Regional power supply stability judgment method, medium and system

Abstract

The invention provides a regional power supply stability determination method, medium and system, and belongs to the technical field of regional power supply, and the method comprises the steps: collecting the power supply data of each node in a region in real time, including power supply parameters such as voltage, current and power factor, load parameters such as load power and change rate, and environmental parameters such as temperature and humidity; through data cleaning and feature extraction, a regional power supply stability judgment equation set considering all features is established, and a stability evaluation vector is calculated. And performing parameter fitting on the equation set by using historical data to obtain a stability evaluation vector at the current moment, and inputting the stability evaluation vector into a pre-trained stability judgment model to obtain a power supply stability evaluation result. According to the method, the medium and the system provided by the invention, mathematical modeling and machine learning technologies are adopted, and the technical problem that the power supply stability state of the regional power grid is difficult to accurately judge in real time due to the lack of deep analysis on a power supply stability formation mechanism in the existing power supply stability evaluation method is solved.

Description

Technical Field

[0001] The present invention belongs to the field of regional power supply technology, and in particular, relates to a method, medium and system for determining regional power supply stability. Background Art

[0002] With the continuous expansion of the scale of power systems and the continuous growth of electricity demand, the complexity of power grid operation is also increasing. In the process of power grid operation, power supply stability is one of the key factors to ensure the safe and reliable operation of the power grid. Power supply stability refers to the ability of the power system to maintain key operating parameters such as voltage, frequency, and phase angle within the allowable range under the influence of various interferences and faults. Good power supply stability can not only ensure the quality of power consumption for end users, but also effectively avoid the occurrence of cascading failures, reduce the risk of power grid accidents, and thus improve the overall reliability of the power system.

[0003] The existing power supply stability assessment methods are mainly the following:

[0004] 1) Analysis method based on the dynamic model of the power system. This method establishes corresponding mathematical models, such as power flow equations, frequency stability equations, voltage stability equations, etc., and evaluates power supply stability by solving these equations. However, this method requires a large amount of system parameter information, has high requirements on parameter accuracy, and has a large amount of calculation, which is not conducive to real-time evaluation.

[0005] 2) Evaluation method based on index system. This method evaluates the power supply quality of the power grid by establishing a comprehensive index system, such as voltage deviation index, frequency deviation index, power factor index, etc. However, this method ignores the coupling relationship between power supply, load and environmental factors, and cannot fully reflect the dynamic characteristics of power supply stability.

[0006] 3) Data-driven analysis methods. This type of method uses massive amounts of power system operation data and adopts machine learning, pattern recognition and other technologies to analyze and predict power supply stability. However, this method often lacks in-depth analysis of the formation mechanism of power supply stability and it is difficult to give the physical meaning of power supply stability assessment.

[0007] In summary, the existing power supply stability assessment methods lack in-depth analysis of the formation mechanism of power supply stability, making it difficult to accurately and real-timely judge the power supply stability of the regional power grid. Therefore, there is an urgent need for a new power supply stability assessment method that can ensure the accuracy of the assessment while also being real-time and interpretable. Summary of the invention

[0008] In view of this, the present invention provides a method, medium and system for determining regional power supply stability, which can solve the technical problems that existing power supply stability assessment methods lack in-depth analysis of the power supply stability formation mechanism and are difficult to accurately and in real time determine the power supply stability status of the regional power grid.

[0009] The present invention is achieved in that:

[0010] A first aspect of the present invention provides a method for determining regional power supply stability, comprising the following steps:

[0011] S10. Real-time collection of power supply data of each node in the area, including power supply parameters, load parameters and environmental parameters. The power supply parameters include voltage, current, frequency, power factor and harmonic content; the load parameters include load power, load type, load change rate and peak-to-valley difference; the environmental parameters include temperature, humidity, wind speed, rainfall and lightning activity index;

[0012] S20, cleaning and normalizing the collected power supply data, removing abnormal values, and unifying the data format;

[0013] S30, extracting time-related features from the power supply data to obtain a power supply feature group, a load feature group, and an environmental feature group, wherein the power supply feature group includes voltage fluctuation rate, frequency deviation, power factor change rate, and harmonic distortion rate; the load feature group includes load change rate, load distribution uniformity, peak-to-valley ratio, and load prediction deviation; and the environmental feature group includes temperature change rate, humidity influence coefficient, and comprehensive index of meteorological conditions;

[0014] S40, establishing a stability judgment equation group of regional power supply taking into account all power supply feature groups, load feature groups and environmental feature groups, for calculating a stability evaluation vector, wherein the stability judgment equation group includes a voltage stability equation, a frequency stability equation, a power balance equation, a harmonic influence equation and an environmental influence equation; the stability evaluation vector includes multiple numerical indicators and range indicators, for comprehensively reflecting the stability state of regional power supply;

[0015] S50, performing parameter fitting on the stability judgment equation group using the power supply characteristic group, the load characteristic group and the environment characteristic group according to the time sequence to obtain a fitting equation group;

[0016] S60, using the fitting equation group to calculate the real-time power supply data collected at the current moment, and calculate the stability evaluation vector at the current moment;

[0017] S70: Input the stability evaluation vector at the current moment into a pre-trained stability judgment model to obtain an evaluation result of the regional power supply stability at the current moment.

[0018] The stability judgment equation group is specifically described as follows:

[0019] 1. Voltage stability equation:

[0020]

[0021] Where V is the node voltage amplitude; t is the time; P i is the i-th active load; Q j is the jth reactive load; δ is the voltage phase angle; Y is the admittance matrix; T is the temperature; H is the humidity; W is the wind speed; F v is a nonlinear function describing the relationship between voltage and network parameters; G v It is a function that describes the effect of environmental factors on voltage;

[0022] Among them, F v and G v The specific form is:

[0023]

[0024] G v (T,H,W)=k1(T-T0) 2 +k2(H-H0)+k3W 1.5 ;

[0025] In the formula, G ik and B ik are the elements of the admittance matrix Y, k1, k2, k3 are unknown coefficients, T0 and H0 are the reference temperature and humidity.

[0026] 2. Frequency stability equation:

[0027]

[0028] Where ω is the system angular frequency; ζ is the damping ratio; ω n is the natural frequency; ω0 is the rated angular frequency; M is the equivalent moment of inertia; is the mechanical power input of the i-th generator is the electrical power output of the i-th generator; D is the load-frequency characteristic coefficient; K f (T,H) is the function of the influence of temperature and humidity on frequency;

[0029] Among them, K f The specific form of (T,H) is:

[0030]

[0031] In the formula, a1, a2, T1 and σ T It is a parameter to be determined.

[0032] 3. Power balance equation:

[0033]

[0034] In the formula, is the output power of the i-th power source; is the power consumption of the jth load; P loss is the system power loss; J is the system equivalent moment of inertia; K p (V) is the function of the influence of voltage on power balance;

[0035] Among them, P loss and K p The calculation method of (V) is:

[0036]

[0037] K p (V) = k p (V-V0) 3 +k q (V-V0);

[0038] In the formula, G ij and B ij is the element of the admittance matrix, θ ij is the phase angle difference between nodes i and j, k p and k q is the unknown coefficient, V0 is the rated voltage.

[0039] 4. Harmonic influence equation:

[0040]

[0041] Where, THD is the total harmonic distortion; h is the harmonic propagation speed; D h is the harmonic diffusion coefficient; I h is the harmonic current source; Z h is the harmonic impedance; S h is the harmonic source function; R h is the harmonic attenuation function;

[0042] Among them, S h and R h The specific form is:

[0043]

[0044]

[0045] Where n is the harmonic order, β1, β2, α1 and α2 are unknown parameters.

[0046] 5. Environmental impact equation:

[0047]

[0048] Where, E is the comprehensive index of environmental impact; λ is the environmental impact attenuation coefficient; γ is the influence function of meteorological factors; η is the feedback function of the power system operation status to the environment; R is the rainfall; L is the lightning activity index;

[0049] Among them, the specific forms of γ and η are:

[0050] γ(T,H,W,R,L)=c1T 2 +c2H+c3W 1.5 +c4R+c5 ln(1+L);

[0051]

[0052] In the formula, c1 to c5 and d1 to d3 are unknown coefficients, V n 、f n and S n They are rated voltage, frequency and apparent power respectively.

[0053] Among them, the numerical solution includes node voltage amplitude, system angular frequency, total harmonic distortion rate, and comprehensive environmental impact index;

[0054] Among them, the range solution includes: voltage stability margin, frequency stability margin, power balance margin, harmonic limit margin, and environmental impact safety range.

[0055] Furthermore, the stability judgment model adopts a support vector machine model structure.

[0056] The steps of training the stability judgment model include:

[0057] Construction of training data set: Using the historical power supply data accumulated in steps S10 to S50 and combining it with expert experience, the power supply stability status at each moment (such as stable, warning, unstable, etc.) is marked to form a training data set.

[0058] Model structure design: Select a suitable machine learning model as the stability judgment model, such as support vector machine (SVM), random forest, neural network, etc. According to the characteristics and requirements of the training data, design the input layer (stability evaluation vector), hidden layer structure and output layer (stability state) of the model.

[0059] Model training and validation: Use the constructed training data set to train the model and evaluate its performance using methods such as cross-validation. Adjust the model's hyperparameters, such as learning rate, regularization coefficient, etc., until the model performs optimally on the validation set.

[0060] Model performance metrics: The performance of the model can be evaluated using metrics such as accuracy, F1 score, and area under the receiver operating characteristic (ROC) curve (AUC).

[0061] Furthermore, the evaluation result of the regional power supply stability at the current moment is the stability state of the regional power supply system at the current moment, which includes at least two states: stable and unstable.

[0062] Specifically, step S10 includes: deploying a series of smart meters and sensor devices at key nodes of the regional power grid, such as the output end of the substation, the intermediate node of the main line, and the large load end, to collect the power supply parameters (voltage, current, frequency, power factor, harmonic content, etc.), load parameters (load power, load type, load change rate, peak-to-valley difference, etc.) and environmental parameters (temperature, humidity, wind speed, rainfall, lightning activity index, etc.) of the nodes in real time; these collected data are uploaded to the central server through a wired or wireless communication network, and the data is time-synchronized to ensure that the time correspondence between the data of different nodes is accurate; at the same time, it is also necessary to use statistical analysis, machine learning and other methods to detect and eliminate outliers in the collected data to ensure the reliability of the data and provide a reliable data basis for subsequent analysis and judgment.

[0063] Wherein, the step S20 specifically includes: firstly, converting data of different units and dimensions into unified physical quantities and units, such as converting voltage from volts (V) to per unit machine value (pu); then using methods such as minimum-maximum normalization and Z-score normalization to normalize the data value range of different features to the interval [0,1] to eliminate the influence of dimension and magnitude differences; then, for abnormal data points that deviate too much from the normal value range, using methods such as the 3σ criterion based on statistical distribution and clustering-based outlier detection to eliminate them; finally, for data missing in the acquisition process, using methods such as linear interpolation and spline interpolation to fill in data. Through the above data cleaning and normalization processing, the unit differences, magnitude differences and abnormal value interference in the original data can be eliminated, making the data comparable and usable, creating conditions for subsequent feature extraction and model training.

[0064] Among them, the step S30 specifically includes: extracting power supply characteristics such as voltage fluctuation rate, frequency deviation, power factor change rate, harmonic distortion rate from power supply parameters; extracting load characteristics such as load change rate, load distribution uniformity, peak-to-valley ratio, load prediction deviation from load parameters; extracting environmental characteristics such as temperature change rate, humidity influence coefficient, meteorological condition comprehensive index from environmental parameters. Through these feature extractions, the dynamic stability of the regional power supply system can be quantified from different angles (power supply, load, environment), providing necessary feature vectors for subsequent comprehensive evaluation.

[0065] Wherein, the step S40 specifically includes: establishing a stability judgment equation group consisting of a voltage stability equation, a frequency stability equation, a power balance equation, a harmonic influence equation, and an environmental influence equation. These equations not only describe the nonlinear dynamic relationship between power supply parameters, load parameters, and environmental parameters, but also consider the impact of various factors on power supply stability, providing a solid mathematical model foundation for comprehensively evaluating the stability of regional power supply.

[0066] Wherein, the step S50 specifically includes: firstly, extracting the power supply feature group, the load feature group and the environment feature group from the regional power supply data collected in steps S10 and S20 in chronological order to form a time series data set; then, using these time series data, adopting nonlinear least squares method or robust regression method, etc., to perform parameter fitting on each stability judgment equation in step S40 to obtain the parameter value of each equation; finally, cross-validating the fitting results to evaluate the prediction ability of the model, if the fitting effect is not good, it is necessary to adjust the feature selection or fitting algorithm. Through this step, a group of stability judgment equations that can reflect the dynamic characteristics of regional power supply is obtained.

[0067] Wherein, the step S60 specifically includes: firstly, substituting the real-time power supply data (voltage, current, frequency, power factor, harmonic content, load power, load change rate, temperature, humidity, etc.) collected in step S10 into the stability judgment equation group obtained by fitting in step S50, and calculating the numerical solution of each component equation at the current moment; then, according to the calculation results, constructing the stability evaluation vector at the current moment, including the numerical solutions of node voltage amplitude, system angular frequency, total harmonic distortion rate, comprehensive environmental impact index, and range solutions of voltage stability margin, frequency stability margin, power balance margin, harmonic limit margin, environmental impact safety range, etc. Through this step, an evaluation vector that comprehensively reflects the current regional power supply stability state is obtained.

[0068] Among them, the specific implementation method of step S70 includes: first, using the historical power supply data accumulated in steps S10 to S50, combined with expert experience, marking the power supply stability status (such as stable, warning, unstable, etc.) at each moment to form a training data set; then selecting a suitable machine learning model, such as support vector machine (SVM), random forest, neural network, etc., designing the input layer (stability evaluation vector), hidden layer structure and output layer (stability status) of the model, and using cross-validation and other methods to train and evaluate the model until the model performs optimally on the validation set; finally, the stability evaluation vector at the current moment calculated in step S60 is sent as input data to the pre-trained stability judgment model, and the model will output the stability status of the regional power supply system at the current moment based on the input, and further convert it into a numerical stability index between 0-100.

[0069] Among them, among the power supply parameters collected in step S10, voltage parameters include node voltage amplitude; current parameters include phase current amplitude and phase angle; frequency parameters include system angular frequency; power parameters include active power, reactive power and apparent power; harmonic parameters include the content of each harmonic voltage and current and the total harmonic distortion rate.

[0070] Among them, among the load parameters collected in step S10, the load power parameters include active power and reactive power; the load type parameters include linear load, nonlinear load, inductive load and capacitive load; the load change rate parameters include the first-order derivative and the second-order derivative of the load power; the peak-to-valley difference parameter includes the difference between the maximum value and the minimum value of the load power.

[0071] Among the environmental parameters collected in step S10, the temperature parameter includes the ambient temperature; the humidity parameter includes the ambient humidity; the wind speed parameter includes the wind speed; the rainfall parameter includes the local rainfall; and the lightning activity index parameter includes the local lightning activity frequency.

[0072] Among them, in the stability judgment equation group established in step S40, the voltage stability equation describes the nonlinear dynamic relationship between the node voltage amplitude and time, active load, reactive load, voltage phase angle and admittance matrix, and takes into account the influence of ambient temperature, humidity and wind speed; the frequency stability equation describes the dynamic change of the system angular frequency, including the damping term, the natural frequency term, the generator power difference term and the load-frequency characteristic term, and takes into account the influence of temperature and humidity on the frequency.

[0073] The stability judgment model trained in step S70 adopts the support vector machine (SVM) algorithm, wherein the input layer receives the stability evaluation vector calculated in step S60, the hidden layer adopts the Gaussian radial basis kernel function, and the output layer gives the stability status of the regional power supply system at the current moment (such as stable, warning, unstable). During the model training process, the 5-fold cross-validation method was adopted, and finally the classification accuracy rate of more than 90% was achieved on the validation set.

[0074] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are executed, they are used to execute the above-mentioned method for determining regional power supply stability.

[0075] A third aspect of the present invention provides a regional power supply stability determination system, which includes the above-mentioned computer-readable storage medium.

[0076] Compared with the prior art, the method, medium and system for determining regional power supply stability provided by the present invention have the following beneficial effects:

[0077] 1. A stability judgment equation group that comprehensively considers power supply characteristics, load characteristics and environmental characteristics is established. This equation group not only describes the dynamic relationship between power supply parameters such as voltage, frequency, power balance, and harmonics, but also considers the impact of factors such as ambient temperature, humidity, and wind speed on power supply stability, thereby being able to more comprehensively reflect the stability characteristics of the regional power supply system.

[0078] 2. The parameter fitting method of time series data is used to realize the dynamic identification of the parameters of the stability judgment equation group. Compared with the existing evaluation method based on static model, this method can capture the dynamic characteristics of the power supply system changing over time, thereby improving the accuracy and sensitivity of the evaluation.

[0079] 3. Based on the pre-trained machine learning model, a rapid stability assessment of real-time collected data is achieved. The model takes the comprehensive stability assessment vector calculated in step S60 as input, and can quickly output the stability state of the regional power supply system at the current moment, providing timely and reliable decision support for power grid dispatchers.

[0080] 4. This method not only gives a comprehensive evaluation result of regional power supply stability, but also can analyze and obtain multiple specific indicators such as voltage stability margin, frequency stability margin, power balance margin, harmonic limit margin, environmental impact safety range, etc. These indicators help power grid operation and maintenance personnel to accurately locate the root cause of power supply anomalies and take targeted measures to adjust and improve them.

[0081] In general, the regional power supply stability determination method proposed in the present invention makes full use of the power system operation data, adopts mathematical modeling and machine learning technology, realizes the comprehensive consideration of power supply, load and environmental factors, and has higher accuracy, real-time and interpretability than the existing methods. It solves the technical problem that the existing power supply stability assessment methods lack in-depth analysis of the power supply stability formation mechanism and it is difficult to accurately and in real time judge the power supply stability state of the regional power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 A flow chart of the method provided by the present invention. DETAILED DESCRIPTION

[0083] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0084] like Figure 1 FIG. 1 is a flow chart of a method for determining regional power supply stability provided by the present invention. The method comprises the following steps:

[0085] S10. Real-time collection of power supply data of each node in the area, including power supply parameters, load parameters and environmental parameters. Power supply parameters include voltage, current, frequency, power factor and harmonic content; load parameters include load power, load type, load change rate and peak-to-valley difference; environmental parameters include temperature, humidity, wind speed, rainfall and lightning activity index;

[0086] S20, cleaning and normalizing the collected power supply data, removing abnormal values, and unifying the data format;

[0087] S30, extracting time-related features of the power supply data to obtain a power supply feature group, a load feature group, and an environmental feature group. The power supply feature group includes voltage fluctuation rate, frequency deviation, power factor change rate, and harmonic distortion rate; the load feature group includes load change rate, load distribution uniformity, peak-to-valley ratio, and load prediction deviation; and the environmental feature group includes temperature change rate, humidity influence coefficient, and meteorological condition comprehensive index;

[0088] S40, establishing a stability judgment equation group of regional power supply considering all power supply feature groups, load feature groups and environmental feature groups, for calculating a stability evaluation vector, the stability judgment equation group including a voltage stability equation, a frequency stability equation, a power balance equation, a harmonic influence equation and an environmental influence equation; the stability evaluation vector includes multiple numerical indicators and range indicators, for comprehensively reflecting the stability state of regional power supply;

[0089] S50, performing parameter fitting on the stability judgment equation group using the power supply characteristic group, the load characteristic group, and the environment characteristic group according to the time sequence to obtain a fitting equation group;

[0090] S60, using the fitting equation group to calculate the real-time power supply data collected at the current moment, and calculate the stability evaluation vector at the current moment;

[0091] S70: Input the stability evaluation vector at the current moment into a pre-trained stability judgment model to obtain an evaluation result of the regional power supply stability at the current moment.

[0092] The specific implementation methods of the above steps are described in detail below:

[0093] Step S10: Real-time collection of power supply data of each node in the area

[0094] In order to achieve real-time determination of regional power supply stability, it is first necessary to collect power supply related data of each node in the region. The specific implementation method is as follows:

[0095] 1. Multi-point power supply data collection: deploy a series of smart meters and sensor devices at key nodes of the regional power grid, such as substation output terminals, main line intermediate nodes, and large load terminals. These devices can collect node power supply parameters (voltage, current, frequency, power factor, harmonic content, etc.), load parameters (load power, load type, load change rate, peak-to-valley difference, etc.) and environmental parameters (temperature, humidity, wind speed, rainfall, lightning activity index, etc.) in real time.

[0096] 2. Data communication and storage: The data of each collection node is uploaded to the central server through a wired or wireless communication network. The central server is responsible for receiving, storing and managing these collected data to form a large database of regional power supply.

[0097] 3. Data synchronization and time alignment: Considering that the clocks of each collection node may be biased, it is necessary to synchronize the collected data to ensure that the time correspondence between the data of different nodes is accurate. This can be achieved through the Global Positioning System (GPS) or other time synchronization technologies.

[0098] 4. Outlier detection and elimination: In the process of data collection and transmission, some outliers or noise data will inevitably appear. Therefore, it is necessary to use statistical analysis, machine learning and other methods to detect outliers in the data and eliminate these abnormal data to ensure the reliability of the data.

[0099] In general, the purpose of step S10 is to establish a regional power supply big data platform to collect various power supply related data in real time and efficiently, and provide a reliable data basis for subsequent analysis and judgment.

[0100] Step S20: Power supply data cleaning and normalization

[0101] The original power supply data collected from step S10 often has different units, dimensions and numerical ranges, which makes it difficult to use directly for subsequent analysis. Therefore, data cleaning and normalization are required, as follows:

[0102] 1. Data unit and dimension conversion: First, convert data of different units and dimensions into unified physical quantities and units, such as converting voltage from volts (V) to per unit (pu). This can be achieved through pre-defined conversion formulas.

[0103] 2. Normalization of data value range: In order to eliminate the influence of dimension and magnitude differences, the data value range of different features needs to be normalized to the interval [0,1]. Commonly used normalization methods include minimum-maximum normalization and Z-score normalization.

[0104] 3. Outlier elimination: During the data cleaning process, for abnormal data points that deviate too much from the normal value range, a special outlier elimination algorithm needs to be used to eliminate them. Methods such as the 3σ criterion based on statistical distribution and outlier detection based on clustering can be used.

[0105] 4. Data filling and interpolation: For missing data that occurs during the collection process, appropriate interpolation methods need to be used to fill in the data, such as linear interpolation, spline interpolation, etc.

[0106] Through the above data cleaning and normalization processing, the unit differences, magnitude differences and outlier interference in the original data can be eliminated, making the data comparable and usable, creating conditions for subsequent feature extraction and model training.

[0107] Step S30: Extraction of power supply characteristics, load characteristics and environmental characteristics

[0108] After completing data preprocessing, it is necessary to extract key characteristic indicators that can reflect the regional power supply stability from the original power supply data, including power supply characteristics, load characteristics and environmental characteristics. The specific method is as follows:

[0109] 1. Power supply feature extraction:

[0110] (1) Voltage fluctuation rate: Calculate the standard deviation of the voltage amplitude at each node within a certain time window to reflect the voltage stability.

[0111] (2) Frequency deviation: Calculate the deviation of system frequency from the rated frequency, reflecting the frequency stability.

[0112] (3) Power factor change rate: Calculate the rate of change of power factor over time to characterize the dynamic change of reactive power.

[0113] (4) Harmonic distortion rate: Calculate the total harmonic distortion rate of voltage and current to reflect the degree of harmonic pollution.

[0114] 2. Load feature extraction:

[0115] (1) Load change rate: Calculate the rate of change of load power per unit time to characterize the dynamic characteristics of the load.

[0116] (2) Load distribution uniformity: Calculate the standard deviation of the load power at each node to reflect the uniformity of load distribution in the area.

[0117] (3) Peak-to-valley ratio: Calculates the ratio of the maximum to minimum load power to characterize the load fluctuation.

[0118] (4) Load prediction deviation: Predict the current load based on historical load data, and calculate the deviation between the predicted value and the actual value to reflect the accuracy of the load prediction.

[0119] 3. Environmental feature extraction:

[0120] (1) Temperature change rate: Calculate the rate of change of ambient temperature per unit time to reflect the dynamic change of temperature.

[0121] (2) Humidity influence coefficient: Based on the impact of temperature and humidity on the insulation performance of power equipment, calculate the influence coefficient of humidity on power supply stability.

[0122] (3) Comprehensive index of meteorological conditions: It takes into account environmental factors such as temperature, humidity, wind speed, rainfall, and lightning activity to construct an index that reflects the comprehensive impact of the environment.

[0123] Through the above feature extraction, the dynamic stability of the regional power supply system can be quantified from different angles (power supply, load, environment), providing the necessary feature vectors for subsequent comprehensive evaluation.

[0124] Step S40: Establishing a set of judgment equations for regional power supply stability

[0125] In order to comprehensively evaluate the stability of regional power supply, it is necessary to establish a stability judgment equation group that considers power supply characteristics, load characteristics and environmental characteristics, mainly including the following component equations:

[0126] 1. Voltage stability equation:

[0127]

[0128] This equation describes the node voltage amplitude V with time t and active load P i , reactive load Q j As well as the nonlinear dynamic relationship between the voltage phase angle δ and the admittance matrix Y, the influence of ambient temperature T, humidity H and wind speed W on the voltage is also considered. v and G v is a nonlinear function that describes these relationships.

[0129] 2. Frequency stability equation:

[0130]

[0131] This equation describes the dynamic change of the system angular frequency ω, including the damping term, the natural frequency term, the generator power difference term and the load-frequency characteristic term. f It is a function that describes the effect of temperature and humidity on frequency.

[0132] 3. Power balance equation:

[0133]

[0134] This equation describes the balance between the generated power, load power and system power loss, taking into account the influence of the system equivalent moment of inertia J and the voltage V on the power balance K p .

[0135] 4. Harmonic influence equation:

[0136]

[0137] The equation describes the total harmonic distortion rate THD with time t and space The propagation and evolution process of the harmonics includes the harmonic diffusion effect, the harmonic source term S h and harmonic attenuation term R h .

[0138] 5. Environmental impact equation:

[0139]

[0140] The equation describes the dynamic changes of the comprehensive environmental impact index E, including the attenuation term of environmental impact, the influence function γ of meteorological factors, and the feedback function η of the power system operation status on the environment.

[0141] By establishing such a set of equations that comprehensively considers power supply characteristics, load characteristics and environmental characteristics, the stability state of the regional power supply system can be fully reflected, providing a solid mathematical model foundation for subsequent stability assessment.

[0142] Step S50: fitting the parameters of the equation system based on the time series

[0143] In order to obtain the stability evaluation result at the current moment, it is necessary to perform parameter fitting on the stability judgment equation group established in step S40 according to the historical collected data. The specific method is as follows:

[0144] 1. Time series data preparation:

[0145] From the regional power supply data collected in steps S10 and S20, power supply feature groups, load feature groups and environmental feature groups are extracted in chronological order to form a time series data set.

[0146] 2. Stability equation parameter fitting:

[0147] By using time series data, nonlinear least squares method or robust regression method, etc., the voltage stability equation, frequency stability equation, power balance equation, harmonic influence equation and environmental influence equation in step S40 are fitted with parameters to obtain parameter values ​​of each equation.

[0148] 3. Evaluation of fitting results:

[0149] The fitting results were cross-validated to evaluate the predictive ability of the model. The mean square error (MSE), coefficient of determination (R 2 ) and other indicators to measure the quality of the fitting effect. If the fitting effect is not good, it is necessary to adjust the feature selection or fitting algorithm.

[0150] Through this step, a set of stability judgment equations that can reflect the dynamic characteristics of regional power supply is obtained, in which the parameter values ​​of each equation are obtained based on the fitting of historical acquisition data. This provides a reliable mathematical model for subsequent stability evaluation.

[0151] Step S60: Substituting real-time data into the calculation stability evaluation vector

[0152] With the stability judgment equation group obtained by fitting in step S50, the stability evaluation vector at the current moment can be calculated using the power supply data collected in real time. The specific steps are as follows:

[0153] 1. Real-time data substitution calculation:

[0154] The real-time power supply data (voltage, current, frequency, power factor, harmonic content, load power, load change rate, temperature, humidity, etc.) collected in step S10 is substituted into the stability judgment equation group obtained by fitting in step S50, and the numerical solution of each component equation at the current moment is calculated.

[0155] 2. Stability evaluation vector construction:

[0156] According to the calculation results, the stability evaluation vector at the current moment is constructed, which includes:

[0157] -Numerical solution: node voltage amplitude, system angular frequency, total harmonic distortion rate, comprehensive environmental impact index, etc.

[0158] -Scope solution: voltage stability margin, frequency stability margin, power balance margin, harmonic limit margin, environmental impact safety range, etc.

[0159] Through this step, an evaluation vector that comprehensively reflects the current regional power supply stability status is obtained, providing necessary input for subsequent stability judgment.

[0160] Step S70: Stability determination based on pre-trained model

[0161] With the stability evaluation vector calculated in step S60, the last step is to use the pre-trained stability judgment model to evaluate the power supply stability at the current moment. The specific implementation method is as follows:

[0162] 1. Stability judgment model training:

[0163] (1) Construction of training data set: Using the historical power supply data accumulated in steps S10 to S50 and combining it with expert experience, the power supply stability status at each moment (such as stable, warning, unstable, etc.) is marked to form a training data set.

[0164] (2) Model structure design: Select a suitable machine learning model as the stability judgment model, such as support vector machine (SVM), random forest, neural network, etc. According to the characteristics and requirements of the training data, design the input layer (stability evaluation vector), hidden layer structure and output layer (stability state) of the model.

[0165] (3) Model training and validation: Use the constructed training data set to train the model and evaluate its performance using methods such as cross-validation. Adjust the model's hyperparameters, such as learning rate and regularization coefficient, until the model performs optimally on the validation set.

[0166] (4) Model performance indicators: Indicators such as accuracy, F1 score, and area under the receiver operating characteristic (ROC) curve (AUC) can be used to evaluate the performance of the model.

[0167] 2. Real-time stability judgment:

[0168] (1) Stability evaluation vector input: The stability evaluation vector at the current moment calculated in step S60 is sent as input data to the pre-trained stability judgment model.

[0169] (2) Stability status output: The model will output the stability status of the regional power supply system at the current moment, such as "stable", "warning" or "unstable" based on the input stability assessment vector.

[0170] (3) Stability index calculation: The output stability state can be further converted into a numerical stability index between 0 and 100 to more intuitively reflect the stability of the power supply system. For example, the stable state corresponds to an index of 100, the warning state corresponds to an index of 50, and the unstable state corresponds to an index of 0.

[0171] Through this step, using the pre-trained stability judgment model and combining it with the real-time collected power supply data, the stability status of the regional power supply system at the current moment can be quickly and accurately evaluated, providing a reliable decision-making basis for power grid dispatchers.

[0172] Specifically, the explanation and solution process of the stability judgment equation group are described in detail as follows:

[0173] 1. Voltage stability equation:

[0174]

[0175] Where V is the node voltage amplitude; t is the time; P i is the i-th active load; Q j is the jth reactive load; δ is the voltage phase angle; Y is the admittance matrix; T is the temperature; H is the humidity; W is the wind speed; F v is a nonlinear function describing the relationship between voltage and network parameters; G v It is a function that describes the effect of environmental factors on voltage;

[0176] Among them, F v and G v The specific form is:

[0177]

[0178] G v (T,H,W)=k1(T-T0) 2 +k2(H-H0)+k3W 1.5 ;

[0179] In the formula, G ik and B ik are the elements of the admittance matrix Y, k1, k2, k3 are unknown coefficients, T0 and H0 are the reference temperature and humidity.

[0180] 2. Frequency stability equation:

[0181]

[0182] Where ω is the system angular frequency; ζ is the damping ratio; ω n is the natural frequency; ω0 is the rated angular frequency; M is the equivalent moment of inertia; P mi is the mechanical power input of the i-th generator; P ei is the electrical power output of the i-th generator; D is the load-frequency characteristic coefficient; K f (T,H) is the function of the influence of temperature and humidity on frequency;

[0183] Among them, K f The specific form of (T,H) is:

[0184]

[0185] In the formula, a1, a2, T1 and σ T It is a parameter to be determined.

[0186] 3. Power balance equation:

[0187]

[0188] In the formula, is the output power of the i-th power source; is the power consumption of the jth load; P loss is the system power loss; J is the system equivalent moment of inertia; K p (V) is the function of the influence of voltage on power balance;

[0189] Among them, P loss and K p The calculation method of (V) is:

[0190]

[0191] K p (V) = k p (V-V0) 3 +k q (V-V0);

[0192] In the formula, G ij and B ij is the element of the admittance matrix, θ ij is the phase angle difference between nodes i and j, k p and k q is the unknown coefficient, V0 is the rated voltage.

[0193] 4. Harmonic influence equation:

[0194]

[0195] Where, THD is the total harmonic distortion; h is the harmonic propagation speed; D h is the harmonic diffusion coefficient; I h is the harmonic current source; Z h is the harmonic impedance; S h is the harmonic source function; R h is the harmonic attenuation function;

[0196] Among them, S h and R h The specific form is:

[0197]

[0198] Where n is the harmonic order, β1, β2, α1 and α2 are unknown parameters.

[0199] 5. Environmental impact equation:

[0200]

[0201] Where, E is the comprehensive index of environmental impact; λ is the environmental impact attenuation coefficient; γ is the influence function of meteorological factors; η is the feedback function of the power system operation status to the environment; R is the rainfall; L is the lightning activity index;

[0202] Among them, the specific forms of γ and η are:

[0203] γ(T,H,W,R,L)=c1T 2 +c2H+c3W 1.5 +c4R+c5 ln(1+L);

[0204]

[0205] In the formula, c1 to c5 and d1 to d3 are unknown coefficients, V n 、f n and S n They are rated voltage, frequency and apparent power respectively.

[0206] Here are the numerical and range solutions obtained:

[0207] The numerical solution includes node voltage amplitude, system angular frequency, total harmonic distortion rate, and comprehensive environmental impact index;

[0208] The range solution includes: voltage stability margin, frequency stability margin, power balance margin, harmonic limit margin, and environmental impact safety range.

[0209] Here is a description of the solution process for the system of equations:

[0210] 1. Voltage stability equation

[0211] Solution steps:

[0212] (1) Discretize the equation and use the forward difference method:

[0213]

[0214] (2) After sorting, we get:

[0215]

[0216] (3) Iterative solution:

[0217] a. Initialize V(0), δ(0), Y(0), T(0), H(0), W(0)

[0218] b. For each time step t, calculate F v and G v

[0219] c. Calculate load changes and

[0220] d. Calculate V(t+Δt)

[0221] e. Repeat steps bd until the desired time range is reached

[0222] Solution representation:

[0223] V(t)=f V (t,P i (t),Q j (t),δ(t),Y(t),T(t),H(t),W(t))

[0224] Meaning: V(t) represents the voltage amplitude of the node at time t, which is a function of various input parameters. Variable and parameter description:

[0225] V: node voltage amplitude, obtained by voltage transformer measurement

[0226] t: time, provided by the system clock

[0227] P i : The i-th active load, measured by a power meter

[0228] Q j : The jth reactive load, measured by a power meter

[0229] δ: voltage phase angle, obtained through phasor measurement device

[0230] Y: Admittance matrix, calculated based on system topology and line parameters:

[0231]

[0232] where y ij is the branch admittance between nodes i and j

[0233] T: Temperature, measured by temperature sensor

[0234] H: Humidity, measured by humidity sensor

[0235] W: Wind speed, measured by anemometer

[0236] F v : A nonlinear function describing the relationship between voltage and network parameters, as mentioned above

[0237] G v : A function that describes the effect of environmental factors on voltage, as described above and The sensitivity of voltage to active and reactive power is obtained through power flow calculation:

[0238] where f V It is a function that describes the relationship between voltage and power, P i 0 and is the current operating point 2. Frequency stability equation

[0239] Solution steps:

[0240] (1) Convert the second-order differential equation into a system of first-order differential equations:

[0241] make but:

[0242]

[0243] (2) Solved using the fourth-order Runge-Kutta method

[0244] For each time step t n to n+1 =t n +h:

[0245] k1=hf(t n ,x n )

[0246]

[0247] k4=hf(t n +h,x n +k3)

[0248]

[0249] where f is the function on the right side of the differential equation.

[0250] (3) Iterate the solution until the required time range is reached

[0251] Solution representation:

[0252]

[0253] Meaning: ω(t) represents the system angular frequency at time t, which is a function of various input parameters.

[0254] Variable and parameter description:

[0255] ω: System angular frequency, measured by a frequency meter

[0256] ζ: Damping ratio, obtained through system identification

[0257] ω n : Natural frequency, calculated by system parameters: Where K is the system equivalent stiffness

[0258] ω0: rated angular frequency, system setting value (such as 2π·50rad / s)

[0259] M: Equivalent moment of inertia, calculated based on system generator parameters: Among them, M i is the moment of inertia of the ith generator

[0260] The mechanical power input of the i-th generator is obtained through the turbine control system

[0261] The electrical power output of the i-th generator is measured by a power meter

[0262] D: Load-frequency characteristic coefficient, obtained by fitting historical data:

[0263] K f (T,H): The effect of temperature and humidity on frequency, as mentioned above

[0264] 3. Power balance equation

[0265] Solution steps:

[0266] (1) Discretize the equation and use the central difference method:

[0267]

[0268] (2) After sorting, we get:

[0269]

[0270] (3) Iterative solution:

[0271] a. Initialize ω(0), V(0),

[0272] b. For each time step t, calculate P loss (t) and K p (V(t))

[0273] c. Calculate ω 2 (t+Δt)

[0274] d. Update

[0275] e. Repeat steps bd until the desired time range is reached

[0276] Solution representation:

[0277]

[0278] Meaning: ω(t) represents the system angular frequency at time t, which is a function of generated power, load power and voltage.

[0279] Variable and parameter description:

[0280] The output power of the i-th power source is measured by a power meter

[0281] The power consumption of the jth load is measured by a power meter

[0282] P loss : System power loss, calculated as before

[0283] J: System equivalent moment of inertia, calculated as described above

[0284] K p (V): The function of voltage's influence on power balance, as mentioned above

[0285] V: node voltage amplitude, measured by voltage transformer

[0286] 4. Harmonic influence equation

[0287] Solution steps:

[0288] (1) Discretize the equation and use the finite difference method:

[0289] Spatial discretization (central difference):

[0290]

[0291] Time discretization (forward difference):

[0292]

[0293] (2) Substitute the discretized equation into the original equation:

[0294]

[0295] (3) After sorting, we get:

[0296]

[0297] (4) Iterative solution:

[0298] a. Initialize THD(0,x),v h (x),I h (0,x),Z h (0,x),V(0,x),f(0,x)

[0299] b. For each time step t and spatial point x, calculate S h and R h

[0300] c. Calculate THD(t+Δt,x)

[0301] d. Update boundary conditions

[0302] e. Repeat steps bd until the desired time and space range is reached.

[0303] Solution representation:

[0304] THD(t,x)=f THD (t,x,I h (t,x),Z h (t,x),V(t,x),f(t,x))

[0305] Meaning: THD(t,x) represents the total harmonic distortion at time t and position x and is a function of various input parameters.

[0306] Variable and parameter description:

[0307] THD: Total Harmonic Distortion, measured by a harmonic analyzer

[0308] v h : Harmonic propagation speed, measured experimentally or calculated theoretically

[0309] D h : Harmonic diffusion coefficient, measured experimentally or calculated theoretically

[0310] I h : Harmonic current source, measured by harmonic analyzer

[0311] Z h : Harmonic impedance, measured by network analyzer or calculated theoretically

[0312] S h : Harmonic source function, as mentioned above

[0313] R h : Harmonic attenuation function, as mentioned above

[0314] V: voltage amplitude, measured by voltage transformer

[0315] f: system frequency, measured by a frequency meter

[0316] 5. Environmental Impact Equation

[0317] Solution steps:

[0318] (1) Discretize the equation and use the Euler method:

[0319]

[0320] (2) After sorting, we get:

[0321] E(t+Δt)=E(t)+Δt[-λE(t)+γ(T(t),H(t),W(t),R(t),L(t))+

[0322] η(V(t),f(t),P(t),Q(t))]

[0323] (3) Iterative solution:

[0324] a. Initialize E(0), T(0), H(0), W(0), R(0), L(0), V(0), f(0), P(0), Q(0)

[0325] b. For each time step t, calculate γ and η

[0326] c. Calculate E(t+Δt)

[0327] d. Repeat steps bc until the desired time range is reached

[0328] Solution representation:

[0329] E(t)=f E (t,T(t),H(t),W(t),R(t),L(t),V(t),f(t),P(t),Q(t))

[0330] Meaning: E(t) represents the comprehensive index of environmental impact at time t, which is a function of various input parameters.

[0331] Variable and parameter description:

[0332] E: Comprehensive environmental impact index, calculated through comprehensive calculation of multiple environmental monitoring indicators

[0333] λ: Environmental impact attenuation coefficient, obtained by fitting historical data:

[0334] Where T is the characteristic time

[0335] γ: The influence function of meteorological factors, as mentioned above

[0336] η: The feedback function of the power system operating state to the environment, as mentioned above

[0337] T: temperature, measured by a temperature sensor

[0338] H: Humidity, measured by humidity sensor

[0339] W: wind speed, measured by an anemometer

[0340] R: rainfall, measured by a rain gauge

[0341] L: Lightning activity index, obtained through the lightning location system

[0342] V: System voltage, measured by voltage transformer

[0343] f: system frequency, measured by a frequency meter

[0344] P: System active power, measured by a power meter

[0345] Q: System reactive power, measured by a power meter

[0346] The coefficients in the γ and η functions (c1 to c5 and d1 to d3) can be obtained by the following steps:

[0347] 1. Collect historical data, including environmental impact index E and all input variables

[0348] 2. Use a multivariate nonlinear regression method, such as the Levenberg-Marquardt algorithm, to fit the data:

[0349] Minimize the objective function:

[0350] Where N is the number of data points and f is the combined function of γ and η

[0351] 3. Solve the optimization problem to get the best fitting coefficient

[0352] Now, we have a complete set of equations, their solution steps and variable descriptions. This set of equations can fully describe the stability state of the regional power supply system, taking into account the influence of voltage, frequency, power balance, harmonics and environmental factors. By numerically solving these equations, we can obtain the changes of various state variables of the system over time, thereby evaluating the stability of the system.

[0353] The following is a description of the formulas for the obtained numerical solutions and range solutions:

[0354] Node voltage amplitude: V(t) = f V (t,P i (t),Q j (t),δ(t),Y(t),T(t),H(t),W(t))

[0355] System angular frequency:

[0356] Total harmonic distortion: THD(t,x)=f THD (t,x,I h (t,x),Z h (t,x),V(t,x),f(t,x))

[0357] Comprehensive index of environmental impact: E(t)=f E (t,T(t),H(t),W(t),R(t),L(t),V(t),f(t),P(t),Q(t))

[0358] These numerical solutions have definite values ​​at each time point and can be directly used to calculate specific numerical indicators.

[0359] The following is the obtained range solution and the formula description of the range solution:

[0360] Voltage stability margin: ΔV(t)=|V(t)-V critical |, where V critical is the critical voltage value

[0361] Frequency stability margin: Δω(t)=|ω(t)-ω limit |, where ω limit is the frequency limit

[0362] Power balance margin: ΔP(t)=|P total (t)-P load (t)|-P reserve , where P reserve is the required spare capacity

[0363] Harmonic limit margin: ΔTHD(t) = THD limit -THD(t), where THD limit is the harmonic distortion limit

[0364] Environmental impact safety range: ΔE(t)=E threshold -E(t), where E threshold is the environmental impact threshold

[0365] These range solutions represent the gap between the system state and a certain limit or threshold, and can be used to assess whether the system is within the safe operating range. For example, if ΔV(t)>0, it means that the voltage is within the stable range; if ΔTHD(t)>0, it means that the harmonic distortion rate does not exceed the limit.

[0366] In general, this regional power supply stability determination method based on mathematical modeling and machine learning makes full use of the real-time power supply big data, and establishes a stability evaluation system that comprehensively reflects power supply, load and environmental factors through steps such as feature extraction, parameter fitting and model training. This method can provide real-time and accurate power supply stability diagnosis information for power grid dispatching, help to timely discover and warn power supply anomalies, and improve the safety and reliability of the power system.

[0367] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are executed, they are used to execute the above-mentioned method for determining regional power supply stability.

[0368] A third aspect of the present invention provides a regional power supply stability determination system, which includes the above-mentioned computer-readable storage medium.

[0369] Specifically, the principle of the present invention is: first, through real-time monitoring of key nodes of the regional power grid, a large amount of operating data including power supply parameters, load parameters and environmental parameters are collected; secondly, these raw data are preprocessed, including data cleaning, normalization and other steps, to eliminate dimensional differences and noise interference, and lay the foundation for subsequent feature extraction and modeling; then, from the three perspectives of power supply, load and environment, key characteristic indicators reflecting the stability of regional power supply are extracted, such as voltage fluctuation rate, frequency deviation, load change rate, temperature change rate, etc.; next, based on these characteristic indicators, a stability judgment equation group considering the coupling relationship between power supply, load and environmental factors is established, covering voltage stability, frequency stability, power balance, harmonic influence and environmental influence and other aspects; then, the parameters of these equation groups are fitted using historical operating data to obtain a mathematical model that can reflect the dynamic characteristics of regional power supply; finally, the real-time collected power supply data is substituted into the fitted model, the stability evaluation vector at the current moment is calculated, and the pre-trained machine learning model is used to classify and judge it, and a real-time evaluation result of the stability state of the regional power supply is given.

[0370] This approach is mainly based on the following considerations:

[0371] 1. The operational stability of the power system is a complex nonlinear dynamic process, which is affected by the coupling of multiple factors such as power supply, load and environment. It is difficult for a traditional single model to fully reflect this complex internal mechanism. The stability judgment equation group proposed in this invention, through mathematical modeling, deeply depicts the relationship between these factors, and provides a more comprehensive theoretical basis for power supply stability assessment.

[0372] 2. The operating state of the power system is time-varying, and it is difficult for static models to capture this dynamic characteristic. The present invention adopts a parameter fitting method for time series data, which can realize dynamic identification of the parameters of the equation group, thereby more accurately reflecting the real-time stable state of regional power supply.

[0373] 3. Stability assessment based on real-time monitoring data requires fast and efficient computing capabilities. The present invention uses a pre-trained machine learning model to quickly map the stability assessment vector to the power supply stability state, meeting the needs of real-time grid dispatch. At the same time, the model is also highly interpretable, not only giving a comprehensive assessment result, but also analyzing the specific causes of unstable power supply.

[0374] 4. The data sources required by the present invention are rich, including both the operating data of the power supply system itself and the monitoring data of environmental factors. This comprehensive data support makes the method highly applicable and universal, and can be applied to power systems of different scales and topologies.

[0375] In order to better understand and implement the present invention, an embodiment of a specific application scenario of the present invention is provided below: The reliability of power supply in a certain area of ​​the power grid has always been a key issue of concern to the local power grid dispatching center. The power grid in this area has a wide coverage area, complex load types, and is greatly affected by the surrounding environment (such as climate, geography, etc.). The power supply stability fluctuates from time to time, and problems such as low voltage, abnormal frequency, and severe harmonic pollution often occur, which brings many inconveniences to the operation of the power grid and the electricity consumption of users. In order to comprehensively and accurately evaluate the power supply stability of the regional power grid, the power grid dispatching center decided to adopt the regional power supply stability determination method proposed in the present invention. The specific implementation process is as follows:

[0376] Step S10: Real-time collection of power supply data of each node in the area

[0377] The power grid dispatching center has deployed a number of smart meters and various sensor equipment at the outlet of the main substation of the regional power grid, the middle section of the 10kV trunk line, and the distribution stations of some large industrial enterprises to monitor the power supply parameters, load parameters and environmental parameters in real time.

[0378] Power supply parameter collection, including:

[0379] (1) Voltage parameters: The phase voltage amplitude at the output end of the main transformer and key nodes, the single-phase voltage fluctuation rate is about 2%.

[0380] (2) Current parameters: the phase current amplitude and phase angle of the main transformer and the line, and the three-phase imbalance rate is about 5%.

[0381] (3) Frequency parameters: System frequency, daily variation range is between 49.8 and 50.2 Hz.

[0382] (4) Power parameters: The active power of the main transformer is about 40MW, the reactive power is about 10MVar, and the power factor is between 0.85 and 0.90.

[0383] (5) Harmonic parameters: The total harmonic distortion rate of voltage (THDV) is about 4%, the total harmonic distortion rate of current (THDI) is about 8%, and the main harmonics are the 5th, 7th and 11th harmonics.

[0384] Load parameter collection includes:

[0385] (1) Load power: The active power of large industrial enterprises in the region is about 15 MW, and the reactive power is about 4 MVar.

[0386] (2) Load type: The loads of industrial enterprises are mainly inductive loads, with a small amount of nonlinear loads.

[0387] (3) Load variation rate: The fluctuation range of an enterprise's production load within a day is approximately ±20%.

[0388] (4) Load peak-to-valley difference: The peak load is about 18MW, the valley load is about 12MW, and the peak-to-valley difference is about 1.5.

[0389] Environmental parameter collection includes:

[0390] (1) Temperature: The daily range of ambient temperature in the area is between 15 and 35°C.

[0391] (2) Humidity: The ambient humidity fluctuates between 40% and 80%.

[0392] (3) Wind speed: The average wind speed in the area is about 3m / s.

[0393] (4) Rainfall: The average monthly rainfall is about 100 mm.

[0394] (5) Lightning activity: Lightning activity in this area is relatively frequent, with a monthly average lightning index of approximately 15 times / km2.

[0395] These real-time monitoring data are transmitted to the data center of the power grid dispatching center through wired / wireless communication networks, forming a power supply big data platform for the regional power grid.

[0396] Step S20: Power supply data cleaning and normalization

[0397] The power grid dispatching center preprocesses the collected raw power supply data as follows:

[0398] 1. Data unit and dimension conversion: convert voltage from volts (V) to per unit unit value (pu), convert current from amperes (A) to per unit unit value (pu), convert active power from megawatts (MW) to per unit unit value (pu), convert reactive power from megavars (MVar) to per unit unit value (pu), etc.

[0399] 2. Normalization of data value range: Use the minimum-maximum normalization method to normalize various parameter data to the [0,1] interval, such as normalizing the voltage amplitude from the nominal value range [0.95,1.05]pu to [0,1].

[0400] 3. Outlier elimination: For abnormal data points that deviate from the normal value by more than 3 standard deviations, the 3σ criterion is used to eliminate them. For example, if the voltage amplitude at a certain moment is 1.12pu, which exceeds the normal range of [0.95, 1.05]pu by 3 standard deviations, it is judged as an outlier and eliminated.

[0401] 4. Data interpolation: For missing data at individual time points, linear interpolation is used to fill in the missing data. For example, if the load power data at a certain moment is missing, linear interpolation is performed based on the load data at the previous and next moments.

[0402] Through the above data preprocessing, the original power supply data has been standardized, creating conditions for subsequent feature extraction and model training.

[0403] Step S30: Extraction of power supply characteristics, load characteristics and environmental characteristics

[0404] Based on the preprocessed power supply big data, the power grid dispatching center extracted the following key characteristic indicators:

[0405] Power supply characteristics:

[0406] (1) Voltage fluctuation rate: Calculate the standard deviation of the voltage amplitude of the main transformer and key nodes within a 10-minute time window to reflect voltage stability. The voltage fluctuation rate at the main transformer outlet is about 0.8%, and the voltage fluctuation rate at the intermediate node of the 10kV trunk line is about 1.2%.

[0407] (2) Frequency deviation: Calculate the deviation of system frequency relative to the rated frequency of 50 Hz. The maximum deviation does not exceed ±0.4 Hz, and the frequency stability is good.

[0408] (3) Power factor change rate: Calculate the first-order derivative of the power factor to reflect the dynamic change of reactive power. The power factor change rate is approximately ±0.02 / min.

[0409] (4) Harmonic distortion rate: Calculate the total harmonic distortion rate of voltage and current. THDV is about 4%, THDI is about 8%, and the harmonic pollution level is medium.

[0410] Load characteristics:

[0411] (1) Load change rate: Calculate the change rate of load power of large industrial enterprises in the area within 1 minute, which is approximately ±2% / min.

[0412] (2) Load distribution uniformity: Calculate the standard deviation of the load power at each node to reflect the uniformity of load distribution in the area. The standard deviation of load power is about 2MW.

[0413] (3) Peak-to-valley ratio: The ratio of the maximum load to the minimum load in the calculation area is about 1.5, and the load fluctuation is large.

[0414] (4) Load forecast deviation: The current load is predicted based on historical load data, and the deviation between the predicted value and the measured value is calculated, which is approximately ±5%. The load forecast accuracy is relatively high.

[0415] Environmental characteristics:

[0416] (1) Temperature change rate: Calculate the rate of change of ambient temperature within 1 hour, which is approximately ±2℃ / h.

[0417] (2) Humidity influence coefficient: Based on the influence of temperature and humidity on the insulation performance of power equipment, the influence coefficient of humidity on power supply stability is calculated to be approximately 0.1.

[0418] (3) Comprehensive index of meteorological conditions: This index takes into account factors such as temperature, humidity, wind speed, rainfall, and lightning activity to construct an index that reflects the comprehensive impact of the environment. The value ranges from 0 to 100, and is currently about 75.

[0419] The above characteristic indicators can comprehensively and quantitatively reflect the power supply dynamic characteristics of the regional power grid and provide necessary input for subsequent comprehensive evaluation.

[0420] Step S40: Establishing a set of judgment equations for regional power supply stability

[0421] Based on the power supply characteristics, load characteristics and environmental characteristics extracted in step S30, the power grid dispatching center established the following stability judgment equation group:

[0422] 1. Voltage stability equation:

[0423]

[0424] Where V is the node voltage amplitude; t is the time; P i , Q j are the i-th active load and the j-th reactive load respectively; δ is the voltage phase angle; Y is the admittance matrix; T is the temperature; H is the humidity; and W is the wind speed.

[0425]

[0426] G v (T,H,W)=0.01(T-25) 2 +0.005(H-60)+0.03W 1.5

[0427] 2. Frequency stability equation:

[0428]

[0429] Among them, ω is the system angular frequency; ζ is the damping ratio; ω n is the natural frequency; ω0 is the rated angular frequency; M is the equivalent moment of inertia; are the mechanical power and electrical power of the i-th generator respectively; D is the load-frequency characteristic coefficient.

[0430]

[0431] 3. Power balance equation:

[0432]

[0433] in, is the output power of the i-th power source; is the power consumption of the jth load; P loss is the system power loss; J is the system equivalent moment of inertia.

[0434]

[0435] K p (V) = 0.1 (V-1) 3 +0.05(V-1)

[0436] 4. Harmonic influence equation:

[0437]

[0438] Among them, THD is the total harmonic distortion; v h is the harmonic propagation speed; D h is the harmonic diffusion coefficient; I h is the harmonic current source; Z h is the harmonic impedance.

[0439]

[0440] R h (V,f)=0.05V 0.8 +0.02f 1.2

[0441] 5. Environmental impact equation:

[0442]

[0443] Among them, E is the comprehensive index of environmental impact; γ is the influence function of meteorological factors; η is the feedback function of the power system operation status to the environment; R is the rainfall; L is the lightning activity index.

[0444] γ(T,H,W,R,L)=0.01T 2 +0.005H+0.03W 1.5 +0.02R+0.01ln(1+L)

[0445]

[0446] The above equations comprehensively describe the coupling relationship between power supply parameters, load parameters and environmental parameters, providing a solid mathematical model foundation for subsequent stability evaluation.

[0447] Step S50: fitting the parameters of the equation system based on the time series

[0448] The power grid dispatching center uses the power supply operation data of the past year to perform parameter fitting on the stability judgment equation group established in step S40, as follows:

[0449] 1. Time series data preparation:

[0450] From the power supply big data accumulated in steps S10 and S20, the corresponding power supply characteristics, load characteristics and environmental characteristics data are extracted in chronological order to form a time series data set with a time granularity of 1 minute. These data include the main transformer voltage amplitude, system frequency, active power, reactive power, THDV, THDI, load power, ambient temperature, ambient humidity, etc.

[0451] 2. Equation system parameter fitting:

[0452] Using the above time series data, the nonlinear least square method is used to perform parameter fitting on each equation in step S40. For example, the coefficient G in the voltage stability equation is ik , B ik , the coefficients ζ and ω in the frequency stability equation n , a1, a2, coefficient k in the power balance equation p , k q ,The coefficients β1, β2, α1, α2 in the harmonic influence equation, ,and the coefficients c1 to c5, d1 to d3 in the environmental influence equation, ,are obtained through fitting calculation.

[0453] 3. Evaluation of fitting results:

[0454] The above fitting results were cross-validated by 5-fold to evaluate the prediction ability of the model. The average MSE of the voltage stability equation was 0.0012, the average MSE of the frequency stability equation was 0.0028, the average MSE of the power balance equation was 0.0041, the average MSE of the harmonic influence equation was 0.0032, and the average MSE of the environmental influence equation was 0.0027. 2 They are all above 0.92, indicating that the fitting effect is good and can accurately capture the dynamic characteristics of the power supply system.

[0455] Through this step, the power grid dispatching center obtains a set of stability judgment equations that can reflect the dynamic characteristics of the power supply of the regional power grid.

[0456] Step S60: Substituting real-time data into the calculation stability evaluation vector

[0457] With the stability judgment equation group obtained by fitting in step S50, the power grid dispatching center calculates the power supply stability evaluation vector of the regional power grid in real time, as follows:

[0458] 1. Real-time data substitution calculation:

[0459] The real-time power supply data collected in step S10, such as the main transformer voltage amplitude of 1.02pu, the system frequency of 50.1Hz, the active power of 38MW, the reactive power of 9MVar, the THDV of 4.2%, the THDI of 7.5%, as well as the ambient temperature of 28°C, the ambient humidity of 60%, the wind speed of 3m / s, etc., are substituted into the equations established in step S40 to calculate the voltage stability index, frequency stability index, power balance index, harmonic impact index and environmental impact index at the current moment.

[0460] 2. Stability evaluation vector construction:

[0461] According to the above calculation results, the stability evaluation vector at the current moment is constructed, including:

[0462] -Numerical solution: voltage amplitude 1.02pu, system frequency 50.1Hz, THDV 4.2%, environmental impact index 82

[0463] - Range solution: voltage stability margin 0.97~1.03pu, frequency stability margin 49.8~50.2Hz, power balance margin 95%~105%, harmonic limit margin 3%~7%, environmental impact safety range 70~90

[0464] These indicators comprehensively reflect the current power supply stability of the regional power grid.

[0465] Step S70: Stability determination based on pre-trained model

[0466] Finally, the power grid dispatching center uses the pre-trained stability judgment model to analyze the stability evaluation vector calculated in step S60 and gives the current power supply stability evaluation result of the regional power grid.

[0467] 1. Stability judgment model training:

[0468] (1) Construction of training data set: Using the power supply operation data of the past year and combining it with expert experience, the power supply status at each moment is labeled as "stable", "warning", and "unstable" to form a training data set.

[0469] (2) Model structure design: Support vector machine (SVM) is selected as the stability judgment model. The input layer accepts the stability evaluation vector calculated in step S60, the hidden layer uses the radial basis kernel function, and the output layer gives the classification result of the current power supply status.

[0470] (3) Model training and validation: Using the constructed training data set, the SVM model was trained and its performance was evaluated using the 5-fold cross-validation method. After adjusting the parameters, the classification accuracy of the model on the validation set reached 95%.

[0471] 2. Real-time stability judgment:

[0472] The stability evaluation vector calculated in step S60 is input into the pre-trained SVM model, and the model quickly gives the power supply stability evaluation result of the regional power grid at the current moment:

[0473] -Voltage stability: Good

[0474] -Frequency stability: Good

[0475] -Power balance: Good

[0476] -Harmonic impact: Average

[0477] -Environmental impact: Good

[0478] The comprehensive evaluation result is "stable". At the same time, the model also gives specific indicator values ​​for various aspects, providing a detailed decision-making basis for subsequent power grid dispatching.

[0479] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for determining regional power supply stability, characterized in that: The following steps are involved: S10. Real-time collection of power supply data of each node in the area, including power supply parameters, load parameters and environmental parameters. The power supply parameters include voltage, current, frequency, power factor and harmonic content; the load parameters include load power, load type, load change rate and peak-to-valley difference; the environmental parameters include temperature, humidity, wind speed, rainfall and lightning activity index; S20, cleaning and normalizing the collected power supply data, removing abnormal values, and unifying the data format; S30, extracting time-related features from the power supply data to obtain a power supply feature group, a load feature group, and an environmental feature group, wherein the power supply feature group includes voltage fluctuation rate, frequency deviation, power factor change rate, and harmonic distortion rate; the load feature group includes load change rate, load distribution uniformity, peak-to-valley ratio, and load prediction deviation; and the environmental feature group includes temperature change rate, humidity influence coefficient, and comprehensive index of meteorological conditions; S40, establishing a stability judgment equation group of regional power supply taking into account all power supply feature groups, load feature groups and environmental feature groups, for calculating a stability evaluation vector, wherein the stability judgment equation group includes a voltage stability equation, a frequency stability equation, a power balance equation, a harmonic influence equation and an environmental influence equation; the stability evaluation vector includes multiple numerical indicators and range indicators, for comprehensively reflecting the stability state of regional power supply; S50, performing parameter fitting on the stability judgment equation group using the power supply characteristic group, the load characteristic group and the environment characteristic group according to the time sequence to obtain a fitting equation group; S60, using the fitting equation group to calculate the real-time power supply data collected at the current moment, and calculate the stability evaluation vector at the current moment; S70: Input the stability evaluation vector at the current moment into a pre-trained stability judgment model to obtain an evaluation result of the regional power supply stability at the current moment.

2. A method for determining regional power supply stability according to claim 1, characterized in that: The voltage stability equation is specifically expressed as: Where V is the node voltage amplitude; t is time; P i is the i-th active load; Q j is the jth reactive load; δ is the voltage phase angle; Y is the admittance matrix; T is the temperature; H is the humidity; W is the wind speed; F v is a nonlinear function describing the relationship between voltage and network parameters; G v It is a function that describes the effect of environmental factors on voltage; Among them, F v and G v The specific form is: G v (T,H,W)=k i (T-T o ) z +k2(H-H o )+k3W i5 : In the formula, G ik and B ik are the elements of the admittance matrix Y, k1, k2, k3 are unknown coefficients, T0 and H0 are the reference temperature and humidity.

3. A method for determining regional power supply stability according to claim 2, characterized in that: The frequency stability equation is specifically expressed as: Where ω is the system angular frequency; ζ is the damping ratio; ω n is the natural frequency; ω0 is the rated angular frequency; M is the equivalent moment of inertia; is the mechanical power input of the i-th generator; is the electrical power output of the i-th generator; D is the load-frequency characteristic coefficient; K f (T,H) is the function of the influence of temperature and humidity on frequency; Among them, K f The specific form of (T,H) is: In the formula, a1, a2, T1 and σ T It is a parameter to be determined.

4. A method for determining regional power supply stability according to claim 3, characterized in that: The power balance equation is specifically expressed as: In the formula, is the output power of the i-th power source; is the power consumption of the jth load; P loss is the system power loss; J is the system equivalent moment of inertia; K p (V) is the function of the influence of voltage on power balance; Among them, P loss and K p The calculation method of (V) is: K p (V)=k p (V-V0) 3 +k q (V-V0); In the formula, G ij and B ij is the element of the admittance matrix, θ ij is the phase angle difference between nodes i and j, k p and k q is the unknown coefficient, V0 is the rated voltage.

5. A method for determining regional power supply stability according to claim 4, characterized in that: The harmonic influence equation is specifically expressed as: Where, THD is the total harmonic distortion; h is the harmonic propagation speed; D h is the harmonic diffusion coefficient; I h is the harmonic current source; Z h is the harmonic impedance; S h is the harmonic source function; R h is the harmonic attenuation function; Among them, S h and R h The specific form is: Where n is the harmonic order, β1, β2, α1 and α2 are unknown parameters.

6. A method for determining regional power supply stability according to claim 5, characterized in that: The environmental impact equation is specifically expressed as: Where, E is the comprehensive index of environmental impact; λ is the environmental impact attenuation coefficient; γ is the influence function of meteorological factors; η is the feedback function of the power system operation status to the environment; R is the rainfall; L is the lightning activity index; Among them, the specific forms of γ and η are: γ(T,H,W,R,L)=c1T 2 +c2H+c3W 1.5 +c4R+c5ln(1+L); Where c1 to c5 and d1 to d3 are unknown coefficients, V n 、f n and S n They are rated voltage, frequency and apparent power respectively.

7. A method for determining regional power supply stability according to claim 6, characterized in that: The stability judgment model adopts a support vector machine model structure.

8. A method for determining regional power supply stability according to claim 7, characterized in that: The evaluation result of the regional power supply stability at the current moment is the stability state of the regional power supply system at the current moment, which includes at least two states: stable and unstable.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program instructions, and when the program instructions are executed, they are used to execute the method for determining regional power supply stability according to any one of claims 1 to 8.

10. A regional power supply stability determination system, characterized in that: A computer-readable storage medium comprising the computer-readable storage medium of claim 9.

Images