Power supply scheme optimization management method and system based on dynamic sensing

Abstract

The application discloses a power distribution network power supply scheme optimization management method and system based on dynamic sensing, and belongs to the technical field of power supply management. The system comprises a dynamic sensing module, a power distribution analysis module, a power supply optimization module and an execution management module; the dynamic sensing module is used for collecting GIS maps and power consumption data, and operation parameters and environmental parameters of all power supply objects; a three-dimensional scene is built and dynamically mapped; the power distribution analysis module divides and identifies domains according to the relative positions of the power supply objects in the three-dimensional scene, constructs three-dimensional air flow fields for the respective domains and predicts future wind field data, and fits power relationship formulas for the power supply objects; the power supply optimization module fits power consumption curves through the power consumption data, constructs power generation reference curves in combination with the future wind field data and the power relationship formulas; different power generation strategies are established and schemes are randomly built, the stability indexes of the schemes are calculated to set the power supply scheme; and the execution management module adjusts according to the power supply scheme and records the operation parameters and the environmental parameters.

Description

Technical Field

[0001] This invention relates to the field of power supply management technology, specifically to a method and system for optimizing power supply schemes in distribution networks based on dynamic sensing. Background Technology

[0002] With the continuous increase in the penetration rate of renewable energy in the power system, the scale of grid connection of fluctuating power sources, represented by wind power, is constantly expanding, posing unprecedented challenges to the real-time balancing capability, operational stability, and overall intelligent control level of the distribution network. The power grid operation mode is undergoing a profound transformation from the traditional source-following-load dynamic to source-load interaction.

[0003] Currently, existing technologies have limitations in addressing this complex and dynamic scenario. First, the perception and modeling of environmental factors and equipment status are fragmented, often relying on simplified or static correlation models, making it difficult to depict the real-time coupling and spatiotemporal evolution of multiple physical fields. Second, the description of the dynamic response of power generation units is too simplistic, failing to fully consider the complex nonlinear interaction mechanisms between their core control parameters and environmental variables. Finally, the optimization decision-making process often focuses on single objectives such as short-term economic efficiency or power point tracking, lacking forward-looking and coordinated consideration of the long-term operational stability of equipment and the overall flexibility of system behavior, resulting in a difficulty in balancing technical feasibility and engineering robustness. There is an urgent need to construct a new management paradigm capable of accurately perceiving, predicting in advance, and collaboratively optimizing the demand for distributed generation units and grid load. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for optimizing and managing power distribution network schemes based on dynamic sensing, so as to solve the problems mentioned in the background art.

[0005] To address the aforementioned technical problems, this invention provides a method for optimizing and managing power distribution network schemes based on dynamic sensing, comprising:

[0006] S100 collects GIS maps and electricity consumption data, as well as the operating parameters and environmental parameters of all power supply objects within the distribution network coverage area; it uses digital twin technology to build a three-dimensional scene based on the GIS map and dynamically maps it in combination with real-time operating parameters;

[0007] S200. Based on the relative positions of each power supply object in the 3D scene and environmental parameters, divide the recognition domain for each power supply object, construct a 3D airflow field for each recognition domain and predict future wind field data; and fit the power relationship formula based on the operating parameters and environmental parameters of each power supply object.

[0008] S300: First, fit the electricity consumption data to obtain the electricity consumption curve, then combine the future wind farm data and power relationship formula to construct the power generation reference curve and set the time point; establish different power generation strategies for each time point and randomly form a scheme; calculate the stability index of each scheme, and select the scheme with the largest stability index to set the power supply scheme.

[0009] S400: Each power supply object is adjusted according to the power supply plan, and operating parameters and environmental parameters are recorded in real time.

[0010] Furthermore, in step S100, the GIS map is used to describe the relative location and distribution of each power supply object within the distribution network coverage area; the power consumption data includes the total power consumption at different times.

[0011] The distribution network coverage area refers to the physical area jointly covered by the power supply service and data collection range of the distribution network; the power supply object refers to wind turbines.

[0012] The operating parameters include a 3D model, as well as blade pitch angles and power generation at different times; the environmental parameters include 3D wind speed vectors at different times.

[0013] Building a 3D scene and performing dynamic mapping specifically includes:

[0014] S101. Extract the 3D model of each power supply object, analyze the appearance size parameters and feature contour data of each 3D model; connect to the GIS map through digital twin technology, and generate a 3D scene using a 3D engine;

[0015] S102. Based on the relative positions and distribution of each power supply object within the distribution network coverage area in the GIS map, load each three-dimensional model at the corresponding position in the three-dimensional scene, and dynamically map the real-time blade pitch angle and power generation on the three-dimensional model.

[0016] A high-fidelity dynamic digital twin foundation was established. Static geographic information was deeply integrated with the dynamic operating status of equipment to generate a three-dimensional virtual scene that is synchronized with the physical world in real time and can be interactively analyzed.

[0017] It changes the traditional monitoring model that relies mainly on reports and data points, and achieves panoramic transparency and visual perception of power supply objects, providing a unified spatiotemporal context and accurate data source for all subsequent advanced analyses.

[0018] Further, step S200 includes:

[0019] S201. Obtain the real-time environmental parameters of each power supply object and calculate the attenuation distance using the three-dimensional wind speed vector; divide the recognition domain for each power supply object in the three-dimensional scene according to the attenuation distance, and fuse all intersecting and overlapping recognition domains;

[0020] S202. The Kriging spatial interpolation algorithm and the Navier-Stokes equations are used to simplify the model. The CFD physical simulation model is used to fuse discrete point data based on environmental parameters to simulate and construct three-dimensional airflow fields for each identification domain and update them in real time.

[0021] Discrete point data refers to a set of observation sampling points that are isolated and discontinuously distributed in three-dimensional space, specifically a comprehensive expression of wind speed, wind direction, time, and spatial location;

[0022] S203. Perform empirical mode decomposition on each three-dimensional airflow field, extract the intrinsic mode functions that characterize different spatiotemporal scales, and select the dominant modes with high energy proportions as feature bases.

[0023] S204. Input the feature base and its corresponding time coefficient sequence into the trained long short-term memory and convolutional neural network coupled model. The coupled model learns the evolution law of spatiotemporal features.

[0024] S205. Using the current modal coefficients and environmental boundary conditions as initial inputs, a multi-step recursive prediction is performed through a coupled model to obtain the future... The sequence of modal coefficients at each time point within a second;

[0025] S206. Perform linear reconstruction of the predicted mode coefficient sequence and the feature basis to generate the future... The three-dimensional flow field data corresponding to each second is used as future wind field data;

[0026] A hybrid prediction framework combining CFD simulation, empirical mode decomposition, and LSTM-CNN coupled model is adopted.

[0027] First, CFD is used to obtain high-resolution spatial flow field details. Then, EMD is used to decompose the complex turbulence into intrinsic mode functions of different scales and extract the dominant characteristic basis.

[0028] Finally, a coupled model is used to capture temporal evolution using LSTM and extract spatial features using CNN. The temporal coefficients of the feature basis are recursively predicted and then reconstructed to obtain the future wind field.

[0029] Breaking through the limitations of traditional single numerical weather prediction or statistical methods, it can provide high-precision three-dimensional wind flow evolution information at the wind turbine location scale, from minutes to seconds.

[0030] S207. Obtain the operating parameters and environmental parameters of each power supply object, take the power generation at the same time as the dependent variable, and the blade pitch angle and three-dimensional wind speed vector as the independent variables, and fit the power relationship for each power supply object.

[0031] Further, step S201 includes:

[0032] S2011, According to the power supply object Real-time environmental parameters, analyzing three-dimensional wind speed vectors Calculate the attenuation distance in three dimensions. The formula is as follows:

[0033] ;

[0034] In the formula, The preset wind speed threshold, The preset empirical attenuation coefficient reflects the influence of environmental turbulence dissipation and surface roughness;

[0035] Flat and open terrain: Complex and rough terrain: ;

[0036] Input the real-time wind speed vector, and perform logarithmic and proportional calculations by combining the preset wind speed threshold with empirical coefficients that reflect the effects of environmental turbulence dissipation and surface roughness.

[0037] Based on the logarithmic wind speed profile and turbulent kinetic energy dissipation theory of fluid mechanics, the attenuation trend and spatial scale of wind speed in the direction of the dominant flow behind the rough element or in the wake region under neutral atmospheric conditions are dynamically quantified, thus providing a dynamic boundary basis for the physical-driven three-dimensional simulation domain division.

[0038] S2012, Obtain the power supply object in the 3D scene Coordinate axes of location ,by With the corresponding spatial location as the center point, axes with a length of twice the attenuation distance are set in the three-dimensional directions respectively;

[0039] S2013. Using the three axes as a reference, construct a three-dimensional convex polyhedral space as the power supply object. The recognition domain, and the recognition domain can completely enclose the smallest circumscribed space region of the three axes;

[0040] S2014. In the 3D scene, divide the recognition domain for each power supply object and merge all intersecting and overlapping recognition domains into one recognition domain. The merged recognition domain serves as the common recognition domain for all power supply objects before fusion.

[0041] By using the attenuation distance formula derived from the logarithmic wind speed profile, the spatial range of the wake effect and wind speed effect of each wind turbine is dynamically defined.

[0042] Compared to a fixed radius range, it can more physically reflect the actual spatial scale of wake effects and wind speed influence effects under different wind speeds and terrains, and fuses overlapping domains, thus constructing a dynamic computing environment that conforms to physical laws for subsequent wind flow simulation.

[0043] Further, step S207 includes:

[0044] S2071, According to the power supply object In historical periods The internal operating parameters and environmental parameters are set evenly. At each historical moment, the power generation, blade pitch angle, and three-dimensional wind speed vector are obtained;

[0045] S2072, Power generation at the same historical moment As the dependent variable, the blade pitch angle and three-dimensional wind speed vector As independent variables, the dependent variable and all independent variables at each historical moment are packaged into a sample;

[0046] S2073, This The independent variables in each sample are used as input values ​​for the electricity relationship formula, and the difference between the output value and the dependent variable for each sample is used as the deviation value; the electricity relationship formula is as follows:

[0047] ;

[0048] In the formula, The comprehensive efficiency coefficient covers the inherent losses of mechanical transmission, generator and converter, and is a constant or slowly time-varying parameter;

[0049] Air density, unit: kg / m³, is used as an input environmental physical quantity. The swept area of ​​the wind turbine is measured in m² and is determined by the model of the wind turbine.

[0050] The effective wind speed is the vector composite wind speed, in m / s, and is calculated using the following formula:

[0051] ;

[0052] Three-dimensional wind speed vector Including axial, lateral and vertical components ; This is the unit vector along the wind turbine axis, and its value is determined by the yaw angle.

[0053] The effective wind speed is synthesized by vector synthesis. The three-dimensional wind speed vector is synthesized into a single scalar through weighted norm, which not only highlights the dominant axial component, but also quantifies the complex influence of wind conditions in multiple directions.

[0054] and These are weighting coefficients, used to quantify the contributions of axial wind speed and lateral / vertical wind speed, respectively. Typically... ;

[0055] The core independent variable of the pitch angle adjustment function is expressed as follows:

[0056] ;

[0057] and These are the fitting coefficients. The first term... The nonlinear decay of aerodynamic efficiency during pitch control; the second term Capture the linear regulation effect that may exist under specific working conditions;

[0058] It breaks through the traditional lookup table or fixed function form, and uses the superposition of exponential and linear terms to flexibly characterize the pitch angle adjustment characteristics in the whole working condition range;

[0059] This is the coupling correction function, used to characterize the interaction between the wind speed vector and the pitch angle, and its expression is:

[0060] ;

[0061] It is the magnitude of the lateral component of the wind speed vector in the wind turbine plane, used to reflect yaw error or the intensity of non-axial incoming flow. The reference wind speed is used to achieve saturation characteristics. The coupling strength coefficient is obtained through data fitting. The function is used to describe the smooth transition of the coupling effect between the pitch angle and the lateral wind speed;

[0062] The explicit introduction of a coupling term between wind speed and blade pitch angle can characterize the power modulation phenomenon caused by their interaction.

[0063] This framework establishes a high-order nonlinear mapping relationship between power generation, wind speed vector, and blade pitch angle.

[0064] The system is constructed by combining physical constants including overall efficiency, air density, and swept area; a vector weighting function that synthesizes three-dimensional wind speed into an effective scalar; an exponential-linear combination function that characterizes the nonlinear adjustment effect of pitch angle on aerodynamic efficiency; and a coupling function that describes the interaction between wind speed and pitch angle.

[0065] Breaking away from traditional lookup table or fixed function models, this approach uses a fusion of mechanism and data to accurately characterize the complex response characteristics of wind turbine power to environmental inputs and control commands across the entire operating range, providing a core model foundation for power prediction and optimized control.

[0066] S2074. The sum of the deviations of all samples is used as the deviation index. The minimum deviation index is obtained by adjusting the coefficients. After fixing the values ​​of the coefficients, the trained power supply object is obtained. The electrical relationship formula;

[0067] S2075, and so on, to fit the trained power relationship for each power supply object.

[0068] It enables advanced prediction of complex airflow environments and accurate quantitative modeling of wind turbine response.

[0069] Further, step S300 includes:

[0070] S301, Extracting historical time periods The electricity consumption data is collected and processed according to a preset cycle. The process is divided into several parts, and curves are plotted based on the changes in total power consumption in each cycle. The curves of all cycles are then fitted to obtain the power consumption curve.

[0071] S302. Create a line graph with time as the horizontal axis and power as the vertical axis, and map it onto the electricity consumption curve, setting the rate of increase. and decline Based on the electricity consumption curve, it fluctuates upwards. downward fluctuation Establish the first trend interval for the boundary;

[0072] S303, Analyzing future wind field data Substituting the three-dimensional wind speed vector changes of each power supply object within a second into the electrical relationship formula of each power supply object, we obtain the future... A model showing the relationship between power generation and blade pitch angle at different times within a second;

[0073] S304. By adjusting the blade pitch angle in the relational model, obtain the maximum and minimum power generation at each time point, and calculate the sum of the maximum power generation of all power supply objects at the same time point, as well as the sum of the minimum power generation.

[0074] S305, Mark the relative position of the current moment within the period and map it onto the line chart as the starting point, based on future... The sum of the maximum power generation and the sum of the minimum power generation at each time point within a second are used as boundaries to establish the second trend interval;

[0075] S306. Take the area where the first trend interval and the second trend interval in the line graph intersect as the power generation reference interval, construct a power generation reference curve within the power generation reference interval, and set time points evenly on the power generation reference curve at preset intervals.

[0076] S307. Establish different power generation strategies for each time point, and construct different schemes based on the power generation strategies at each time point. Calculate the stability index of each scheme, and select the scheme with the highest stability index as the power supply scheme.

[0077] Further, step S306 includes:

[0078] S3061. Within the power generation reference interval, uniformly set sampling points and analyze the time and power corresponding to each sampling point. Group sampling points at the same time into the same category, and multiply the number of sampling points in all categories to obtain the result. ;

[0079] S3062, Quantity of Statistical Categories and establish Each group contains [number] items. The sampling points come from different classes, and the sampling points in different combinations are not completely the same;

[0080] S3063. Calculate the standard deviation of the power corresponding to all sampling points in each combination, mark all sampling points in the combination with the smallest standard deviation, and use a smoothing algorithm to construct a power generation reference curve based on all marked sampling points.

[0081] By fitting the demand range using historical electricity consumption data and combining it with future wind farm forecasts and wind turbine models to calculate the supply range, the intersection of the two forms the power generation reference range, ensuring the feasibility of the plan.

[0082] Within this range, by optimizing the combination of sampling points, a power generation reference curve with the most stable power change is generated, aiming to smooth out fluctuations.

[0083] Further, step S307 includes:

[0084] S3071, Obtaining Time Points power And each power supply object at a given time point A model showing the relationship between power generation and blade pitch angle was developed, allowing different power generation rates to be obtained by adjusting the blade pitch angle.

[0085] S3072, Establishment Each power generation strategy includes different blade pitch angles set for each power supply object, and the sum of the power generation corresponding to the blade pitch angles of all power supply objects within each power generation strategy is equal to... ;

[0086] In each power generation strategy, the order of the power supply objects is consistent, and all blade pitch angles are included in a one-to-one correspondence according to the order of the power supply objects.

[0087] S3073, establish for each time point The result is obtained by multiplying the number of power generation strategies at all time points. Count the number of all time points. and establish One option;

[0088] S3074, Randomly place within each scheme There are several power generation strategies from different points in time. All power generation strategies are arranged in chronological order. The power generation strategies within different schemes are not exactly the same.

[0089] S3075, Analysis Scheme All power generation strategies are analyzed, and the blade pitch angles set for each power supply object at the corresponding time point for each power generation strategy are analyzed. Weights are assigned to each power supply object. Calculation scheme stability index :

[0090] ;

[0091] In the formula, For the first The power supply object is in the first The blade pitch angle set at each time point, For the first The power supply object is in the first The blade pitch angle set at each time point. For the first The maximum adjustable blade pitch angle for each power supply object;

[0092] The stability index is used to quantify and evaluate the operational stability of a multi-period power supply scheme for wind turbine generators.

[0093] By calculating the blade pitch angle adjustment range of each power supply object in the scheme between adjacent execution time points, a weighted penalty calculation based on a negative exponential function is performed, and the penalty values ​​of all power supply objects and all time periods are summed to obtain the final exponent;

[0094] The engineering objectives of reducing mechanical wear and impact on equipment and preventing the risk of increased long-term variables during large-scale adjustments of a small number of power supply targets are transformed into calculable mathematical optimization objectives.

[0095] The more power supply objects are adjusted, the smaller the magnitude, the shorter the time consumed and the lower the risk. The higher the index value, the smoother and more continuous the pitch angle action sequence of all wind turbines in the scheme, thereby effectively extending the service life of key mechanical components and optimizing the total life cycle operating cost while meeting the power generation requirements;

[0096] And so on, calculating the stability index for each scheme.

[0097] The smoothness of the scheme is quantified by defining a stability index. An exponential decay function is used to penalize the pitch angle variation of all wind turbines and all adjacent time points in the scheme.

[0098] The smaller the change, the higher the index value. This index can automatically avoid aggressive strategies that, while meeting power requirements, would lead to frequent and drastic changes in the pitch angle, thereby selecting the optimal power supply scheme that minimizes the mechanical impact on the wind turbine drive system and pitch mechanism, achieving the best balance between power generation benefits and equipment lifespan.

[0099] Further, in step S400, the blade pitch angle set by each power supply object at each time point in the power supply scheme is analyzed, and each power supply object adjusts the blade pitch angle to the corresponding set value when it arrives at each time point.

[0100] Real-time recording of blade pitch angle, power generation, and three-dimensional wind speed vector serves as operational and environmental parameters.

[0101] The present invention also provides a power distribution network optimization management system based on dynamic perception, including a dynamic perception module, a power distribution analysis module, a power supply optimization module and an execution management module.

[0102] The dynamic sensing module is used to collect GIS maps and electricity consumption data, as well as the operating parameters and environmental parameters of all power supply objects. It builds a 3D scene and dynamically maps it.

[0103] Collect geographic information system maps, total power grid consumption data, and real-time operating parameters and environmental parameters of all wind turbines.

[0104] By using digital twin technology, GIS maps are combined with 3D models of each wind turbine to build a 3D virtual scene of the power distribution network coverage area, and real-time data is dynamically mapped onto the corresponding model to realize a digital mirror of the physical world.

[0105] Construct a high-fidelity, real-time 3D digital twin environment that reflects the state of the physical world. This environment intuitively displays the geographical distribution of wind farms and the real-time status of equipment, providing a unified, accurate, and dynamic data foundation and visualization platform for subsequent refined analysis and optimization. It is the basis for achieving dynamic perception.

[0106] The power distribution analysis module divides the identification domains according to the relative positions of each power supply object in the three-dimensional scene. Each identification domain constructs a three-dimensional airflow field and predicts future wind field data, thus fitting the power relationship formula for each power supply object.

[0107] The attenuation distance is calculated based on the location of the wind turbine and the wind speed. A three-dimensional space for the wind speed influence is defined for each wind turbine, and adjacent domains are merged.

[0108] Within the identification domain, real-time monitoring data is fused, and a three-dimensional airflow field is constructed using computational fluid dynamics simulation. A coupled model of empirical mode decomposition and LSTM-CNN is then used to predict short-term future wind field data.

[0109] Meanwhile, a nonlinear power relationship is fitted for each wind turbine, with wind speed vector and pitch angle as independent variables and power generation as dependent variable. The formula introduces vector synthesis effective wind speed, pitch angle adjustment function and wind speed-pitch angle coupling correction function.

[0110] It enables advanced and refined prediction of complex wind flow in wind farms and performs high-precision mathematical modeling of the relationship between environmental input and equipment response.

[0111] This method breaks through the traditional approach of simplifying wind turbine power estimation, significantly improving the ability to describe the complex wind flow interaction and nonlinear power generation characteristics, and serves as the core basis for optimization decision-making.

[0112] The power supply optimization module fits power consumption curves to power consumption data and constructs a power generation reference curve by combining future wind farm data and power relationship formulas. Different power generation strategies are established and schemes are randomly generated; the stability index of each scheme is calculated to determine the power supply scheme.

[0113] Based on historical data, a typical power consumption curve of the power grid is fitted and its fluctuation range is established. Combining future wind farm data and the power relationship formula of each wind turbine, the total power generation potential range of the entire field in the next S seconds can be calculated, thus forming a power generation reference range.

[0114] Within the intersecting interval, aiming for stable power output, a power generation reference curve is generated, and multiple future time points are set on it. For each time point, the module assigns different pitch angle combinations to each wind turbine, generating multiple power generation strategies.

[0115] Subsequently, strategies at various time points are randomly combined to form numerous complete power supply scheme candidates. Finally, a stability index is defined to evaluate each scheme. This index quantifies the smoothness of pitch angle adjustment of all wind turbines at all time points in the scheme through an exponential decay function, and the scheme with the highest index is selected as the final power supply scheme to be implemented.

[0116] It achieves multi-timescale and multi-objective collaborative optimization. It can automatically and intelligently select the solution with the least mechanical impact on the wind turbine and the most stable operation from numerous possible solutions that meet the power demand of the grid. It effectively balances the economics of power generation with the safety of equipment, improving the robustness and lifespan of the entire wind power distribution network system.

[0117] The execution management module adjusts according to the power supply scheme and records operating parameters and environmental parameters.

[0118] The optimized power supply scheme is then distributed to each wind turbine for execution. The core operation involves analyzing the blade pitch angle that each turbine should be set at each future time point in the scheme, and controlling the turbine to adjust the pitch angle to the target value when these points are actually reached.

[0119] This completes the final closed loop from perception, analysis, decision-making to control, ensuring the implementation of optimization strategies. Simultaneously, new operating parameters of the wind turbine and environmental parameters are continuously recorded during execution. This data can be fed back to the aforementioned modules for updating the model and iterative optimization, thus forming a dynamic optimization management system with self-learning and continuous improvement capabilities.

[0120] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0121] Comprehensive Dynamic Perception and Visual Modeling: This solution integrates Geographic Information Systems (GIS) and Digital Twin technology to construct a three-dimensional dynamic scene synchronized with the physical world in real time. This enables panoramic and transparent perception of the operating status and environmental parameters of all power supply objects within the distribution network coverage area. Compared to the discrete, two-dimensional monitoring methods in existing technologies, this approach provides a unified, high-fidelity spatiotemporal data foundation, which can intuitively reflect the spatial correlation and interaction effects between devices, laying a precise perceptual foundation for advanced analysis and decision-making.

[0122] High-precision windflow prediction and accurate response modeling: A hybrid prediction framework combining computational fluid dynamics simulation, empirical mode decomposition, and deep learning is employed to achieve refined prediction of three-dimensional windflow fields at the turbine location level and in the short term. Simultaneously, by integrating mechanistic and data-driven power relations, vector-synthesized effective wind speed, nonlinear pitch angle adjustment functions, and wind speed-pitch angle coupled correction functions are introduced. This overcomes the limitations of traditional simplified models or fixed power curves, significantly improving the modeling accuracy and prediction reliability of turbine power characteristics under complex wind conditions.

[0123] Multi-objective collaborative optimization and operational stability assurance: Based on historical power consumption curves and future power generation potential ranges, the scheme generates a smooth power generation reference curve. Within this framework, it automatically selects power supply schemes through random combination and stability index evaluation. The stability index quantifies the continuity and smoothness of pitch angle action, incorporating the reduction of equipment mechanical shock and fatigue wear into the optimization objectives. This achieves proactive collaborative optimization of equipment operational stability and life-cycle cost while meeting grid power demand, overcoming the shortcomings of existing technologies that often focus on economics while neglecting engineering robustness. Attached Figure Description

[0124] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0125] Figure 1 This is a flowchart illustrating the power distribution network optimization management method based on dynamic sensing according to the present invention.

[0126] Figure 2 This is a schematic diagram of the power generation reference range of the present invention;

[0127] Figure 3 This is a schematic diagram of the structure of the power distribution network optimization management system based on dynamic sensing of the present invention;

[0128] Figure 4 This is a bar chart comparing the core performance indicators of the present invention;

[0129] Figure 5 This is a line graph showing the 72-hour power generation tracking error of the present invention;

[0130] Figure 6 This is a bar chart comparing the core performance indicators of the present invention under different wind conditions. Detailed Implementation

[0131] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0132] Example 1: Please refer to Figure 1 This invention provides a method for optimizing and managing power distribution network schemes based on dynamic sensing, including:

[0133] S100 collects GIS maps and electricity consumption data, as well as the operating parameters and environmental parameters of all power supply objects within the distribution network coverage area; it uses digital twin technology to build a three-dimensional scene based on the GIS map and dynamically maps it in combination with real-time operating parameters.

[0134] In practice, GIS maps are used to describe the relative location and distribution of various power supply objects within the distribution network coverage area. Electricity consumption data includes the total power consumption at different times.

[0135] The distribution network coverage area refers to the physical area jointly covered by the power supply service and data collection range of the distribution network. The power supply object refers to wind turbines.

[0136] The operating parameters include a 3D model, as well as blade pitch angles and power generation at different times. The environmental parameters include 3D wind speed vectors at different times.

[0137] Building a 3D scene and performing dynamic mapping specifically includes:

[0138] S101. Extract the 3D models of each power supply object and analyze the appearance, size parameters, and feature contour data of each 3D model. Integrate with a GIS map using digital twin technology and generate a 3D scene using a 3D engine.

[0139] S102. Analyze the relative positions and distribution of each power supply object in the distribution network coverage area in the GIS map, load each three-dimensional model at the corresponding position in the three-dimensional scene, and dynamically map the real-time blade pitch angle and power generation on the three-dimensional model.

[0140] In the implementation process, a high-fidelity dynamic digital twin foundation is established. Static geographic information is deeply integrated with the dynamic operating status of equipment to generate a three-dimensional virtual scene that is synchronized with the physical world in real time and can be interactively analyzed.

[0141] It changes the traditional monitoring model that relies mainly on reports and data points, and achieves panoramic transparency and visual perception of power supply objects, providing a unified spatiotemporal context and accurate data source for all subsequent advanced analyses.

[0142] S200. Analyze the relative positions of each power supply object in the 3D scene to divide the recognition domain. Construct a 3D airflow field for each recognition domain and predict future wind field data. Analyze the operating parameters and environmental parameters of each power supply object and fit the power relationship formula. Specifically, this includes:

[0143] S201. Obtain real-time environmental parameters for each power supply object and calculate the attenuation distance using a three-dimensional wind speed vector. Based on the attenuation distance, divide the recognition domain for each power supply object in the three-dimensional scene and fuse all intersecting and overlapping recognition domains. Specifically, this includes:

[0144] S2011, Analyze the power supply object Real-time environmental parameters, analyzing three-dimensional wind speed vectors Substitute into the formula to calculate the attenuation distance in three dimensions. The formula is as follows:

[0145] ;

[0146] In the formula, The preset wind speed threshold, The preset empirical attenuation coefficient reflects the influence of environmental turbulence dissipation and surface roughness.

[0147] Flat and open terrain: Complex and rough terrain: .

[0148] In the specific implementation process, the real-time wind speed vector is input, and logarithmic and proportional calculations are performed by combining the preset wind speed threshold with the empirical coefficients that reflect the influence of environmental turbulence dissipation and surface roughness.

[0149] Based on the logarithmic wind speed profile and turbulent kinetic energy dissipation theory of fluid mechanics, the attenuation trend and spatial scale of wind speed in the direction of the dominant flow behind the rough element or in the wake region under neutral atmospheric conditions are dynamically quantified, thus providing a dynamic boundary basis for the physical-driven three-dimensional simulation domain division.

[0150] S2012, Obtain the power supply object in the 3D scene Coordinate axes of location ,by With the corresponding spatial location as the center point, axes with a length of twice the attenuation distance are set in the three-dimensional directions.

[0151] S2013. Using the three axes as a reference, construct a three-dimensional convex polyhedral space as the power supply object. The recognition domain is such that it can completely enclose the smallest circumscribed space region of the three axes.

[0152] S2014. In the 3D scene, divide the recognition domain for each power supply object and merge all intersecting and overlapping recognition domains into one recognition domain. The merged recognition domain serves as the common recognition domain for all power supply objects before fusion.

[0153] In the specific implementation process, the spatial range (identification domain) of the wake influence and wind speed influence of each wind turbine is dynamically defined by using the attenuation distance formula derived from the logarithmic wind speed profile.

[0154] Compared to a fixed radius range, it can more physically reflect different wind speeds and terrain (through coefficients). This method reflects the actual spatial scale of the wake effect and wind speed influence, and merges the overlapping domains to construct a dynamic computing environment that conforms to physical laws for subsequent wind flow simulation.

[0155] S202. The model is simplified by using the Kriging spatial interpolation algorithm and the Navier-Stokes equation. The CFD physical simulation model is used to fuse discrete point data based on environmental parameters to simulate and construct a three-dimensional airflow field for each identification domain and update it in real time.

[0156] Discrete point data refers to a set of isolated, discontinuously distributed observation sampling points in three-dimensional space, specifically a comprehensive expression of wind speed, wind direction, time, and spatial location. It is obtained by analyzing the fixed position of the sensor based on the three-dimensional model provided by the operating parameters, combined with the three-dimensional wind speed vector collected by the sensor. The simplified Navier-Stokes equation model (CFD) serves as the physical kernel and prediction engine, using the discrete point measured data as a correction benchmark. Kriging spatial interpolation is used as the parameter correlation between the data and the physical field to achieve dynamic construction and updating of the three-dimensional airflow field.

[0157] S203. Perform empirical mode decomposition on each three-dimensional airflow field, extract the intrinsic mode functions that characterize different spatiotemporal scales, and select the dominant modes with high energy proportions as feature bases.

[0158] Empirical Mode Decomposition (EMD) is an adaptive time-frequency signal processing method that decomposes complex signals into a finite number of intrinsic mode functions (IMFs) through a selection process. It includes:

[0159] 1. Identifying extreme points and constructing the envelope:

[0160] First, identify all local maxima and local minima in the original data sequence. Then, use a cubic spline interpolation function to connect all the maxima to form an upper envelope and connect all the minima to form a lower envelope.

[0161] 2. Calculate the mean envelope and extract candidate components:

[0162] Calculate the average of the upper and lower envelopes to obtain the mean envelope. Subtract this mean envelope from the original data sequence to obtain a new candidate data sequence (intermediate signal).

[0163] 3. Determining IMF criteria and iterative screening:

[0164] The candidate data sequence is checked to see if it meets the two conditions of the Intrinsic Mode Function (IMF): first, the number of extreme points and zero-crossing points must be equal or differ by at most one throughout the entire data range; second, at any given time point, the average of the envelopes of the local maxima and the local minima must be zero. If these conditions are not met, the candidate sequence is used as a new input, and steps 1 and 2 are repeated until the IMF conditions are met. This process is called "screening".

[0165] 4. Separate the IMF component from the updated residual signal:

[0166] Once a candidate sequence satisfies the IMF condition, it is identified as the first intrinsic mode function component. This IMF component is then subtracted from the original signal (or the current residual signal) to obtain a new residual signal.

[0167] 5. Decompose cyclically until termination:

[0168] The remaining signal is used as the new data to be decomposed, and steps 1 to 4 are repeated to extract the subsequent IMF components. When the final remaining signal becomes a monotonic function or its amplitude is less than the preset error, and IMF can no longer be extracted from it, the entire EMD decomposition process stops, and the remaining signal is the residual term.

[0169] S204. Input the feature base and its corresponding time coefficient sequence into the trained long short-term memory and convolutional neural network coupled model. The coupled model learns the evolution law of spatiotemporal features.

[0170] The training methods, which are known to those skilled in the art, generally include:

[0171] The historical time coefficient sequence is used to construct supervised learning samples. The input time step (e.g., coefficients of the previous 5 time steps) and the output time step (e.g., coefficients of the next 2 time steps) are set, and training samples are extracted by using a sliding window.

[0172] Before inputting the data into the network, the time coefficient sequence must be normalized (e.g., by MinMaxScaler) to scale it to the [0,1] interval to accelerate model convergence and prevent gradient explosion.

[0173] Mean squared error (MSE) is typically used as the loss function to evaluate the error between the predicted and true coefficients; the optimizer often uses the Adam algorithm, which combines momentum and adaptive learning rate and can converge quickly and stably.

[0174] S205. Using the current modal coefficients and environmental boundary conditions as initial inputs, a multi-step recursive prediction is performed through a coupled model to obtain the future... The sequence of modal coefficients at each time point within a second.

[0175] S206. Perform linear reconstruction of the predicted mode coefficient sequence and the feature basis to generate the future... The three-dimensional flow field data corresponding to each second is used as future wind field data.

[0176] In the specific implementation process, a hybrid prediction framework of CFD simulation + Empirical Mode Decomposition (EMD) + LSTM-CNN coupled model is adopted.

[0177] First, CFD is used to obtain high-resolution spatial flow field details. Then, EMD is used to decompose the complex turbulence into intrinsic mode functions (IMFs) at different scales, and the dominant characteristic basis is extracted.

[0178] Finally, a coupled model using LSTM to capture temporal evolution and CNN to extract spatial features is used to recursively predict the temporal coefficients of the feature base and then reconstruct the future wind field.

[0179] Breaking through the limitations of traditional single numerical weather prediction (WRF) or statistical methods, it can provide high-precision three-dimensional wind flow evolution information at the wind turbine location scale, down to the minute to second level.

[0180] S207. Obtain the operating parameters and environmental parameters of each power supply object, using the power generation at the same moment as the dependent variable and the blade pitch angle and three-dimensional wind speed vector as independent variables, and fit the power relationship formula for each power supply object. Specifically, this includes:

[0181] S2071, Obtain the power supply target In historical periods The internal operating parameters and environmental parameters are set evenly. At each historical moment, the power generation, blade pitch angle, and three-dimensional wind speed vector are obtained.

[0182] S2072, Power generation at the same historical moment As the dependent variable, the blade pitch angle and three-dimensional wind speed vector As independent variables, the dependent variable and all independent variables at each historical moment are packaged into a sample.

[0183] S2073, This The independent variables in each sample are used as input values ​​for the electricity relationship formula, and the difference between the output value and the dependent variable for each sample is used as the deviation value. The electricity relationship formula is as follows:

[0184] ;

[0185] In the formula, The comprehensive efficiency coefficient covers the inherent losses of mechanical transmission, generator and converter, and is a constant or slowly time-varying parameter.

[0186] Air density, unit: kg / m³, is used as an input environmental physical quantity. The swept area of ​​the wind turbine is measured in m², and is determined by the model of the wind turbine.

[0187] The effective wind speed is the vector composite wind speed, in m / s, and is calculated using the following formula:

[0188] ;

[0189] Three-dimensional wind speed vector Including axial, lateral and vertical components ; This is the unit vector along the wind turbine axis, and its value is determined by the yaw angle.

[0190] In the specific implementation process, the effective wind speed is synthesized by vector synthesis. The three-dimensional wind speed vector is synthesized into a single scalar through weighted norm, which not only highlights the dominant axial component, but also quantifies the complex influence of wind conditions in multiple directions.

[0191] and These are weighting coefficients that quantify the contributions of axial wind speed (the primary energy source) and lateral / vertical wind speed (reflecting the effects of turbulence, shear, and yaw error), respectively. .

[0192] The pitch angle adjustment function (dimensionless), with the core independent variable, is expressed as follows:

[0193] ;

[0194] and These are the fitting coefficients. The first term... The nonlinear decay of aerodynamic efficiency during simulated pitch control (primary term); the second term Capture the linear adjustment effect (correction term) that may exist under specific operating conditions.

[0195] Breaking away from traditional lookup tables or fixed function forms, it uses the superposition of exponential and linear terms to flexibly characterize the pitch angle adjustment characteristics across the entire operating range.

[0196] The coupling correction function (dimensionless) is used to characterize the interaction between the wind speed vector and the pitch angle, and its expression is:

[0197] ;

[0198] It is the magnitude of the lateral component of the wind speed vector in the wind turbine plane, used to reflect yaw error or the intensity of non-axial incoming flow. The reference wind speed (such as the rated wind speed) is used to achieve saturation characteristics. The function smooths out power at high wind speeds, avoiding overestimation. The coupling strength coefficient is obtained through data fitting. The function is used to describe the smooth transition of the coupling effect between the pitch angle and the lateral wind speed (such as dynamic inflow and wake distortion).

[0199] The explicit introduction of a coupling term between wind speed and pitch angle can characterize the power modulation phenomenon caused by their interaction (such as the change in pitch efficiency at large yaw angles).

[0200] This framework establishes a high-order nonlinear mapping relationship between power generation, wind speed vector, and blade pitch angle.

[0201] The system is constructed by combining physical constants including overall efficiency, air density, and swept area; a vector weighting function that synthesizes three-dimensional wind speed into an effective scalar; an exponential-linear combination function that characterizes the nonlinear adjustment effect of pitch angle on aerodynamic efficiency; and a coupling function that describes the interaction between wind speed and pitch angle.

[0202] Breaking away from traditional lookup table or fixed function models, this approach uses a fusion of mechanism and data to accurately characterize the complex response characteristics of wind turbine power to environmental inputs and control commands across the entire operating range, providing a core model foundation for power prediction and optimized control.

[0203] S2074. The sum of the deviations of all samples is used as the deviation index. The minimum deviation index is obtained by adjusting the coefficients. After fixing the values ​​of the coefficients, the trained power supply object is obtained. The electrical relationship formula.

[0204] S2075, and so on, to fit the trained power relationship for each power supply object.

[0205] It enables advanced prediction of complex airflow environments and accurate quantitative modeling of wind turbine response.

[0206] S300: First, obtain the electricity consumption curve by fitting the electricity consumption data. Then, combine future wind farm data and power relationship formulas to construct a power generation reference curve and set time points. Establish different power generation strategies based on each time point and randomly construct schemes.

[0207] Calculate the stability index for each scheme and select the scheme with the highest stability index to set the power supply scheme. Specifically, this includes:

[0208] S301, Extracting historical time periods The electricity consumption data is collected and processed according to a preset cycle. The system is divided into several parts, and curves are plotted based on the changes in total power consumption within each cycle. The curves of all cycles are then fitted to obtain the power consumption curve.

[0209] S302. Create a line graph with time as the horizontal axis and power as the vertical axis, and map it onto the electricity consumption curve, setting the rate of increase. and decline Based on the electricity consumption curve, it fluctuates upwards. downward fluctuation Establish the first trend interval for the boundary.

[0210] S303, Analyzing future wind field data Substituting the three-dimensional wind speed vector changes of each power supply object within a second into the electrical relationship formula of each power supply object, we obtain the future... A model showing the relationship between power generation and blade pitch angle at different times within a second.

[0211] S304. By adjusting the blade pitch angle in the relational model, obtain the maximum and minimum power generation at each time point, and calculate the sum of the maximum and minimum power generation of all power supply objects at the same time point.

[0212] S305, Mark the relative position of the current moment within the period and map it onto the line chart as the starting point, based on future... The sum of the maximum power generation and the sum of the minimum power generation at each time point within a second are used as boundaries to establish the second trend interval.

[0213] S306, please refer to Figure 2 The intersection area of ​​the first and second trend intervals in the line chart is used as the power generation reference interval. A power generation reference curve is constructed within this interval, and time points are evenly set on the power generation reference curve at preset intervals. Specifically, this includes:

[0214] S3061. Within the power generation reference interval, uniformly set sampling points and analyze the time and power corresponding to each sampling point. Group sampling points at the same time into the same category, and multiply the number of sampling points in all categories to obtain the result. .

[0215] S3062, Quantity of Statistical Categories and establish Each group contains [number] items. The sampling points come from different classes, and the sampling points in different combinations are not completely the same.

[0216] S3063. Calculate the standard deviation of the power corresponding to all sampling points in each combination, mark all sampling points in the combination with the smallest standard deviation, and use a smoothing algorithm to construct a power generation reference curve based on all marked sampling points.

[0217] In the specific implementation process, the demand range is fitted by historical electricity consumption data, and the supply range is calculated by combining future wind farm forecasts and wind turbine models. The intersection of the two constitutes the power generation reference range, which ensures the feasibility of the plan.

[0218] Within this range, by optimizing the combination of sampling points, a power generation reference curve with the most stable power change is generated, aiming to smooth out fluctuations.

[0219] S307. Establish different power generation strategies for each time point, and construct different schemes based on the power generation strategies at each time point. Calculate the stability index for each scheme, and select the scheme with the highest stability index as the power supply scheme. Specifically, this includes:

[0220] S3071, Obtaining Time Points power And each power supply object at a given time point A model showing the relationship between power generation and blade pitch angle was developed, allowing for the acquisition of different power generation rates by adjusting the blade pitch angle.

[0221] S3072, Establishment Each power generation strategy includes different blade pitch angles set for each power supply object, and the sum of the power generation corresponding to the blade pitch angles of all power supply objects within each power generation strategy is equal to... .

[0222] In the specific implementation process, the power supply objects in each power generation strategy are arranged in the same order, and all blade pitch angles are included in a one-to-one correspondence according to the power supply object arrangement order.

[0223] S3073, establish for each time point The result is obtained by multiplying the number of power generation strategies at all time points. Count the number of all time points. and establish One option.

[0224] S3074, Randomly place within each scheme There are several power generation strategies from different points in time. All power generation strategies are arranged in chronological order, and the power generation strategies within different schemes are not exactly the same.

[0225] S3075, Analysis Scheme All power generation strategies are analyzed, and the blade pitch angles set for each power supply object at the corresponding time point for each power generation strategy are analyzed. Weights are assigned to each power supply object. Calculation scheme stability index :

[0226] ;

[0227] In the formula, For the first The power supply object is in the first The blade pitch angle set at each time point, For the first The power supply object is in the first The blade pitch angle set at each time point. For the first The maximum adjustable blade pitch angle for each power supply object.

[0228] The stability index is used to quantify and evaluate the operational stability of a multi-period power supply scheme for wind turbine generators.

[0229] By calculating the blade pitch angle adjustment range of each power supply object in the scheme between adjacent execution time points, a weighted penalty calculation based on a negative exponential function is performed, and the penalty values ​​of all power supply objects and all time periods are summed to obtain the final exponent.

[0230] The engineering objectives of reducing mechanical wear and impact on equipment, and preventing the risk of increased intra-term variables during large-scale adjustments of a small number of power supply targets, are transformed into calculable mathematical optimization objectives.

[0231] In practical implementation, the more power supply objects are adjusted and the smaller the adjustment range, the shorter the time consumed and the lower the risk. The higher the index value, the smoother and more continuous the overall pitch angle action sequence of all wind turbines in the scheme, thereby effectively extending the service life of key mechanical components and optimizing the total life cycle operating cost while meeting the power generation requirements.

[0232] And so on, calculating the stability index for each scheme.

[0233] The smoothness of the scheme is quantified by defining a stability index. An exponential decay function is used to penalize the pitch angle variation of all wind turbines and all adjacent time points in the scheme.

[0234] The smaller the change, the higher the index value. This index can automatically avoid aggressive strategies that, while meeting power requirements, would lead to frequent and drastic changes in the pitch angle, thereby selecting the optimal power supply scheme that minimizes the mechanical impact on the wind turbine drive system and pitch mechanism, achieving the best balance between power generation benefits and equipment lifespan.

[0235] S400: Each power supply object is adjusted according to the power supply plan, and operating parameters and environmental parameters are recorded in real time.

[0236] In the specific implementation process, the blade pitch angle set by each power supply object at each time point in the power supply scheme is analyzed, and each power supply object adjusts the blade pitch angle to the corresponding set value when it arrives at each time point.

[0237] Real-time recording of blade pitch angle, power generation, and three-dimensional wind speed vector serves as operational and environmental parameters.

[0238] Example 2: Please refer to Figure 3 The present invention also provides a power distribution network optimization management system based on dynamic perception, including a dynamic perception module, a power distribution analysis module, a power supply optimization module and an execution management module.

[0239] The dynamic sensing module is used to collect GIS maps and electricity consumption data, as well as the operating parameters and environmental parameters of all power supply objects. It builds a 3D scene and dynamically maps it.

[0240] In the specific implementation process, geographic information system (GIS) maps, total power grid consumption data, and real-time operating parameters (such as blade pitch angle and power generation) and environmental parameters (three-dimensional wind speed vector) of all wind turbines (power supply objects) are collected.

[0241] By using digital twin technology, GIS maps are combined with 3D models of each wind turbine to build a 3D virtual scene of the power distribution network coverage area. Real-time data (such as pitch angle and power) are dynamically mapped onto the corresponding model to realize a digital mirror of the physical world.

[0242] Construct a high-fidelity, real-time 3D digital twin environment that reflects the state of the physical world. This environment intuitively displays the geographical distribution of wind farms and the real-time status of equipment, providing a unified, accurate, and dynamic data foundation and visualization platform for subsequent refined analysis and optimization. It is the basis for achieving dynamic perception.

[0243] The power distribution analysis module divides the identification domains according to the relative positions of each power supply object in the three-dimensional scene. Each identification domain constructs a three-dimensional airflow field and predicts future wind field data, thus fitting the power relationship formula for each power supply object.

[0244] In the specific implementation process, the attenuation distance is calculated based on the location of the wind turbine and the wind speed, a three-dimensional space (identification domain) of wind speed influence is defined for each wind turbine, and adjacent domains are merged.

[0245] Within the identification domain, real-time monitoring data is fused, and a three-dimensional airflow field is constructed using computational fluid dynamics (CFD) simulation. A coupled model of empirical mode decomposition and LSTM-CNN is then used to predict future wind field data in the short term (S seconds).

[0246] Meanwhile, a nonlinear power relationship is fitted for each wind turbine, with wind speed vector and pitch angle as independent variables and power generation as dependent variable. The formula introduces vector synthesis effective wind speed, pitch angle adjustment function and wind speed-pitch angle coupling correction function.

[0247] It enables advanced and refined prediction of complex wind flow (especially wake effect) in wind farms, and performs high-precision mathematical modeling of the relationship between environmental input (wind) and equipment response (power generation).

[0248] This method breaks through the traditional approach of simplifying wind turbine power estimation, significantly improving the ability to describe the complex wind flow interaction and nonlinear power generation characteristics, and serves as the core basis for optimization decision-making.

[0249] The power supply optimization module fits power consumption curves to power consumption data and constructs a power generation reference curve by combining future wind farm data and power relationship formulas. Different power generation strategies are established and schemes are randomly generated; the stability index of each scheme is calculated to determine the power supply scheme.

[0250] In the specific implementation process, a typical power consumption curve of the power grid is fitted based on historical data, and its fluctuation range is established. Combining future wind farm data and the power relationship formula of each wind turbine, the total power generation potential range of the entire field in the next S seconds can be calculated, thus forming a power generation reference range.

[0251] Within the intersecting interval, aiming for stable power output, a power generation reference curve is generated, and multiple future time points are set on it. For each time point, the module assigns different pitch angle combinations to each wind turbine, generating multiple power generation strategies.

[0252] Subsequently, strategies at various time points are randomly combined to form numerous complete power supply scheme candidates. Finally, a stability index is defined to evaluate each scheme. This index quantifies the smoothness of pitch angle adjustment of all wind turbines at all time points in the scheme through an exponential decay function, and the scheme with the highest index is selected as the final power supply scheme to be implemented.

[0253] It achieves collaborative optimization across multiple time scales and objectives (demand tracking and smooth operation). It can automatically and intelligently select the option with the least mechanical impact on the wind turbine and the most stable operation from numerous possible solutions that meet the grid's power demands. This effectively balances the economics of power generation with the safety of equipment, improving the robustness and lifespan of the entire wind power distribution network system.

[0254] The execution management module adjusts according to the power supply scheme and records operating parameters and environmental parameters.

[0255] In the actual implementation process, the optimized power supply scheme is distributed to each wind turbine for execution. The core action is to analyze the blade pitch angle that each wind turbine should be set at each future time point in the scheme, and when the actual time reaches these points, control the wind turbine to adjust the pitch angle to the target value.

[0256] This completes the final closed loop from perception, analysis, decision-making to control, ensuring the implementation of optimization strategies. Simultaneously, new operating parameters of the wind turbine and environmental parameters are continuously recorded during execution. This data can be fed back to the aforementioned modules for updating the model and iterative optimization, thus forming a dynamic optimization management system with self-learning and continuous improvement capabilities.

[0257] Example 3: This example verifies the practical application effect of the dynamic sensing-based power distribution network optimization management method of the present invention compared with the existing wind power distribution network management technology through comparative experiments.

[0258] The analysis specifically examines the accuracy of wind farm forecasting, power supply matching, equipment operational stability, and lifespan protection.

[0259] Experimental Scenario: The distribution network coverage area of ​​a mountainous wind farm in East China was selected, with a physical area of ​​5.2 km² and complex, rough terrain. The empirical attenuation coefficient K = 0.0025. Fifteen 2.5MW permanent magnet direct-drive wind turbines were deployed within the farm, serving as the core power supply targets for the distribution network.

[0260] Experimental period: 72 consecutive hours, covering three typical wind conditions: clear weather, gusts, and weak turbulence; data sampling period per_sam=15min, future wind field prediction duration S=300 seconds.

[0261] Existing technologies: adopt traditional two-dimensional data monitoring, wind field prediction based on WRF numerical weather prediction, wind turbine fixed power curve modeling, and extensive control strategies that only aim at power balance.

[0262] This invention employs a full-process management method that combines digital twin 3D perception, CFD-EMD-LSTM-CNN hybrid wind field prediction, high-precision power relationship formulas, and stability index optimization.

[0263] Evaluation indicators include:

[0264] Mean absolute error (MAE) of wind field prediction: reflects the accuracy of wind field prediction, measured in m / s.

[0265] Average error in power generation tracking: reflects the degree of matching between the power supply scheme and the electricity demand, in kW.

[0266] Average adjustment range of wind turbine blade pitch angle: reflects the smoothness of equipment operation, in degrees.

[0267] The average stability index of the scheme: quantifies the stability of the system operation; the higher the value, the more stable the system.

[0268] Equivalent fatigue loss rate of pitch mechanism: reflects the effectiveness of equipment life protection, unit % / 72h.

[0269] Activate existing technology systems: collect two-dimensional geographic data, conventional meteorological data and wind turbine operation data, predict wind field through WRF model, generate power supply scheme based on fixed power curve, and execute pitch angle control.

[0270] The system of this invention is launched: the dynamic perception module collects GIS maps and 3D operation / environment parameters, builds a digital twin 3D scene and maps it in real time.

[0271] The power distribution analysis module divides the wind turbine identification domain, constructs a three-dimensional airflow field, predicts wind field data for the next 300 seconds through a hybrid model, and fits a high-precision power relationship formula for each wind turbine.

[0272] The power supply optimization module fits the power consumption curve, constructs a power generation reference range, and selects the optimal power supply scheme through the stability index.

[0273] The execution management module adjusts the propeller pitch angle according to the plan and records the operation data in real time.

[0274] Two sets of experimental data were collected every 15 minutes for 72 consecutive hours. The mean values ​​of each indicator were calculated, as shown in Table 1.

[0275] Table 1

[0276] Please see Figure 4 This invention employs a hybrid prediction framework combining CFD simulation, empirical mode decomposition, and LSTM-CNN coupled models. Compared to the existing WRF single numerical prediction, it can accurately capture the three-dimensional turbulence and wake effects under complex mountainous terrain. The average absolute error of wind field prediction is reduced from 1.28 m / s to 0.42 m / s, and the prediction accuracy is improved by 67.19%, providing accurate environmental data support for subsequent power supply optimization.

[0277] This invention uses power consumption curve fitting and power generation reference interval construction, combined with high-precision power relationship formulas to calculate the power generation potential of wind turbines. Compared with the fixed power curve modeling of existing technologies, the average error of power generation tracking is reduced from 186.5kW to 52.3kW, and the matching degree between power supply and power demand is improved by 71.96%, effectively avoiding the problem of power imbalance in the distribution network.

[0278] This invention uses the stability index as the core optimization objective and quantifies the penalty effect of the pitch angle adjustment range through a negative exponential function to select the smoothest control scheme. The average blade pitch angle adjustment range is reduced from 8.7° to 2.1°, the average stability index of the scheme is increased from 0.32 to 0.89, the wind turbine's operational stability is improved by 178.13%, and the mechanical shock to the equipment is significantly reduced.

[0279] Smooth adjustment of pitch angle directly reduces fatigue wear of pitch mechanism. During the experimental period, the equivalent fatigue loss rate of pitch mechanism decreased from 4.12% to 0.87%, a reduction of 78.88%, which can effectively extend the service life of key wind turbine components and reduce the operation and maintenance cost of distribution network throughout its entire life cycle.

[0280] Please see Figure 5 The horizontal axis represents the experimental run time (0-72h), divided into 15-minute sampling periods, with a total of 288 sampling points covering the entire experimental period.

[0281] The vertical axis represents the power generation tracking error (kW), reflecting the deviation between the real-time power supply scheme and the electricity demand.

[0282] The red curve represents the real-time error of the prior art, and the green curve represents the real-time error of the present invention. The red dashed line represents the average error of the prior art (186.5kW), and the green dashed line represents the average error of the present invention (52.3kW).

[0283] The standard deviation of the error in the existing technology is set at 40 (large fluctuation), while that in this invention is set at 10 (small fluctuation).

[0284] The power tracking error of existing technologies exhibits large and irregular fluctuations, with the error value varying drastically between 50-300kW, which can easily lead to power imbalance in the distribution network.

[0285] The error of this invention is always maintained within a narrow range of 10-100kW, with gentle fluctuations and extremely low overall value, achieving real-time and accurate matching between power supply and power demand.

[0286] During periods of complex wind conditions such as gusts and weak turbulence (e.g., 20-30h, 50-60h), the error of existing technologies increases sharply, while the error of this invention only fluctuates slightly. This is because the dynamic sensing module of this invention realizes a digital mirror of the physical world, the power distribution analysis module can predict wind field changes in advance, and the power supply optimization module can dynamically adjust the power generation strategy based on real-time wind field data, exhibiting extremely strong adaptability to complex wind conditions.

[0287] The stability of power generation tracking error directly determines the stability of distribution network operation. The low error and low fluctuation characteristics of this invention effectively avoid voltage and frequency fluctuations caused by power imbalance in the distribution network, and solve the distribution network operation risks caused by the extensive control of existing technologies.

[0288] Please see Figure 6 The left subplot shows the average adjustment range of blade pitch angle under different wind conditions, and the right subplot shows the predicted wind field MAE under different wind conditions.

[0289] The horizontal axis represents three typical wind conditions (sunny day, gusts, and weak turbulence), while the vertical axis represents the pitch angle adjustment range (°) and the wind field prediction MAE (m / s).

[0290] The red bars represent existing technical indicators, while the green bars represent the indicators of this invention. The data is broken down according to the overall performance improvement ratio, which is consistent with the actual working conditions under different wind conditions.

[0291] Left sub-figure: Average adjustment range of blade pitch angle under different wind conditions.

[0292] Sunny day: Existing technology 6.2°, this invention 1.4°, an improvement of 77.42%; Gust: Existing technology 10.5°, this invention 2.6°, an improvement of 75.24%; Weak turbulence: Existing technology 9.4°, this invention 2.3°, an improvement of 77.66%.

[0293] Existing technologies require the largest pitch angle adjustment range under gust conditions because gusts have highly variable wind speeds, and existing technologies cannot predict wind field changes in advance, so they can only passively adjust the pitch angle drastically. In contrast, this invention can accurately predict the wind field 300 seconds in advance, and formulate control strategies in advance to achieve small and gradual adjustments to the pitch angle, maintaining smooth operation even under gust conditions.

[0294] Under all three wind conditions, the pitch angle adjustment range of the present invention remained within 2.6°, indicating that the optimization strategy of the present invention has all-weather adaptability and is not affected by the complexity of the wind conditions, thus ensuring the stable operation of the wind turbine at all times.

[0295] Right side subplot: Wind field prediction MAE under different wind conditions.

[0296] Sunny day: Existing technology 0.90m / s, this invention 0.30m / s, an improvement of 66.67%; Gust: Existing technology 1.50m / s, this invention 0.50m / s, an improvement of 66.67%; Weak turbulence: Existing technology 1.40m / s, this invention 0.45m / s, an improvement of 67.86%.

[0297] Under conditions of gusts and weak turbulence, the prediction error of existing technologies is significantly higher than that under clear skies. This is because the WRF numerical prediction of existing technologies cannot capture small-scale, highly dynamic wind flow changes. The CFD-EMD-LSTM-CNN hybrid prediction framework of this invention first obtains high-resolution spatial flow field details through CFD, then decomposes turbulence features through EMD, and finally predicts spatiotemporal evolution through a coupled model, maintaining high prediction accuracy even under complex wind conditions.

[0298] The wind field prediction MAE of this invention is below 0.5 m / s under all three wind conditions, providing accurate environmental data support for subsequent power supply optimization and forming the basis for accurate tracking of power generation and stable equipment operation.

[0299] Compared with existing wind power distribution network management technologies, this invention achieves four technological breakthroughs: precise dynamic sensing, high-precision wind farm prediction, coordinated power supply optimization, and stable equipment operation.

[0300] Overcoming the shortcomings of existing technologies such as fragmented perception and two-dimensional monitoring, a high-fidelity three-dimensional dynamic scene is constructed through digital twin technology to achieve panoramic and transparent perception of the power distribution network.

[0301] Overcoming the shortcomings of existing wind field prediction technologies, which rely on coarse and single models, a hybrid prediction framework is used to achieve high-precision three-dimensional wind field prediction at the turbine location level and within seconds.

[0302] Overcoming the shortcomings of existing technologies in power generation modeling, such as simplification and fixed power curves, this paper uses a power relationship formula that integrates mechanism and data to accurately characterize the power generation characteristics of wind turbines under all operating conditions.

[0303] Overcoming the shortcomings of existing technologies that focus solely on power balance and have only a single optimization objective, this approach incorporates stable equipment operation and lifespan protection into the optimization objectives through a stability index, thereby achieving a synergistic balance between power generation revenue and equipment safety.

[0304] This invention can simultaneously meet the engineering requirements of power distribution network reliability, operational stability, and equipment durability under complex wind conditions, effectively solving the core technical defects of existing technologies in wind power distribution network management.

[0305] In summary, compared with existing wind power distribution network management technologies, this invention achieves four technological breakthroughs: precise dynamic sensing, high-precision wind farm prediction, coordinated power supply optimization, and stable equipment operation. Under complex wind conditions, it can simultaneously meet the engineering requirements of power supply reliability, operational stability, and equipment durability of the distribution network, and solves the technical defects of existing technologies such as fragmented sensing, coarse modeling, and single optimization.

[0306] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0307] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing and managing power distribution network schemes based on dynamic sensing, characterized in that: The method includes: S100 collects GIS maps and electricity consumption data, as well as the operating parameters and environmental parameters of all power supply objects within the distribution network coverage area; it uses digital twin technology to build a three-dimensional scene based on the GIS map and dynamically maps it in combination with real-time operating parameters; S200. Based on the relative positions of each power supply object in the three-dimensional scene and environmental parameters, divide the identification domain for each power supply object, construct a three-dimensional airflow field for each identification domain, and predict future wind field data based on the three-dimensional airflow field; and obtain the power relationship formula by fitting the operating parameters and environmental parameters of each power supply object. S300: First, fit the electricity consumption data to obtain the electricity consumption curve, then combine the future wind farm data and power relationship formula to construct the power generation reference curve and set the time point; establish different power generation strategies for each time point and randomly form a scheme; calculate the stability index of each scheme, and select the scheme with the largest stability index to set the power supply scheme. S400: Each power supply object is adjusted according to the power supply plan, and operating parameters and environmental parameters are recorded in real time.

2. The method for optimizing and managing power distribution network schemes based on dynamic sensing according to claim 1, characterized in that: In step S100, the power consumption data includes the total power consumption at different times; the operating parameters include the three-dimensional model, as well as the blade pitch angle and power generation at different times; and the environmental parameters include the three-dimensional wind speed vector at different times. Building a 3D scene and performing dynamic mapping specifically includes: S101. Extract the three-dimensional model of each power supply object and analyze the appearance dimension parameters and feature contour data of each three-dimensional model; By integrating digital twin technology into GIS maps, a 3D engine is used to generate 3D scenes; S102. Based on the relative positions and distribution of each power supply object within the distribution network coverage area in the GIS map, load each three-dimensional model at the corresponding position in the three-dimensional scene, and dynamically map the real-time blade pitch angle and power generation on the three-dimensional model.

3. The method for optimizing and managing power distribution network schemes based on dynamic sensing according to claim 1, characterized in that: Step S200 includes: S201. Obtain the real-time environmental parameters of each power supply object and calculate the attenuation distance using the three-dimensional wind speed vector; divide the recognition domain for each power supply object in the three-dimensional scene according to the attenuation distance, and fuse all intersecting and overlapping recognition domains; S202. The Kriging spatial interpolation algorithm and the Navier-Stokes equations are used to simplify the model. The CFD physical simulation model is used to fuse discrete point data based on environmental parameters to simulate and construct three-dimensional airflow fields for each identification domain and update them in real time. S203. Perform empirical mode decomposition on each three-dimensional airflow field, extract the intrinsic mode functions that characterize different spatiotemporal scales, and select the dominant modes with high energy proportions as feature bases. S204. Input the feature base and its corresponding time coefficient sequence into the trained long short-term memory and convolutional neural network coupled model. The coupled model learns the evolution law of spatiotemporal features. S205. Using the current modal coefficients and environmental boundary conditions as initial inputs, a multi-step recursive prediction is performed through a coupled model to obtain the future... The sequence of modal coefficients at each time point within a second; S206. Perform linear reconstruction of the predicted mode coefficient sequence and the feature basis to generate the future... The three-dimensional flow field data corresponding to each second is used as future wind field data; S207. Obtain the operating parameters and environmental parameters of each power supply object, take the power generation at the same time as the dependent variable, and the blade pitch angle and three-dimensional wind speed vector as the independent variables, and fit the power relationship for each power supply object.

4. The method for optimizing and managing power distribution network schemes based on dynamic sensing according to claim 3, characterized in that: S201 includes: S2011, According to the power supply object Three-dimensional wind speed vector of real-time environmental parameters Calculate the attenuation distance in three dimensions. The formula is as follows: ； In the formula, The preset wind speed threshold, The preset empirical attenuation coefficient; S2012, Obtain the power supply object in the 3D scene Coordinate axes of location ,by With the corresponding spatial location as the center point, axes with a length of twice the attenuation distance are set in the three-dimensional directions respectively; S2013. Using the three axes as a reference, construct a three-dimensional convex polyhedral space as the power supply object. The recognition domain, and the recognition domain can completely enclose the smallest circumscribed space region of the three axes; S2014. In the 3D scene, divide the recognition domain for each power supply object and merge all intersecting and overlapping recognition domains into one recognition domain. The merged recognition domain serves as the common recognition domain for all power supply objects before fusion.

5. The method for optimizing and managing power distribution network schemes based on dynamic sensing according to claim 3, characterized in that: Step S207 includes: S2071, According to the power supply object In historical periods The internal operating parameters and environmental parameters are set evenly. At each historical moment, the power generation, blade pitch angle, and three-dimensional wind speed vector are obtained; S2072, Power generation at the same historical moment As the dependent variable, the blade pitch angle and three-dimensional wind speed vector As independent variables, the dependent variable and all independent variables at each historical moment are packaged into a sample; S2073, will The independent variables in each sample are used as input values ​​for the electricity relationship formula, and the difference between the output value and the dependent variable for each sample is used as the deviation value; the electricity relationship formula is as follows: ； In the formula, The overall efficiency coefficient, air density, The area swept by the wind turbine; The effective wind speed is calculated using the vector synthesis formula: ； The unit vector along the wind turbine axis. and These are the weighting coefficients; The pitch angle adjustment function is expressed as follows: ； and These are the fitting coefficients; The coupling correction function is expressed as follows: ； Let be the magnitude of the lateral component of the wind speed vector in the wind turbine plane. For reference wind speed, The coupling strength coefficient is... The function is used to describe the smooth transition of the coupling effect between the pitch angle and the lateral wind speed; S2074. The sum of the deviations of all samples is used as the deviation index. The minimum deviation index is obtained by adjusting the coefficients. After fixing the values ​​of the coefficients, the trained power supply object is obtained. The electrical relationship formula; S2075, and so on, to fit the trained power relationship for each power supply object.

6. The method for optimizing and managing power distribution network schemes based on dynamic sensing according to claim 1, characterized in that: Step S300 includes: S301, Extracting historical time periods The electricity consumption data is collected and processed according to a preset cycle. The process is divided into several parts, and curves are plotted based on the changes in total power consumption in each cycle. The curves of all cycles are then fitted to obtain the power consumption curve. S302. Create a line graph with time as the horizontal axis and power as the vertical axis, and map it onto the electricity consumption curve, setting the rate of increase. and decline Based on the electricity consumption curve, it fluctuates upwards. downward fluctuation Establish the first trend interval for the boundary; S303, Analyzing future wind field data Substituting the three-dimensional wind speed vector changes of each power supply object within a second into the electrical relationship formula of each power supply object, we obtain the future... A model showing the relationship between power generation and blade pitch angle at different times within a second; S304. By adjusting the blade pitch angle in the relational model, obtain the maximum and minimum power generation at each time point, and calculate the sum of the maximum power generation of all power supply objects at the same time point, as well as the sum of the minimum power generation. S305, Mark the relative position of the current moment within the period and map it onto the line chart as the starting point, based on future... The sum of the maximum power generation and the sum of the minimum power generation at each time point within a second are used as boundaries to establish the second trend interval; S306. Take the area where the first trend interval and the second trend interval in the line graph intersect as the power generation reference interval, construct a power generation reference curve within the power generation reference interval, and set time points evenly on the power generation reference curve at preset intervals. S307. Establish different power generation strategies for each time point, and construct different schemes based on the power generation strategies at each time point; calculate the stability index of each scheme, and select the scheme with the largest stability index as the power supply scheme.

7. The method for optimizing and managing power distribution network schemes based on dynamic sensing according to claim 6, characterized in that: Step S306 includes: S3061. Evenly distribute sampling points within the power generation reference interval, and analyze the time and power corresponding to each sampling point; group sampling points at the same time into the same category, and multiply the number of sampling points in all categories to obtain the result. ; S3062, Quantity of Statistical Categories and establish Each group contains [number] items. The sampling points come from different classes, and the sampling points in different combinations are not completely the same; S3063. Calculate the standard deviation of the power corresponding to all sampling points in each combination, mark all sampling points in the combination with the smallest standard deviation, and use a smoothing algorithm to construct a power generation reference curve based on all marked sampling points.

8. The method for optimizing and managing power distribution network schemes based on dynamic sensing according to claim 6, characterized in that: Step S307 includes: S3071, Obtaining Time Points power And each power supply object at a given time point A model showing the relationship between power generation and blade pitch angle was developed, allowing different power generation rates to be obtained by adjusting the blade pitch angle. S3072, Establishment Each power generation strategy includes different blade pitch angles set for each power supply object, and the sum of the power generation corresponding to the blade pitch angles of all power supply objects within each power generation strategy is equal to... ; S3073, establish for each time point The result is obtained by multiplying the number of power generation strategies at all time points. Count the number of all time points. and establish One option; S3074, Randomly place within each scheme There are several power generation strategies from different points in time. All power generation strategies are arranged in chronological order. The power generation strategies within different schemes are not exactly the same. S3075, Analysis Scheme All power generation strategies are analyzed, and the blade pitch angles set for each power supply object at the corresponding time point for each power generation strategy are analyzed; weights are assigned to each power supply object. Calculation scheme stability index : ； In the formula, For the first The power supply object is in the first The blade pitch angle set at each time point, For the first The power supply object is in the first The blade pitch angle set at each time point; For the first The maximum adjustable blade pitch angle for each power supply object; And so on, calculating the stability index for each scheme.

9. The method for optimizing and managing power distribution network schemes based on dynamic sensing according to claim 1, characterized in that: In step S400, the blade pitch angle set by each power supply object at each time point in the power supply scheme is analyzed, and each power supply object adjusts the blade pitch angle to the corresponding set value when it arrives at each time point.

10. A distribution network power supply scheme optimization management system based on dynamic perception, applied to the distribution network power supply scheme optimization management method based on dynamic perception as described in claim 1, characterized in that: The system includes a dynamic sensing module, a power distribution analysis module, a power supply optimization module, and an execution management module; The dynamic sensing module is used to collect GIS maps and electricity consumption data, as well as the operating parameters and environmental parameters of all power supply objects; it builds a 3D scene and dynamically maps it. The power distribution analysis module divides the identification domain according to the relative position of each power supply object in the three-dimensional scene. Each identification domain constructs a three-dimensional airflow field and predicts future wind field data, and fits the power relationship formula for each power supply object. The power supply optimization module fits the power consumption curve through power consumption data, and constructs a power generation reference curve by combining future wind farm data and power relationship formula; it establishes different power generation strategies and randomly sets up schemes, calculates the stability index of each scheme, and thus sets the power supply scheme. The execution management module adjusts according to the power supply scheme and records operating parameters and environmental parameters.

Images