Analysis and early warning method and system for electric energy quality steady-state index

Abstract

The invention provides an analysis and early warning method and system for an electric energy quality steady-state index, and relates to the technical field of electric energy quality, and the method comprises the steps: collecting electric energy quality data and environment data of a power distribution system, fusing the data, and carrying out the dimension reduction of the collected data through principal component analysis; constructing a power quality prediction model based on the long and short term memory network, and predicting power quality data of the next time step by using the data of the first N time steps; calculating a change rate in combination with the predicted value, setting a fuzzy rule, and evaluating a current early warning state; and when the early warning state is in a relatively high degree, further analyzing the environment data to calculate a current electric energy quality safety index, and if the current electric energy quality safety index is lower than a set threshold value, giving out danger early warning. Through multi-level data analysis and fuzzy logic reasoning, the method can accurately evaluate the steady-state index change and potential risk of the electric energy quality, and provides technical support for safe operation and quick response of a power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of power quality, and in particular to an analysis and early warning method and system for power quality steady-state indicators. Background Art

[0002] Power quality refers to a comprehensive indicator of whether the electric energy provided by the power system to users meets certain quality requirements, covering multiple aspects such as voltage, frequency, waveform and reliability. With the increasing dependence of modern society on electricity, power quality issues have gradually become the core content of power grid operation and power users. Especially in the context of smart grids, renewable energy access and complex power loads, power quality issues have become more complex and diverse. For example, the access of intermittent energy sources such as photovoltaic power generation and wind power generation will cause voltage fluctuations, and the large-scale use of nonlinear loads may cause harmonic pollution. These problems will affect the safe operation of the power system and the normal use of user equipment.

[0003] With the increasing complexity of power systems and the diversification of load demands, power quality issues have gradually become an important factor affecting the safe and stable operation of power systems. Steady-state indicators of power quality (such as voltage deviation, frequency deviation, and harmonic distortion rate) are crucial in the operation of power grids. Their fluctuations may lead to abnormal equipment operation, reduced system efficiency, and even safety hazards. However, due to the existence of various disturbance factors in the power system (such as load fluctuations, distributed energy access, faults, etc.), power quality indicators have significant uncertainty and nonlinear characteristics, which makes it difficult for traditional prediction and early warning methods to accurately capture the dynamic changes of indicators. Especially when the steady-state indicators deviate from normal values ​​or mutate, traditional methods often fail to issue effective early warnings in time, resulting in uncontrollable system risks.

[0004] The prior art has the following deficiencies: Existing power quality monitoring and early warning methods mostly use a single indicator or static evaluation method, which fails to fully integrate the dynamic relationship between power quality data and environmental data, and it is difficult to effectively capture the changing trend under the combined effect of multiple factors. In addition, traditional methods have limited use of time series information, ignoring the regularity of the evolution of power quality indicators over time, resulting in low accuracy of the prediction model. The technical means for early warning are also relatively simple, often based on fixed thresholds to judge the state, lacking flexibility and adaptability, and unable to effectively distinguish between different degrees of potential risks. These shortcomings limit the application effect of the power quality monitoring system, and it is difficult to meet the needs of modern power grids for efficient, accurate and intelligent monitoring.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a method and system for analyzing and warning steady-state indicators of power quality, so as to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions: A method for analyzing and early warning of steady-state indicators of power quality, the specific steps of which include: Step 1: Collect power quality data of the power distribution system and environmental data of the power grid in real time, the power quality data includes voltage, power frequency and total harmonic distortion rate data, and the environmental data includes temperature, humidity and wind speed data; Step 2: Combine the power quality data and environmental data into a comprehensive data set according to the timestamp, and perform principal component analysis on the comprehensive data set to obtain the feature set after dimensionality reduction; Step 3: Take the power quality data and environmental data of the first N time steps in the data feature matrix after dimension reduction as input, and the power quality data of the next time step as the label, and build a power quality prediction model based on the long short-term memory network model; Step 4: Input the power quality data and environmental data of the current time step and the previous N-1 time steps in the data feature matrix after dimension reduction into the power quality prediction model to obtain the power quality data prediction value, calculate the rate of change of the power quality data, and set fuzzy rules to determine the current warning level; Step 5: Analyze the power quality data predicted by the model and the current environmental data to determine whether an early warning is needed.

[0008] Furthermore, the specific logic for collecting power quality data and environmental data is as follows: Voltage sensors, frequency sensors and harmonic analyzers are installed at the outgoing line of the power station, voltage sensors and harmonic analyzers are installed on the busbar of the substation, and voltage sensors, frequency sensors and harmonic analyzers are installed at the low-voltage output end of the distribution transformer. The average of the voltage and total harmonic distortion rate collected at the outgoing line of the power station, the busbar of the substation and the distribution transformer is used as the voltage and total harmonic distortion rate of the distribution system, and the average of the frequency data collected at the outgoing line of the power station and the low-voltage output end of the distribution transformer is used as the operating frequency of the distribution system.

[0009] Furthermore, the specific logic for obtaining the feature set after dimensionality reduction is: The collected voltage, power frequency and total harmonic distortion rate data are standardized. After standardization, the power quality data and environmental data of each time step are a vector. All vectors are integrated into a data matrix, and the covariance matrix of the data matrix is ​​calculated. The formula is: ; in, is the covariance matrix, is the total amount of data, is the original data matrix, is the transpose of the original data matrix; Perform eigenvalue decomposition on the covariance matrix and find the eigenvalues ​​and corresponding eigenvectors: ; Among them, the eigenvalue Indicates The variance of the principal components, the eigenvector Indicates The direction of the principal components, represents the index of the principal component, is the number of principal components; Combine all eigenvectors into an eigenvector matrix: ; in, is the eigenvector matrix, eigenvalue Arranged in descending order, the larger the eigenvalue, the higher the importance of the corresponding eigenvector; The formula for calculating the cumulative variance contribution is: ; in, For the front The cumulative variance contribution of the principal components; Set the threshold to 95% and filter out the minimum value of all cumulative variance contribution rates that is not less than the threshold , That is, the minimum number of principal components is required to project the original data to the front In the principal component space, the data after dimension reduction is calculated and expressed as: ; in, is the data feature matrix after dimension reduction, It is from the previous A matrix of eigenvectors.

[0010] Furthermore, the specific logic for constructing the power quality prediction model is as follows: The data feature matrix after dimensionality reduction is divided into training set, validation set and test set with a division ratio of 7:1.5:1.5. The input sequence length of the long short-term memory network is set to 10 time steps, each time step contains the extracted principal component, and the power quality data and environmental data of the first 10 time steps in the data feature matrix after dimensionality reduction are input. The power quality data of the next time step in the data feature matrix after dimensionality reduction is used as the label. The long short-term memory network model is selected to train the power quality prediction model. After the training is completed, the test set is input into the trained power quality prediction model for testing. The mean square error is selected as the loss function. When the loss of the validation set decreases by less than 0.001 in 10 consecutive rounds, it indicates that the training of the power quality prediction model is completed.

[0011] Furthermore, the specific logic for determining the current warning level is as follows: The input variables are defined to include the deviation degree and the rate of change of the predicted value. The formula for calculating the deviation degree is: ; in, is the degree of deviation, is the next time step to be predicted Power quality data values, For the Normal reference values ​​of various power quality data values, is the total number of types of power quality data, and , An index of the type of power quality data; The formula used to calculate the rate of change is: ; in, is the rate of change, is the current time step Power quality data values, is the last time step Types of power quality data values; Define the value range of the input variable, the value range of the deviation degree is The range of the rate of change is ,in, is the maximum possible value of the deviation, is the maximum possible value of the rate of change, and the output variable is defined as the warning level, with a value range of [0,1]; Define the degree of deviation of the fuzzy set: The fuzzy set of the deviation degree of the predicted value is defined as: negative large, negative small, zero, positive small, positive large, and the membership function of each fuzzy set is defined as: Negative big: ; Negative small: ; zero: ; True small: ; Zhengda: ; in, They are the fuzzy membership functions corresponding to the fuzzy sets negative large, negative small, zero, positive small, and positive large; The fuzzy set of change rate is defined as: fast drop, slow drop, zero, slow rise, fast rise, and the membership function of each fuzzy set is defined as: Quick drop: ; Slow descent: ; zero: ; Slow rise: ; Fast rise: ; in, They are the fuzzy membership functions corresponding to the fuzzy sets fast drop, slow drop, zero, slow rise, and fast rise respectively; The actual value of the input variable is mapped to the fuzzy set, and then the membership of the input variable is calculated. The activation rule is set in the form of IF-THEN rules, which is expressed as follows: Rule 1: If the degree of deviation is large and negative, and the rate of change is fast decreasing, then the warning level is high; Rule 2: If the degree of deviation is large and negative, and the rate of change is zero, then the warning level is medium; Rule 3: If the degree of deviation is small and negative, and the rate of change is fast increasing, then the warning level is medium; Rule 4: If the degree of deviation is zero and the rate of change is zero, then the warning level is low; Rule 5: If the degree of deviation is large and positive, and the rate of change is fast increasing , then the warning level is high; Rule 6: If the deviation is negative and large, and the rate of change is rising slowly, then the warning level is medium; Rule 7: If the deviation is negative and small, and the rate of change is falling quickly, then the warning level is medium; Rule 8: If the deviation is negative and small, and the rate of change is falling slowly, then the warning level is medium; Rule 9: If the deviation is zero, and the rate of change is rising slowly, then the warning level is low; Rule 10: If the deviation is positive and small, and the rate of change is rising quickly, then the warning level is medium; Rule 11: If the deviation is positive and large, and the rate of change is zero, then the warning level is high; The centroid method is then used to transform the fuzzy output value into the actual warning level, and the warning level is divided into normal state , slight abnormality , there is an abnormal state , high risk , if the output value is If the output value is within the range, no warning is required. If it is within the range, take emergency response and proceed to the next step.

[0012] Furthermore, the specific logic for calculating the power quality safety index at the current time step is: When taking emergency response steps, the power quality safety index of the current time step is calculated using the formula: ; in, is the power quality safety index, and its value range is , is the collected temperature, is the collected humidity, is the collected wind speed, is the temperature reference value, is the humidity reference value, are the indexes of deviation degree and rate of change, respectively. are the adjustment parameters for temperature, humidity and wind speed, respectively. are the weight coefficients of their respective items, ,and ; The power quality safety index of the current time step is compared with the preset safety threshold. If the power quality safety index of the current time step is higher than the preset safety threshold, it means that the current power quality is in a normal state. If the power quality safety index of the current time step is lower than the preset safety threshold, it means that the current power quality is in an abnormal state, and a danger warning is issued.

[0013] The present invention further provides an analysis and early warning system for steady-state indicators of power quality, wherein the analysis and early warning system for steady-state indicators of power quality is used to implement the above-mentioned analysis and early warning method for steady-state indicators of power quality, comprising: A data acquisition module is used to collect power quality data of the power distribution system and environmental data of the power grid in real time, wherein the power quality data includes voltage, power frequency and total harmonic distortion rate data, and the environmental data includes temperature, humidity and wind speed data; The data fusion and dimension reduction module is used to merge the power quality data and environmental data into a comprehensive data set according to the timestamp, and perform principal component analysis on the comprehensive data set to obtain the feature set after dimension reduction; A prediction model building module is used to take the power quality data and environmental data of the first N time steps in the data feature matrix after dimension reduction as input, and the power quality data of the next time step as a label, and build a power quality prediction model based on the long short-term memory network model; The dynamic change analysis module is used to input the power quality data and environmental data of the current time step and the previous N-1 time steps in the data feature matrix after dimension reduction into the power quality prediction model to obtain the power quality data prediction value, calculate the change rate of the power quality data, and set fuzzy rules to determine the current warning level; The comprehensive analysis and early warning decision module is used to analyze the power quality data predicted by the model and the current environmental data to determine whether an early warning is needed.

[0014] In the above technical solution, the technical effects and advantages provided by the present invention are: This scheme realizes the accurate quantification of the trend of power quality index changes and the scientific division of dynamic warning levels by constructing a prediction and warning model of steady-state power quality indicators based on fuzzy logic control. This method significantly improves the accuracy of power quality trend prediction by introducing the fusion analysis of environmental data and power quality data, and using the long short-term memory network (LSTM) to mine time series characteristics. Compared with the traditional fixed threshold warning method, this scheme has significant advantages: fuzzy logic control can effectively handle the ambiguity of the deviation degree and change rate between the predicted value and the reference value, and provide more accurate warning results; by establishing a rule base, the complex power quality characteristics are dynamically mapped with the warning level, so that the system can quickly adjust the warning level according to the real-time state changes, improving the adaptability and response speed of the system; this scheme can achieve stable operation in scenarios with large uncertainty, significantly reducing the false alarm rate and missed alarm rate, thereby ensuring the safety and reliability of system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the overall method flow of the present invention; Figure 2 Schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0016] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0017] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0018] Example: See also Figure 1 , the present invention provides a technical solution: A method for analyzing and early warning of steady-state indicators of power quality, the specific steps of which include: Step 1: Collect power quality data of the power distribution system and environmental data of the power grid in real time, the power quality data includes voltage, power frequency and total harmonic distortion rate data, and the environmental data includes temperature, humidity and wind speed data; In this embodiment, the specific logic for collecting power quality data and environmental data is as follows: Install voltage sensors, frequency sensors and harmonic wave analyzers at the outgoing line of the power station, install voltage sensors and harmonic wave analyzers on the busbar of the substation, and install voltage sensors, frequency sensors and harmonic wave analyzers at the low-voltage output end of the distribution transformer. The average of the voltage and total harmonic distortion rate collected at the outgoing line of the power station, the busbar of the substation and the distribution transformer is used as the voltage and total harmonic distortion rate of the distribution system. The average of the frequency data collected at the outgoing line of the power station and the low-voltage output end of the distribution transformer is used as the power frequency of the distribution system. The collected power quality data and environmental data are aligned according to the timestamps so that each record corresponds to the same time point. The Z-score of each data is calculated. When the absolute value of the Z-score of the data When the data is considered as an outlier, it is removed and the missing values ​​are filled using linear interpolation, using the mean of the previous and next timestamps.

[0019] By installing voltage, frequency and harmonic sensors at different key nodes of the power grid (such as the outlet of power stations, the bus terminals of substations and the low-voltage output terminals of distribution transformers), the key indicators of power quality of the distribution system (voltage, power frequency and total harmonic distortion rate) can be fully collected, ensuring the spatial coverage and representativeness of the data. Taking the mean of data from multiple nodes as the overall indicator of the system helps to smooth the impact of local fluctuations on the overall analysis and improve the stability and accuracy of the data. At the same time, environmental data (temperature, humidity and wind speed) can reflect the external conditions under which the power grid operates, providing a basis for the subsequent analysis of the relationship between power quality and environmental factors.

[0020] The collected power quality data and environmental data are aligned according to the timestamp to ensure that each record corresponds to the same time point, which is convenient for the fusion and subsequent analysis of multi-source data. This time synchronization processing avoids the data mismatch problem caused by inconsistent sampling time and provides a reliable basis for steady-state index analysis. The Z-score method is used to identify outliers, which can effectively eliminate the interference of abnormal data on the analysis results and ensure the quality of the data. Specifically, the Z-score judgment is based on statistical principles and can dynamically adapt to the distribution characteristics of the data instead of relying on a fixed threshold, which has strong applicability and robustness. For the missing data that appears after the outliers are eliminated, the linear interpolation method is used to fill it, and the mean of the previous and next timestamps is used for estimation, which can ensure the continuity of the data while avoiding the introduction of significant deviations, thus laying a high-quality data foundation for subsequent time series analysis and prediction. The implementation of this step not only ensures the integrity and accuracy of the steady-state indicators of power quality from the perspective of acquisition scope and data processing, but also improves the reliability and time series continuity of the data through reasonable data cleaning and interpolation methods, which provides a solid foundation for the subsequent power quality early warning model based on time series analysis, making the analysis results more convincing and practical.

[0021] Step 2: Combine the power quality data and environmental data into a comprehensive data set according to the timestamp, and perform principal component analysis on the comprehensive data set to obtain the feature set after dimensionality reduction; In this embodiment, the specific logic for obtaining the feature set after dimensionality reduction is: The collected voltage, power frequency and total harmonic distortion rate data are standardized. After standardization, the power quality data and environmental data of each time step are a vector. All vectors are integrated into a data matrix, and the covariance matrix of the data matrix is ​​calculated. The formula is: ; in, is the covariance matrix, is the total amount of data, is the original data matrix, is the transpose of the original data matrix; Perform eigenvalue decomposition on the covariance matrix and find the eigenvalues ​​and corresponding eigenvectors: ; Among them, the eigenvalue Indicates The variance of the principal components, the eigenvector Indicates The direction of the principal components, represents the index of the principal component, is the number of principal components; Combine all eigenvectors into an eigenvector matrix: ; in, is the eigenvector matrix, eigenvalue Arranged in descending order, the larger the eigenvalue, the higher the importance of the corresponding eigenvector; The formula for calculating the cumulative variance contribution is: ; in, For the front The cumulative variance contribution of the principal components; Set the threshold to 95% and filter out the minimum value of all cumulative variance contribution rates that is not less than the threshold , That is, the minimum number of principal components is required to project the original data to the front In the principal component space, the data after dimension reduction is calculated and expressed as: ; in, is the data feature matrix after dimension reduction, It is from the previous A matrix of eigenvectors.

[0022] The principal component analysis is used to reduce the dimensionality of the comprehensive data set. By calculating the covariance matrix and decomposing the eigenvalues, the original high-dimensional data is converted into a low-dimensional principal component space, which can effectively extract the main features in the data and remove redundant information and noise. The sorting and selection of eigenvalues ​​and eigenvectors are based on the distribution law of the variance contribution rate in the data. By calculating the cumulative variance contribution rate, it is ensured that the selected principal components can retain the information of the original data to the maximum extent, while significantly reducing the data dimension, reducing the computational complexity, and improving the efficiency of subsequent analysis and modeling. Standardization processing ensures the uniformity of data of different dimensions and avoids the problem of analytical distortion caused by differences in data scales; the covariance matrix captures the linear relationship between different indicators and provides comprehensive correlation information for feature extraction; the threshold (such as 95%) is set by the cumulative variance contribution rate, taking into account both information retention and dimensionality reduction effects, ensuring that the extracted principal components are representative and do not lose key information. Finally, the original data is projected into a low-dimensional space composed of the selected principal components. The feature set obtained after dimensionality reduction greatly reduces data redundancy, while fully retaining the key feature change trend, and inputting more concise and efficient feature data for subsequent power quality analysis and abnormal warning models. This dimensionality reduction process not only improves the operating efficiency of the analysis model, but also significantly reduces the risk of overfitting, making the warning results more stable and reliable.

[0023] Step 3: Take the power quality data and environmental data of the first N time steps in the data feature matrix after dimension reduction as input, and the power quality data of the next time step as the label, and build a power quality prediction model based on the long short-term memory network model; In this embodiment, the specific logic for constructing the power quality prediction model is: The feature matrix of the reduced-dimensional data is normalized using Min-Max to normalize all eigenvalues ​​to range, and then divided into training set, validation set and test set, the division ratio is 7:1.5:1.5, the input sequence length of the long short-term memory network is set to 10 time steps, each time step contains the extracted principal component, the power quality data and environmental data of the first 10 time steps in the data feature matrix after dimensionality reduction are input, and the power quality data of the next time step in the data feature matrix after dimensionality reduction is used as the label, and the long short-term memory network model is selected to train the power quality prediction model. After the training is completed, the test set is input into the trained power quality prediction model for testing, and the mean square error is selected as the loss function. When the loss of the validation set decreases by less than 0.001 in 10 consecutive rounds, it indicates that the training of the power quality prediction model is completed.

[0024] With the time series modeling capability of the long short-term memory network (LSTM), the dynamic changes of power quality and environmental data can be effectively captured. LSTM can handle long-term dependency problems in time series data, and is particularly suitable for modeling time series data such as power quality that is greatly affected by historical status. The feature matrix after dimensionality reduction not only retains the main information of power quality data, but also integrates the characteristics of environmental data, which can fully reflect the operation status of the power grid. By inputting the data of the first 10 time steps to predict the power quality data and environmental data of the next time step, short-term prediction of power quality can be achieved, providing a reliable basis for power grid operation status monitoring and early warning. Using Min-Max normalization to normalize the eigenvalues ​​to the range of [0,1] can accelerate network training, avoid gradient vanishing or explosion problems, and improve the convergence speed and prediction accuracy of the model. The data is divided into training set, validation set and test set, and a reasonable ratio (7:1.5:1.5) is set to ensure the adequacy and generalization ability of model training. Using mean square error (MSE) as the loss function can directly reflect the deviation between the predicted value and the true value. Minimizing the loss function helps improve the prediction accuracy of the model. Controlling the training process by the decrease in the validation set loss (training is stopped when the decrease is less than 0.001 in 10 consecutive rounds) can effectively avoid problems such as overfitting and reduce unnecessary computing resource consumption.

[0025] Step 4: Input the power quality data and environmental data of the current time step and the previous N-1 time steps in the data feature matrix after dimension reduction into the power quality prediction model to obtain the power quality data prediction value, calculate the rate of change of the power quality data, and set fuzzy rules to determine the current warning level; In this embodiment, the specific logic for determining the current warning level is: The input variables are defined to include the deviation degree and the rate of change of the predicted value. The formula for calculating the deviation degree is: ; in, is the degree of deviation, is the next time step to be predicted Power quality data values, For the Normal reference values ​​of various power quality data values, is the total number of types of power quality data, and , An index of the type of power quality data; The formula used to calculate the rate of change is: ; in, is the rate of change, is the current time step Power quality data values, is the last time step Types of power quality data values; Define the value range of the input variable, the value range of the deviation degree is The range of the rate of change is ,in, is the maximum possible value of the deviation, is the maximum possible value of the rate of change, and the output variable is defined as the warning level, with a value range of [0,1]; Define the degree of deviation of the fuzzy set: The fuzzy set of the deviation degree of the predicted value is defined as: negative large, negative small, zero, positive small, positive large, and the membership function of each fuzzy set is defined as: Negative big: ; Negative small: ; zero: ; True small: ; Zhengda: ; in, They are the fuzzy membership functions corresponding to the fuzzy sets negative large, negative small, zero, positive small, and positive large; The fuzzy set of change rate is defined as: fast drop, slow drop, zero, slow rise, fast rise, and the membership function of each fuzzy set is defined as: Quick drop: ; Slow descent: ; zero: ; Slow rise: ; Fast rise: ; in, They are the fuzzy membership functions corresponding to the fuzzy sets fast drop, slow drop, zero, slow rise, and fast rise respectively; The actual value of the input variable is mapped to the fuzzy set, and then the membership of the input variable is calculated. The activation rule is set in the form of IF-THEN rules, which is expressed as follows: Rule 1: If the degree of deviation is large and negative, and the rate of change is fast decreasing, then the warning level is high; Rule 2: If the degree of deviation is large and negative, and the rate of change is zero, then the warning level is medium; Rule 3: If the degree of deviation is small and negative, and the rate of change is fast increasing, then the warning level is medium; Rule 4: If the degree of deviation is zero and the rate of change is zero, then the warning level is low; Rule 5: If the degree of deviation is large and positive, and the rate of change is fast increasing, then the warning level is high; Rule 6: If the degree of deviation is large and positive, and the rate of change is fast increasing, then the warning level is high Rule 7: If the deviation degree is negative and small, and the rate of change is slowly rising, then the warning level is medium; Rule 8: If the deviation degree is negative and small, and the rate of change is slowly falling, then the warning level is medium; Rule 9: If the deviation degree is zero and the rate of change is slowly rising, then the warning level is low; Rule 10: If the deviation degree is positive and small, and the rate of change is quickly rising, then the warning level is medium; Rule 11: If the deviation degree is positive and large, and the rate of change is zero, then the warning level is high; Then use the centroid method to convert the fuzzy output value into the actual warning level, and divide the warning level into normal state , slight abnormality , there is an abnormal state , high risk , if the output value is If the output value is within the range, no warning is required. If it is within the range, take emergency response and proceed to the next step.

[0026] By considering the two input variables of the deviation degree and the rate of change of the predicted value at the same time, the abnormal degree of power quality can be evaluated more comprehensively, not only focusing on the deviation of the indicator from the normal reference value at a certain time step, but also considering the potential risks brought by the trend of indicator changes. This multi-dimensional analysis method significantly improves the accuracy of early warning. Fuzzy logic can deal with the problems of uncertainty and nonlinear relationship of input variables very well. For example, the deviation degree and the rate of change may not be linearly related, but fuzzy rules can effectively describe the complex relationship between them and provide clear decision support. By fuzzifying the input variables and setting IF-THEN rules, the state of power quality can be clearly divided into different early warning levels (normal, slightly abnormal, abnormal, high risk). This hierarchical early warning level can help power grid operators quickly understand the system operation status and take corresponding response measures. According to the division of early warning levels, the system can distinguish between high-risk states that require emergency response (output value above 0.6) and normal states that do not require early warning (output value below 0.6). This mechanism can effectively reduce false alarms and missed alarms and improve the response efficiency of the system.

[0027] Step 5: Analyze the power quality data predicted by the model and the current environmental data to determine whether an early warning is needed; In this embodiment, the specific logic for calculating the power quality safety index at the current time step is: When taking emergency response steps, the power quality safety index of the current time step is calculated using the formula: ; in, is the power quality safety index, and its value range is , is the collected temperature, is the collected humidity, is the collected wind speed, is the temperature reference value, is the humidity reference value, are the indexes of deviation degree and rate of change, respectively. are the adjustment parameters for temperature, humidity and wind speed, respectively. are the weight coefficients of their respective items, ,and ; The power quality safety index of the current time step is compared with the preset safety threshold. If the power quality safety index of the current time step is higher than the preset safety threshold, it means that the current power quality is in a normal state. If the power quality safety index of the current time step is lower than the preset safety threshold, it means that the current power quality is in an abnormal state, and a danger warning is issued.

[0028] The degree of deviation indicates the deviation between the predicted power quality data and the normal reference value. The absolute value is used to ensure non-negative; the absolute value is used and the exponent is introduced. , emphasizing that the impact of the degree of deviation on the safety index is nonlinear. The greater the degree of deviation, the faster the safety index decreases. , the influence of the degree of deviation on the safety index is amplified, which is applicable to the danger when the intensity deviation is large. When , the influence of the deviation degree on the safety index is weakened, which is suitable for tolerance of small deviations. Here it can be set to , moderately amplifying the impact of the degree of deviation. The rate of change also introduces an exponential , making the influence of the change rate present nonlinear characteristics, allowing more flexible adjustment of its influence, Resti, change rate Reflects the changing trend of power quality data, rapid changes (large ) may bring higher risks, index When the effect is amplified, the slow change (small ) or changes tend to be stable, the index To reduce its influence, you can set it to , that is, the effect on the rate of change is slightly higher than the linear relationship, but not too sensitive. Through the exponential function, the effect of temperature is introduced in a nonlinear form. When the temperature exceeds the reference value When the safety index drops rapidly, the temperature The impact on power grid equipment is proportional to the deviation between the temperature and the reference value. The greater the deviation, the greater the impact and the lower the safety index. The form of can reflect the nonlinear growth characteristics of this temperature effect. Usually the reference value of temperature can be set to 25℃. In cold areas, the reference value can be appropriately lowered, such as 15℃. In high temperature areas, the reference value can be appropriately increased, and the comparative reference value is 30℃. The adjustment parameter should be positively correlated with the severity of the influencing factor. Here, the range of the temperature adjustment parameter is When the temperature reference value is set to 25℃, the temperature adjustment parameter can be set to 6; when the temperature reference value is set to 15℃, the temperature adjustment parameter can be set to 5; when the temperature reference value is set to 30℃, the temperature adjustment parameter can be set to 7. Compare the collected humidity data with the normal humidity reference value. By comparison, the nonlinear effect of humidity is also introduced. The greater the deviation between humidity and the reference value, the lower the safety index. Usually the reference value of humidity can be set to 50%. In humid areas (such as southern coastal areas or rainy seasons), it can be set to 60%, and in dry areas (such as northern winter or desert areas), it can be set to 40%. The range of humidity adjustment parameters here is When the humidity reference value is set to 50%, the humidity adjustment parameter can be set to 12; when the humidity reference value is set to 40%, the humidity adjustment parameter can be set to 10; when the humidity reference value is set to 60%, the humidity adjustment parameter can be set to 15. The exponential function introduces the nonlinear effect of wind speed. An increase in wind speed indicates a decrease in the safety index. The impact may be small in the low wind speed range, but when the wind speed exceeds a certain critical value (such as strong wind weather), the impact on the power grid increases rapidly. The wind speed adjustment parameter can be set to 5. The adjustment parameter is used to control the nonlinearity of the impact amplitude, reflecting the sensitivity of different physical factors to power quality. The adjustment parameter can coordinate the comprehensive impact of temperature, humidity, and wind speed on power quality, avoiding the excessive dominance of a single factor on the formula result, thereby more accurately quantifying the comprehensive impact of environmental factors on the power quality safety index.

[0029] The degree of deviation directly reflects the gap between the power quality data and the normal reference value. The greater the deviation, the more unstable the power quality, so it has the greatest impact on the safety index. The stability of power quality is the basis for ensuring the normal operation of the power system. The rate of change describes the volatility of power quality over time. A high rate of change usually means system instability, which may lead to equipment failure or power service interruption. Although the rate of change is also important, in some cases, the impact of the degree of deviation may be more significant, so the weight is slightly lower than the degree of deviation. Temperature affects the performance and safety of power equipment. Excessive temperature may cause equipment overheating, failure or reduced efficiency. Therefore, the impact of temperature on power quality is significant, but it may be more affected by environmental conditions than the degree of deviation and rate of change. Therefore, the weight is lower than the degree of deviation and rate of change. High humidity may cause equipment short circuit or corrosion, affecting power quality. However, compared with the degree of deviation, rate of change and temperature, humidity is usually a minor factor, especially in modern power equipment, where the impact of humidity may be mitigated by design and materials, so the weight is lower. Wind speed has a relatively small impact on power quality, especially in most power facilities. Although excessive wind speed may affect the stability of equipment (such as wind turbines), it is usually considered a minor factor in the overall power quality assessment and therefore has the least weight. This constraint ensures that the power quality safety index is a normalized indicator, ensuring that the total contribution of the weighted combination of these factors is 100%.

[0030] See also Figure 2 The present invention further provides an analysis and early warning system for steady-state indicators of power quality, wherein the analysis and early warning system for steady-state indicators of power quality is used to implement the above-mentioned analysis and early warning method for steady-state indicators of power quality, including: A data acquisition module is used to collect power quality data of the power distribution system and environmental data of the power grid in real time, wherein the power quality data includes voltage, power frequency and total harmonic distortion rate data, and the environmental data includes temperature, humidity and wind speed data; The data fusion and dimension reduction module is used to merge the power quality data and environmental data into a comprehensive data set according to the timestamp, and perform principal component analysis on the comprehensive data set to obtain the feature set after dimension reduction; A prediction model building module is used to take the power quality data and environmental data of the first N time steps in the data feature matrix after dimension reduction as input, and the power quality data of the next time step as a label, and build a power quality prediction model based on the long short-term memory network model; The dynamic change analysis module is used to input the power quality data and environmental data of the current time step and the previous N-1 time steps in the data feature matrix after dimension reduction into the power quality prediction model to obtain the power quality data prediction value, calculate the change rate of the power quality data, and set fuzzy rules to determine the current warning level; The comprehensive analysis and early warning decision module is used to analyze the power quality data predicted by the model and the current environmental data to determine whether an early warning is needed.

[0031] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0032] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0033] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0034] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for analyzing and warning steady-state indicators of power quality, characterized in that: The specific steps include: Step 1: Collect power quality data of the power distribution system and environmental data of the power grid in real time, the power quality data includes voltage, power frequency and total harmonic distortion rate data, and the environmental data includes temperature, humidity and wind speed data; Step 2: Combine the power quality data and environmental data into a comprehensive data set according to the timestamp, and perform principal component analysis on the comprehensive data set to obtain the feature set after dimensionality reduction; Step 3: Take the power quality data and environmental data of the first N time steps in the data feature matrix after dimension reduction as input, and the power quality data of the next time step as the label, and build a power quality prediction model based on the long short-term memory network model; Step 4: Input the power quality data and environmental data of the current time step and the previous N-1 time steps in the data feature matrix after dimension reduction into the power quality prediction model to obtain the power quality data prediction value, calculate the rate of change of the power quality data, and set fuzzy rules to determine the current warning level; Step 5: Analyze the power quality data predicted by the model and the current environmental data to determine whether an early warning is needed.

2. The method for analyzing and warning of steady-state indicators of power quality according to claim 1, characterized in that: The specific logic for collecting power quality data and environmental data is as follows: Voltage sensors, frequency sensors and harmonic analyzers are installed at the outgoing line of the power station, voltage sensors and harmonic analyzers are installed on the busbar of the substation, and voltage sensors, frequency sensors and harmonic analyzers are installed at the low-voltage output end of the distribution transformer. The average of the voltage and total harmonic distortion rate collected at the outgoing line of the power station, the busbar of the substation and the distribution transformer is used as the voltage and total harmonic distortion rate of the distribution system, and the average of the frequency data collected at the outgoing line of the power station and the low-voltage output end of the distribution transformer is used as the operating frequency of the distribution system.

3. The method for analyzing and warning of steady-state indicators of power quality according to claim 2, characterized in that: The specific logic for obtaining the feature set after dimensionality reduction is: The collected voltage, power frequency and total harmonic distortion rate data are standardized. After standardization, the power quality data and environmental data of each time step are a vector. All vectors are integrated into a data matrix, and the covariance matrix of the data matrix is ​​calculated. The formula is: ； in, is the covariance matrix, is the total amount of data, is the original data matrix, is the transpose of the original data matrix; Perform eigenvalue decomposition on the covariance matrix and find the eigenvalues ​​and corresponding eigenvectors: ； Among them, the eigenvalue Indicates The variance of the principal components, the eigenvector Indicates The direction of the principal components, represents the index of the principal component, is the number of principal components; Combine all eigenvectors into an eigenvector matrix: ； in, is the eigenvector matrix, eigenvalue Arranged in descending order, the larger the eigenvalue, the higher the importance of the corresponding eigenvector; The formula for calculating the cumulative variance contribution is: ； in, For the front The cumulative variance contribution of the principal components; Set the threshold to 95% and filter out the minimum value of all cumulative variance contribution rates that is not less than the threshold , That is, the minimum number of principal components is required to project the original data to the front In the principal component space, the data after dimension reduction is calculated and expressed as: ； in, is the data feature matrix after dimension reduction, It is from the previous A matrix of eigenvectors.

4. The method for analyzing and warning of steady-state indicators of power quality according to claim 3 is characterized in that: The specific logic for building the power quality prediction model is: The data feature matrix after dimensionality reduction is divided into training set, validation set and test set with a division ratio of 7:1.5:1.

5. The input sequence length of the long short-term memory network is set to 10 time steps, each time step contains the extracted principal component, and the power quality data and environmental data of the first 10 time steps in the data feature matrix after dimensionality reduction are input. The power quality data of the next time step in the data feature matrix after dimensionality reduction is used as the label. The long short-term memory network model is selected to train the power quality prediction model. After the training is completed, the test set is input into the trained power quality prediction model for testing. The mean square error is selected as the loss function. When the loss of the validation set decreases by less than 0.001 in 10 consecutive rounds, it indicates that the training of the power quality prediction model is completed.

5. The method for analyzing and early warning of steady-state indicators of power quality according to claim 4, characterized in that: The specific logic for determining the current warning level is: The input variables are defined to include the deviation degree and the rate of change of the predicted value. The formula for calculating the deviation degree is: ； in, is the degree of deviation, is the predicted next time step Power quality data values, For the Normal reference values ​​of various power quality data values, is the total number of types of power quality data, and , An index of the type of power quality data; The formula used to calculate the rate of change is: ； in, is the rate of change, is the current time step Power quality data values, is the last time step Types of power quality data values; Define the value range of the input variable, the value range of the deviation degree is The range of the rate of change is ,in, is the maximum possible value of the deviation, is the maximum possible value of the rate of change, and the output variable is defined as the warning level, with a value range of [0,1]; Define the degree of deviation of the fuzzy set: The fuzzy set of the deviation degree of the predicted value is defined as: negative large, negative small, zero, positive small, positive large, and the membership function of each fuzzy set is defined as: Negative big: ； Negative small: ； zero: ； True small: ； Zhengda: ； in, They are the fuzzy membership functions corresponding to the fuzzy sets negative large, negative small, zero, positive small, and positive large; The fuzzy set of change rate is defined as: fast drop, slow drop, zero, slow rise, fast rise, and the membership function of each fuzzy set is defined as: Quick drop: ； Slow descent: ； zero: ； Slow rise: ； Fast rise: ； in, They are the fuzzy membership functions corresponding to the fuzzy sets fast drop, slow drop, zero, slow rise, and fast rise respectively; The actual value of the input variable is mapped to the fuzzy set, and then the membership of the input variable is calculated. The activation rule is set in the form of IF-THEN rules, which is expressed as follows: Rule 1: If the degree of deviation is large and negative, and the rate of change is fast decreasing, then the warning level is high; Rule 2: If the degree of deviation is large and negative, and the rate of change is zero, then the warning level is medium; Rule 3: If the degree of deviation is small and negative, and the rate of change is fast increasing, then the warning level is medium; Rule 4: If the degree of deviation is zero and the rate of change is zero, then the warning level is low; Rule 5: If the degree of deviation is large and positive, and the rate of change is fast increasing , then the warning level is high; Rule 6: If the deviation is negative and large, and the rate of change is rising slowly, then the warning level is medium; Rule 7: If the deviation is negative and small, and the rate of change is falling quickly, then the warning level is medium; Rule 8: If the deviation is negative and small, and the rate of change is falling slowly, then the warning level is medium; Rule 9: If the deviation is zero, and the rate of change is rising slowly, then the warning level is low; Rule 10: If the deviation is positive and small, and the rate of change is rising quickly, then the warning level is medium; Rule 11: If the deviation is positive and large, and the rate of change is zero, then the warning level is high; The centroid method is then used to transform the fuzzy output value into the actual warning level, and the warning level is divided into normal state , slight abnormality , there is an abnormal state , high risk , if the output value is If the output value is within the range, no warning is required. If it is within the range, take emergency response and proceed to the next step.

6. The method for analyzing and warning of steady-state indicators of power quality according to claim 5, characterized in that: The specific logic for calculating the power quality safety index at the current time step is: When taking emergency response steps, the power quality safety index of the current time step is calculated using the formula: ； in, is the power quality safety index, with a value range of , is the collected temperature, is the collected humidity, is the collected wind speed, is the temperature reference value, is the humidity reference value, are the indexes of deviation degree and rate of change, respectively. are the adjustment parameters for temperature, humidity and wind speed, respectively. are the weight coefficients of their respective items, ,and ; The power quality safety index of the current time step is compared with the preset safety threshold. If the power quality safety index of the current time step is higher than the preset safety threshold, it means that the current power quality is in a normal state. If the power quality safety index of the current time step is lower than the preset safety threshold, it means that the current power quality is in an abnormal state, and a danger warning is issued.

7. An analysis and early warning system for steady-state indicators of power quality, characterized in that: The power quality steady-state indicator analysis and early warning system is used to implement the power quality steady-state indicator analysis and early warning method according to any one of claims 1 to 6, comprising: A data acquisition module is used to collect power quality data of the power distribution system and environmental data of the power grid in real time, wherein the power quality data includes voltage, power frequency and total harmonic distortion rate data, and the environmental data includes temperature, humidity and wind speed data; The data fusion and dimension reduction module is used to merge the power quality data and environmental data into a comprehensive data set according to the timestamp, and perform principal component analysis on the comprehensive data set to obtain the feature set after dimension reduction; A prediction model building module is used to take the power quality data and environmental data of the first N time steps in the data feature matrix after dimension reduction as input, and the power quality data of the next time step as a label, and build a power quality prediction model based on the long short-term memory network model; The dynamic change analysis module is used to input the power quality data and environmental data of the current time step and the previous N-1 time steps in the data feature matrix after dimension reduction into the power quality prediction model to obtain the power quality data prediction value, calculate the change rate of the power quality data, and set fuzzy rules to determine the current warning level; The comprehensive analysis and early warning decision module is used to analyze the power quality data predicted by the model and the current environmental data to determine whether an early warning is needed.

Images