Multi-source dynamic power grid harmonic intelligent early warning method and system based on robust optimization

Abstract

The invention discloses a multi-source dynamic power grid harmonic intelligent early warning method and system based on robust optimization, and relates to the technical field of electric power system monitoring, and the method comprises the following specific steps: data acquisition in a multi-source injection scene: through a distributed power supply monitoring device, a load monitoring device, a voltage transformer and a current transformer equipment, carrying out the multi-source injection scene; the output data of the distributed power supply and the real-time power data of the load are respectively collected, the change rule of the harmonic source can be accurately described by establishing the probability distribution models of the output and load change of the distributed power supply, and when the harmonic impedance of the power grid is dynamically monitored, the probability models are introduced and different actual operation conditions are simulated. The method comprehensively considers the influence of the uncertainty of the harmonic source on the harmonic impedance, obtains a harmonic impedance estimated value after the uncertainty factor is fully considered, achieves the dynamic and high-precision monitoring of the harmonic impedance of the power grid along with the change of time and working conditions, and provides a reliable basis for the accurate evaluation of the operation state of the power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of power system monitoring, and in particular to a multi-source dynamic power grid harmonic intelligent early warning method and system based on robust optimization. Background Art

[0002] With the large-scale construction of AC transmission networks above 750 kV and the widespread application of intelligent dispatching systems, the multi-source injection characteristics of distributed power sources (such as photovoltaic and wind power) and nonlinear loads (such as frequency conversion equipment and arc furnaces) have posed severe challenges to the security and defense systems of large-scale power grids.

[0003] Traditional harmonic impedance monitoring and early warning methods have obvious limitations and fail to fully consider the uncertainty factors of harmonic sources in multi-source injection scenarios. These methods are usually calculated and analyzed based on certain operating conditions and parameters, ignoring the randomness and volatility of distributed power output and load changes. Due to the lack of effective processing of harmonic source uncertainty, traditional methods have deviations in the assessment of power grid harmonic risks. When the harmonic source fluctuates greatly, the system cannot capture the changes in harmonic impedance in a timely and accurate manner, and thus cannot issue early warning information in a timely manner. This lag and inaccuracy increases the risk of power grid failures caused by harmonic problems, making it difficult for the power grid to operate safely and stably when facing complex and changeable operating conditions.

[0004] In view of the above problems, it is necessary to optimize the existing methods of dynamic monitoring and intelligent early warning of power grid harmonic impedance. By establishing a probability model, dynamically monitoring harmonic impedance and designing strategies and implementing intelligent early warning based on robust optimization theory, high-precision monitoring and reliable early warning of power grid harmonic impedance can be achieved to ensure the safe and stable operation of the power grid. Therefore, the development of a harmonic impedance dynamic monitoring and intelligent early warning technology that can comprehensively handle multi-source uncertainties and adapt to the characteristics of power grids above 750 kV is of great significance to improving the security protection capabilities of large-scale power grids and supporting the efficient operation of intelligent dispatching systems. Summary of the Invention

[0005] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide a multi-source dynamic power grid harmonic intelligent early warning method and system based on robust optimization. It can establish a probability model and a robust optimization strategy by fully considering the uncertainty of the harmonic source in the multi-source injection scenario. In terms of data acquisition, it uses a variety of monitoring devices to comprehensively collect distributed power output data, real-time load power data, and voltage and current data of each node in the power grid. By establishing a harmonic source uncertainty probability model, it accurately describes the uncertainty of distributed power output and load changes. In the harmonic impedance monitoring link, the probability model is combined to dynamically calculate the harmonic impedance estimation value considering uncertainty. Based on the robust optimization theory, monitoring and early warning are designed, and an optimization model is constructed to determine reasonable monitoring parameters and early warning thresholds. Ultimately, high-precision monitoring and reliable early warning of the power grid harmonic impedance are achieved, ensuring the safe and stable operation of the power grid in the multi-source injection scenario, and providing strong technical support for the stable operation of the power system.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: On the one hand, a multi-source dynamic power grid harmonic intelligent early warning method based on robust optimization, the method comprising the following specific steps:

[0007] Data acquisition in multi-source injection scenarios: Distributed power monitoring devices, load monitoring devices, voltage transformers, and current transformers are used to collect distributed power output data, real-time load power data, and voltage and current data at each grid node.

[0008] Establishment of a probability model for harmonic source uncertainty: Analyze the collected distributed power output data and load change data separately to determine their probability distribution type. Estimate the corresponding distribution parameters based on historical data and establish a probability distribution model that can describe the uncertainty of distributed power output and load changes.

[0009] Dynamic monitoring of grid harmonic impedance: The voltage and current data of each node in the grid are used to extract the harmonic voltage and current components. The harmonic impedance of the grid at different times is dynamically calculated based on the definition of harmonic impedance. The established harmonic source uncertainty probability model is integrated into the calculation process of harmonic impedance, comprehensively considering the impact of harmonic source uncertainty on harmonic impedance.

[0010] Design monitoring and early warning systems based on robust optimization theory: Construct an optimization model targeting harmonic impedance monitoring accuracy and early warning reliability, taking into account constraints related to harmonic source uncertainty. Use a robust optimization algorithm to solve the optimization model, and determine reasonable monitoring parameters and early warning thresholds based on the solution.

[0011] Intelligent early warning of power grid harmonics: The harmonic impedance calculated by real-time monitoring is compared with the determined early warning threshold. When the harmonic impedance exceeds the early warning threshold, the system issues an early warning message.

[0012] Furthermore, in the step of establishing the harmonic source uncertainty probability model, the collected distributed power output data and load change data are analyzed separately to determine their probability distribution types. Based on the historical distributed power output data, the parameter estimation method is used to determine the specific parameter values corresponding to the distribution type, and a probability distribution model that can characterize the distributed power output is established, thereby reflecting the uncertainty of the distributed power output. The time series analysis method is used to analyze the load change data to determine the probability distribution type of the load change data. According to the characteristics of the load change data itself and the results of the analysis, a calculation method adapted thereto is used to estimate the relevant parameters of the probability distribution and construct a probability distribution model of the load change.

[0013] Furthermore, in the step of establishing the harmonic source uncertainty probability model, a probability distribution model capable of describing the output of distributed power sources is established, and the model formula is: f(x, y; θ) = f X|Y (x|y;θ X|Y )·f Y (y;θ Y ), where x represents the ratio of the output of distributed generation to the rated output, y represents the load power, θ=(θ X|Y ,θ Y ) is the model parameter set, f Y (y;θ Y ) is the marginal probability distribution of load power, and the formula is: Among them, μ is the mean parameter of the normal distribution, σ is the standard deviation parameter of the normal distribution, which is calculated by historical load data, f X|Y (x|y;θ X|Y ) is the conditional probability distribution of distributed power output under the given load power y, taking into account the temporal and spatial correlation between meteorological factors and load, and its formula is: X|Y in, is the signal variance, which affects the overall variation of the conditional probability distribution, l is the length scale parameter, which controls the degree of influence of distance on the correlation in the covariance function, Observation noise variance takes into account the impact of noise in actual measurements on the model. is the sample mean of the distributed power output, k(y) is the covariance vector of y and the historical load power data, which is used to measure the correlation between the current load power y and the historical data, and K is the covariance matrix, which reflects the correlation structure between the historical load power data.

[0014] Furthermore, in the step of establishing the harmonic source uncertainty probability model, a probability distribution model of load change is established, and its model parameters are: Among them, π k is the weight of the kth mixture component, satisfying And π=(π1,π2,…) obeys Dirichlet distribution DP(α,G0), y represents the load power, μ k is the mean parameter of the kth Gaussian distribution mixture component. is the variance parameter of the k-th Gaussian distribution mixture component, As the base distribution, it provides the basic distribution form for generating mixture components of the Dirichlet process.

[0015] Furthermore, in the step of dynamically monitoring the harmonic impedance of the power grid, the voltage and current data of each node of the power grid are obtained to extract the harmonic voltage and current components of each order. According to the definition of harmonic impedance, the voltage and current components corresponding to each extracted harmonic are calculated, and the specific value of each harmonic impedance is obtained by calculating their ratio. The calculation formula is: Among them, Z h Indicates the hth harmonic impedance, U h It represents the hth harmonic voltage, which is the voltage value corresponding to the hth harmonic obtained by processing the collected grid node voltage data. h It represents the hth harmonic current, which is the current value corresponding to the hth harmonic extracted after processing the collected grid node current data.

[0016] Furthermore, in the step of dynamically monitoring the harmonic impedance of the power grid, the harmonic impedance of the power grid at different times is dynamically calculated based on the definition of harmonic impedance. Specifically, different combinations of distributed power output and load change data that conform to the two probability distributions are randomly generated multiple times. For each set of data combination, the harmonic impedance under this operating condition is calculated. Through a large number of calculations and analyses, the impact of harmonic source uncertainty on harmonic impedance is comprehensively considered, and the harmonic impedance estimation value that fully considers the uncertainty factors is obtained, thereby realizing dynamic monitoring of the power grid harmonic impedance that changes with time and operating conditions.

[0017] Furthermore, in the step of dynamically monitoring the harmonic impedance of the power grid, for each set of data combinations, the harmonic impedance under such working conditions is calculated. Specifically, it is assumed that N simulations are performed, and each simulation generates a set of distributed power output P according to the probability model of distributed power output and load changes. DG,i (i=1,2,…,N) and load power P L ,i, calculate the hth harmonic impedance Z h,i , and obtain the estimated value of harmonic impedance considering uncertainty where γ i It is the correction coefficient introduced in each simulation based on the current simulated distributed power output and load power combination and the historical operation data of the power grid.h,i It represents the hth harmonic impedance obtained by the ith simulation, where N is the number of simulations, which is determined according to the required accuracy and computing resources. In the i-th simulation, P DG,i and P L,i Under certain specific working conditions, the voltage and current data of the grid nodes are processed to extract the hth harmonic voltage U h and current I h Calculate Z h,i , is the estimated value of the hth harmonic impedance after considering the uncertainty, γ i is the correction coefficient for each simulation, which is dynamically generated based on the machine learning model.

[0018] Furthermore, in the design of monitoring and early warning steps based on robust optimization theory, an optimization model with harmonic impedance monitoring accuracy and early warning reliability as the goals is constructed, and the relevant constraints of harmonic source uncertainty are considered. The robust optimization algorithm is used to solve the optimization model, and its formula is: Among them, J is the objective function value, w1 and w2 are weight coefficients, which are set according to the degree of attention paid to harmonic impedance monitoring accuracy and early warning reliability. w1, w2 ≥ 0 and w1 + w2 = 1. The system is run with the weight coefficients for experimental testing. During the test, the relevant data of harmonic impedance monitoring accuracy and early warning reliability are recorded. The recorded data are analyzed. If it is found that the monitoring accuracy does not meet the expected requirements, that is, The value of is large, so the value of w1 can be increased. The new w1 is calculated by the formula: in is the weight coefficient before adjustment, Δw is the adjustment step size (such as Δw = 0.1), and H is the total number of harmonics considered, is the estimated hth harmonic impedance, Z h is the actual hth harmonic impedance, R is the warning reliability index, which is evaluated by historical warning situations and has a value range of [0, 1]. and They are the minimum and maximum values of the distributed power output, which are determined according to the equipment characteristics and operating conditions. DG is the actual output of distributed power generation, and They are the minimum and maximum values of load power, respectively, which are determined by the type and operation law of the load. L is the actual load power.

[0019] On the other hand, a multi-source dynamic power grid harmonic intelligent early warning system based on robust optimization is proposed, which includes the following components:

[0020] Data acquisition module: used to collect output data of distributed power sources, real-time power data of loads, and voltage and current data of each node in the power grid in a multi-source injection scenario;

[0021] Probabilistic model building module: Based on the data obtained by the data acquisition module, a probability distribution model of distributed power output and load changes is established to describe the uncertainty of harmonic sources;

[0022] Harmonic impedance monitoring module: Based on the voltage and current data obtained by the data acquisition module, it extracts the harmonic components and calculates the harmonic impedance. In combination with the probability model established by the probability model building module, it comprehensively considers the impact of harmonic source uncertainty on harmonic impedance.

[0023] Robust strategy design module: Based on robust optimization theory, it builds an optimization model, considers the impact of harmonic source uncertainty, determines reasonable monitoring parameters and warning thresholds, and designs robust monitoring and warning strategies;

[0024] Intelligent early warning module: compares the real-time harmonic impedance monitored by the harmonic impedance monitoring module with the early warning threshold determined by the robust strategy design module, and issues an early warning message when the harmonic impedance exceeds the early warning threshold.

[0025] Compared with the existing technology, the multi-source dynamic power grid harmonic intelligent early warning method and system based on robust optimization have the following beneficial effects:

[0026] 1. By establishing a probability distribution model for distributed power output and load changes, the present invention can accurately describe the variation patterns of harmonic sources. When dynamically monitoring the harmonic impedance of the power grid, these probability models are introduced and different actual operating conditions are simulated. The impact of harmonic source uncertainty on harmonic impedance is comprehensively considered, and an estimated value of harmonic impedance is obtained after fully accounting for uncertainty factors. This enables dynamic and high-precision monitoring of the changes in power grid harmonic impedance over time and under different operating conditions, providing a reliable basis for accurate assessment of the power grid's operating status.

[0027] 2. The present invention constructs an optimization model aimed at improving the accuracy of harmonic impedance monitoring and enhancing the reliability of early warning results, and considers various related factors of harmonic source uncertainty as constraints. By solving the model, reasonable monitoring parameters and early warning thresholds are determined, so that the intelligent early warning module can obtain the harmonic impedance value obtained by dynamic monitoring in real time and accurately compare it with the early warning threshold. When the harmonic impedance exceeds the early warning threshold, it can quickly issue an early warning information according to a pre-set communication method, effectively solving the problem of not being able to issue early warning information in a timely and accurate manner when the harmonic source fluctuates greatly, and greatly improving the safety and stability of power grid operation.

[0028] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0030] Figure 1 This is a flowchart of a multi-source dynamic power grid harmonic intelligent early warning method based on robust optimization;

[0031] Figure 2 The diagram shows the structure of a multi-source dynamic power grid harmonic intelligent early warning system based on robust optimization. DETAILED DESCRIPTION

[0032] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0033] Example 1

[0034] A cross-regional power grid connects power generation areas in multiple different regions, including areas with centralized access to distributed power sources such as large wind farms and photovoltaic power stations, as well as areas with various load types, such as industrial concentration areas and urban commercial and residential mixed power consumption areas. The distributed power sources in different regions are affected by local natural conditions, and their output has significant uncertainty. Various load types also present complex and changeable power consumption patterns due to differences in regional economic structure and user electricity consumption habits.

[0035] High-precision distributed power monitoring devices are installed near each distributed power generation area in the cross-regional power grid, such as each wind turbine in a large wind farm and each photovoltaic array in a photovoltaic power plant. These devices, with a 12kHz sampling frequency that meets the Nyquist sampling theorem, collect real-time data such as wind speed, direction, rotation speed, and real-time power generation of wind turbines, as well as light intensity, panel temperature, and real-time power generation of photovoltaic power plants. At the same time, the monitoring devices are regularly maintained and calibrated to ensure the accuracy and reliability of the data.

[0036] Load monitoring devices are strategically deployed across various load areas, including factories in industrial clusters, urban commercial centers, and residential areas, based on load distribution and importance. Real-time power data for each load type is collected at a sampling frequency of 12kHz. For large industrial users and critical commercial facilities, the density of monitoring points is increased to fully capture dynamic load changes.

[0037] High-precision voltage and current transformers are installed at key nodes in interregional power grids, such as the ends of interregional tie lines and busbars in large substations. Voltage and current data at each node are collected at a sampling frequency of 12kHz. Using well-shielded cables and a reliable data transmission network, the data is accurately transmitted to the data processing center, preventing signal interference and data loss.

[0038] For distributed power output data, we conduct an in-depth analysis of the historical output data collected from large wind farms and photovoltaic power stations. Using data mining and machine learning algorithms, combined with local meteorological data (such as the long-term statistical distribution of wind speed and light intensity) and the characteristic parameters of power generation equipment, we establish a probability distribution model that can characterize the output of distributed power. The model formula is: f(x, y; θ) = f X|Y (x|y;θ X|Y )·f Y (y;θ Y ), where x represents the ratio of the output of distributed generation to the rated output, y represents the load power, θ=(θ X|Y ,θ Y ) is the model parameter set, f Y (y;θ Y ) is the marginal probability distribution of load power, and the formula is: Among them, μ is the mean parameter of the normal distribution, σ is the standard deviation parameter of the normal distribution, which is calculated by historical load data, f X|Y (x|y;θ X|Y ) is the conditional probability distribution of distributed generation output under the given load power y, taking into account the temporal and spatial correlation between meteorological factors and load, and its formula is: in, is the signal variance, which affects the overall variation of the conditional probability distribution, l is the length scale parameter, which controls the degree of influence of distance on correlation in the covariance function, Observation noise variance takes into account the impact of noise in actual measurements on the model. is the sample mean of the distributed power output, k(y) is the covariance vector of y and the historical load power data, which is used to measure the correlation between the current load power y and the historical data, and K is the covariance matrix, which reflects the correlation structure between the historical load power data.

[0039] For load change data, we use time series analysis and spectrum analysis methods, combined with factors such as the industry characteristics of loads in different regions and user electricity usage habits, to build a probability distribution model for load changes. The model parameters are: Among them, π k is the weight of the kth mixture component, satisfying And π=(π1,π2,…) obeys Dirichlet distribution DP(α,G0), y represents the load power, μ k is the mean parameter of the kth Gaussian distribution mixture component. is the variance parameter of the k-th Gaussian distribution mixture component, As the base distribution, it provides the basic distribution form for generating mixture components of the Dirichlet process.

[0040] The collected voltage and current data of each node of the power grid are transmitted to the high-performance data processing unit of the data processing center. The data are processed using the fast Fourier transform (FFT) algorithm to extract the harmonic voltage and current components. In order to improve the calculation accuracy, the signal is preprocessed to reduce the spectrum leakage error. According to the definition formula of harmonic impedance (where Z h Indicates the hth harmonic impedance, U h Indicates the hth harmonic voltage, I h The harmonic impedance of each order is calculated, and the probability model of distributed photovoltaic output and load change is introduced. 1000 Monte Carlo simulations are performed. In each simulation, the proportion of the distributed photovoltaic power station output to the rated output x is generated according to the probability distribution f(x; α, β) of the distributed photovoltaic output, and the output P of the distributed photovoltaic power station is obtained. DG , P DG =x×P rated ,,P rated is the rated power of the photovoltaic power station. According to the probability distribution of load changes, g(y; μ, σ) generates the load power P L , and obtain a set of distributed photovoltaic output P D G,i and load power P L ,i(i=1,2,…,1000), calculate the hth harmonic impedance Z h,i , combined with the correction factor γ determined based on the historical operation data of the power grid i , through the formula Get an estimate of harmonic impedance taking uncertainty into account Realize dynamic monitoring of power grid harmonic impedance.

[0041] Constructing an optimization model with harmonic impedance monitoring accuracy and early warning reliability as the goals (Where J is the objective function value to be minimized, w1 and w2 are weight coefficients, initially set based on expert experience w1 = 0.6, w2 = 0.4, and w1, w2 ≥ 0 and w1 + w2 = 1, H is the total number of harmonics considered, is the estimated hth harmonic impedance, Z h is the actual hth harmonic impedance, R is the warning reliability index, which can be evaluated by historical warning situations, etc., and its value range is [0, 1]). The range of distributed photovoltaic output and load changes is considered as a constraint condition, that is, (in and They are the minimum and maximum output values of distributed photovoltaic power stations, which are determined according to their equipment characteristics and operating conditions. DG is the actual output of the distributed photovoltaic power station. and They are the minimum and maximum values of load power, respectively, which are determined by the type and operation law of the load. L is the actual load power), and at the same time consider the influence of factors such as the topology of the power grid and equipment parameters on harmonic impedance, incorporate these factors into the constraints of the optimization model, and use the robust optimization algorithm to solve the model. In the solution process, parallel computing technology is used to improve the calculation efficiency, and the weight coefficient is adjusted according to the test results. By conducting simulation tests in the actual power grid, the monitoring accuracy and early warning reliability data under different weight coefficients are collected, and the optimal weight coefficient is determined using data analysis and optimization algorithms. Specifically, the weight coefficient is used to run the system for experimental testing. During the test, the harmonic impedance monitoring accuracy and early warning reliability data are recorded, and the recorded data are analyzed. If it is found that the monitoring accuracy does not meet the expected requirements, that is, The value of is large, so the value of w1 can be increased. The new w1 is calculated by the formula: in is the weight coefficient before adjustment, Δw is the adjustment step size (such as Δw = 0.1), and Finally, reasonable monitoring parameters (such as maintaining a sampling frequency of 10kHz and a monitoring time interval of 5 minutes) and early warning thresholds are determined. In order to improve the accuracy of the early warning, the early warning thresholds are dynamically adjusted and updated in real time according to the operating status of the power grid and load changes.

[0042] The intelligent early warning module obtains the monitored harmonic impedance value in real time When the harmonic impedance exceeds the threshold, as compared with the established warning threshold, the operation and maintenance personnel of the cross-regional power grid are notified through various communication methods (such as text messages, emails, instant messaging software, etc.). The notification content includes detailed information such as the specific value of the harmonic impedance, the number of times it exceeds the threshold, the possible location of the harmonic source (such as the specific power generation area or load area), and the frequency component of the harmonics. At the same time, an audible and visual alarm prompt is issued on the large screen of the power grid monitoring center and in professional monitoring software to attract the attention of the operation and maintenance personnel. The intelligent warning module also provides a detailed harmonic risk analysis report, including an assessment of the impact of harmonics on cross-regional power grid equipment (such as transformers and cables), the types of grid failures that may occur (such as equipment overheating, malfunction of protection devices, etc.), and targeted response measures (such as adjusting the output control strategy of distributed power sources and optimizing load distribution, etc.), to help operation and maintenance personnel take effective measures to reduce harmonic risks in a timely manner and ensure the safe and stable operation of the cross-regional power grid.

[0043] Example 2

[0044] A certain industrial park has a large-scale wind farm connected to the power grid. The wind farm consists of multiple wind turbines distributed in a specific area around the park. The wind farm is significantly affected by factors such as wind speed, wind direction, and seasonal changes. At the same time, there are a large number of industrial loads in the park, covering multiple industries such as electronics manufacturing, mechanical processing, and chemicals. The operation of these industrial loads is highly volatile. The harmonic characteristics generated by production equipment in different industries vary, and their electricity demand changes continuously with production plans and market conditions. The access of the wind farm makes the already complex grid harmonic situation even more difficult to predict and control, posing a potential threat to the safe operation of power equipment and the stability of power supply in the park.

[0045] Distributed power monitoring devices are installed near each wind turbine in the wind farm. These devices are equipped with high-precision sensors that can accurately measure wind speed, wind direction, turbine speed, and real-time power generation in real time. Data is collected at an 8kHz sampling frequency. To ensure data continuity and stability, a data redundancy backup mechanism is implemented. Regular on-site inspections and remote monitoring of the monitoring devices are also conducted to promptly detect and address potential equipment failures and data anomalies. Load monitoring devices are strategically located in various industrial load areas within the industrial park based on load distribution and importance. For large industrial plants and critical production equipment, the density of monitoring points is increased to ensure comprehensive capture of real-time load power changes. Real-time power data for industrial loads is also collected at an 8kHz sampling frequency, and load data from different industries is categorized, stored, and managed. High-precision voltage and current transformers are installed at key grid nodes, such as the incoming and outgoing lines of substations and branch points of important distribution lines. These transformers undergo rigorous calibration and testing to ensure measurement accuracy meets requirements.

[0046] An in-depth and detailed statistical analysis of wind farm output data is conducted. Advanced data mining algorithms and signal processing techniques are used, combined with the region's meteorological data (such as the statistical distribution of historical wind speed and wind direction) and the wind farm's unit characteristic parameters (such as rated power, blade size, etc.). Probability distribution models are established for wind turbines of different models and operating states. A comprehensive analysis of industrial load change data is conducted, using a combination of time series analysis, spectrum analysis and other methods. Combined with the production process and electricity consumption patterns of various industries, based on historical load data, statistical inference and parameter estimation methods are adopted. Consideration is given to the impact of factors such as different time periods (such as weekdays, weekends, holidays), different production shifts, and changes in market demand on the load. According to industry classification and differences in load characteristics, probability distribution models are established for different types of industrial loads to more accurately describe the uncertainty of industrial load changes.

[0047] The collected voltage and current data are transmitted via a high-speed fiber optic network to a data processing center equipped with a high-performance server cluster and professional data processing software. The center uses the Fast Fourier Transform (FFT) algorithm to process the input data, extract the harmonic components, and calculate the harmonic impedance of each order based on the definition of harmonic impedance. A probabilistic model of wind farm output and industrial load changes is introduced, and 800 simulations are performed. During the simulation process, the mutual coupling relationship between wind farms and industrial loads, as well as the impact of changes in the grid topology on harmonic propagation, are considered. Through random sampling and correlation analysis, a reasonable combination of wind farm output and industrial load power data is generated. The corresponding harmonic impedance is calculated for each simulation and combined with a correction coefficient determined based on the historical operation of the grid and fault records. By establishing a correction coefficient prediction model based on machine learning and comprehensively considering the impact of multiple factors on the correction coefficient, an estimated harmonic impedance that takes into account uncertainty is obtained, realizing dynamic monitoring of the grid's harmonic impedance. At the same time, the monitoring results are analyzed and visualized in real time, allowing staff to intuitively understand the harmonic status of the grid.

[0048] An optimization model is constructed with the goal of improving the accuracy of harmonic impedance monitoring and enhancing the reliability of early warning. The weight coefficients are set (initially w1=0.7 and w2=0.3 according to the grid planning and operation objectives), where w1 focuses on the accuracy of harmonic impedance monitoring and w2 focuses on the reliability of early warning. The range of wind farm output and industrial load changes is considered as a constraint condition. At the same time, factors such as grid equipment parameters (such as transformer ratio, line impedance, etc.), operation mode (such as ring network operation, open-loop operation) and configuration of harmonic control equipment are included in the constraint conditions. The model is solved using a robust optimization algorithm, and parallel computing and intelligent optimization are adopted. Search strategy is used to improve solution efficiency and search accuracy. During the solution process, the weight coefficient is adjusted in real time according to changes in the grid operation status, such as new industrial loads, maintenance and overhaul of wind farm units, etc. By establishing a mapping relationship between the weight coefficient and the grid operation parameters, the weight coefficient is adaptively adjusted using a machine learning algorithm, and finally reasonable monitoring parameters (such as a sampling frequency of 8kHz and a monitoring time interval of 10 minutes) and early warning thresholds are determined. In order to ensure the scientificity and rationality of the early warning thresholds, combined with the safe operation standards of the grid and the tolerance capacity of the equipment, verification and optimization are carried out through a combination of simulation and actual testing.

[0049] The intelligent early warning module obtains the harmonic impedance value in real time and compares it with the early warning threshold. When the harmonic impedance exceeds the early warning threshold, the power management personnel in the park are notified through system pop-up prompts and high-decibel sound alarms. The system pop-up window displays in detail the specific value of the harmonic impedance, the time when the threshold is exceeded, the relevant harmonic source information and the possible impact on the power grid equipment. The sound alarm uses different frequencies and rhythms to distinguish different levels of harmonic risks. At the same time, the intelligent early warning module also sends a detailed harmonic risk analysis report to relevant managers via email and instant messaging software. The report content includes the frequency component, amplitude, propagation path analysis, potential hazard assessment of equipment and targeted response measures, etc., to help managers take effective measures to reduce harmonic risks in a timely manner, such as adjusting the operating mode of industrial loads and starting harmonic control equipment.

[0050] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A multi-source dynamic power grid harmonic intelligent early warning method based on robust optimization, characterized in that: The method comprises the following specific steps: Data acquisition in multi-source injection scenarios: Distributed power monitoring devices, load monitoring devices, voltage transformers, and current transformers are used to collect distributed power output data, real-time load power data, and voltage and current data at each grid node. Establishment of a probability model for harmonic source uncertainty: Analyze the collected distributed power output data and load change data separately to determine their probability distribution type. Estimate the corresponding distribution parameters based on historical data and establish a probability distribution model that can describe the uncertainty of distributed power output and load changes. Dynamic monitoring of grid harmonic impedance: The voltage and current data of each node in the grid are used to extract the harmonic voltage and current components. The harmonic impedance of the grid at different times is dynamically calculated based on the definition of harmonic impedance. The established harmonic source uncertainty probability model is integrated into the calculation process of harmonic impedance, comprehensively considering the impact of harmonic source uncertainty on harmonic impedance. Design monitoring and early warning systems based on robust optimization theory: Construct an optimization model targeting harmonic impedance monitoring accuracy and early warning reliability, taking into account constraints related to harmonic source uncertainty. Use a robust optimization algorithm to solve the optimization model, and determine reasonable monitoring parameters and early warning thresholds based on the solution. Intelligent early warning of power grid harmonics: The harmonic impedance calculated by real-time monitoring is compared with the determined early warning threshold. When the harmonic impedance exceeds the early warning threshold, the system issues an early warning message.

2. The method for intelligent early warning of multi-source dynamic power grid harmonics based on robust optimization according to claim 1, characterized in that: In the step of establishing the harmonic source uncertainty probability model, the collected distributed power output data and load change data are analyzed separately to determine their probability distribution types. Based on the historical distributed power output data, the parameter estimation method is used to determine the specific parameter values corresponding to the distribution type, and a probability distribution model that can characterize the distributed power output is established, thereby reflecting the uncertainty of the distributed power output. The time series analysis method is used to analyze the load change data to determine the probability distribution type of the load change data. According to the characteristics of the load change data itself and the results of the analysis, a calculation method adapted thereto is used to estimate the relevant parameters of the probability distribution and construct a probability distribution model of the load change.

3. The method for intelligent early warning of multi-source dynamic power grid harmonics based on robust optimization according to claim 1, characterized in that: In the step of establishing the harmonic source uncertainty probability model, a probability distribution model capable of describing the output of distributed power sources is established, and the model formula is: f(x, y; θ) = f X|Y (x|y;θ X|Y )·f Y (y;θ Y ), where x represents the ratio of the output of distributed generation to the rated output, y represents the load power, θ=(θ X|Y ,θ Y ) is the model parameter set, f Y (y;θ Y ) is the marginal probability distribution of load power, and the formula is: Among them, μ is the mean parameter of the normal distribution, σ is the standard deviation parameter of the normal distribution, which is calculated by historical load data, f X|Y (x|y;θ X|Y ) is the conditional probability distribution of distributed generation output under the given load power y, taking into account the temporal and spatial correlation between meteorological factors and load, and its formula is: in, is the signal variance, which affects the overall variation of the conditional probability distribution, l is the length scale parameter, which controls the degree of influence of distance on correlation in the covariance function, Observation noise variance takes into account the impact of noise in actual measurements on the model. is the sample mean of the distributed power output, k(y) is the covariance vector of y and the historical load power data, which is used to measure the correlation between the current load power y and the historical data, and K is the covariance matrix, which reflects the correlation structure between the historical load power data.

4. The method for intelligent early warning of multi-source dynamic power grid harmonics based on robust optimization according to claim 1, characterized in that: In the step of establishing the harmonic source uncertainty probability model, a probability distribution model of load change is established, and its model parameters are: Among them, π k is the weight of the kth mixture component, satisfying And π=(π1,π2,…) obeys Dirichlet distribution DP(α,G0), y represents the load power, μ k is the mean parameter of the kth Gaussian distribution mixture component. is the variance parameter of the k-th Gaussian distribution mixture component, As the base distribution, it provides the basic distribution form for generating mixture components of the Dirichlet process.

5. The method for intelligent early warning of multi-source dynamic power grid harmonics based on robust optimization according to claim 1, characterized in that: In the step of dynamically monitoring the harmonic impedance of the power grid, the voltage and current data of each node of the power grid are obtained to extract the voltage and current components of each harmonic. According to the definition of harmonic impedance, the voltage and current components corresponding to each extracted harmonic are calculated. By calculating their ratios, the specific values of each harmonic impedance are obtained. The calculation formula is: Among them, Z h Indicates the hth harmonic impedance, U h It represents the hth harmonic voltage, which is the voltage value corresponding to the hth harmonic obtained by processing the collected grid node voltage data. h It represents the hth harmonic current, which is the current value corresponding to the hth harmonic extracted after processing the collected grid node current data.

6. The method for intelligent early warning of multi-source dynamic power grid harmonics based on robust optimization according to claim 1, characterized in that: In the step of dynamically monitoring the harmonic impedance of the power grid, the harmonic impedance of the power grid at different times is dynamically calculated based on the definition of harmonic impedance. Specifically, different combinations of distributed power output and load change data that conform to the two probability distributions are randomly generated multiple times. For each set of data combination, the harmonic impedance under this operating condition is calculated. Through a large number of calculations and analyses, the impact of harmonic source uncertainty on harmonic impedance is comprehensively considered to obtain a harmonic impedance estimate that fully considers the uncertainty factors, thereby realizing dynamic monitoring of the power grid harmonic impedance that changes over time and operating conditions.

7. The method for intelligent early warning of multi-source dynamic power grid harmonics based on robust optimization according to claim 6, characterized in that: In the step of dynamically monitoring the harmonic impedance of the power grid, for each set of data combinations, the harmonic impedance under this working condition is calculated. Specifically, it is assumed that N simulations are performed, and each simulation generates a set of distributed power output P according to the probability model of distributed power output and load changes. DG,i (i=1,2,…,N) and load power P L ,i, calculate the hth harmonic impedance Z h,i , and obtain the estimated value of harmonic impedance considering uncertainty where γ i It is the correction coefficient introduced in each simulation based on the current simulated distributed power output and load power combination and the historical operation data of the power grid. h,i It represents the hth harmonic impedance obtained by the ith simulation, where N is the number of simulations, which is determined according to the required accuracy and computing resources. In the i-th simulation, P DG,i and P L,i Under certain specific working conditions, the voltage and current data of the grid nodes are processed to extract the hth harmonic voltage U h and current I h Calculate Z h,i , is the estimated value of the hth harmonic impedance after considering the uncertainty, γ i is the correction coefficient for each simulation, which is dynamically generated based on the machine learning model.

8. The method for intelligent early warning of multi-source dynamic power grid harmonics based on robust optimization according to claim 5, characterized in that: In the step of designing monitoring and early warning based on robust optimization theory, an optimization model with harmonic impedance monitoring accuracy and early warning reliability as the goals is constructed, and the relevant constraints of harmonic source uncertainty are considered. The robust optimization algorithm is used to solve the optimization model, and its formula is: Where J is the objective function value, w1 and w2 are weight coefficients, which are set according to the importance attached to harmonic impedance monitoring accuracy and warning reliability, w1, w2 ≥ 0 and w1 + w2 = 1, H is the total number of harmonics considered, is the estimated hth harmonic impedance, Z h is the actual hth harmonic impedance, R is the warning reliability index, which is evaluated by historical warning situations and has a value range of [0, 1]. and They are the minimum and maximum values of the distributed power output, which are determined according to the equipment characteristics and operating conditions. DG is the actual output of distributed power generation, and They are the minimum and maximum values of load power, respectively, which are determined by the type and operation law of the load. L is the actual load power.

9. A multi-source dynamic power grid harmonic intelligent early warning system based on robust optimization, the system being applicable to a multi-source dynamic power grid harmonic intelligent early warning method based on robust optimization according to any one of claims 1 to 8, characterized in that: The system includes the following components: Data acquisition module: used to collect output data of distributed power sources, real-time power data of loads, and voltage and current data of each node in the power grid in a multi-source injection scenario; Probabilistic model building module: Based on the data obtained by the data acquisition module, a probability distribution model of distributed power output and load changes is established to describe the uncertainty of harmonic sources; Harmonic impedance monitoring module: Based on the voltage and current data obtained by the data acquisition module, it extracts the harmonic components and calculates the harmonic impedance. In combination with the probability model established by the probability model building module, it comprehensively considers the impact of harmonic source uncertainty on harmonic impedance. Robust strategy design module: Based on robust optimization theory, it builds an optimization model, considers the impact of harmonic source uncertainty, determines reasonable monitoring parameters and warning thresholds, and designs robust monitoring and warning strategies; Intelligent early warning module: compares the real-time harmonic impedance monitored by the harmonic impedance monitoring module with the early warning threshold determined by the robust strategy design module, and issues an early warning message when the harmonic impedance exceeds the early warning threshold.

Images