High-permeability new energy power distribution network collaborative optimization method, system and device and storage medium
By identifying path loss in foggy environments, establishing a channel fading model, adjusting the response sequence of energy storage units, and optimizing energy storage and topology configuration, the problem of dispatch delay in mountainous wind power microgrids under foggy conditions was solved, improving operational reliability and economy.
Patent Information
- Application Number
- CN202511632637.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-10
- Publication Date
- 2025-12-05
- Estimated Expiration
- 2045-11-10
AI Technical Summary
Traditional methods struggle to handle real-time changes caused by factors such as fog in the dynamic environment of mountainous wind power microgrids, leading to a disconnect between predictions and actual operation, affecting the timeliness and accuracy of dispatch instructions, and consequently impacting the stability and efficiency of the microgrid.
By acquiring wind farm operation data and signal strength, identifying path loss in foggy environments, establishing a channel fading model, analyzing the potential delay of energy storage scheduling commands, adjusting the response sequence of energy storage units, generating an energy storage and topology configuration optimization scheme, and combining a multi-objective optimization algorithm for optimized scheduling.
It effectively solves the problem of energy storage dispatch delay caused by unstable wireless communication in wind farms in hilly and foggy environments, and improves the reliability and economy of wind farm operation.
Smart Images

Figure CN121076935A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of information technology, in particular to a high-penetration new energy distribution network collaborative optimization method, system, device and storage medium. BACKGROUND
[0002] The application of virtual reality technology in the field of new energy microgrid is increasingly attracting attention. It provides a new perspective for the operation optimization of complex power grids by building immersive scenes, especially in the scenario of mountain wind power microgrid with complex geographical environment and variable climate. The stable operation of the microgrid is directly related to the efficient use of new energy and the reliability of regional energy supply. Virtual reality technology can significantly improve the decision-making efficiency of operation and maintenance personnel through visualization and interaction. However, existing solutions have obvious shortcomings in dealing with uncertainty in complex scenarios. Traditional methods rely on static models or single data analysis, which are difficult to cope with real-time changes caused by dynamic environments such as hilly fog, leading to prediction deviation and actual operation disconnection, affecting the timeliness and accuracy of scheduling instructions. In the hilly fog scenario, the uncertainty of new energy output is the primary challenge. This uncertainty is first reflected in the fact that fog weather often leads to deviations in wind power output prediction, which directly affects the power balance of the microgrid. More complex is that due to the complex terrain of the mountains and the obstruction of the forest, the fluctuation of wireless signals disturbed by the fog is further intensified. This instability of the signal directly leads to the delay of the dispatching instruction transmission of the energy storage system, and the instruction delay in turn exacerbates the incoordination between the microgrid topology configuration and the energy storage scheduling, ultimately leading to reduced operation efficiency and even threatened system stability.
[0003] Therefore, from output prediction deviation to signal fluctuation, to scheduling delay, this series of closely related technical bottlenecks collectively hinder the effective implementation of dynamic optimization solutions. How to integrate scene analysis and probabilistic power flow data in the virtual reality environment to dynamically present optimization solutions to assist operation and maintenance personnel in real-time adjusting energy storage and topology configuration has become a key problem to improve the operation efficiency and user trust of mountain wind power microgrids. SUMMARY
[0004] In view of the above problems, the present application is proposed.
[0005] Therefore, the technical problem solved by the present application is to solve the problem that traditional methods rely on static models or single data analysis, which are difficult to cope with real-time changes caused by dynamic environments such as hilly fog, leading to prediction deviation and actual operation disconnection, affecting the timeliness and accuracy of scheduling instructions.
[0006] To solve the above technical problems, the present application provides the following technical solutions: a high-penetration new energy distribution network collaborative optimization method, comprising:
[0007] Obtaining wind farm operation data and signal strength, performing data processing on the wind farm operation data and signal strength to obtain a signal attenuation trend;
[0008] Fitting the signal attenuation trend to obtain a signal attenuation characteristic, identifying a thin fog environment path loss according to the signal attenuation characteristic to obtain a wind power output probability distribution function;
[0009] Obtaining an output fluctuation range according to the wind power output probability distribution function, predicting an output deviation scenario, calculating a wireless signal fluctuation probability for each scenario, and extracting a signal fluctuation distribution characteristic;
[0010] Establishing a channel fading model according to the signal fluctuation distribution characteristic and the thin fog environment path loss, outputting a wireless signal propagation delay, obtaining a transmission time sequence, and obtaining a potential delay interval of energy storage system scheduling instructions according to the transmission time sequence;
[0011] Simulating scheduling disorder and power regulation instruction conflict in the potential delay interval, identifying distribution deviation of microgrid topology configuration and energy storage scheduling, and generating scheduling disorder evaluation data;
[0012] Extracting power flow fluctuations from the scheduling disorder evaluation data, and combining historical wind power output data to judge the power flow fluctuations;
[0013] According to the judgment result, adjusting the response order of the energy storage unit, generating a power regulation instruction sequence, calculating the adjusted power flow distribution data, and generating energy storage and topology configuration optimization data.
[0014] As a preferred scheme of the high-penetration new energy distribution network collaborative optimization method, the wind power output probability distribution function comprises:
[0015] Obtaining a signal sequence of a wind farm, preprocessing the signal sequence, and establishing a signal amplitude observation matrix;
[0016] Normalizing the signal amplitude observation matrix, and fitting to generate a signal amplitude probability distribution function by the least square method;
[0017] Calculating the scattering coefficient according to the signal amplitude probability distribution function, and combining the distance attenuation parameter to generate an electromagnetic wave energy loss matrix;
[0018] Establishing a path loss mapping relationship according to the electromagnetic wave energy loss matrix, calculating a transmission path loss probability distribution, and generating a wind power output probability function.
[0019] The beneficial effects of the preferred technical scheme are that the wind power output uncertainty is quantified through signal processing and modeling, accurate basis is provided for optimal scheduling, and system operation reliability and economy are improved.
[0020] As a preferred scheme of the high-penetration new energy distribution network collaborative optimization method, the wind power output probability distribution function further comprises:
[0021] The actual output record of the wind farm is obtained, and the actual output record of the wind farm is decomposed by wavelet to obtain a smooth output curve;
[0022] The smooth output curve is segmented according to a fixed time window, and the peak point set and the valley point set in the time window are calculated to obtain an output fluctuation frequency characteristic;
[0023] A training sample matrix is generated according to the output fluctuation frequency characteristic and the environmental parameter sequence, a time sequence characteristic is extracted, and a wind power output probability function is obtained.
[0024] The beneficial effects of the preferred technical scheme are that the wind power output uncertainty is accurately quantified, and the accuracy and reliability of the scheduling optimization are enhanced.
[0025] As a preferred scheme of the high-penetration new energy distribution network collaborative optimization method, the signal fluctuation distribution characteristic extraction comprises:
[0026] The wind power output probability function is randomly sampled by using a Monte Carlo method to obtain an output deviation prediction data set;
[0027] The fluctuation amplitude threshold is determined according to the output deviation prediction data set, the output fluctuation scene is divided, and a scene occurrence probability value is generated;
[0028] An electromagnetic wave propagation path graph is established according to the forest coverage density graph and the terrain elevation data, and signal transmission loss distribution data is calculated;
[0029] The signal transmission loss distribution data is processed, and a signal fluctuation characteristic sequence data is reconstructed and generated.
[0030] The beneficial effects of the preferred technical scheme are that the signal fluctuation characteristics are accurately described through random sampling and feature extraction, and data support is provided for subsequent optimization.
[0031] As a preferred scheme of the high-penetration new energy distribution network collaborative optimization method, the potential delay interval of the energy storage system scheduling instruction comprises:
[0032] The upper and lower limits of the signal amplitude threshold are set according to the historical signal fluctuation statistical value, the mean value and the standard deviation of the real-time signal are calculated, and it is judged whether the signal fluctuation exceeds the upper and lower limits of the signal amplitude threshold;
[0033] For the signal segment exceeding the upper and lower limits of the signal amplitude threshold, a channel fading model is established in combination with the signal fluctuation distribution characteristic and the thin fog path loss;
[0034] Output wireless signal propagation delay, determine the transmission time sequence of energy storage scheduling instruction;
[0035] According to the wireless signal transmission delay, the sending and receiving time stamps of the energy storage scheduling instruction are obtained, and the instruction transmission delay time sequence is generated;
[0036] Data processing is performed on the instruction transmission delay time sequence to generate a smooth delay sequence;
[0037] The statistical parameters of the smooth delay sequence are calculated to generate a delay time probability distribution function;
[0038] Error analysis logs are extracted from the probability distribution function, error type correlation is analyzed, delay influencing factors are identified, and potential delay intervals are generated.
[0039] As a preferred scheme of the high-penetration new energy distribution network collaborative optimization method, the dispatching disorder evaluation data includes:
[0040] Obtain the time stamp information of the energy storage scheduling instruction sequence, extract the power regulation direction parameter and the regulation amplitude parameter, and generate the instruction execution sequence;
[0041] Generate a disturbance execution sequence according to the instruction execution sequence;
[0042] Scan the disturbance execution sequence, extract the regulation direction sequence, and judge the power conflict point of adjacent instructions;
[0043] According to the time sequence deviation and instruction conflict characteristics, output the energy storage scheduling disorder evaluation data.
[0044] As a preferred scheme of the high-penetration new energy distribution network collaborative optimization method, the energy storage and topology configuration optimization data includes:
[0045] According to the energy storage scheduling disorder evaluation data, extract the delay-induced power flow fluctuation characteristics;
[0046] Combine historical wind power output data to determine whether the power flow fluctuation exceeds the preset range;
[0047] If it exceeds the preset range, obtain the energy storage unit charging and discharging power data and response time data to generate an energy storage characteristic database;
[0048] Calculate the response time weight coefficient according to the energy storage characteristic database to generate a response time sequence mapping table;
[0049] Use the piecewise linear programming method to process the response time sequence mapping table, set the constraint condition, and generate the power regulation instruction;
[0050] According to the power regulation instruction, a power flow distribution is calculated, a power flow distribution data set is generated, a network state evaluation index is constructed, and energy storage and topology configuration optimization data are generated.
[0051] The application provides a high-penetration new energy power distribution network collaborative optimization system.
[0052] To solve the above technical problems, the application provides the following technical scheme: a high-penetration new energy power distribution network collaborative optimization system, comprising:
[0053] A data processing module acquires wind farm operation data and signal strength, processes the wind farm operation data and signal strength, and obtains a signal attenuation trend;
[0054] An identification module fits the signal attenuation trend, obtains signal attenuation characteristics, identifies thin fog environment path loss according to the signal attenuation characteristics, and obtains a wind power output probability distribution function;
[0055] A calculation module obtains an output fluctuation range according to the wind power output probability distribution function, predicts an output deviation scenario, calculates wireless signal fluctuation probability for each scenario, and extracts signal fluctuation distribution characteristics;
[0056] A model construction module establishes a channel fading model according to the signal fluctuation distribution characteristics and the thin fog environment path loss, outputs wireless signal propagation delay, obtains a transmission time sequence, and obtains a potential delay interval of an energy storage system scheduling instruction according to the transmission time sequence;
[0057] A simulation module simulates scheduling disorder and power regulation instruction conflict in the potential delay interval, identifies allocation deviation of microgrid topology configuration and energy storage scheduling, and generates scheduling disorder evaluation data;
[0058] A judgment module extracts power flow fluctuation from the scheduling disorder evaluation data, and judges the power flow fluctuation in combination with historical wind power output data;
[0059] A collaborative optimization module adjusts the response order of an energy storage unit according to the judgment result, generates a power regulation instruction sequence, calculates adjusted power flow distribution data, and generates energy storage and topology configuration optimization data.
[0060] The application provides a computer device, comprising a memory and a processor, and the memory stores a computer program.
[0061] The application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the steps of the high-penetration new energy power distribution network collaborative optimization method.
[0062] The application discloses a high-permeability new energy power distribution network collaborative optimization method, extracts Rayleigh fading parameters and signal attenuation characteristics by collecting wind speed, mist density and other data, identifies mist environment path loss, and determines a wind power output probability distribution function. According to signal fluctuation characteristics, a channel fading model is established, potential delay of energy storage scheduling instructions is analyzed, and scheduling disorder problems are evaluated. For power flow fluctuations caused by the delay, the application adjusts the response order of the energy storage unit, generates a power regulation scheme, and applies a multi-objective optimization algorithm to obtain an energy storage and topology configuration optimization scheme. The application effectively solves the problem of energy storage scheduling delay caused by unstable wireless communication of wind farms in hilly mist environments, and improves the reliability and economy of wind farm operation. BRIEF DESCRIPTION OF DRAWINGS
[0063] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.
[0064] Figure 1 A high-permeability new energy power distribution network collaborative optimization method provided by an embodiment of the application is shown in the overall flowchart. DETAILED DESCRIPTION
[0065] In order to make the above-mentioned purposes, features and advantages of the application more obvious and easy to understand, the specific embodiments of the application will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are only a part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor should be within the protection scope of the application.
[0066] Embodiment 1, refer to Figure 1 For an embodiment of the application, the embodiment provides a high-permeability new energy power distribution network collaborative optimization method, which comprises the following steps.
[0067] S100: Obtain wind farm operation data and signal intensity, and perform data processing on the wind farm operation data and signal intensity to obtain a signal attenuation trend;
[0068] In the embodiment of the application, wind farm operation data in a hilly mist environment is obtained, wind speed, mist density and signal intensity of a forest sheltered area are collected by a sensor, and signal attenuation trend is obtained by filtering and noise reduction processing;
[0069] Specifically, the wind speed data in the ground-to-impeller height interval is measured by a height-adjustable wind speed sensor at each fan, and a wind speed spatial distribution sampling point set is established according to the data recorded by the fan blade speed and the wind speed sensor, and Gaussian filtering is performed on the sampling point set to eliminate abnormal data fluctuations, and the filter window size is set to three times the sampling time interval to obtain a stable wind speed data set. According to the atmospheric visibility and relative humidity parameters obtained by the meteorological sensor in the wind farm area, the thin fog concentration data at different height positions are measured by using a scattering type concentration meter, and the scattering attenuation numerical matrix of electromagnetic waves by thin fog is obtained through Mie scattering theory. According to the geographical information measurement data of the forest covered area near the fan tower, a three-dimensional topographic model is constructed, and the terrain elevation point cloud data in the forest sheltered area is obtained by using a laser radar scanning method, and the regional sheltering intensity distribution map is obtained through the point cloud density. The stable wind speed data set, the scattering attenuation numerical matrix and the regional sheltering intensity distribution map are processed in time synchronization and space registration, and a comprehensive environmental impact factor data set is constructed by using a weighted average method. The signal strength data collected by the fan sensor is processed by wavelet transform, db4 wavelet basis function (Morse series wavelet basis function) is selected for three-layer decomposition of the signal, and high-frequency noise components are removed by soft threshold method. The signal data matrix after noise reduction is processed by singular value decomposition method to extract features, and the singular value vector corresponding to the eigenvalue with a cumulative contribution rate of 95% is selected. According to the comprehensive environmental impact factor data set and the signal feature vector, a signal attenuation prediction model is constructed, and the signal attenuation trend curve of each monitoring area is calculated.
[0070] The collection of wind farm operation data relies on virtual reality technology, which can monitor and analyze data in real time through an immersive interface, ensuring that operation and maintenance personnel can intuitively understand the power grid operation status under complex environment. Data collection can be triggered by sensor network or remote instruction, and the specific method can be determined flexibly according to the actual scene.
[0071] It should be noted that the height-adjustable wind speed sensor is arranged at each wind turbine position of the wind farm, and the wind speed in the range from the ground to the height of the impeller is measured, the sampling frequency is 10 Hz, and the sampling time is not less than 10 minutes. Based on the wind speed data recorded synchronously by the wind turbine blade speed, a spatial distribution sampling point set is constructed, Gaussian filtering is used to eliminate random fluctuations caused by turbulence, the filter window size is three times the sampling time interval, a stable wind speed data set is generated, the data smoothing degree is significantly improved, and the standard deviation is reduced by about 50%; the scattering type concentration meter collects thin fog concentration data at different heights, combines the atmospheric visibility and humidity parameters provided by the meteorological sensor, calculates the scattering attenuation of electromagnetic waves based on the Mie scattering theory, and generates a scattering attenuation numerical matrix. Actual measurement shows that the thin fog concentration is higher within 30 meters from the ground, resulting in about 30% of the total attenuation of electromagnetic wave energy loss; the laser radar scans the forest sheltered area to obtain point cloud data, and the point cloud density reaches 50 points per square meter within 300 meters. Based on the point cloud density, the regional sheltering intensity is calculated to generate a sheltering intensity distribution map, which clearly reflects the influence of forest height and leaf area on signal propagation, and the sheltering attenuation accounts for about 45% of the total loss; the wind speed data set, the scattering attenuation numerical matrix and the sheltering intensity distribution map are time and space registered, the data is integrated by using a weighted average method, and a comprehensive environmental impact factor data set is generated. The embodiments of the present application do not excessively limit the registration algorithm, and the skilled in the art can select a suitable method according to actual needs; the signal strength data is decomposed into three layers by using a db4 wavelet base function, high-frequency noise is removed by using a soft threshold denoising method, and the signal-to-noise ratio is improved by about 6 decibels. The noise reduction signal characteristics are extracted by singular value decomposition, the vector corresponding to the cumulative contribution rate of 95% of the characteristic value is selected, a noise reduction signal data matrix is generated, and the correlation coefficient between the reconstructed signal and the original signal is above 0.95.
[0072] S200: fitting the signal attenuation trend to obtain the signal attenuation characteristic, identifying the thin fog environment path loss according to the signal attenuation characteristic, and obtaining the wind power output probability distribution function;
[0073] S300: obtaining the output fluctuation range according to the wind power output probability distribution function, predicting the output deviation scene, calculating the wireless signal fluctuation probability of each scene, and extracting the signal fluctuation distribution characteristic;
[0074] S400: establishing a channel fading model according to the signal fluctuation distribution characteristic and the thin fog environment path loss, outputting the wireless signal propagation delay, obtaining the transmission time sequence, and obtaining the potential delay interval of the energy storage system scheduling instruction according to the transmission time sequence;
[0075] S500: simulating scheduling disorder and power regulation instruction conflict in the potential delay interval, identifying the distribution deviation of the microgrid topology configuration and the energy storage scheduling, and generating scheduling disorder evaluation data;
[0076] S600: Extract power flow fluctuation from dispatch sequence evaluation data, combine historical wind power output data to judge power flow fluctuation;
[0077] Specifically, the dispatch sequence evaluation data and microgrid topology data are collected, the node power injection vector and admittance matrix are established, and the initial power flow distribution is obtained by solving the power flow equation through Newton iteration method. The time series reconstruction is performed on the power flow calculation results, the differential method is used to calculate the power change rate sequence and voltage deviation sequence, and the abnormal fluctuation points are eliminated through median filtering. The filtered sequence is segmented by using a sliding time window, the window length is set as a multiple of the sampling period, and the mean and standard deviation in the window are calculated as statistical features. The statistical feature sequence is wavelet transformed, db4 wavelet basis function is selected for three-layer decomposition, high-frequency coefficients and low-frequency coefficients are extracted, and the fluctuation characteristic curve is reconstructed. The wind power output data is read from the historical database, the least square method is used to fit the power change trend curve, and the deviation sequence of the actual power and the trend curve is calculated. According to the deviation sequence, an adaptive threshold is set, the threshold parameter is dynamically adjusted with the deviation standard deviation, and the power curve is divided into steady-state segment and fluctuation segment. The frequency spectrum analysis is performed on the fluctuation segment data by using fast Fourier transform, the amplitude and phase characteristics of the main frequency components are extracted, and the fluctuation feature vector is constructed. The fluctuation feature vector is compared with the preset fluctuation range, the fluctuation overrun probability is calculated by kernel density estimation, and the overrun evaluation result of the power flow fluctuation is obtained.
[0078] It should be noted that the dispatching disorder evaluation data and the micro-grid topology information are collected, the node power injection vector and the admittance matrix are constructed, and the number of nodes is usually 10 to 50. Newton iteration method is used to solve the power flow equation, and after 4 to 6 iterations, the error is less than 0.0001 per unit, the initial power flow distribution is generated, which reflects the power balance and network connection characteristics of each node, and lays a foundation for subsequent fluctuation analysis; the time series reconstruction is performed on the power flow distribution data, the differential method is used to calculate the power change rate and the voltage deviation sequence, and under normal working conditions, the power change rate is not more than 20% of the rated power per minute, and the voltage deviation is controlled within the range of plus or minus 5% of the nominal value. The instantaneous fluctuation caused by measurement noise is eliminated through 5-point median filtering, and the standard deviation of the filtered sequence is reduced by about 35%, and a smooth sequence is generated to provide reliable data for feature extraction; the smooth sequence is processed by using a 10-minute sliding time window, the step is 5 minutes, and the overlap rate is 50%, the mean, standard deviation, skewness and kurtosis in the window are calculated, and the fast and slow fluctuation characteristics are captured. Using db4 wave basis function for three-layer decomposition, high-frequency coefficients reflect short-term fluctuations above 0.1 Hz, and low-frequency coefficients correspond to long-term trends below 0.01 Hz, and the fluctuation characteristic curve is reconstructed. The 60-minute trend curve is extracted from the historical wind power output data, the least square method is used for fitting, the deviation sequence of the actual power and the trend is calculated, the adaptive threshold is set to 2 times the standard deviation of the deviation, and the stable state and fluctuation section are dynamically divided to adapt to the fluctuation characteristics of different wind conditions; the fast Fourier transform is applied to the fluctuation section data, the amplitude and phase of the main frequency components are extracted, and the frequency range includes weather scale changes of 0.001 to 0.01 Hz, turbulence fluctuations of 0.01 to 0.1 Hz and noise components above 0.1 Hz. Construct a fluctuation feature vector, combine the preset fluctuation range, select a Gaussian kernel through kernel density estimation, optimize the bandwidth through cross-validation, and calculate the fluctuation overrun probability. The results show that the overrun probability increases from 5% to 15% under strong wind conditions, and the proportion of high-frequency components increases, reflecting the significant impact of turbulence on the stability of the power grid, providing accurate basis for optimizing scheduling.
[0079] S700: According to the judgment result, the response order of the energy storage unit is adjusted, the power regulation instruction sequence is generated, the adjusted power flow distribution data is calculated, and the energy storage and topology configuration optimization data are generated.
[0080] It should be noted that through multi-source data fusion and dynamic modeling, the uncertainty of wind power output is accurately quantified, the energy storage scheduling and micro-grid topology configuration are optimized, the operation reliability and economy of the new energy distribution network are significantly improved, and the scheduling delay problem under complex environment is effectively solved.
[0081] In the embodiment of the application, the step S200 includes the following sub-steps A1-A4.
[0082] In A1: the signal sequence of the wind farm is acquired, the signal sequence is preprocessed, and a signal amplitude observation matrix is established;
[0083] In A2: The signal amplitude observation matrix is normalized, and the signal amplitude probability distribution function is generated by fitting the least squares method.
[0084] In A3: The scattering coefficient is calculated based on the signal amplitude probability distribution function, and the electromagnetic wave energy loss matrix is generated by combining the distance attenuation parameter;
[0085] In A4: Establish the path loss mapping relationship based on the electromagnetic wave energy loss matrix, calculate the transmission path loss probability distribution, and generate the wind power output probability function.
[0086] In this embodiment of the invention, there are two ways to obtain the wind power output probability function;
[0087] Specifically, the first method for obtaining the wind power output probability function involves preprocessing the instantaneous signal data sequence collected from the wind farm, removing abnormal fluctuations through median filtering to obtain a stable signal amplitude time series, and establishing a signal amplitude observation matrix. Normalizing the signal amplitude observation matrix yields a standardized signal amplitude sequence, expressed as: ;
[0088] in, For the normalized first Each signal amplitude value, The first in the original signal amplitude observation matrix Amplitude value, This represents the minimum value of the signal amplitude. This represents the maximum value of the signal amplitude.
[0089] Normalizing the signal amplitude value to between 0 and 1 facilitates subsequent probability distribution fitting calculations.
[0090] By fitting the Rayleigh probability density function using the least squares method, the fading depth parameter and the signal amplitude probability distribution function are extracted and expressed as: ;
[0091] in, Here are the parameters of the Rayleigh distribution. The total number of observed data points. For the first The empirical cumulative distribution function value for each observation point Rayleigh distribution in the th The theoretical cumulative distribution function value for each observation point For the first One signal amplitude observation value.
[0092] The optimal parameters are estimated by fitting the probability distribution function of the signal amplitude using the least squares method and minimizing the squared error between the observed values and the theoretical distribution.
[0093] An exponential function is used to fit the variation of the normalized signal amplitude sequence with propagation distance. The signal attenuation coefficient and characteristic distance parameters are extracted to establish a distance attenuation function. The Mie scattering coefficient is calculated based on the particle size distribution and dielectric constant of the fog particles. Combined with the distance attenuation function, an electromagnetic wave energy loss matrix is constructed, yielding the scattering loss distribution function, expressed as: ;
[0094] in, Distance Path loss at the location, For reference distance Path loss at the location, This is the path loss index. The Mie scattering coefficient is... For transmission distance, For reference distance.
[0095] A path loss mapping relationship is established based on the electromagnetic wave energy loss matrix and the scattering loss distribution function. The total loss probability distribution along the signal transmission path is estimated using a Bayesian probability estimation method. The path loss probability distribution is randomly sampled using the Monte Carlo method. Combined with the characteristics of the wind turbine power curve, the output power under different loss conditions is obtained. The wind power output probability density distribution function, i.e., the wind power output probability function, is obtained through kernel density estimation, expressed as: ;
[0096] in, Wind power output is less than or equal to the threshold The cumulative probability, For the number of samples taken in Monte Carlo, For the first The wind power output value of the second sampling This is an indicator function; it returns 1 when the condition is met and 0 otherwise. The number of samples.
[0097] In this embodiment of the invention, after completing steps A1-A4, step S200 also includes steps A5-A7;
[0098] In A5: Obtain the actual power output record of the wind farm, perform wavelet decomposition on the actual power output record of the wind farm, and obtain a smooth power output curve;
[0099] In A6: the smooth output curve is segmented according to a fixed time window, the peak point set and the valley point set in the time window are calculated, and the output fluctuation frequency feature is obtained;
[0100] In A7: a training sample matrix is generated according to the output fluctuation frequency feature and the environmental parameter sequence, the time sequence feature is extracted, and the wind power output probability function is obtained;
[0101] Specifically, the second method for obtaining the wind power output probability function is that the wind farm monitoring device collects wind turbine group output data and meteorological environmental parameters, linear interpolation processing is performed on the original data to complete the missing values, and the median filter is used to eliminate abnormal fluctuation points, so as to obtain the preprocessed output time sequence. The preprocessed output time sequence is wavelet decomposed, a db4 wavelet base function is selected for 3-layer decomposition, the high-frequency coefficients are denoised through a soft threshold method, and a smooth output curve is reconstructed. The smooth output curve is segmented by using a fixed time window, the window length is set to an integer multiple of the sampling period, the peak point set and the valley point set in each window are used to statistically obtain the output fluctuation frequency feature. The wind speed and pressure data collected by the meteorological station are used to establish an environmental parameter sequence, the periodical feature of the wind speed sequence is extracted through Fourier transform, and the change feature of the pressure sequence is extracted through trend decomposition. The output fluctuation frequency feature and the environmental parameter sequence are aligned according to the time stamp, a training sample matrix is constructed, and the time sequence feature of the sample data is extracted by using a long short-term memory network. The probability density estimator is trained based on the extracted time sequence feature, the probability distribution function of the wind power output is obtained through the kernel density estimation method, the probability density curve of the output uncertainty is obtained, and the wind power output probability function is obtained.
[0102] It should be noted that the signal attenuation characteristics and the analysis of the wind power output uncertainty rely on the virtual reality platform, the signal attenuation trend and the output fluctuation law are intuitively presented through the immersive scene, and the operation and maintenance personnel can easily master the power grid operation state under complex environment. The present application generates high-precision path loss and output distribution characteristics through multi-dimensional data processing and probability modeling, and provides a reliable basis for subsequent optimization scheduling. The specific implementation mode can be flexibly adjusted according to the actual scene, and the embodiments of the present application do not excessively limit the specific algorithm details.
[0103] It should also be noted that the instantaneous signal data sequence is collected by the wind farm sensor network, which contains the Rayleigh fading characteristics caused by thin fog and terrain shielding. The median filtering method is used for signal sequence preprocessing, and the filter window size is set to five times the sampling period to remove abnormal values caused by device jitter or electromagnetic interference, which are usually more than 3 times the normal signal. The fluctuation amplitude of the filtered signal is reduced by about 40%, generating a stable signal amplitude time series, which provides the basis for subsequent analysis. The signal amplitude observation matrix is normalized, and 90% of the normalized amplitude values are distributed in the 0.3 to 0.8 interval. The least squares method is used to fit the Rayleigh distribution probability density function, and the fading depth parameter is extracted, with the shape parameter ranging from 1.2 to 1.8. The larger the parameter value, the more serious the signal fading. Further, the exponential function is used to fit the variation of signal amplitude with propagation distance. Based on the test data of 20 measurement points with a spacing of 50 meters, the signal attenuation coefficient is calculated to be between 0.02 and 0.05, and the characteristic distance parameter is between 300 and 500 meters, reflecting the propagation distance required for signal strength to decay to the initial value 1 / e. According to the particle size distribution of thin fog, 2 to 8 microns, particle number density 500 to 2000 per cubic centimeter, and dielectric constant, real part 5 to 7, the scattering coefficient is calculated to be between 0.01 and 0.03 based on Mie scattering theory. Combined with the distance attenuation function, the electromagnetic wave energy loss matrix is constructed, and the scattering loss distribution function is generated. Based on the electromagnetic wave energy loss matrix, the total loss probability distribution on the transmission path is calculated using the Bayesian probability estimation method, with a mean value of 35 to 50 decibels and a standard deviation of 8 to 12 decibels. The distribution curve shows a right-skewed characteristic, reflecting the combined effect of thin fog and terrain on signal propagation. Through the Monte Carlo method, the path loss probability distribution is randomly sampled 10000 times, and combined with the wind turbine power curve characteristics, the output power distribution under different loss conditions is calculated. The kernel density estimation is used to generate the wind power probability density function, which shows a double-peak characteristic, with the main peak corresponding to 75% to 85% of the rated power and the secondary peak corresponding to 45% to 55%. This reflects the significance of output fluctuations in a thin fog environment. The distribution standard deviation increases by about 30%, and the 95% confidence interval width is about 50% of the rated power, providing accurate basis for subsequent dispatching optimization. The wind farm monitoring device collects wind turbine output data and meteorological environmental parameters, including wind speed and pressure sequences. For data missing, random missing accounts for 3% to 5%, and continuous missing lasts for 10 to 30 minutes. Linear interpolation is used to complete the missing values, with an interpolation error controlled within 5%. The median filtering method is used to process abnormal fluctuations, with a filter window of 5 data points, and the standard deviation is reduced by about 40%, generating a preprocessed output time series for subsequent feature extraction. The db4 wavelet basis function is used for three-level decomposition of the preprocessed output time series, separating high-frequency rapid fluctuations and low-frequency trend components. The soft threshold method is used to process high-frequency coefficients, with a threshold set to 1.5 times the standard deviation, and the smoothed output curve is reconstructed.The power fluctuation frequency is counted by the peak and valley point set in the window, the time interval is logarithmically normally distributed, and the mean value is between 15 and 25 minutes; the wind speed and air pressure data are collected by the weather station, the Fourier transform is used to extract the wind speed day cycle of 24 hours, the amplitude is 2 to 4 meters per second, and the seasonal cycle is 365 days, the amplitude is 3 to 6 meters per second. The linear trend change rate of 0.1 to 0.3 hundred pascals per hour, the fluctuation period of 12 to 24 hours and the amplitude of 2 to 5 hundred pascals are extracted by the trend decomposition method. The output fluctuation frequency characteristics and the environmental parameter sequence are aligned according to the time stamp to construct a training sample matrix, the time span is 1 year, and the sampling interval is 10 minutes; the long short-term memory network is used to process the training sample matrix, the network contains three layers of structure, the input is 24 hours of historical data, and the time sequence correlation characteristics of output fluctuation and environmental parameters are extracted. The kernel density estimation is used to train the probability density estimator, and the wind power output probability density function is generated, which presents a double-peak feature, the main peak value corresponds to 70% to 80% of the rated power, the secondary peak value corresponds to 30% to 40%, and the 95% confidence interval width is about 50% of the rated power. The embodiment of the present application improves the output prediction accuracy and reduces the uncertainty influence in the thin fog environment by multi-dimensional feature extraction and probability modeling, and provides reliable support for optimizing energy storage scheduling and topology configuration.
[0104] In the embodiment of the present application, the above step S300 comprises the following sub-steps B1-B4.
[0105] In B1, the wind power output probability function is randomly sampled by using the Monte Carlo method to obtain the output deviation prediction data set.
[0106] In B2, the fluctuation amplitude threshold is determined according to the output deviation prediction data set, the output fluctuation scene is divided, and the scene occurrence probability value is generated.
[0107] In B3, the electromagnetic wave propagation path graph is established according to the forest coverage density graph and the terrain elevation data, and the signal transmission loss distribution data is calculated.
[0108] In B4, the signal transmission loss distribution data is processed to reconstruct the signal fluctuation characteristic sequence data.
[0109] Specifically, the output fluctuation interval is obtained according to the wind power output probability distribution function, a Monte Carlo method is used for random sampling of the probability distribution, a plurality of sets of output deviation prediction data are generated, the mean and standard deviation of each set of data are obtained to obtain the fluctuation range characteristic value. The output deviation prediction data is normalized, the threshold value of the fluctuation amplitude is determined by the quantile method, the output fluctuation scene is divided according to the threshold value, and the occurrence probability and duration of each scene are obtained. The spatial distribution of the shielding area is obtained by using the forest coverage density diagram, and the electromagnetic wave propagation path diagram is established by combining the terrain elevation data obtained by the laser radar, and the signal transmission loss distribution under each scene is obtained. The wireless signal data in the shielding area is preprocessed, the abnormal values are removed by median filtering, and the missing data points are completed by linear interpolation to obtain a continuous signal intensity sequence. The preprocessed signal sequence is decomposed by 4 layers by selecting the Haar wavelet basis function, the high-frequency coefficients are denoised by the soft threshold method, and the signal fluctuation characteristic sequence is reconstructed. The signal fluctuation characteristic sequence is divided by using the adaptive segmentation method, the amplitude mean, frequency median and duration of each signal are obtained, and the probability density function of the signal fluctuation characteristic is constructed.
[0110] It should be noted that the analysis of the wind power output probability distribution function is combined with virtual reality technology to show the output fluctuation and signal transmission characteristics through an immersive scene, helping operation and maintenance personnel to intuitively understand the power grid operation state under complex environment. The present application accurately quantifies the output deviation and signal fluctuation characteristics by using Monte Carlo sampling, quantile analysis and wavelet transform, generates a propagation path model combined with terrain and vegetation data, and improves the pertinence and reliability of dispatching optimization.
[0111] It should also be noted that based on the wind power output probability density function, a Monte Carlo method is used for 10000 times of random sampling to generate a plurality of sets of output deviation prediction data covering various scenes such as normal operation and reduced output operation. The mean and standard deviation of each set of data are calculated to determine the output fluctuation range, the standard deviation reflects the fluctuation intensity, the main peak value corresponds to 70% to 80% of the rated power, and the secondary peak value corresponds to 40% to 50%. Through normalization processing, 90% of the data fluctuation amplitude is distributed between 0.2 and 0.8, the quantile method is used to set 0.3 and 0.7 as the threshold value, the output fluctuation is divided into small, medium and large fluctuation scenes, and the small fluctuation accounts for 60%, and the duration is usually within 30 minutes, providing a basis for subsequent scene analysis.
[0112] The terrain elevation data of the forest sheltered area is obtained by laser radar scanning, the average height of the forest is 15-25 meters, the crown density is 0.8-1.2 per square meter, and the electromagnetic wave propagation path model is constructed in combination with the coverage density map. Based on the output fluctuation scene, the signal transmission loss distribution under each scene is calculated, the loss value is significantly affected by the height and density of the trees, and the typical loss range is 30-45 decibels. The propagation path model quantifies the attenuation effect of the shelter on the electromagnetic wave through numerical simulation, and generates loss distribution data to support signal fluctuation analysis.
[0113] The wireless signal data of the sheltered area is preprocessed, the median filter is used to remove abnormal values caused by device jitter or external interference, the amplitude of the abnormal value is more than 2.5 times of the normal value, the filter window is 5 data points, and the signal standard deviation is reduced by about 35%. For data missing, linear interpolation is used to complete, the error is controlled within 5%, and a continuous signal strength sequence is generated. The Haar wavelet basis function is selected for four-layer decomposition, the high-frequency mutation and low-frequency trend components are separated, the soft threshold method is used to process the high-frequency coefficients, the threshold is set to 1.5 times of the coefficient standard deviation, and the signal fluctuation characteristic sequence is reconstructed to retain the fast change characteristics.
[0114] The signal fluctuation characteristic sequence is divided by using an adaptive segmentation method, when the mean difference of adjacent data points is more than 2 times of the standard deviation, it is identified as a segmentation point, the duration is between 5 and 20 minutes, and the amplitude change range is 20%-60% of the peak value. The amplitude mean, frequency median and duration of each segment are calculated, and the probability density function of the signal fluctuation characteristics is constructed. Analysis shows that the signal fluctuation frequency during the day is higher than that at night, the period is 15-40 minutes, the amplitude change range increases by about 25% during the day, and the energy proportion of the high-frequency component rises to 35%, reflecting the dynamic influence of thin fog and terrain on signal transmission stability, providing accurate data support for optimization scheduling.
[0115] In the embodiment of the present application, the above step S400 includes the following sub-steps C1-C7.
[0116] In C1, the signal amplitude threshold upper and lower limits are set according to the historical signal fluctuation statistical value, the mean and standard deviation of the real-time signal are calculated, and it is judged whether the signal fluctuation exceeds the signal amplitude threshold upper and lower limits;
[0117] In C2, for the signal segment exceeding the signal amplitude threshold upper and lower limits, a channel fading model is established in combination with the signal fluctuation distribution characteristics and the thin fog path loss;
[0118] In C3, the wireless signal propagation delay is output, and the transmission time sequence of the energy storage scheduling instruction is determined;
[0119] In C4, the sending and receiving time stamps of the energy storage scheduling instruction are obtained according to the wireless signal transmission delay, and the instruction transmission delay time sequence is generated;
[0120] In C5: data processing is performed on the instruction transmission delay time sequence to generate a smooth delay sequence;
[0121] In C6: statistical parameters of the smooth delay sequence are calculated to generate a delay time probability distribution function;
[0122] In C7: error analysis logs are extracted from the probability distribution function, error type correlation is analyzed, delay influencing factors are identified, and potential delay intervals are generated.
[0123] Specifically, signal amplitude threshold upper and lower limits are set according to historical signal fluctuation statistical values, the mean and standard deviation of real-time signals are calculated through a sliding window method, and it is determined whether the signal fluctuation exceeds the signal amplitude threshold upper and lower limits, which is represented as:
[0124] wherein, is the standard deviation of wind power output deviation prediction data, is the total number of deviation prediction data, is the th output deviation prediction value, is the mean of all deviation prediction values, is the data serial number.
[0125] For the signal segment exceeding the signal amplitude threshold upper and lower limits, wavelet transform is used to extract fluctuation amplitude and frequency characteristics, and the fluctuation probability distribution function is calculated through maximum likelihood estimation, which is represented as:
[0126] wherein, is the probability density function of signal fluctuation amplitude, is the signal amplitude value at time , is the mean of signals in the sliding window, is the standard deviation of signals in the sliding window, is an exponential function.
[0127] A channel state feature vector is constructed combining the fluctuation probability distribution function and Rayleigh distribution parameters, and a channel attenuation coefficient is calculated through path loss data of the thin fog environment, which is represented as:
[0128] wherein, is the channel attenuation coefficient, is the basic path loss at the reference distance, is the transmission distance, is the reference distance, is the weight coefficients of the eigenvectors, for the first channel state eigenvector, dimension of the eigenvector.
[0129] The recursive neural network is trained using the channel state eigenvector, the input features include signal amplitude, frequency, and attenuation coefficient, and the output is a predicted sequence of channel states. The wireless signal propagation delay is calculated according to the predicted channel state sequence, and the probability density estimation method is used to construct the delay distribution function, which is expressed as: ;
[0130] wherein, is the cumulative distribution function of the transmission delay, is the upper limit value of the delay, is the integral variable representing a specific delay value, is the mean parameter of the delay distribution, is the standard deviation parameter of the delay distribution, is the exponential function.
[0131] The energy storage control instructions are prioritized according to the response time requirements, and the transmission time intervals of instructions with different priorities are obtained by combining the delay distribution function. The transmission time intervals are randomly sampled using the Monte Carlo method, and the delay sequence of the instruction transmission is obtained through the cumulative distribution function.
[0132] It should be noted that the construction of the channel fading model utilizes virtual reality technology to real-time display the signal fluctuation and transmission delay characteristics through an immersive interface, helping operation and maintenance personnel to quickly assess the wireless communication state. This step optimizes the transmission strategy by extracting signal fluctuation features and predicting the delay distribution, combined with instruction priority, to improve the efficiency of energy storage scheduling.
[0133] It should be noted that according to historical signal data statistics, the signal amplitude standard deviation is in the range of 2 to 5 decibels, the upper threshold is set to the mean plus 3 times the standard deviation, and the lower threshold is set to the mean minus 2.5 times the standard deviation. The mean and standard deviation of the real-time signal are calculated using a 60-second sliding window to determine whether it exceeds the upper and lower limits of the signal amplitude threshold. The signal segment that exceeds the threshold reflects abnormal fluctuations caused by thin fog or obstruction, triggering subsequent feature extraction and time delay analysis to ensure timely detection of communication anomalies. For the signal segment that exceeds the threshold, a db4 wavelet basis function is used for three-layer decomposition to extract the fluctuation amplitude and frequency characteristics. The amplitude feature is calculated by the energy distribution of the wavelet coefficient, and 90% of the energy is concentrated in the first two layers. The frequency feature is obtained by reconstructing the signal with zero crossing rate. The fluctuation probability distribution function is calculated using maximum likelihood estimation to reflect the dynamic change law of signal strength, providing data support for channel modeling. Combined with the fluctuation probability distribution function and the Rayleigh distribution parameters, the shape parameter is between 1.2 and 1.8, and a multi-dimensional channel state feature vector containing amplitude, frequency and thin fog path loss is constructed. The path loss increases exponentially with distance, and the attenuation coefficient is between 0.2 and 0.5. A long short-term memory neural network is used, with 300 seconds of historical features as input, the network contains two layers of 64 neuron hidden layers, the training data covers various weather conditions in a year, and the prediction accuracy of the channel state reaches more than 85%, generating a state prediction sequence; based on the channel state prediction sequence, a probability density estimation is used to construct a time delay distribution function. Under normal conditions, the time delay is normally distributed, with a mean of 20 to 50 milliseconds; when the channel deteriorates, it is right skewed, with a mean of 80 to 150 milliseconds. Energy storage commands are divided into high, medium and low priority according to response time, requiring less than 100 milliseconds, 200 milliseconds and 500 milliseconds respectively. Monte Carlo method is used for 10000 times sampling to simulate the transmission process under different conditions, the high priority command has a 95% time delay of less than 80 milliseconds, the medium priority command has a 150 milliseconds time delay, and the low priority command has a 400 milliseconds time delay. Combined with the cumulative distribution function, the tail characteristics of the time delay distribution are analyzed to optimize the command transmission strategy and ensure that the high priority command still maintains a relatively small time delay under harsh conditions, about 30%, improving the scheduling reliability.
[0134] Specifically, the sending timestamp and receiving timestamp of the energy storage dispatching command are collected, and the time reference is calibrated through a time synchronization server to obtain the actual delay time sequence of the command transmission. Anomaly detection is performed on the delay time sequence, and abnormal delay points are identified by the three-sigma method. The abnormal values are smoothed by median filtering. Statistical features are extracted from the smoothed delay sequence, including mean, variance, skewness and kurtosis, to construct the probability distribution function of the delay time as follows: ;
[0135] wherein, is the delay time probability density, is a mean parameter of the delay time, is a standard deviation parameter of the delay time, is a standard deviation parameter of the delay time, is an exponential function.
[0136] The instruction parsing log is read from the receiving end database, the parsing error code and error description are extracted, an error type dictionary is established, and the occurrence frequency of each type of error is counted. The association rule mining method is used to analyze the association relationship between error types, the minimum support and confidence thresholds are set, and the frequent error patterns are extracted. The delay time characteristics and error pattern characteristics are combined to construct a feature vector, and the random forest classifier is used to identify the key delay influencing factors. According to the weight of the identified delay influencing factors, the confidence interval of the delay distribution is obtained by the kernel density estimation method, and then the potential delay interval range of the energy storage scheduling instruction is obtained.
[0137] It should be noted that the delay distribution and error patterns are displayed through the immersive interface, which helps the operation and maintenance personnel to monitor the communication performance in real time. This step accurately quantifies the instruction transmission delay and its influencing factors, improves the reliability and real-time performance of the scheduling instruction, and provides support for power grid operation optimization in complex environments.
[0138] It should be noted that by collecting the sending and receiving time stamps of the energy storage scheduling instructions, time calibration is performed relying on the network time protocol server, the synchronization accuracy is better than 1 millisecond, and the actual delay time sequence is calculated. The delay sequence presents periodic characteristics, reflecting the dynamic changes of network communication. The three standard deviation method is used to detect abnormal delay points. Abnormal values are usually caused by network delay or communication interruption, which are more than 3 times the normal value. The normal delay range is 50 to 150 milliseconds, and the abnormal value can reach 500 milliseconds or more. The 5-point median filter is used to smooth the abnormal points, and the standard deviation of the filtered sequence is reduced by about 30%, which provides stable data for subsequent analysis; the mean, variance, skewness and kurtosis of the smoothed delay sequence are calculated. The measured data show that the delay mean is about 80 milliseconds under good communication state, the standard deviation is less than 20 milliseconds, the skewness is close to 0, and the kurtosis is between 2.8 and 3.2, which is approximately normally distributed. The kernel density estimation method is used, the Gaussian kernel function is selected, the bandwidth is determined by the Silverman empirical formula, the delay probability distribution function is generated, and the distribution characteristics of the delay are clearly reflected, which provides a basis for confidence interval calculation; the instruction analysis log is extracted from the receiving end database to construct an error type dictionary, which includes data verification error, format analysis error and timeout error, accounting for about 25%, 15% and 10% respectively. Data verification error is mainly caused by bit flip, format analysis error is related to mismatched instruction version, and timeout error reflects network congestion. The Apriori algorithm is used for association rule mining, the minimum support is set to 0.1 and the minimum confidence is set to 0.6, the association probability of data verification error and timeout error reaches 0.7, indicating that when the network quality decreases, the error types are coupled with each other, increasing the risk of instruction transmission; the delay statistical characteristics and error mode are combined into a 10-dimensional feature vector, including delay mean, standard deviation, error probability, etc., which is input into the random forest classifier for training. The classifier uses 10-fold cross-validation to optimize the parameters, and the accuracy is more than 85%, and the weight analysis shows that network delay and signal quality are the main delay factors. Combined with the kernel density estimation result, the delay interval under 95% confidence level is calculated, the range is 2.5 times the standard deviation of the mean, and the upper limit of the delay of high-priority instructions is controlled within 200 milliseconds. Seasonal analysis shows that the delay mean increases by about 30% under winter fog weather, and the distribution tail is heavier, so additional time allowance needs to be reserved to ensure the real-time performance of the scheduling instruction.
[0139] In the embodiment of the application, the above step S500 comprises the following sub-steps D1-D4.
[0140] In D1, the time stamp information of the energy storage scheduling instruction sequence is acquired, the power regulation direction parameter and the regulation amplitude parameter are extracted, and the instruction execution sequence is generated.
[0141] In D2, the disturbance execution sequence is generated according to the instruction execution sequence.
[0142] In D3: scan the execution sequence of the disturbance, extract the adjustment direction sequence, and judge the power conflict point of the adjacent instructions;
[0143] In D4: according to the time sequence deviation and instruction conflict characteristics, output the energy storage scheduling disorder evaluation data.
[0144] Specifically, the energy storage scheduling instruction sequence is time stamped and sorted, the power adjustment direction and adjustment amplitude parameters of the instructions are extracted, and an instruction execution sequence database is established. According to the potential delay interval, the time sampling interval is set, the Monte Carlo method is used to generate multiple sets of delay time sequences, and the execution sequence after the disturbance is obtained through instruction time sequence rearrangement. An instruction comparison matrix is established for the execution sequence after the disturbance, and according to the time interval and power change characteristics between adjacent instructions, the time sequence deviation of instruction execution is obtained. The fixed time window scanning method is used to detect the power adjustment instruction conflict, the window sliding step and overlap rate are set, and the adjustment direction sequence of the instructions in the window is extracted. According to the adjustment direction sequence, the instruction conflict rule table is constructed, the direction opposite and amplitude overlap judgment threshold are set, and the conflict point of the power adjustment instruction is identified. The topology connection data is obtained from the microgrid controller, the capacity parameters and power limit of the energy storage unit are extracted, and the energy storage resource allocation constraint condition is established. The support vector regression is used to obtain the allocation deviation of the energy storage scheduling, the time sequence deviation and instruction conflict characteristics are combined, and the scheduling disorder evaluation result is output through the multilayer perception machine.
[0145] It should be noted that the timestamp information is obtained from the energy storage scheduling instruction sequence, the precision is better than 1 millisecond, the power regulation direction and amplitude parameter are extracted, the direction includes charging and discharging, and the amplitude range covers 10% to 90% of the rated power. Based on the timestamp sorting, the instruction execution sequence database is constructed, the execution time, regulation target and priority of each instruction are recorded, the data integrity of subsequent analysis is ensured, and the basis for out-of-sequence evaluation is provided; according to the potential delay interval, the sampling interval is set to 1 / 10 of the minimum delay, and when the typical interval is 50 to 200 milliseconds, the sampling interval is 5 milliseconds. 10000 groups of delay time sequences are generated by using Monte Carlo method to simulate the influence of communication delay on instruction execution order, and the disturbance execution sequence is generated by time sequence rearrangement. Statistics show that 90% of the time sequence deviation is 50 to 150 milliseconds, and the maximum deviation is not more than 3 times of the original interval, reflecting the disturbance degree of delay on the execution order; the regulation direction sequence in the window is extracted by scanning the disturbance execution sequence with a fixed time window of 60 seconds, a sliding step of 30 seconds and an overlap rate of 50%. The instruction comparison matrix is constructed, the matrix elements represent the time interval and power change of adjacent instructions, the number of direction changes is normally not more than 4 times, and too many changes indicate control oscillation. Set the conflict rules: the direction conflict occurs when the directions of adjacent instructions are opposite and the interval is less than 20 seconds, the amplitude conflict occurs when the amplitude overlap degree is more than 70%, the power conflict point is identified, and the time sequence deviation of instruction execution is quantified. The topology connection data is obtained from the microgrid controller, the capacity and power limit of the energy storage unit are extracted, the charging and discharging power is controlled within 85% of the rated power, and the state of charge is maintained at 20% to 90%. Support vector regression is used, radial basis function kernel is selected, parameters are optimized by cross-validation, energy storage allocation deviation is calculated, and the correlation coefficient is above 0.75. Combined with the time sequence deviation and conflict characteristics, a 16-dimensional feature vector is constructed, input into a multilayer perceptron with two hidden layers, the number of neurons is 32 and 16 respectively, and the scheduling out-of-sequence evaluation value is output. The results show that 15% of the instructions have out-of-sequence, 3% are serious out-of-sequence, especially when the load changes rapidly, the delay and frequent instruction switching superimpose, causing the deviation of multi-unit collaborative scheduling to intensify, and the scheduling strategy needs to be optimized to improve the system stability.
[0146] In the embodiment of the application, the above step S700 includes the following sub-steps E1-E6.
[0147] In E1: according to the energy storage scheduling out-of-sequence evaluation data, the delay-induced power flow fluctuation characteristics are extracted;
[0148] In E2: combined with the historical wind power output data, it is judged whether the power flow fluctuation exceeds the preset range;
[0149] In E3: if it exceeds the preset range, the charging and discharging power data and response time data of the energy storage unit are obtained, and an energy storage characteristic database is generated;
[0150] In E4: Calculate the response time weighting coefficient based on the energy storage characteristic database and generate a response time series mapping table;
[0151] In E5: a piecewise linear programming method is used to process the response timing mapping table, set constraints, and generate power regulation commands;
[0152] In E6: Calculate power flow distribution based on power regulation commands, generate power flow distribution dataset, construct network status evaluation indicators, and generate energy storage and topology configuration optimization data.
[0153] Specifically, the charging and discharging power and response time data of energy storage units are collected to construct an energy storage characteristic database. Response time weighting coefficients are obtained using the analytic hierarchy process (AHP). Energy storage units are prioritized based on these weighting coefficients, and a response time sequence mapping table is established using the scheduling disorder assessment results to generate a new response order. For the adjusted response order, piecewise linear programming is used to solve the power allocation problem, setting upper and lower power limits and ramp rate constraints to obtain power adjustment commands for each time period. The energy storage power adjustment commands are used to obtain a new power flow distribution. Node voltage and branch power are solved using power flow equations to generate a power flow distribution dataset. Network status assessment indicators are constructed based on the power flow distribution data, including voltage deviation, network loss, and energy storage lifetime indicators, with weighting coefficients set for each indicator. A multi-objective optimization function is established based on the assessment indicators, setting variable constraints for energy storage capacity and topology. A non-dominated sorting genetic algorithm is used to solve the optimization problem. Pareto optimality is assessed on the optimization results, and the optimal solution set satisfying the constraints is selected to generate an energy storage capacity configuration scheme and a network topology scheme. The multi-objective optimization function is expressed as follows: ;
[0154] in, To synthesize the objective function value, , , These are the weighting coefficients for voltage deviation, grid loss, and energy storage life, respectively. The total number of nodes. For the first The actual voltage value of each node For reference voltage value, The total number of branch roads, For the first Power loss of the branch circuit This represents the total number of energy storage devices. For the first Capacity decay of an energy storage device For the first The rated lifespan of each energy storage device.
[0155] It should be noted that the charging and discharging power and response time data of the energy storage unit are collected, the response time of the lithium battery is 50 to 100 milliseconds, the response time of the flywheel energy storage is less than 20 milliseconds, and the response time of the super capacitor is about 10 milliseconds. A database of energy storage characteristics including power, response speed and reliability is constructed. The weight coefficient of response time is calculated by using the analytic hierarchy process. The response speed, power fluctuation and scheduling reliability are considered, and the weights are 0.5, 0.3 and 0.2 respectively. The priority order is generated by the judgment matrix. When the scheduling disorder degree exceeds 0.6, the response order needs to be completely rearranged; when the disorder degree is between 0.3 and 0.6, only part of the unit is adjusted, the response time mapping table is formed, and the execution efficiency of the instruction is optimized; based on the response time mapping table, the power distribution is solved by using piecewise linear programming, the power upper limit is set to 90% of the rated power, the lower limit is 15%, and the climbing rate is limited to 30% per minute, to ensure the safe operation of the energy storage. After generating the power regulation instruction, the power flow distribution is calculated by using Newton-Raphson method, the convergence accuracy is 0.0001 per unit, the iteration is completed within 5 times, the node voltage is maintained between 0.95 and 1.05 per unit, and the network loss rate is less than 3%, to generate the power flow distribution dataset, which provides a basis for network state evaluation; according to the power flow distribution data, the network state evaluation index is constructed, including voltage deviation, network loss and energy storage life, and the weights are 0.4, 0.3 and 0.3 respectively, reflecting the operation quality, economy and maintenance cost. A multi-objective optimization function is established, the energy storage capacity and topological structure are set as constraint variables, the non-dominated sorting genetic algorithm is used, the population size is 200, the evolution algebra is 500, the crossover probability is 0.8, the mutation probability is 0.1, and 10 to 20 Pareto optimal solutions are selected. In the optimization scheme, the fast response energy storage accounts for 30% of the total capacity, the medium speed response energy storage accounts for 50%, and the slow speed response energy storage accounts for 20%. The fast response energy storage is preferentially configured in the load intensive area to improve the scheduling flexibility, and the slow response energy storage is configured in the load stable area to reduce the maintenance cost, so as to optimize the operation efficiency of the power grid.
[0156] The above is a schematic scheme of the high-penetration new energy distribution network collaborative optimization method of the embodiment. It should be noted that the technical scheme of the high-penetration new energy distribution network collaborative optimization system belongs to the same concept as the technical scheme of the high-penetration new energy distribution network collaborative optimization method described above. The details of the technical scheme of the high-penetration new energy distribution network collaborative optimization system in the embodiment are not described in detail and can be referred to the description of the technical scheme of the high-penetration new energy distribution network collaborative optimization method.
[0157] The high-penetration new energy distribution network collaborative optimization system in the embodiment comprises:
[0158] The data processing module acquires wind farm operation data and signal strength, processes the wind farm operation data and signal strength, and obtains a signal attenuation trend.
[0159] The identification module fits the signal attenuation trend to obtain a signal attenuation characteristic, identifies a thin fog environment path loss according to the signal attenuation characteristic, and obtains a wind power output probability distribution function;
[0160] The calculation module obtains an output fluctuation range according to the wind power output probability distribution function, predicts an output deviation scenario, calculates a wireless signal fluctuation probability for each scenario, and extracts a signal fluctuation distribution feature;
[0161] The model construction module establishes a channel fading model according to the signal fluctuation distribution feature and the thin fog environment path loss, outputs a wireless signal propagation delay, obtains a transmission time sequence, and obtains a potential delay interval of an energy storage system scheduling instruction according to the transmission time sequence;
[0162] The simulation module simulates scheduling disorder and power regulation instruction conflict in the potential delay interval, identifies a distribution deviation of microgrid topology configuration and energy storage scheduling, and generates scheduling disorder evaluation data;
[0163] The judgment module extracts power flow fluctuation from the scheduling disorder evaluation data, and judges the power flow fluctuation in combination with historical wind power output data;
[0164] The cooperative optimization module adjusts a response order of an energy storage unit according to the judgment result, generates a power regulation instruction sequence, calculates adjusted power flow distribution data, and generates energy storage and topology configuration optimization data.
[0165] The embodiment also provides a computer device suitable for a high-penetration new energy power distribution network cooperative optimization method, which comprises a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the high-penetration new energy power distribution network cooperative optimization method.
[0166] The embodiment also provides a storage medium having a computer program stored thereon, and the program is executed by a processor to realize the high-penetration new energy power distribution network cooperative optimization method.
[0167] The storage medium proposed in the embodiment and the high-penetration new energy power distribution network cooperative optimization method proposed in the above embodiment belong to the same inventive concept, and the technical details not described in detail in the embodiment can be referred to the above embodiment, and the embodiment has the same beneficial effects as the above embodiment.
[0168] From the above description of the embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software and necessary universal hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a floppy disk, a Read Only Memory (ROM), a Random Access Memory (RAM), a FLASH, a hard disk, or an optical disc, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods of various embodiments of the present application.
[0169] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application, and although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, and all of them should be covered in the scope of the claims of the present application.
Claims
1. A high-permeability new energy power distribution network collaborative optimization method, characterized in that, The method comprises the following steps: Obtain wind farm operation data and signal strength, process the wind farm operation data and signal strength, and obtain signal attenuation trend; Fit the signal attenuation trend to obtain signal attenuation characteristics, identify thin fog environment path loss according to the signal attenuation characteristics, and obtain wind power output probability distribution function; Obtain output fluctuation range according to the wind power output probability distribution function, predict output deviation scenarios, calculate wireless signal fluctuation probability of each scenario, and extract signal fluctuation distribution characteristics; Establish a channel fading model according to the signal fluctuation distribution characteristics and the thin fog environment path loss, output wireless signal propagation delay, obtain transmission time sequence, and obtain potential delay interval of energy storage system scheduling instruction according to the transmission time sequence; Simulate scheduling disorder and power regulation instruction conflict in the potential delay interval, identify allocation deviation of microgrid topology configuration and energy storage scheduling, and generate scheduling disorder evaluation data; Extract power flow fluctuation from the scheduling disorder evaluation data, and judge the power flow fluctuation in combination with historical wind power output data; According to the judgment result, adjust the response order of the energy storage unit, generate the power regulation instruction sequence, calculate the adjusted power flow distribution data, and generate the energy storage and topology configuration optimization data.
2. The high-permeability new energy power distribution network collaborative optimization method of claim 1, wherein, The wind power output probability distribution function comprises: Obtain the signal sequence of the wind farm, preprocess the signal sequence, and establish a signal amplitude observation matrix; Normalize the signal amplitude observation matrix, and generate a signal amplitude probability distribution function by least square fitting; Calculate the scattering coefficient according to the signal amplitude probability distribution function, and generate an electromagnetic wave energy loss matrix in combination with the distance attenuation parameter; Establish a path loss mapping relationship according to the electromagnetic wave energy loss matrix, calculate the transmission path loss probability distribution, and generate the wind power output probability function.
3. The high-permeability new energy power distribution network collaborative optimization method of claim 1, wherein, The wind power output probability distribution function further comprises: Obtain the actual output record of the wind farm, perform wavelet decomposition on the actual output record of the wind farm, and obtain a smooth output curve; Segment the smooth output curve according to a fixed time window, calculate the peak point set and the valley point set in the time window, and obtain the output fluctuation frequency characteristics; Generate a training sample matrix according to the output fluctuation frequency characteristics and the environmental parameter sequence, extract the time sequence characteristics, and obtain the wind power output probability function.
4. The high-permeability new energy power distribution network collaborative optimization method of claim 1, wherein, The extraction of signal fluctuation distribution characteristics comprises: Randomly sample the wind power output probability function by using the Monte Carlo method to obtain an output deviation prediction data group; Determine the fluctuation amplitude threshold value according to the output deviation prediction data group, divide the output fluctuation scenarios, and generate scenario occurrence probability values; Establish an electromagnetic wave propagation path diagram according to the forest coverage density diagram and the terrain elevation data, and calculate signal transmission loss distribution data; Process the signal transmission loss distribution data, and reconstruct the signal fluctuation characteristic sequence data.
5. The high-permeability new energy power distribution network collaborative optimization method of claim 4, wherein, The potential delay interval of the energy storage system scheduling instruction comprises: Set the upper and lower limits of the signal amplitude threshold value according to the historical signal fluctuation statistical value, calculate the mean value and standard deviation of the real-time signal, and judge whether the signal fluctuation exceeds the upper and lower limits of the signal amplitude threshold value; For the signal segment exceeding the upper and lower limits of the signal amplitude threshold value, establish a channel fading model in combination with the signal fluctuation distribution characteristics and the thin fog path loss; The output wireless signal propagation delay determines the transmission time sequence of the energy storage scheduling instruction; The transmission and reception time stamps of the energy storage scheduling instruction are obtained according to the wireless signal transmission delay, and the instruction transmission delay time sequence is generated; The instruction transmission delay time sequence is subjected to data processing to generate a smooth delay sequence; The statistical parameters of the smooth delay sequence are calculated to generate a delay time probability distribution function; From the probability distribution function, the error analysis log is extracted, the error type correlation is analyzed, the delay influencing factors are identified, and the potential delay interval is generated.
6. The high-permeability new energy power distribution network collaborative optimization method of claim 5, wherein, The scheduling disorder evaluation data includes: Obtain the time stamp information of the energy storage scheduling instruction sequence, extract the power regulation direction parameter and the regulation amplitude parameter, and generate the instruction execution sequence; According to the instruction execution sequence, a disturbance execution sequence is generated; The disturbance execution sequence is scanned to extract the regulation direction sequence, and the power conflict point of adjacent instructions is judged; According to the time sequence deviation and the instruction conflict characteristics, the energy storage scheduling disorder evaluation data is output.
7. The high-permeability new energy power distribution network collaborative optimization method of claim 6, wherein, The energy storage and topology configuration optimization data includes: According to the energy storage scheduling disorder evaluation data, the delay-induced power flow fluctuation characteristics are extracted; Combined with historical wind power output data, it is judged whether the power flow fluctuation exceeds the preset range; If it exceeds the preset range, the energy storage unit charging and discharging power data and response time data are obtained to generate an energy storage characteristic database; According to the energy storage characteristic database, the response time weight coefficient is calculated to generate a response time sequence mapping table; The response time sequence mapping table is processed by using the piecewise linear programming method, the constraint conditions are set, and the power regulation instruction is generated; According to the power regulation instruction, the power flow distribution is calculated, the power flow distribution data set is generated, the network state evaluation index is constructed, and the energy storage and topology configuration optimization data is generated.
8. A high-penetration new energy power distribution network collaborative optimization system, applying a high-penetration new energy power distribution network collaborative optimization method according to any one of claims 1-7, characterized in that, It includes: A data processing module obtains wind farm operation data and signal strength, processes the wind farm operation data and signal strength, and obtains a signal attenuation trend; An identification module fits the signal attenuation trend to obtain signal attenuation characteristics, identifies the thin fog environment path loss based on the signal attenuation characteristics, and obtains a wind power output probability distribution function; A calculation module obtains an output fluctuation range based on the wind power output probability distribution function, predicts an output deviation scenario, calculates the wireless signal fluctuation probability of each scenario, and extracts signal fluctuation distribution characteristics; A model construction module establishes a channel fading model based on the signal fluctuation distribution characteristics and the thin fog environment path loss, outputs the wireless signal propagation delay, obtains the transmission time sequence, and obtains the potential delay interval of the energy storage system scheduling instruction based on the transmission time sequence; A simulation module simulates scheduling disorder and power regulation instruction conflict in the potential delay interval, identifies the allocation deviation of microgrid topology configuration and energy storage scheduling, and generates scheduling disorder evaluation data; A judgment module extracts power flow fluctuations from the scheduling disorder evaluation data and judges the power flow fluctuations in combination with historical wind power output data; A collaborative optimization module adjusts the response order of the energy storage unit based on the judgment result, generates a power regulation instruction sequence, calculates the adjusted power flow distribution data, and generates energy storage and topology configuration optimization data. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor implements the steps of the high-penetration new energy distribution network collaborative optimization method in any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the high-penetration new energy distribution network collaborative optimization method in any one of claims 1 to 7.
Citation Information
Patent Citations
Wind power generation system peak regulation capacity assessment method
CN107480833A
Power system optimization dispatching method and device, computer device and storage medium
CN110783963A
Power system steady-state power flow risk identification and prevention method, storage medium and equipment
CN111740415A
Building light storage and charging integrated energy management system
CN120222459A
Power capacity evaluation method based on dynamic confidence interval and double-layer optimization
CN120257820A
Cited By
Dry powder extinguishing agent production line working condition data processing method and system
CN121614799A
