Power system automation fault alarm system based on power output calculation prediction

Abstract

The present invention discloses an automatic fault alarm system for an electric power system based on power output calculation prediction, including a data acquisition module, a data storage module, a data analysis module and an automatic interaction module. The data analysis module obtains each group of voltage operation data, and constructs a prediction reference model to calculate the voltage fault prediction reference value of each transmission line. The automatic interaction module obtains each voltage fault prediction reference value, constructs a system interaction model, outputs the overall prediction interaction value, compares it with the preset basic warning threshold, and performs fault judgment alarm. Big data technology and model intelligent calculation are applied to power system fault alarm, and potential power system faults are automatically identified and alarmed by real-time monitoring and prediction of the operating status of the power system.

Description

Technical Field

[0001] The present invention relates to the technical field of power monitoring and alarming, and in particular to an automatic fault alarm system for a power system based on power output calculation prediction. Background Art

[0002] The power system is the fundamental infrastructure of modern society, and its stable operation is crucial to ensuring its normal functioning. However, due to the complexity of the power system, power output failures are inevitable. Traditional fault detection methods rely primarily on manual inspections and scheduled checks, which are not only inefficient but also prone to missing early warnings of faults.

[0003] Therefore, a new power system automatic fault alarm system is needed that can monitor and predict potential faults in real time. Summary of the Invention

[0004] The purpose of this section is to summarize some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of this application to avoid obscuring the purpose of this section, the abstract and the title of the invention, and such simplifications or omissions should not be used to limit the scope of the present invention.

[0005] In view of the problems existing in the above-mentioned existing power output fault detection method, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is to solve the problems that the existing power output fault detection method is prone to miss the early warning of the fault on the one hand, and is inefficient on the other hand.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: an automatic fault alarm system for an electric power system based on power output calculation prediction, comprising the following components: a data acquisition module, configured at the power output end of each transmission line, for real-time acquisition of voltage operation data of the current power output end of the transmission line, and uploading the corresponding voltage operation data to the data storage module; a data storage module, wirelessly connected to the data acquisition module, for acquiring and storing each group of voltage operation data; a data analysis module, embedded in the data storage module, for acquiring each group of voltage operation data, and for constructing a prediction reference model to calculate a voltage fault prediction reference value for each transmission line; an automatic interaction module, wirelessly connected to the data analysis module, for acquiring each of the voltage fault prediction reference values, for constructing a system interaction model, for inputting each of the voltage fault prediction reference values, for outputting an overall prediction interaction value, for comparing A fault judgment alarm is issued for the preset basic warning threshold; wherein, the data acquisition module is also embedded with a data preprocessing unit, which preprocesses each group of voltage operation data after obtaining the data, and uploads each group of preprocessed voltage operation data to the data storage module; data preprocessing specifically includes: data cleaning, data conversion and data dimension reduction; wherein, calculating the voltage fault prediction reference value of each transmission line specifically includes the following steps: S1: real-time acquisition of voltage operation data α at the power output end of the current transmission line; S2: built-in MCU microprocessor synchronously outputs time-voltage curve graph; S3: obtains pressure difference change information in the real-time output time-voltage curve graph; S4: constructs the prediction reference model, inputs the pressure difference change information, and outputs the corresponding voltage fault prediction reference value; wherein, the constructed prediction reference model is specifically:

[0008]

[0009] Wherein, δ is the voltage fault prediction reference value of the corresponding transmission line, U1, U2, ..., U n is the real-time voltage value corresponding to each selected point on the time-voltage curve, n is the number of selected points, D1, D2, ..., D n is the derivative value corresponding to each selected point on the time-voltage curve, U Q1 is the real-time voltage value corresponding to the maximum derivative point on the time-voltage curve, U Q2 is the real-time voltage value corresponding to the minimum point of the derivative on the time-voltage curve, and dx is the integral operation; wherein the constructed system interaction model is specifically as follows:

[0010]

[0011] Where η is the overall predicted interaction value, δ1, δ2, ..., δ mis the voltage fault prediction reference value obtained for each transmission line, and m is the number of transmission lines in the power system.

[0012] As a preferred solution of the power system automation fault alarm system based on power output calculation prediction described in the present invention, the preset basic warning threshold is set to 2.360 or 2.361 or 2.362.

[0013] As a preferred solution of the power system automation fault alarm system based on power output calculation prediction described in the present invention, wherein: the overall predicted interaction value is output, compared with the preset basic warning threshold, and a fault judgment alarm is performed thereon: when the overall predicted interaction value is not lower than the preset basic warning threshold, a fault alarm is performed.

[0014] This invention provides an automated power system fault alarm system based on power output computational prediction. It applies big data technology and model-based intelligent computing to power system fault alarms. By monitoring and predicting the power system's operating status in real time, it automatically identifies and issues warnings about potential power system faults. Compared to existing technologies, this invention offers the following innovative features:

[0015] 1. Real-time monitoring: Compared with traditional periodic inspections and manual inspections, the system of the present invention can collect and analyze the operating data of the power system in real time, greatly improving the real-time nature and accuracy of monitoring.

[0016] 2. Predictive analysis: Through big data technology and model intelligent calculation, the system of the present invention can predict the future output of the power system, thereby detecting potential faults in advance.

[0017] 3. Automatic early warning: Through the fault early warning module, the system of the present invention can automatically identify and warn of potential power system faults without manual intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0019] Figure 1 This is a system module diagram of the power system automation fault alarm system based on power output calculation prediction provided by the present invention.

[0020] Figure 2 Flowchart of the method for calculating the voltage fault prediction reference value of each transmission line provided by the present invention. DETAILED DESCRIPTION

[0021] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0022] Due to the complexity of power systems, power output failures are inevitable. Traditional fault detection methods rely primarily on manual inspections, including specialized training, infrared measurements, device alarms, continuous power outage detection, scheduled inspections, and electrical tests. These methods are not only inefficient but also prone to missing early warnings of faults.

[0023] Therefore, please refer to Figure 1 The present invention provides an automatic fault alarm system for a power system based on power output calculation prediction, comprising the following components:

[0024] The data acquisition module 100 is configured at the power output end of each transmission line, and is used to collect the voltage operation data of the power output end of the current transmission line in real time, and upload the corresponding voltage operation data to the data storage module 200;

[0025] It should be noted that the data acquisition module 100 can generally be configured at the power output end of the transmission line, and a conventional and durable voltage monitoring sensor can be selected. After monitoring the voltage, the data is uploaded to the data storage module 200 through the built-in microprocessor. The data acquisition module 100 and the data storage module 200 involved in the present invention are both direct applications of existing conventional components, and will not be elaborated here.

[0026] Furthermore, the data acquisition module 100 is also embedded with a data preprocessing unit, which preprocesses each set of voltage operation data after acquiring the data, and uploads the preprocessed voltage operation data to the data storage module 200;

[0027] Data preprocessing specifically includes: data cleaning, data conversion and data dimensionality reduction.

[0028] Specifically:

[0029] 1. Data cleaning is an important step in data preprocessing. It involves a series of cleaning operations on the data to eliminate noise, missing values, outliers, and redundant data, thereby providing high-quality data for data analysis or machine learning model training. Specifically, data cleaning techniques include the following aspects:

[0030] Handling missing values: Missing values ​​may be caused by errors in the data collection process, natural loss of data, etc. There are many ways to handle missing values, such as filling the mean or median, using interpolation, or simply deleting records with missing values.

[0031] Detecting and handling outliers: Outliers can be caused by errors in the data collection process, special circumstances, etc. After detecting outliers, it may be necessary to identify and possibly fix or exclude the unreasonable values. This can be done by removing outliers or using smoothing techniques.

[0032] Deduplication: Duplicate records can lead to data inconsistency or redundancy, so duplicate records need to be removed to avoid data clutter. This can be achieved by using unique identifiers or using deduplication algorithms.

[0033] In addition to these, other common cleaning techniques include normalization, standardization, and transcoding. Normalization adjusts the data range to a certain level, typically to facilitate subsequent data analysis and machine learning modeling. Standardization converts data to a uniform scale, typically to enhance comparability between different features. Transcoding converts unstructured data into structured values ​​or labels for easier processing and analysis.

[0034] In short, the goal of data cleaning is to ensure the quality and consistency of data in order to provide reliable and accurate data support for subsequent data analysis and machine learning modeling.

[0035] 2. Data conversion is a key step in data preprocessing, which involves converting raw data into a format suitable for analysis or modeling. The following are some specific techniques for data type conversion and character encoding conversion during data extraction:

[0036] ①Data type conversion:

[0037] Numeric conversion: Converts a string value to a numeric type (such as an integer or floating-point number) for mathematical operations. For example, convert the string "123" to the integer 123 or the floating-point number 123.0.

[0038] Date and time conversion: Convert date and time from string format to date and time objects for date calculations and time series analysis. This usually involves parsing date and time formats (such as "YYYY-MM-DD HH:MM:SS") and converting them to a standardized date and time format.

[0039] Categorical conversion: Convert categorical data to numerical types for numerical calculations. This can be achieved through label encoding or one-hot encoding.

[0040] ②Character encoding conversion:

[0041] Character set conversion: Convert data from one character set (such as ASCII) to another character set (such as UTF-8) to ensure data consistency across different systems or applications.

[0042] Language encoding conversion: If the data involves multiple languages, you may need to convert the string from one language encoding (such as ISO-8859-1) to another language encoding (such as UTF-8) to support multi-language processing.

[0043] ③Data formatting:

[0044] Remove leading and trailing whitespace: Delete whitespace characters on both sides of a string, such as spaces, tabs, and newlines.

[0045] String length normalization: Trim or pad strings to a specific length to unify the dimensions of different data items.

[0046] Data segmentation: For text data, sentences or paragraphs are divided into words, phrases, or sentence fragments for lexical analysis and natural language processing.

[0047] ④Data normalization / standardization:

[0048] Normalization: Scaling numerical data to a fixed range, such as [0, 1] or [-1, 1]. This can be achieved by scaling the maximum and minimum values ​​of the data to the target range.

[0049] Standardization: Transform the data into a form with zero mean and unit variance by subtracting the mean and dividing by the standard deviation. This helps enhance comparisons between different features.

[0050] The choice and implementation of data transformation techniques depends on the characteristics of the data, the analysis objectives, and the subsequent processing requirements. Proper data transformation can improve data quality and provide a better foundation for data analysis and machine learning models.

[0051] 3. Data dimensionality reduction is a technique that simplifies a dataset by reducing its dimensions, thereby reducing the complexity of data processing and analysis while preserving as much useful information as possible. The following is a detailed explanation of two commonly used data dimensionality reduction methods: PCA (Principal Component Analysis) and LDA (Latent Dirichlet Allocation):

[0052] ① Principal Component Analysis (PCA):

[0053] -PCA is an unsupervised learning algorithm that aims to transform data into a set of linearly uncorrelated principal components through an orthogonal transformation.

[0054] The steps include:

[0055] Compute the covariance matrix of the data.

[0056] Compute the eigenvalues ​​and eigenvectors of the covariance matrix.

[0057] The eigenvector corresponding to the largest eigenvalue is selected as the first principal component, and the eigenvector corresponding to the second largest eigenvalue is selected as the second principal component, and so on, until the required number of principal components is reached.

[0058] Transform the data using these eigenvectors as the new axes.

[0059] PCA can be used for data visualization and feature extraction because it can identify the main directions of variation in the data and retain most of the information in these directions.

[0060] ② Latent Dirichlet Allocation (LDA):

[0061] LDA is a supervised learning algorithm that aims to project high-dimensional data into a lower-dimensional space while preserving maximum separability between different classes.

[0062] The steps include:

[0063] Calculate the intra-class scatter and inter-class scatter. The intra-class scatter represents the distribution within the data of the same category, while the inter-class scatter represents the distribution between different categories.

[0064] Find a transformation matrix that projects the data into a new space where the intra-class divergence is minimized and the inter-class divergence is maximized.

[0065] Transform the data using this transformation matrix.

[0066] LDA is often used for data dimensionality reduction in multi-class classification problems because it helps to distinguish data of different categories and makes data of different categories more separable in low-dimensional space.

[0067] The choice of using PCA or LDA depends on the characteristics of the data and the analysis objectives. PCA is suitable for unsupervised data dimensionality reduction, while LDA is suitable for supervised data dimensionality reduction, especially when it is necessary to maintain class separability in a low-dimensional space. Data dimensionality reduction can significantly improve model training speed and predictive performance while reducing the risk of overfitting.

[0068] The data storage module 200 is wirelessly connected to the data acquisition module 100 to obtain and store each set of voltage operation data;

[0069] The data analysis module 300 is embedded in the data storage module 200, obtains each set of voltage operation data, and constructs a prediction reference model to calculate the voltage fault prediction reference value of each transmission line;

[0070] For further information, see Figure 2 Calculating the voltage fault prediction reference value of each transmission line specifically includes the following steps:

[0071] S1: Real-time acquisition of voltage operation data α at the power output end of the current transmission line;

[0072] S2: Built-in MCU microprocessor synchronously outputs time-voltage curve;

[0073] S3: Obtain voltage difference change information from the real-time output time-voltage curve;

[0074] S4: Build a prediction reference model, input voltage difference change information, and output the corresponding voltage fault prediction reference value.

[0075] Furthermore, the prediction reference model constructed is as follows:

[0076]

[0077] Among them, δ is the reference value for voltage fault prediction of the corresponding transmission line, U1, U2, ..., U n is the real-time voltage value corresponding to each selected point on the time-voltage curve, n is the number of selected points, D1, D2, ..., D n is the derivative value corresponding to each selected point on the time-voltage curve, U Q1 is the real-time voltage value corresponding to the maximum derivative point on the time-voltage curve, U Q2 is the real-time voltage value corresponding to the minimum point of the derivative on the time-voltage curve, and dx is the integral operation.

[0078] It should be noted that the number n of selected points is actually a virtual value and can be infinitely selected. In the specific operation process, it is only necessary to uniformly select a finite number of selected points.

[0079] The automated interaction module 400 is wirelessly connected to the data analysis module 300 to obtain the reference value of each voltage fault prediction, build a system interaction model, input the reference value of each voltage fault prediction, output the overall prediction interaction value, compare it with the preset basic warning threshold, and issue a fault judgment alarm.

[0080] Furthermore, the constructed system interaction model is as follows:

[0081]

[0082] Among them, η is the overall predicted interaction value, δ1, δ2, ..., δ m is the voltage fault prediction reference value obtained for each transmission line, and m is the number of transmission lines in the power system.

[0083] Specifically, the preset basic warning threshold is set to 2.360, 2.361, or 2.362.

[0084] It should be noted that the preset basic warning threshold depends on the calculation model adopted by the user. After selecting the above model, the present invention selects the above three preferred values ​​after verification by a simulation machine and specific experiments. Of course, other values ​​are also acceptable as long as they are not very different from the above three values.

[0085] It should be noted that the overall predicted interaction value is output, compared with the preset basic warning threshold, and a fault judgment alarm is issued specifically as follows: when the overall predicted interaction value is not lower than the preset basic warning threshold, a fault alarm is issued.

[0086] In order to verify the technical effect of the present invention, the first batch of test operations were carried out on a simulation machine, and some data tables are shown in Table 1 below:

[0087] Table 1: Performance data sheet

[0088] Data Item Numerical Real-time monitoring duration 12 months Frequency of power system operation data collection 1 time per second Power output parameters Voltage Historical data analysis duration Past 36 months Fault warning accuracy ≥95％ Fault response time ≤1 minute System operation stability 99.99% uptime System maintenance frequency Once a month User feedback satisfaction ≥90％

[0089] In addition, in order to further discuss the beneficial effects of the technical solution of the present invention, the following specific experimental verification process is given:

[0090] 1. System construction:

[0091] Build a simulated power system, including key components such as generators, transmission lines, and transformers.

[0092] Install sensors and data acquisition equipment to monitor the operating parameters of the power system in real time.

[0093] 2. Data Collection:

[0094] Collect power system operation data in real time.

[0095] The data collection interval is set to a short time interval (such as every second or every minute).

[0096] 3. Model training and testing:

[0097] Using the collected historical data, a predictive model is trained to predict the future output of the power system.

[0098] Test the model's accuracy and reliability to ensure it can effectively identify potential failures.

[0099] 4. Fault simulation:

[0100] Some faults are artificially introduced into the power system, such as short circuit, overload, insulation aging, etc.

[0101] Observe the system's response and alarm capabilities to these faults.

[0102] 5. Performance evaluation:

[0103] Evaluate the system's performance in fault detection and alarm, including the timeliness, accuracy, and completeness of the alarm.

[0104] The data table is shown in Table 2 below:

[0105] Table 2: Test data performance table

[0106]

[0107] Experimental data analysis:

[0108] The test results of Sample A show that the system can collect power system operation data in real time at a frequency of 10 times per second, and the data transmission delay is ≤1 millisecond, ensuring the real-time and accuracy of the data.

[0109] The test results of sample B show that the mean square error of the prediction model is 0.025 and the determination coefficient is 0.98, indicating that the model has high prediction accuracy and can accurately predict the future output of the power system.

[0110] The test results of sample C show that the system can cover a variety of fault types, and the fault detection coverage rate is ≥90%, indicating that the system has comprehensive fault detection capabilities.

[0111] The test results of sample D show that the system's response time to faults is 0.3 seconds on average, with a maximum delay of no more than 0.5 seconds, indicating that the system responds quickly to faults without delay and can issue an alarm in a timely manner.

[0112] The test results of sample E show that the system's alarm accuracy is 98.5% and the false alarm rate is ≤1.5%, indicating that the system's fault alarm accuracy is high and the false alarm rate is low.

[0113] The above data proves the superiority of the power system automation fault alarm system of the present invention in terms of real-time monitoring, prediction analysis, automatic early warning, etc.

[0114] This invention provides an automated power system fault alarm system based on power output computational prediction. It applies big data technology and model-based intelligent computing to power system fault alarms. By monitoring and predicting the power system's operating status in real time, it automatically identifies and issues warnings about potential power system faults. Compared to existing technologies, this invention offers the following innovative features:

[0115] 1. Real-time monitoring: Compared with traditional periodic inspections and manual inspections, the system of the present invention can collect and analyze the operating data of the power system in real time, greatly improving the real-time nature and accuracy of monitoring.

[0116] 2. Predictive analysis: Through big data technology and model intelligent calculation, the system of the present invention can predict the future output of the power system, thereby detecting potential faults in advance.

[0117] 3. Automatic early warning: Through the fault early warning module, the system of the present invention can automatically identify and warn of potential power system faults without manual intervention.

[0118] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An automatic fault alarm system for a power system based on power output calculation prediction, characterized in that: Includes the following components: A data acquisition module (100) is arranged at the power output end of each transmission line, and is used to collect voltage operation data of the power output end of the current transmission line in real time, and upload the corresponding voltage operation data to the data storage module (200); A data storage module (200) is wirelessly connected to the data acquisition module (100) to obtain and store each group of voltage operation data; A data analysis module (300) is embedded in the data storage module (200), acquires each group of voltage operation data, builds a prediction reference model, and calculates a voltage fault prediction reference value for each transmission line; The automated interaction module (400) is wirelessly connected to the data analysis module (300) to obtain each voltage fault prediction reference value, construct a system interaction model, input each voltage fault prediction reference value, output an overall prediction interaction value, compare it with a preset basic warning threshold, and issue a fault judgment alarm; The data acquisition module (100) is also embedded with a data preprocessing unit, which preprocesses each set of voltage operation data after acquiring the data, and uploads the preprocessed voltage operation data to the data storage module (200); Data preprocessing specifically includes: data cleaning, data conversion and data dimension reduction; Calculating the voltage fault prediction reference value of each transmission line specifically includes the following steps: S1: Real-time acquisition of voltage operation data α at the power output end of the current transmission line; S2: Built-in MCU microprocessor synchronously outputs time-voltage curve graph; S3: Obtain the voltage difference change information in the real-time output time-voltage curve diagram; S4: constructing the prediction reference model, inputting the voltage difference change information, and outputting the corresponding voltage fault prediction reference value; The prediction reference model constructed is specifically: Wherein, δ is the voltage fault prediction reference value of the corresponding transmission line, U1, U2, ..., U n is the real-time voltage value corresponding to each selected point on the time-voltage curve, n is the number of selected points, D1, D2, ..., D n is the derivative value corresponding to each selected point on the time-voltage curve, U Q1 is the real-time voltage value corresponding to the maximum derivative point on the time-voltage curve, U Q2 is the real-time voltage value corresponding to the minimum point of the derivative on the time-voltage curve, and dx is the integral operation; The system interaction model constructed is specifically: Where η is the overall predicted interaction value, δ1, δ2, ..., δ m is the voltage fault prediction reference value obtained for each transmission line, and m is the number of transmission lines in the power system.

2. The power system automation fault alarm system based on power output calculation prediction according to claim 1 is characterized in that: The preset basic warning threshold is set to 2.360 or 2.361 or 2.

362.

3. The power system automation fault alarm system based on power output calculation prediction according to claim 2 is characterized in that: Output the overall predicted interaction value, compare it with the preset basic warning threshold, and issue a fault judgment alarm. Specifically, when the overall predicted interaction value is not lower than the preset basic warning threshold, issue a fault alarm.

Images