment i; the larger the capacity of transformer equipment i, the greater the capacity. The larger. This represents the influence coefficient of the transformer ratio of transformer i, determined based on the degree to which the transformer ratio of transformer i deviates from the standard transformer ratio; the greater the degree to which the transformer ratio of transformer i deviates from the standard transformer ratio, the greater the influence coefficient of the transformer ratio. The larger. This represents the influence coefficient of the winding material of transformer i, which is determined based on the winding material of transformer i. This represents the influence coefficient of the cooling method on transformer i, determined based on the cooling efficiency of the cooling method for transformer i; the lower the cooling efficiency of the cooling method for transformer i, the better. The larger. This represents the influence coefficient of the tap changer type of transformer equipment i, which is used to determine the failure rate based on the tap changer type of transformer equipment i. This indicates the load rate of transformer i. This indicates the winding temperature of transformer i. and This represents the lower and upper limits of the winding temperature of transformer i. This represents the no-load loss value of transformer i. This represents the upper limit of the no-load loss of transformer i. This represents the load loss value of transformer i. This represents the upper limit of load loss for transformer i. This represents the partial discharge value of transformer i. - These represent the corresponding parameter weights (those skilled in the art can set the specific values ​​of each parameter weight based on actual scenarios and needs, but this application embodiment does not limit this).

[0068] In some embodiments, the grid topology loss score is determined based on the topology data of the grid to be evaluated. The topology data of the grid to be evaluated includes the number of generating units, the number of transmission lines, the number of transformers, and the connections between the generating units, transmission lines, and transformers.

[0069] The process of determining the topology loss score of a power grid based on the topology data of the power grid to be evaluated is explained in detail below.

[0070] Assume the number of power generation devices is The number of transmission lines is The number of transformer equipment is Define the topological complexity coefficient Used to quantify the impact of the quantity and connection relationship of each power generation device, transmission line, and transformer on losses:

[0071] ;

[0072] In relation to In the calculation formula, , and These are the historical maximum numbers of similar devices. - These represent the corresponding weight coefficients (those skilled in the art can set the specific values ​​of each weight coefficient based on actual scenarios and needs, but this application embodiment does not limit this).

[0073] Further considering the impact of device connectivity on losses, we assume that the connectivity can be abstracted into a graph model. Where V is the set of device nodes and E is the set of connecting edges. Define the line detour coefficient. By calculating the actual transmission path length with shortest path length The ratio is:

[0074] ;

[0075] In relation to In the calculation formula, Indicates the length of the transmission line. This is the shortest path length from the source node to the target node, calculated using graph theory algorithms (such as Dijkstra's algorithm).

[0076] Finally, the power grid topology loss score It can be determined using the following formula:

[0077] ;

[0078] In relation to In the calculation formula, The comprehensive weighting coefficient (those skilled in the art can set the specific value of the comprehensive weighting coefficient based on actual scenarios and needs, and the embodiments of this application are not limited thereto) is used to adjust the overall influence of topology complexity and line tortuosity on the power grid topology loss score.

[0079] In some embodiments, the environmental loss score is determined based on environmental data corresponding to the environments in which the power generation unit, transmission unit, and regulation unit are located.

[0080] The environmental data corresponding to the power generation unit, power transmission unit, and regulation unit are described in detail below.

[0081] In some embodiments, the environmental data corresponding to the power generation unit includes the temperature, humidity, temperature baseline value, humidity baseline value, temperature threshold, humidity threshold, and meteorological disaster index of the environment in which the power generation unit is located.

[0082] The environmental data corresponding to the power transmission unit includes the temperature, humidity, temperature baseline value, humidity baseline value, temperature threshold, humidity threshold, and meteorological disaster index of the environment in which the power transmission unit is located.

[0083] The environmental data corresponding to the control unit includes the temperature, humidity, temperature reference value, humidity reference value, temperature threshold, humidity threshold, and meteorological disaster index of the environment in which the control unit is located.

[0084] The following is a detailed explanation of the process for determining environmental loss scores based on environmental data corresponding to the environments in which the power generation unit, transmission unit, and regulation unit are located.

[0085] ;

[0086] in, Indicates environmental degradation score. , and These represent the weights of the power generation unit, transmission unit, and regulation unit in the environmental loss score (those skilled in the art can set specific values ​​for each weight based on actual scenarios and needs; this application's embodiments do not impose such limitations). This indicates the environmental loss score of the power generation unit. This indicates the environmental loss score of the transmission unit. This indicates the environmental loss score of the regulating unit.

[0087] ;

[0088] In relation to In the calculation formula, , and These represent the corresponding parameter weights (those skilled in the art can set the specific values ​​of each parameter weight based on actual scenarios and needs, but this application embodiment does not limit this). This indicates the temperature of the environment in which the power generation unit is located. and These represent the standard temperature value and the temperature threshold value of the environment in which the power generation unit is located, respectively. This indicates the humidity of the environment in which the power generation unit is located. and These represent the standard humidity value and the humidity threshold value of the environment in which the power generation unit is located, respectively. This indicates the meteorological disaster index of the environment in which the power generation unit is located. The more severe the meteorological disasters suffered by the environment in which the power generation unit is located (for example, if the environment in which the power generation unit is located is extremely cold), the higher the index. The larger.

[0089] It should be noted that, This ensures that the impact of losses is only taken into account when the temperature is above the temperature reference value. This ensures that the impact of losses is only accounted for when the humidity is above the humidity baseline. It directly reflects the impact of extreme weather on the assessment of power generation losses in the power grid.

[0090] ;

[0091] In relation to In the calculation formula, , and These represent the corresponding parameter weights (those skilled in the art can set the specific values ​​of each parameter weight based on actual scenarios and needs, but this application embodiment does not limit this). This indicates the temperature of the environment in which the transmission unit is located. and These represent the standard temperature value and the temperature threshold value of the environment in which the power transmission unit is located, respectively. This indicates the humidity of the environment in which the transmission unit is located. and These represent the standard humidity value and the humidity threshold value of the environment in which the power transmission unit is located, respectively. This indicates the meteorological disaster index of the environment in which the transmission unit is located. The more severe the meteorological disasters in the environment where the transmission unit is located (for example, if the environment where the transmission unit is located is extremely cold), the higher the index. The larger.

[0092] ;

[0093] In relation to In the calculation formula, , and These represent the corresponding parameter weights (those skilled in the art can set the specific values ​​of each parameter weight based on actual scenarios and needs, but this application embodiment does not limit this). This indicates the temperature of the environment in which the regulating unit is located. and These represent the standard temperature value and the temperature threshold value of the environment in which the regulating unit is located, respectively. This indicates the humidity of the environment in which the regulating unit is located. and These represent the standard humidity value and the humidity threshold value of the environment in which the regulating unit is located, respectively. This represents the meteorological disaster index of the environment in which the regulating unit is located. The more severe the meteorological disasters suffered by the environment in which the regulating unit is located (for example, if the environment in which the regulating unit is located is extremely cold), the higher the index. The larger.

[0094] In some embodiments, the management loss score is determined based on the management data corresponding to the power grid to be evaluated.

[0095] The management data includes the accuracy of the power metering devices, maintenance data, and reactive power compensation switching data for the power grid under evaluation. The accuracy of the power metering devices is determined based on their metering error rate. Maintenance data includes the failure rate of the power grid under evaluation and the timeliness of maintenance for the grid. Reactive power compensation switching data includes the power factor, number of switching operations, average transmission distance, and optimal transmission distance of the reactive power compensation switching equipment corresponding to the power grid under evaluation.

[0096] The process of determining management loss scores based on management data is explained in detail below.

[0097] ;

[0098] in, -m indicates a management loss score. This indicates the impact of errors in the electricity metering device on the score. This indicates that maintenance data affects the score. This indicates that reactive power compensation switching data affects the score. - This represents the weighting coefficient of the corresponding parameter (those skilled in the art can set the specific value of the weighting coefficient of each parameter based on the actual scenario and needs, but this application embodiment does not limit this).

[0099] ;

[0100] In relation to In the calculation formula, The weighting coefficient represents the error of the power metering device (those skilled in the art can set the specific value of this parameter weight based on actual scenarios and needs, and this application embodiment does not limit it). This indicates the historical maximum metering error rate of the electricity metering device. This indicates the metering error rate of the power metering device.

[0101] ;

[0102] In relation to In the calculation formula, and These represent the weighting coefficients for the failure rate of the power grid to be evaluated and the timeliness of maintenance for the power grid to be evaluated, respectively (those skilled in the art can set specific values ​​for the weighting coefficients of failure rate and timeliness of maintenance based on actual scenarios and needs, and this application embodiment does not limit this). This indicates the highest historical failure rate of the power grid to be evaluated. This represents the on-time maintenance rate under ideal conditions (e.g., setting...). =100%). This represents the real-time failure rate of the power grid to be evaluated. This indicates the real-time maintenance timeliness rate of the power grid to be evaluated.

[0103] ;

[0104] In relation to In the calculation formula, , and The weighting coefficients for power factor, transmission distance and number of switching are respectively (those skilled in the art can set the specific values ​​of each weighting coefficient based on actual scenarios and needs, and the embodiments of this application are not limited thereto). This indicates the power factor of the reactive power compensation switching equipment. This represents the average transmission distance of the power grid to be evaluated. This represents the optimal transmission distance for the power grid to be evaluated. This represents the historical maximum average transmission distance of the power grid to be evaluated. This indicates the number of times the reactive power compensation switching equipment is switched on and off. This indicates the historical maximum number of switching operations for the reactive power compensation switching equipment.

[0105] S102: If the comprehensive power loss score of the power grid to be evaluated in the first time period is less than or equal to the score threshold, predict the comprehensive power loss score of the power grid to be evaluated in the second time period.

[0106] The start time of the second time period is no earlier than the real-time time.

[0107] In some embodiments, a digital twin model of the power grid to be evaluated can be determined. See also... Figure 3 Digital twin models include device-level twin models, topology-level twin models, environment-level twin models, and management-level twin models.

[0108] Based on equipment-level twin models, historical equipment attribute data and historical operating parameter data corresponding to the generation unit, transmission unit, and regulation unit can be simulated to determine the equipment attribute data and operating parameter data corresponding to the generation unit, transmission unit, and regulation unit in the second time period. Based on the equipment attribute data and operating parameter data corresponding to the generation unit, transmission unit, and regulation unit in the second time period, the equipment loss score of the power grid to be evaluated in the second time period can be determined.

[0109] By simulating the historical topology data of the power grid under evaluation using a topology-level twin model, the topology data of the power grid under evaluation in the second time period can be determined. Based on the topology data of the power grid under evaluation in the second time period, the topology loss score of the power grid under evaluation in the second time period can be determined.

[0110] Simulations based on historical environmental data for the power generation, transmission, and regulation units using environmental twin models allow for the determination of environmental data for each unit in the second time period. Based on this environmental data, the environmental loss score for the power grid under evaluation in the second time period can be determined.

[0111] By simulating historical management data of the power grid under evaluation using a management-level twin model, the management data for the power grid under evaluation in the second time period can be determined. Based on the management data of the power grid under evaluation in the second time period, the management loss score for the power grid under evaluation in the second time period can be determined.

[0112] Based on the equipment loss score, grid topology loss score, environmental loss score, and management loss score of the power grid to be evaluated in the second time period, the comprehensive power loss score of the power grid to be evaluated in the second time period can be determined.

[0113] The methods for determining the equipment loss score, grid topology loss score, environmental loss score, and management loss score of the power grid to be evaluated in the second time period can refer to the descriptions in the foregoing embodiments of the methods for determining the equipment loss score, grid topology loss score, environmental loss score, and management loss score of the power grid to be evaluated in the first time period, and will not be repeated here.

[0114] In some embodiments, the power generation loss assessment method based on a digital twin model provided by the present invention further includes constructing a digital twin model comprising an equipment-level twin model, a topology-level twin model, an environmental-level twin model, and a management-level twin model.

[0115] Specifically, based on the historical equipment attribute data and historical operating parameter data corresponding to the power generation unit, transmission unit, and regulation unit, respectively, deep learning algorithms can be used to construct equipment-level twin models. Specifically, in the data preprocessing stage, attribute data such as rated power and insulation level of the power generation equipment, as well as operating parameters such as real-time current and temperature, can be normalized and denoised, integrating them into a tensor containing three dimensions: equipment type, parameter category, and time series, which serves as the model input.

[0116] The deep learning algorithms include convolutional neural networks (CNNs) and long short-term memory (LSTM) networks. CNNs use multiple layers of convolutional kernels of varying sizes, such as a 3×3×3 local feature extraction kernel and a 1×1×5 global feature fusion kernel, to operate on the multidimensional parameter tensors of the equipment. This captures the coupling relationships between different parameters at the same time, such as the nonlinear correlation between transformer oil temperature and load rate, and outputs feature vectors representing the correlation features between the elements in the equipment's multidimensional parameter matrix. The elements in the multidimensional parameter matrix are determined based on historical equipment attribute data and historical operating parameter data corresponding to the power generation unit, transmission unit, and regulation unit, respectively. The LTM network uses the feature vectors output by the CNN as input and can utilize the collaborative mechanism of forget gates, input gates, and output gates to remember the long-term dependence information of equipment losses over time. For example, it analyzes the relationship between the continuous operating time of the equipment and the loss growth rate, thereby outputting a predictive sequence representing the temporal evolution of equipment losses. This predictive sequence is used to predict the equipment attribute data and operating parameter data corresponding to the power generation unit, transmission unit, and regulation unit at any given time. To improve the dynamic adaptability of the prediction, the equipment-level twin model is fine-tuned online based on real-time monitoring data.

[0117] In some embodiments, a topology-level twin model can be constructed using a graph neural network based on historical topology data corresponding to the power grid to be evaluated. To more accurately reflect the topological characteristics of the power grid, the power grid can first be abstracted as a weighted directed graph, where nodes represent components such as power generation equipment and transmission lines, and edge weights characterize parameters such as line impedance and transmission capacity. Based on this, the graph neural network aggregates information about node neighbors through graph convolution operations, and mines the impact of potential connections between nodes on losses, such as analyzing the correlation between line tortuosity and line loss.

[0118] The graph neural network is used to output a topology feature vector that represents the trend of changes in the power grid topology. To address dynamic changes that may occur during power grid operation, such as line switching and equipment failures, the topology-level twin model can enhance the features of key nodes through an attention mechanism, improving the accuracy of topology prediction. The topology feature vector is used to predict the topology data of the power grid to be evaluated at any given time, including changes in node connectivity and parameter updates.

[0119] In some embodiments, an environmental twin model can be constructed using the random forest algorithm based on historical environmental data corresponding to the power generation unit, transmission unit, and regulation unit, respectively. Considering the complexity of environmental factors, joint feature engineering can be performed on environmental data such as temperature and humidity, combined with equipment operating parameters. For example, the change in equipment insulation resistance under different humidity levels can be calculated. Multiple decision trees can be constructed using the random forest algorithm, leveraging its parallel computing advantages to analyze the nonlinear impact of environmental factors on equipment losses, such as studying the variation law of corona loss in transmission lines under heavy rain.

[0120] The random forest algorithm is used to output environmental feature vectors that characterize the impact of environmental factors on equipment losses. To quantify the influence weight of each environmental factor, the contribution of different factors to losses can be determined by calculating the change in the Gini index when the decision tree node splits. The environmental feature vectors are used to predict the environmental data corresponding to the power generation unit, transmission unit, and regulation unit at any given time. In addition, to cope with sudden extreme environments, the environmental twin model is equipped with a dynamic weight adjustment mechanism.

[0121] In some embodiments, a management-level twin model can be constructed using reinforcement learning algorithms based on historical management data corresponding to the power grid to be evaluated. Management data such as power metering accuracy and maintenance plans can be transformed into a state space, while reactive power compensation switching strategies and equipment maintenance schemes can be defined as action spaces, with grid loss minimization as the reward function. Through continuous interaction between the agent and the digital twin environment, the reinforcement learning algorithm learns optimal management strategies, such as developing differentiated maintenance plans based on equipment health status and load forecasting.

[0122] The reinforcement learning algorithm is used to output a management feature vector representing the impact of management strategies on grid losses. To address the delayed feedback problem in management decisions, an experience replay mechanism and a dual-network structure can be introduced to improve the algorithm's convergence speed. The management feature vector is used to predict the management data of the grid to be evaluated at any given time, supporting real-time decision optimization.

[0123] In some embodiments, device-level twin models, topology-level twin models, environmental-level twin models, and management-level twin models can be integrated to obtain a digital twin model corresponding to the power grid to be evaluated. During the integration process, standardized data interface protocols can be designed to enable real-time data sharing and format conversion between different models, such as passing loss prediction data from the device-level model to the topology-level model for power flow calculation.

[0124] Specifically, the integration includes constructing data interaction interfaces and loss conduction coefficient matrices between various twin models. Through historical operational data and simulation experiments, the quantitative relationships between losses in different dimensions can be determined. For example, the conduction coefficient of a 1°C change in ambient temperature on equipment and topology losses can be clarified. The loss conduction coefficient matrix is ​​used to quantify the degree of mutual influence between losses in different dimensions, thereby achieving multi-dimensional collaborative analysis and prediction of power grid losses.

[0125] In some embodiments, if the overall power loss score of the power grid to be evaluated in the first time period exceeds a scoring threshold, a third early warning message may be issued. This third early warning message indicates that the current overall power loss of the power grid has exceeded the normal operating range, and targeted loss mitigation measures need to be initiated immediately.

[0126] S103: If the difference between the comprehensive power loss score of the power grid to be evaluated in the first time period and the comprehensive power loss score in the second time period is greater than the difference threshold, the first warning information will be issued.

[0127] In some embodiments, the first warning information may include key dimension attribution of loss fluctuations. For example, analysis using a digital twin model may show that "the equipment loss score fluctuation accounts for 65%, mainly due to abnormal bearing temperature of generator set No. 3," and simultaneously output suggested measures, such as "start real-time monitoring using the equipment-level twin model and arrange infrared temperature measurement inspection within 4 hours."

[0128] S104: If the comprehensive power loss score of the power grid to be evaluated in the second time period is greater than the score threshold, a second warning message will be issued.

[0129] In some embodiments, the second early warning information can be associated with the optimization strategy library of the management-level twin model to automatically generate multiple candidate solutions (such as equipment maintenance, topology reconfiguration, and reactive power compensation combination strategies), and mark the expected loss reduction and implementation cost of each solution, such as "Solution 2: Adjust the tap changers of 5 transformers, with an expected loss reduction of 12% and a cost of 50,000 yuan".

[0130] In some embodiments, based on the calculation of the comprehensive power loss score, independent early warning thresholds and response mechanisms can be set for each dimension of the score to help quickly locate abnormal power loss sources.

[0131] Among these features, a health warning threshold can be set. When the equipment loss score exceeds the health warning threshold, a warning message can be triggered. At this time, the digital twin model automatically locates the equipment with excessive loss. For example, the equipment-level twin model analysis shows that "the temperature score of the No. 3 main transformer winding accounts for 40%, exceeding the insulation tolerance threshold," and outputs maintenance suggestions: "Recall the equipment's historical defect data and arrange oil chromatography analysis within 24 hours."

[0132] A topology efficiency warning threshold can also be set. When the topology loss score exceeds the topology efficiency warning threshold, a prompt can be triggered to generate a topology structure warning message. The graph neural network automatically analyzes the topology feature vector and outputs a reconstruction scheme accordingly.

[0133] Environmental impact warning thresholds can also be set. When the environmental loss score exceeds the environmental impact warning threshold, an environmental loss warning message can be triggered. The environmental twin model uses a random forest algorithm to identify the dominant factors. For example, if "heavy rain causes the corona discharge score of the transmission line to account for 55%, and the humidity exceeds the 90% threshold," the model will output a response strategy: "Activate the line anti-pollution flashover plan and dispatch drones to inspect the icing situation." The thresholds are dynamically adjusted based on the meteorological department's disaster warning level, automatically decreasing by 15% when the warning level is orange or higher.

[0134] A management performance warning threshold can also be set. When the management loss score exceeds the management performance warning threshold, a management loss warning message can be triggered. The management-level twin model can analyze data through a reinforcement learning strategy library, such as "reactive power compensation switching frequency reaches 30 times / day (threshold 20 times), power factor is below 0.92," and push optimization solutions: "Enable intelligent switching algorithm, adjust the compensation point to the 220kV bus." The threshold is updated regularly based on power grid management specifications and a historical loss reduction case library, and is automatically lowered by 8% before the quarterly maintenance cycle.

[0135] As can be seen, the early warning information of each dimension in this invention includes specific parameters of the dimension of loss exceeding the standard, the attribution analysis results of the digital twin model, and the associated twin model optimization strategy, forming a closed-loop management of "score monitoring - threshold triggering - model diagnosis - strategy generation".

[0136] In some embodiments, see Figure 4 The power production loss assessment method based on digital twins provided by this invention further includes the following steps:

[0137] S401: When a target warning message is provided, generate a power loss optimization strategy based on the target warning message.

[0138] Among them, the target early warning information is any one of the first early warning information, the second early warning information, and the third early warning information; the power loss optimization strategy includes at least one of the equipment maintenance and adjustment strategy, the power grid topology reconfiguration strategy, and the reactive power compensation adjustment strategy.

[0139] S402: Input the power loss optimization strategy for the target early warning information into the digital twin model for simulation and deduction, and predict the implementation cost, risk probability, and power loss reduction of the power grid to be evaluated.

[0140] Specifically, when inputting power loss optimization strategies targeting early warning information into a digital twin model for simulation, the optimization strategy can be decomposed into a set of quantifiable parameters. Taking equipment maintenance and adjustment strategies as an example, parameters such as the type of equipment to be maintained, maintenance procedure duration, and replacement component model can be input into the equipment-level twin model. Combining historical equipment defect data and real-time operating parameters, the impact of maintenance on the correlation characteristics of equipment parameters is analyzed through convolutional neural networks, and then long short-term memory networks are used to predict the temporal evolution trend of equipment losses after maintenance. For grid topology reconfiguration strategies, the line switching scheme to be adjusted and node load redistribution parameters can be input into the topology-level twin model. The changes in the topology feature vector are calculated through graph neural networks, thereby simulating the power flow distribution and line loss rate changes of the grid after reconfiguration. For reactive power compensation adjustment strategies, parameters such as the switching sequence of compensation equipment and the target power factor can be input into the management-level twin model. Combining the management feature vector output by reinforcement learning algorithms, the impact of compensation strategies on transmission distance loss and power factor loss can be predicted.

[0141] During simulation, digital twin models can quantify the mutual influence of losses in different dimensions through the loss conduction coefficient matrix. For example, the decrease in equipment losses caused by equipment maintenance will affect the calculation of topology losses through the conduction coefficient. The environmental twin model will synchronously update the parameters of the impact of environmental factors on the equipment during maintenance. Finally, through the collaborative calculation of each twin model, the implementation cost (including material costs, labor costs, and power outage loss costs) and risk probability (based on the probability of equipment failure and operational errors from more than 1,000 Monte Carlo simulations) after implementing the optimization strategy will be output, as well as the magnitude of the power loss reduction of the power grid to be evaluated in the next 72 hours (loss curve prediction with hourly time steps) will be provided.

[0142] S403: When the risk probability of implementing the power loss optimization strategy is less than the preset risk probability threshold, the strategy score of the power loss optimization strategy is determined based on the implementation cost, risk probability, and the power loss reduction of the power grid to be evaluated.

[0143] If the probability of risk in implementing the power loss optimization strategy is less than a preset risk probability threshold, the strategy score can be determined based on the following formula:

[0144] ;

[0145] In relation to In the calculation formula, Indicates the strategy score. , and All represent weighting coefficients (those skilled in the art can set specific values ​​for each weighting coefficient based on actual scenarios and needs, for example, setting...). =0.5、 , (The embodiments in this application are not limited). This indicates the predicted decrease in power loss. This indicates the largest single-cycle loss reduction in history. Indicates the predicted probability of risk. This indicates a preset risk probability threshold. Indicates implementation costs, This indicates the lowest historical implementation cost for similar strategies.

[0146] Specifically, the rate of decrease in losses needs to be normalized to the proportion of the largest historical decrease, and the risk probability is determined by 1- This is transformed into a positive indicator, and the implementation cost reflects the economic efficiency by comparing the historical minimum cost with the current cost.

[0147] For example, a certain strategy predicts a loss reduction of 2000 kW·h ( =5000), risk probability 0.2 ( =0.3), implementation cost 150,000 yuan ( If =10), then the strategy score is:

[0148] .

[0149] S404: If the strategy score based on the power loss optimization strategy is greater than the preset strategy score, execute the power loss optimization strategy.

[0150] In other words, when an early warning message is triggered, this invention can generate optimization strategies such as equipment maintenance and adjustment, power grid topology reconfiguration, and reactive power compensation adjustment. It can also simulate and deduce the implementation cost, risk probability, and loss reduction of the strategies through digital twin models. After assessing the feasibility of the strategies, the optimal solution is executed, realizing closed-loop management from early warning to optimization. This can effectively reduce power production losses, improve power grid operating efficiency, and provide data support and technical guarantee for the refined management and scientific decision-making of the power system.

[0151] In some embodiments, during the execution of power loss optimization strategies, IoT sensors can be used to collect real-time operating parameters (such as generator temperature and transmission line loss rate) after equipment maintenance, input them into the digital twin model to update the equipment's multidimensional parameter matrix, correct the predicted loss reduction curve, and ensure that the deviation between the actual loss and the simulation results does not exceed the deviation threshold.

[0152] In some embodiments, if a power grid topology reconfiguration strategy is executed, the management-level twin model can be invoked simultaneously to adjust the reactive power compensation switching strategy and dynamically optimize the compensation capacity based on the power flow distribution after reconfiguration. For example, when the load rate of a certain transmission line exceeds 80%, the reactive power compensation equipment of a nearby substation is automatically triggered to reduce the reactive power loss of the line.

[0153] In some embodiments, after implementing a power loss optimization strategy, a loss assessment report can be generated every 24 hours to compare the actual loss with the simulation prediction. If the report shows that the loss reduction is less than 80% of the predicted value for a preset number of consecutive times (e.g., 3 times), an alternative optimization strategy (such as increasing the maintenance frequency or adjusting the topology reconfiguration scheme) can be automatically activated, and the simulation and strategy scoring can be performed again.

[0154] See Figure 5 The present invention also provides a power production loss assessment system based on digital twins, comprising:

[0155] The real-time scoring module is used to determine the comprehensive power loss score of the power grid under evaluation in the first time period, based on the equipment loss score, grid topology loss score, environmental loss score, and management loss score of the power grid under evaluation in the first time period. The power grid under evaluation includes generation units, transmission units, and regulation units. The end time of the first time period is no later than the real-time time. The equipment loss score is determined based on the equipment attribute data and operating parameter data corresponding to the generation unit, transmission unit, and regulation unit, respectively. The grid topology loss score is determined based on the topology data of the power grid under evaluation. The environmental loss score is determined based on the environmental data corresponding to the environment in which the generation unit, transmission unit, and regulation unit are located. The management loss score is determined based on the management data corresponding to the power grid under evaluation. The management data includes the accuracy of the power metering devices, maintenance data, and reactive power compensation switching data for the power grid under evaluation.

[0156] The prediction and scoring module is used to predict the comprehensive power loss score of the power grid to be evaluated in the second time period if the comprehensive power loss score of the power grid to be evaluated in the first time period is less than or equal to the scoring threshold; the start time of the second time period is no earlier than the real time.

[0157] The early warning module is used to issue a first early warning if the difference between the comprehensive power loss score of the power grid to be evaluated in the first time period and the comprehensive power loss score in the second time period is greater than a difference threshold; and to issue a second early warning if the comprehensive power loss score of the power grid to be evaluated in the second time period is greater than a score threshold.

[0158] In some embodiments, the power production loss assessment system provided by the present invention can be connected to a digital twin model (e.g., Figure 5(as shown), or, the power production loss assessment system provided by the present invention may include a digital twin model.

[0159] Specifically, the power grid to be evaluated may include generation units, transmission units, and regulation units. The digital twin model may include equipment-level twin models, topology-level twin models, environmental-level twin models, and management-level twin models. The prediction and scoring module can be connected to the digital twin model and predict the comprehensive power loss score of the power grid to be evaluated for the corresponding time period based on the data output by the digital twin model. A description of the digital twin model can be found in the foregoing embodiments and will not be repeated here.

[0160] In some embodiments, the prediction scoring module is further configured to:

[0161] Determine the digital twin model corresponding to the power grid to be evaluated. The digital twin model includes equipment-level twin model, topology-level twin model, environment-level twin model, and management-level twin model.

[0162] Simulations were performed using equipment-level twin models to analyze historical equipment attribute data and historical operating parameter data for the generation, transmission, and regulation units, respectively. This determined the corresponding equipment attribute data and operating parameter data for each unit in the second time period. Based on these data, the equipment loss scores for the power grid under evaluation in the second time period were determined.

[0163] Simulations were performed using a topology-level twin model based on historical topology data of the power grid to be evaluated, determining the topology data for the power grid in the second time period. Based on this topology data, a topology loss score for the power grid in the second time period was determined.

[0164] Simulations were performed using historical environmental data corresponding to the power generation unit, transmission unit, and regulation unit based on an environmental twin model to determine the environmental data for each unit in the second time period. Based on this environmental data, the environmental loss score of the power grid under evaluation in the second time period was determined.

[0165] Simulations were performed using a management-level twin model based on historical management data of the power grid to be evaluated, determining the management data for the power grid in the second time period. Based on the management data for the power grid in the second time period, the management loss score for the power grid in the second time period was determined.

[0166] Based on the equipment loss score, grid topology loss score, environmental loss score, and management loss score of the power grid to be evaluated in the second time period, the comprehensive power loss score of the power grid to be evaluated in the second time period is determined.

[0167] In some embodiments, see continue to see Figure 5 The power production loss assessment system provided by the present invention also includes a data acquisition module, which is used to monitor the power grid to be assessed in real time to obtain real-time data for calculating each loss score (e.g., equipment attribute data and operating parameter data corresponding to the power generation unit, transmission unit and regulation unit in the second time period, respectively).

[0168] Specifically, the data acquisition module can also connect to the digital twin model and input the acquired real-time data into the digital twin model. The digital twin model can then store the real-time data input from the data acquisition module as historical data and use it for subsequent prediction outputs.

[0169] In some solutions, multiple embodiments of this application can be combined, and the combined solution can be implemented. Optionally, some operations in the processes of each method embodiment may be combined, and / or the order of some operations may be changed. Furthermore, the execution order between the steps of each process is merely exemplary and does not constitute a limitation on the execution order between steps; other execution orders are also possible. It is not intended to indicate that the execution order is the only possible order in which these operations can be performed. Those skilled in the art will conceive of various ways to reorder the operations described herein. In addition, it should be noted that the process details involved in one embodiment of this document are similarly applicable to other embodiments, or different embodiments may be combined.

[0170] Furthermore, some steps in the method embodiments can be equivalently replaced with other possible steps. Alternatively, some steps in the method embodiments may be optional and can be deleted in certain use cases. Or, other possible steps may be added to the method embodiments. Moreover, the various method embodiments can be implemented individually or in combination.

[0171] Through the above description of the implementation methods, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above.

[0172] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0173] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0174] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, in essence, or the part that contributes, or all or part of the technical solution, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0175] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A digital-twin-based power production loss assessment method, characterized by, The method comprises: determining a comprehensive power loss score of the to-be-evaluated power grid in a first time period based on a device loss score, a power grid topology loss score, an environment loss score and a management loss score corresponding to the to-be-evaluated power grid in the first time period; the to-be-evaluated power grid comprises a power generation unit, a power transmission unit and a regulation unit; the end time of the first time period is not later than a real time; the device loss score is determined based on device attribute data and operation parameter data corresponding to the power generation unit, the power transmission unit and the regulation unit respectively; the power grid topology loss score is determined based on topology structure data of the to-be-evaluated power grid; the environment loss score is determined based on environment data corresponding to the environment in which the power generation unit, the power transmission unit and the regulation unit are respectively located; the management loss score is determined based on management data corresponding to the to-be-evaluated power grid; the management data comprises the accuracy of a power metering device of the to-be-evaluated power grid, maintenance data and reactive power compensation switching data; in the case where the comprehensive power loss score of the to-be-evaluated power grid in the first time period is less than or equal to a score threshold, predicting a comprehensive power loss score of the to-be-evaluated power grid in a second time period; the start time of the second time period is not earlier than the real time; if the difference between the comprehensive power loss score of the to-be-evaluated power grid in the first time period and the comprehensive power loss score of the to-be-evaluated power grid in the second time period is greater than a difference threshold, prompting a first early warning information; if the comprehensive power loss score of the to-be-evaluated power grid in the second time period is greater than the score threshold, prompting a second early warning information; wherein the prediction of the comprehensive power loss score of the to-be-evaluated power grid in the second time period comprises: determining a digital twin model corresponding to the to-be-evaluated power grid; the digital twin model comprises a device-level twin model, a topology-level twin model, an environment-level twin model and a management-level twin model; based on the device-level twin model, simulating historical device attribute data and historical operation parameter data corresponding to the power generation unit, the power transmission unit and the regulation unit respectively to determine device attribute data and operation parameter data corresponding to the power generation unit, the power transmission unit and the regulation unit in the second time period respectively; based on the device attribute data and the operation parameter data corresponding to the power generation unit, the power transmission unit and the regulation unit in the second time period respectively, determining a device loss score of the to-be-evaluated power grid in the second time period; based on the topology-level twin model, simulating historical topology structure data corresponding to the to-be-evaluated power grid to determine topology structure data corresponding to the to-be-evaluated power grid in the second time period; based on the topology structure data corresponding to the to-be-evaluated power grid in the second time period, determining a power grid topology loss score of the to-be-evaluated power grid in the second time period; simulate, based on the environment-level twin model, historical environment data corresponding to the power generation unit, the power transmission unit, and the regulation unit respectively, to determine environment data corresponding to the power generation unit, the power transmission unit, and the regulation unit respectively in the second time period; determine, based on the environment data corresponding to the power generation unit, the power transmission unit, and the regulation unit respectively in the second time period, an environment loss score corresponding to the power grid to be evaluated in the second time period; simulate, based on the management-level twin model, historical management data corresponding to the power grid to be evaluated, to determine management data corresponding to the power grid to be evaluated in the second time period; determine, based on the management data corresponding to the power grid to be evaluated in the second time period, a management loss score corresponding to the power grid to be evaluated in the second time period; determine, based on the equipment loss score, the power grid topology loss score, the environment loss score, and the management loss score corresponding to the power grid to be evaluated in the second time period, a comprehensive power loss score corresponding to the power grid to be evaluated in the second time period.

2. The method of claim 1, wherein, The power generation unit includes at least one power generation device, the power transmission unit includes at least one power transmission line, and the regulation unit includes at least one transformer device. The equipment attribute data corresponding to the power generation unit includes the equipment type, rated output power, operation time length, and service life of each power generation device. The operation parameter data corresponding to the power generation unit includes the real-time output power, operation efficiency, device temperature, vibration frequency, and start-stop number of each power generation device. The equipment attribute data corresponding to the power transmission unit includes the resistivity, cross-sectional area, and length of each power transmission line. The operation parameter data corresponding to the power transmission unit includes the current value, voltage value, line loss rate, temperature distribution data, and corona discharge data of each power transmission line. The equipment attribute data corresponding to the regulation unit includes the capacity, transformation ratio, winding material, cooling method, and tap changer type of each transformer device. The operation parameter data corresponding to the regulation unit includes the load rate, winding temperature, no-load loss data, load loss data, and partial discharge data of each transformer device.

3. The method of claim 2, wherein, The topology structure data of the power grid to be evaluated includes the number of power generation devices, the number of power transmission lines, the number of transformer devices, and the connection relationship between each power generation device, each power transmission line, and each transformer device.

4. The method of claim 3, wherein The environment data corresponding to the power generation unit includes the temperature, humidity, temperature reference value, humidity reference value, temperature threshold value, humidity threshold value, and meteorological disaster index of the environment in which the power generation unit is located. The environment data corresponding to the power transmission unit includes the temperature, humidity, temperature reference value, humidity reference value, temperature threshold value, humidity threshold value, and meteorological disaster index of the environment in which the power transmission unit is located. The environment data corresponding to the regulation unit includes the temperature, humidity, temperature reference value, humidity reference value, temperature threshold value, humidity threshold value, and meteorological disaster index of the environment in which the regulation unit is located.

5. The method of claim 4, wherein, The accuracy of the power metering device is determined based on a metering error rate of the power metering device; the maintenance data includes a failure rate of the to-be-evaluated power grid and a maintenance timeliness rate for the to-be-evaluated power grid; and the reactive power compensation switching data includes a power factor, a switching frequency, an average power transmission distance, and an optimal power transmission distance of a reactive power compensation switching device corresponding to the to-be-evaluated power grid.

6. The method of claim 1, wherein, Before the determining the digital twin model corresponding to the to-be-evaluated power grid, the method further includes: constructing the digital twin model; The constructing the digital twin model includes: based on the historical equipment attribute data and the historical operation parameter data corresponding to the power generation unit, the power transmission unit, and the regulation unit, respectively, a deep learning algorithm is used to construct the equipment-level twin model; the deep learning algorithm includes a convolutional neural network and a long short-term memory network, the convolutional neural network is used to output a feature vector representing the correlation features between elements included in a device multi-dimensional parameter matrix; the elements included in the device multi-dimensional parameter matrix are determined based on the historical equipment attribute data and the historical operation parameter data corresponding to the power generation unit, the power transmission unit, and the regulation unit, respectively; the long short-term memory network is used to output a prediction sequence representing the time evolution law of device loss based on the feature vector representing the correlation features between elements included in the device multi-dimensional parameter matrix; the prediction sequence is used to predict the equipment attribute data and the operation parameter data corresponding to the power generation unit, the power transmission unit, and the regulation unit at any time; based on the historical topology structure data corresponding to the to-be-evaluated power grid, a graph neural network is used to construct the topology-level twin model; the graph neural network is used to output a topology structure feature vector representing the change trend of the power grid topology structure; the topology structure feature vector is used to predict the topology structure data corresponding to the to-be-evaluated power grid at any time; based on the historical environment data corresponding to the power generation unit, the power transmission unit, and the regulation unit, respectively, a random forest algorithm is used to construct the environment-level twin model; the random forest algorithm is used to output an environment feature vector representing the influence of environmental factors on device loss; the environment feature vector is used to predict the environment data corresponding to the power generation unit, the power transmission unit, and the regulation unit at any time; based on the historical management data corresponding to the to-be-evaluated power grid, a reinforcement learning algorithm is used to construct the management-level twin model; the reinforcement learning algorithm is used to output a management feature vector representing the influence of management strategies on power grid loss; the management feature vector is used to predict the management data corresponding to the to-be-evaluated power grid at any time; The equipment-level twin model, the topology-level twin model, the environment-level twin model, and the management-level twin model are integrated to obtain the digital twin model corresponding to the to-be-evaluated power grid; the integration includes constructing a data interaction interface and a loss conduction coefficient matrix between the twin models, and the loss conduction coefficient matrix is used to quantify the mutual influence degree between different dimensions of loss.

7. The method of claim 6, wherein, Further includes: In a case where the comprehensive power loss score of the to-be-evaluated power grid in the first time period is greater than the score threshold, third early warning information is prompted.

8. The method of claim 7, wherein, Also comprising: In a case where the target early warning information is prompted, a power loss optimization strategy for the target early warning information is generated; The target early warning information is any one of the first early warning information, the second early warning information, and the third early warning information; and the power loss optimization strategy comprises at least one of a device maintenance adjustment strategy, a power grid topology reconstruction strategy, and a reactive power compensation adjustment strategy; The power loss optimization strategy for the target early warning information is input to the digital twin model for simulation deduction, to predict an implementation cost, a risk probability, and a power loss reduction amplitude of the to-be-evaluated power grid of executing the power loss optimization strategy; In a case where the risk probability of executing the power loss optimization strategy is less than a preset risk probability threshold, a strategy score of the power loss optimization strategy is determined based on the implementation cost, the risk probability, and the power loss reduction amplitude of the to-be-evaluated power grid of executing the power loss optimization strategy; The power loss optimization strategy is executed based on the strategy score of the power loss optimization strategy being greater than a preset strategy score. 9.A digital-twin-based power production loss assessment system, characterized by, Comprising: A real-time scoring module is configured to determine a comprehensive power loss score of a to-be-evaluated power grid in a first time period based on a device loss score, a power grid topology loss score, an environmental loss score, and a management loss score of the to-be-evaluated power grid in the first time period; the to-be-evaluated power grid comprises a power generation unit, a power transmission unit, and a regulation unit; an end time point of the first time period is not later than a real time point; the device loss score is determined based on device attribute data and operation parameter data corresponding to the power generation unit, the power transmission unit, and the regulation unit; the power grid topology loss score is determined based on topology structure data of the to-be-evaluated power grid; the environmental loss score is determined based on environmental data corresponding to an environment in which the power generation unit, the power transmission unit, and the regulation unit are respectively located; and the management loss score is determined based on management data corresponding to the to-be-evaluated power grid; the management data comprises accuracy of a power metering device for the to-be-evaluated power grid, maintenance data, and reactive power compensation switching data; A prediction scoring module is configured to, in a case where the comprehensive power loss score of the to-be-evaluated power grid in the first time period is less than or equal to a score threshold, predict a comprehensive power loss score of the to-be-evaluated power grid in a second time period; A start time point of the second time period is not earlier than the real time point; An early warning module is configured to, if a difference between the comprehensive power loss score of the to-be-evaluated power grid in the first time period and the comprehensive power loss score of the to-be-evaluated power grid in the second time period is greater than a difference threshold, prompt first early warning information; and if the comprehensive power loss score of the to-be-evaluated power grid in the second time period is greater than the score threshold, prompt second early warning information; In the prediction scoring module is configured to predict the comprehensive power loss score of the to-be-evaluated power grid in the second time period, the prediction scoring module is specifically configured to: determine a digital twin model corresponding to the power grid to be evaluated; the digital twin model comprises a device-level twin model, a topology-level twin model, an environment-level twin model, and a management-level twin model; simulate historical device attribute data and historical operation parameter data corresponding to the power generation unit, the power transmission unit, and the regulation unit based on the device-level twin model, to determine device attribute data and operation parameter data corresponding to the power generation unit, the power transmission unit, and the regulation unit in the second time period, respectively; determine a device loss score corresponding to the power grid to be evaluated in the second time period based on the device attribute data and the operation parameter data corresponding to the power generation unit, the power transmission unit, and the regulation unit in the second time period, respectively; simulate historical topology structure data corresponding to the power grid to be evaluated based on the topology-level twin model, to determine topology structure data corresponding to the power grid to be evaluated in the second time period; determine a power grid topology loss score corresponding to the power grid to be evaluated in the second time period based on the topology structure data corresponding to the power grid to be evaluated in the second time period; simulate historical environment data corresponding to the power generation unit, the power transmission unit, and the regulation unit based on the environment-level twin model, to determine environment data corresponding to the power generation unit, the power transmission unit, and the regulation unit in the second time period, respectively; determine an environment loss score corresponding to the power grid to be evaluated in the second time period based on the environment data corresponding to the power generation unit, the power transmission unit, and the regulation unit in the second time period, respectively; simulate historical management data corresponding to the power grid to be evaluated based on the management-level twin model, to determine management data corresponding to the power grid to be evaluated in the second time period; determine a management loss score corresponding to the power grid to be evaluated in the second time period based on the management data corresponding to the power grid to be evaluated in the second time period; determine a comprehensive power loss score corresponding to the power grid to be evaluated in the second time period based on the device loss score, the power grid topology loss score, the environment loss score, and the management loss score corresponding to the power grid to be evaluated in the second time period.

Images