Frequency power failure early warning system based on time sequence analysis

Abstract

The invention discloses a frequent power failure early warning system based on time sequence analysis, which relates to the technical field of data analysis and mining, and is characterized in that a real-time monitoring module acquires voltage and current data by using a voltage divider type voltage sensor and a Hall effect type current sensor, and acquires the opening and closing states of a circuit breaker in combination with a data acquisition and monitoring control system; the power failure prediction module evaluates the power failure probability based on a long short-term memory network algorithm, and the power failure analysis module sets a risk score threshold, generates a power failure early warning report and sends early warning information to a user through a short message. According to the design scheme, through the real-time monitoring and advanced data analysis technology, the operation state of the power grid is accurately mastered, the accuracy and timeliness of power failure early warning are effectively improved, the frequent power failure early warning system can automatically evaluate the power failure risk and quickly generate an early warning report, the early warning report is directly sent to a user through a short message, and the user experience is improved. The influence of frequent power failure on the life and work of the user is greatly reduced, and the intelligent level of power grid management is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis and mining, and particularly to a frequent power outage warning system based on time series analysis. Background Art

[0002] With the acceleration of the urbanization process and the continuous growth of power demand, the power supply stability and reliability have become the focus of attention from all sectors of society. Frequent power outages not only affect the normal life of residents, but also may cause serious impacts on industrial production, commercial operations, and the operation of critical infrastructure. Therefore, the development and application of a frequent power outage warning system are of great significance. The frequent power outage warning system can monitor the grid status in real time, respond to and process potential power outage risks, thereby effectively reducing the occurrence of power outage accidents. In addition, with the continuous development of advanced technologies such as the Internet of Things, big data, and cloud computing, the functions and efficiency of the frequent power outage warning system will be further improved. Looking ahead, the frequent power outage warning system will play an increasingly important role in ensuring stable power supply, improving grid management level, and promoting the upgrading of the power industry.

[0003] In the prior art, when the current frequent power outage warning system is actually applied, it is difficult to conduct targeted power outage prediction based on the current grid status, resulting in deficiencies in the sending of power outage warning information and subsequent responses. For this reason, a frequent power outage warning system based on time series analysis is proposed to solve the problems of prediction accuracy and timely response in the current frequent power outage warning system. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a frequent power outage warning system based on time series analysis, which solves the problems in the above background art.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A frequent power outage warning system based on time series analysis includes a communication warning module, and the communication warning module is communicatively connected to a real-time monitoring module, a power outage prediction module, and a power outage analysis module; The real-time monitoring module collects voltage data and current data by installing voltage dividers and Hall effect current sensors in the substations and distribution boxes of the power grid, and obtains the breaker opening and closing state data through the data acquisition and monitoring control system in the power grid; The power outage prediction module constructs a power outage prediction model based on the long short-term memory network algorithm to evaluate the power outage possibility and issue a power outage warning; The power outage analysis module sets a risk score threshold, generates a power outage warning report, conducts real-time communication, and sends power outage warning information to users in the form of text messages.

[0006] Further, in the real-time monitoring module, the process of collecting voltage data by installing voltage dividers type voltage sensors in the substations and distribution boxes of the power grid includes: Deploy voltage sensors inside the substations and distribution boxes. The high-voltage end of the voltage divider type sensor is connected to the high-voltage line, and the low-voltage end is connected to the low-voltage measuring electrical equipment. The high voltage is reduced proportionally to a low-voltage signal using a resistor voltage division network, and the low-voltage signal is converted into a digital signal by an analog-to-digital converter; Preprocess the collected voltage data. The preprocessing operations include using a filtering algorithm to remove high-frequency noise introduced by environmental interference, adjusting the readings of the voltage divider type voltage sensors, converting the original analog signal into a unified digital format, and detecting abnormal voltage data points; Calculate the mean, variance, maximum and minimum values of the preprocessed voltage data. Convert the time-domain voltage signal to the frequency domain through Fourier transform, analyze the periodic changes, frequency components, and the change trend of voltage over time in the frequency-domain voltage signal, and mark voltage mutations as potential risks of damage to electrical equipment.

[0007] Further, in the real-time monitoring module, the process of collecting current data by installing Hall effect type current sensors in the substations and distribution boxes of the power grid includes: Adopt a non-intrusive design. Surround the Hall effect type current sensor around the wires inside the substations and distribution boxes. The Hall effect type current sensor senses the magnetic field generated by the current through its internal Hall element, converts the magnetic field intensity into a proportional Hall voltage signal, amplifies and adjusts the Hall voltage through its built-in circuit, and converts the Hall voltage into a digital signal by an analog-to-digital converter; Preprocess the collected current data. The preprocessing operations include using a filtering algorithm to remove high-frequency noise introduced by environmental interference, adjusting the readings of the Hall effect type current sensors, converting the original analog signal into a unified digital format, and detecting abnormal current data points; Calculate the mean, variance, maximum and minimum values of the preprocessed current data. Convert the time-domain current signal to the frequency domain through Fourier transform, analyze the periodic changes, frequency components, and the change trend of current over time in the frequency-domain current signal, and mark current mutations as potential overload and short-circuit risks.

[0008] Further, in the real-time monitoring module, the process of obtaining the opening and closing state data of the circuit breaker through the data acquisition and monitoring control system in the power grid includes: Install remote terminal units in the substations and distribution boxes of the power grid. The remote terminal units are connected to electrical equipment, circuit breakers, and their switch state sensors. Deploy Ethernet to connect the remote terminal units and the communication warning module. Set up the data acquisition and monitoring control system and the human-machine interface in the communication warning module, define data points, set alarm thresholds, and user permissions; The switch status sensor detects the opening and closing status of the circuit breaker and transmits the status information to the remote terminal unit. The remote terminal unit uses the distributed network protocol and the communication network to transmit the switch status data to the communication warning module. After receiving and preliminarily processing the switch status data, the communication warning module stores it in its internal database and displays the real-time status and historical records through the human-machine interface. The operator sends control commands to the remote terminal unit through the human-machine interface; Preprocess the collected circuit breaker opening and closing status data. The preprocessing operations include removing outliers and noise, calibrating the switch status sensor readings and formatting them, detecting and marking abnormal status changes, and adding timestamps to each opening and closing status data record; Calculate the time ratio of the circuit breaker in the open and closed states, analyze the change trend of the circuit breaker opening and closing status over time, mark frequent tripping events and long-time non-operation events, and analyze the change of the circuit breaker opening and closing status in combination with voltage data and current data.

[0009] Furthermore, in the power outage prediction module, the process of constructing the input layer of the power outage prediction model based on the long short-term memory network algorithm to evaluate the power outage probability includes: Construct a long short-term memory network model. The long short-term memory network model includes an input layer, a hidden layer, and an output layer. The input layer receives the voltage data, current data, and circuit breaker opening and closing status data sequence after preprocessing and feature extraction from the real-time monitoring module as the time step data of the input power outage prediction model.

[0010] Furthermore, in the power outage prediction module, the process of constructing the hidden layer of the power outage prediction model based on the long short-term memory network algorithm to evaluate the power outage probability includes: The hidden layer captures the long-term dependencies in the time series data received by the input layer through the gating mechanism of its internal long short-term memory units. The gating mechanism includes a forget gate, an input gate, and an output gate; The forget gate determines the voltage mutations, current mutations, and circuit breaker abnormal opening and closing status information to be discarded from the long short-term memory unit state, calculates a candidate value between 0 and 1 through an activation function, and is used to represent the degree of information retention and discard; The input gate calculates new candidate values and determines the degree of information update through an activation function. The information update includes adding voltage mutations marked as potential damage risks of electrical equipment, current mutations of potential overload and short-circuit risks, and abnormal circuit breaker opening and closing status as risk factors to the power outage prediction analysis; Combining the results of the forget gate and the input gate, update the cell state that stores long-term information. The updated cell state analyzes the change trends of voltage, current, and circuit breaker status, and identifies the patterns leading to power outages; The output gate calculates the risk score for the current time step based on the updated cell state and passes it to the long short-term memory unit of the next time step. If a risk factor appears, the output gate outputs a high risk score and issues a power outage warning.

[0011] Further, in the power outage prediction module, the process of constructing the output layer of the power outage prediction model based on the long short-term memory network algorithm to evaluate the power outage probability includes: The output layer receives the risk score from the hidden layer. The fully connected layer inside the output layer connects the risk score output by the hidden layer and the neurons of the output layer to form a linear combination, integrating the voltage, current, and circuit breaker status information from different time steps. The Sigmoid activation function is used to perform a non-linear transformation on the result of the linear combination, and an outage probability value between 0 and 1 is output.

[0012] Further, in the power outage analysis module, the process of setting the risk score threshold based on the prediction result of the long short-term memory network model includes: The power outage analysis module receives the power outage probability value generated by the output layer and sets a risk score threshold to evaluate the power outage probability. When the risk score reaches 0.3, a primary warning message is sent to remind the operator to pay attention to the change trend of relevant parameters. When the risk score reaches 0.6, an intermediate warning message is sent, suggesting that the operator arrange equipment inspection and adjust the load distribution. When the risk score reaches 0.8, a high-level warning message is sent to start the standby power supply and arrange emergency maintenance. When the risk score approaches 1.0, an extreme warning message is sent to start the emergency plan to ensure the safe operation of critical electrical equipment.

[0013] Further, in the power outage analysis module, the process of generating a power outage warning report based on the risk score warning level includes: Integrate the power outage probability values from the long short-term memory network model, analyze the cause of the power outage in combination with the historical data of voltage, current, and circuit breaker opening and closing states. When voltage mutation, current mutation, and frequent circuit breaker tripping occur simultaneously, the power outage analysis module analyzes the correlation of risk factors and generates a power outage warning report according to the risk assessment level. The power outage warning report includes the output power outage probability and the power outage warning level, lists the main risk factors leading to the power outage, and proposes specific measures to prevent power outages.

[0014] Further, in the power outage analysis module, the process of conducting real-time communication and sending power outage warning information to users in the form of text messages includes: The power outage analysis module uses the MQTT protocol for real-time communication, integrates the warning information and formats it, calls the SMS gateway API interface to send the warning information to the user's mobile phone, regularly monitors the SMS sending performance, collects user feedback, and optimizes the communication mechanism according to actual needs.

[0015] The present invention provides a frequent power outage warning system based on time series analysis, which has the following beneficial effects: First of all, through the ingenious design of the real-time monitoring module, the system can comprehensively and real-time monitor the voltage, current and breaker opening and closing states of key nodes of the power grid. The application of voltage dividers and Hall effect current sensors ensures the high-precision and stability of data acquisition, providing a solid foundation for subsequent power outage prediction. At the same time, the integration of data acquisition and monitoring and control systems enables the system to obtain the real-time operation state of the power grid, discover and handle potential problems in a timely manner. Secondly, the power outage prediction module uses the long short-term memory network algorithm to construct a power outage prediction model, which greatly improves the accuracy and reliability of the prediction. The long short-term memory network has unique advantages in processing time series data and can capture long-term dependencies in the data, thus accurately evaluating the possibility of power outages. This not only helps to formulate countermeasures in advance, reduce the occurrence of power outage accidents, but also effectively extends the service life of power grid equipment and reduces maintenance costs. Finally, the power outage analysis module automatically generates a power outage warning report by setting a risk score threshold and sends warning information to users in the form of text messages immediately. This design not only improves the timeliness of the warning, but also ensures the accurate transmission of information, enabling users to understand potential power outage risks in the first time and take corresponding preventive measures. In addition, the addition of real-time communication functions enables the system to maintain close contact with users, collect user feedback in a timely manner, and continuously optimize and improve the warning mechanism. In summary, the frequent power outage warning system based on time series analysis proposed by the present invention has significant beneficial effects in improving the operation stability of the power grid, reducing power outage risks, and ensuring the safety of user power consumption. It is an important milestone in the intelligent and informatization development of the power industry. Brief Description of the Drawings

[0016] Figure 1 It is a block diagram of a frequent power outage warning system based on time series analysis according to an embodiment of the present application. Detailed Embodiments

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0018] As Figure 1 shown, the present invention provides a technical solution: a frequent power outage warning system based on time series analysis, including a communication warning module, and the communication warning module is communicatively connected to a real-time monitoring module, a power outage prediction module and a power outage analysis module; The real-time monitoring module collects voltage data and current data by installing voltage dividers type voltage sensors and Hall effect type current sensors in the substations and distribution boxes of the power grid, and obtains the opening and closing state data of the circuit breakers through the data acquisition and monitoring control system in the power grid; The power outage prediction module constructs a power outage prediction model based on the long short-term memory network algorithm to evaluate the power outage possibility and issue a power outage warning; The power outage analysis module sets a risk score threshold, generates a power outage warning report, conducts real-time communication, and sends power outage warning information to users in the form of text messages.

[0019] In the real-time monitoring module, the process of collecting voltage data by installing voltage dividers type voltage sensors in the substations and distribution boxes of the power grid includes: Deploy voltage sensors inside the substations and distribution boxes. The high-voltage end of the voltage divider type sensor is connected to the high-voltage line, and the low-voltage end is connected to the low-voltage measuring electrical equipment. The high voltage is reduced proportionally to a low-voltage signal using a resistor voltage division network, and the low-voltage signal is converted into a digital signal by an analog-to-digital converter; Preprocess the collected voltage data. The preprocessing operations include using a filtering algorithm to remove high-frequency noise introduced by environmental interference, adjusting the readings of the voltage dividers type voltage sensors, converting the original analog signal into a unified digital format, and detecting abnormal voltage data points; Calculate the mean, variance, maximum and minimum values of the preprocessed voltage data. Convert the time-domain voltage signal to the frequency domain through Fourier transform, analyze the periodic changes, frequency components and the change trend of voltage over time in the frequency-domain voltage signal, and mark voltage mutations as potential risks of damage to electrical equipment.

[0020] In the real-time monitoring module, the process of collecting current data by installing Hall effect type current sensors in the substations and distribution boxes of the power grid includes: Adopt a non-intrusive design. Surround the Hall effect type current sensor around the wires inside the substations and distribution boxes. The Hall effect type current sensor senses the magnetic field generated by the current through its internal Hall element, converts the magnetic field intensity into a proportional Hall voltage signal, amplifies and adjusts the Hall voltage through its built-in circuit, and converts the Hall voltage into a digital signal by an analog-to-digital converter; Preprocess the collected current data. The preprocessing operations include using a filtering algorithm to remove high-frequency noise introduced by environmental interference, adjusting the readings of the Hall effect type current sensors, converting the original analog signal into a unified digital format, and detecting abnormal current data points; Calculate the mean, variance, maximum and minimum values of the preprocessed current data. Transform the time-domain current signal to the frequency domain through Fourier transform, analyze the periodic changes, frequency components of the frequency-domain current signal, and the changing trend of the current over time. Mark current mutations as potential overload and short-circuit risks.

[0021] In the real-time monitoring module, the process of obtaining the opening and closing state data of the circuit breaker through the data acquisition and monitoring control system in the power grid includes: Install remote terminal units at the substations and distribution boxes of the power grid. The remote terminal units are connected to electrical equipment, circuit breakers, and their switch state sensors. Deploy Ethernet to connect the remote terminal units to the communication warning module. Set up the data acquisition and monitoring control system and the human-machine interface in the communication warning module, define data points, set alarm thresholds and user permissions; The switch state sensor detects the opening and closing state of the circuit breaker and transmits the state information to the remote terminal unit. The remote terminal unit uses the distributed network protocol and the communication network to transmit the switch state data to the communication warning module. After receiving and preliminarily processing the switch state data, the communication warning module stores it in its internal database, displays the real-time state and historical records through the human-machine interface, and the operator sends control commands to the remote terminal unit through the human-machine interface; Preprocess the collected opening and closing state data of the circuit breaker. The preprocessing operations include removing outliers and noise, calibrating and formatting the readings of the switch state sensor, detecting and marking abnormal state changes, and adding timestamps to each record of the opening and closing state data; Calculate the time ratio of the circuit breaker in the open and closed states, analyze the changing trend of the opening and closing state of the circuit breaker over time, mark frequent tripping events and long-time inoperative events, and combine voltage data and current data to analyze the changing state of the opening and closing of the circuit breaker.

[0022] In the power outage prediction module, the process of constructing the input layer of the power outage prediction model based on the long short-term memory network algorithm to evaluate the power outage probability includes: Construct a long short-term memory network model. The long short-term memory network model includes an input layer, a hidden layer, and an output layer. The input layer receives the voltage data, current data, and the sequence of opening and closing state data of the circuit breaker after preprocessing and feature extraction from the real-time monitoring module as the time-step data input to the power outage prediction model.

[0023] In the power outage prediction module, the process of constructing the hidden layer of the power outage prediction model based on the long short-term memory network algorithm to evaluate the power outage probability includes: The hidden layer captures the long-term dependencies in the time-series data received by the input layer through the gating mechanism of its internal long short-term memory units. The gating mechanism includes a forget gate, an input gate, and an output gate; The voltage mutation, current mutation, and breaker abnormal opening and closing state information that the forget gate decides to discard from the long short-term memory cell state are used to calculate a candidate value between 0 and 1 through an activation function, which is used to represent the degree of information retention and discard; The input gate calculates a new candidate value and determines the degree of information update through an activation function. The information update includes adding voltage mutations marked as potential risks of electrical equipment damage, current mutations with potential overload and short-circuit risks, and abnormal circuit breaker opening and closing states as risk factors to the power outage prediction analysis; Combining the results of the forget gate and the input gate, update the cell state that stores long-term information. The updated cell state analyzes the change trends of voltage, current, and breaker states to identify the patterns leading to power outages; Based on the updated cell state, the output gate calculates the risk score at the current time step and passes it to the long short-term memory unit at the next time step. If risk factors appear, the output gate outputs a high risk score and issues a power outage warning.

[0024] In the power outage prediction module, the process of constructing the output layer of the power outage prediction model based on the long short-term memory network algorithm to evaluate the power outage probability includes: The output layer receives the risk score from the hidden layer. The fully connected layer inside the output layer connects the risk score output by the hidden layer and the neurons of the output layer to form a linear combination, integrating the voltage, current, and breaker state information from different time steps, and using the Sigmoid activation function to perform a non-linear transformation on the result of the linear combination, and outputting a power outage probability value between 0 and 1.

[0025] In the power outage analysis module, the process of setting the risk score threshold based on the prediction result of the long short-term memory network model includes: The power outage analysis module receives the power outage probability value generated by the output layer, sets the risk score threshold to evaluate the power outage probability. When the risk score reaches 0.3, it sends a primary warning message to remind the operator to pay attention to the change trend of relevant parameters. When the risk score reaches 0.6, it sends an intermediate warning message, suggesting that the operator arrange equipment inspections and adjust the load distribution. When the risk score reaches 0.8, it sends a high-level warning message, starts the standby power supply and arranges for emergency maintenance. When the risk score approaches 1.0, it sends an extreme warning message and starts the emergency plan to ensure the safe operation of critical electrical equipment.

[0026] In the power outage analysis module, the process of generating a power outage warning report based on the risk score warning level includes: Integrate the power outage probability values from the long short-term memory network model, analyze the reasons for power outages by combining historical data of voltage, current, and the opening and closing states of circuit breakers. When voltage mutation, current mutation, and frequent circuit breaker tripping occur simultaneously, the power outage analysis module analyzes the relevance of risk factors, and generates a power outage warning report according to the risk assessment level. The power outage warning report includes the output power outage probability and the power outage warning level, lists the main risk factors leading to the power outage, and proposes specific measures to prevent power outages.

[0027] In the power outage analysis module, the process of real-time communication and sending power outage warning information to users in the form of text messages includes: The power outage analysis module uses the MQTT protocol for real-time communication, integrates warning information and formats it, calls the SMS gateway API interface to send the warning information to the user's mobile phone, regularly monitors the SMS sending performance, collects user feedback, and optimizes the communication mechanism according to actual needs.

[0028] First, after the system starts, the real-time monitoring module will automatically start working. This module accurately collects the voltage and current data of the power grid by installing voltage dividers and Hall effect current sensors in the substations and distribution boxes of the power grid. At the same time, by connecting to the data acquisition and monitoring control system in the power grid, the real-time monitoring module can also obtain the opening and closing state data of the circuit breakers, which is crucial for judging the operating state of the power grid. Next, the collected data will be transmitted to the power outage prediction module. This module uses an advanced long short-term memory network algorithm to construct a power outage prediction model. This model can accurately evaluate the power outage probability of the power grid based on historical data and real-time monitoring data. Once a power outage risk is predicted, the power outage prediction module will immediately trigger the warning mechanism, providing a valuable time window for subsequent countermeasures. Subsequently, the power outage analysis module will receive the warning signal from the power outage prediction module. This module further analyzes and processes the warning signal according to the preset risk score threshold, and generates a detailed power outage warning report. The report content includes the time, location, possible reasons for the power outage, and recommended countermeasures, etc. Finally, the power outage analysis module will send the power outage warning information to users in the form of text messages through the communication warning module. This step ensures that users can learn about potential power outage risks in a timely manner and take corresponding preventive measures, such as preparing backup power supplies and protecting important equipment. At the same time, the system also supports real-time communication functions. Users can feedback the received warning information to the system by text message reply or other means, so that the system can continuously optimize and improve the warning mechanism. In summary, the usage method of the present invention is simple and clear. Through real-time monitoring, accurate prediction, and timely warning, it effectively improves the stability of the power grid operation and the safety of user power consumption.

[0029] It should be noted that, in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0030] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A frequent power outage warning system based on time series analysis, characterized by: It includes a communication warning module, which is communicatively connected to a real-time monitoring module, a power outage prediction module and a power outage analysis module; The real-time monitoring module collects voltage data and current data by installing voltage divider type voltage sensors and Hall effect type current sensors in the substation and distribution box of the power grid, and obtains the circuit breaker opening and closing state data through the data acquisition and monitoring control system in the power grid; The power outage prediction module constructs a power outage prediction model based on a long short-term memory network algorithm to assess the possibility of power outages and provide power outage warnings; The power outage analysis module sets a risk score threshold, generates a power outage warning report, performs real-time communication, and sends power outage warning information to users in the form of text messages.

2. The frequent power outage early warning system based on time series analysis according to claim 1 is characterized by: In the real-time monitoring module, by installing a voltage divider type voltage sensor in the substation and distribution box of the power grid, the process of collecting voltage data includes: Voltage sensors are deployed in substations and distribution boxes. The high-voltage end of the voltage divider type sensor is connected to the high-voltage line, and the low-voltage end is connected to the low-voltage measurement electrical equipment. The high voltage is proportionally reduced to a low-voltage signal using a resistor voltage divider network. The low-voltage signal is converted into a digital signal by an analog-to-digital converter; Preprocessing the collected voltage data, wherein the preprocessing operation includes using a filtering algorithm to remove high-frequency noise introduced by environmental interference, adjusting the voltage divider type voltage sensor reading, converting the original analog signal into a unified digital format, and detecting abnormal voltage data points; The mean, variance, and maximum and minimum values ​​of the preprocessed voltage data are calculated, and the time domain voltage signal is converted to the frequency domain through Fourier transform. The periodic changes, frequency components, and time trend of the frequency domain voltage signal are analyzed, and voltage mutations are marked as potential damage risks to electrical equipment.

3. The frequent power outage early warning system based on time series analysis according to claim 2 is characterized in that: In the real-time monitoring module, by installing Hall effect current sensors in the substations and distribution boxes of the power grid, the process of collecting current data includes: The non-intrusive design is adopted, and the Hall effect current sensor is wrapped around the wires in the substation and the distribution box. The Hall effect current sensor senses the magnetic field generated by the current through its internal Hall element, converts the magnetic field strength into a proportional Hall voltage signal, amplifies and adjusts the Hall voltage through its built-in circuit, and converts the Hall voltage into a digital signal through the analog-to-digital converter; Preprocessing the collected current data includes using a filtering algorithm to remove high-frequency noise introduced by environmental interference, adjusting the reading of the Hall effect current sensor, converting the original analog signal into a unified digital format, and detecting abnormal current data points; The mean, variance, and maximum and minimum values ​​of the preprocessed current data are calculated, and the time domain current signal is converted to the frequency domain through Fourier transform. The periodic changes, frequency components, and time trend of the frequency domain current signal are analyzed, and the current mutations are marked as potential overload and short circuit risks.

4. The frequent power outage early warning system based on time series analysis according to claim 3 is characterized by: In the real-time monitoring module, the process of obtaining the circuit breaker opening and closing status data through the data acquisition and monitoring control system in the power grid includes: Install remote terminal units in substations and distribution boxes of the power grid. The remote terminal units are connected to power-consuming equipment, circuit breakers and their switch status sensors. Ethernet is deployed to connect the remote terminal units with the communication warning module. The data acquisition and monitoring control system and human-machine interface are set up in the communication warning module, data points are defined, and alarm thresholds and user permissions are set. The switch status sensor detects the open and closed status of the circuit breaker and transmits the status information to the remote terminal unit. The remote terminal unit uses the distributed network protocol and the communication network to transmit the switch status data to the communication warning module. The communication warning module receives and preliminarily processes the switch status data and stores it in its internal database. The real-time status and historical records are displayed through the human-machine interface. The operator sends control commands to the remote terminal unit through the human-machine interface. Preprocessing the collected circuit breaker opening and closing state data, wherein the preprocessing operation includes removing outliers and noise, calibrating and formatting the switch state sensor readings, detecting and marking abnormal state changes, and adding a timestamp to each opening and closing state data record; Calculate the proportion of time the circuit breaker is in the open or closed state, analyze the change trend of the circuit breaker's open or closed state over time, mark frequent tripping events and long-term no-action events, and analyze the changes in the circuit breaker's open or closed state by combining voltage data and current data.

5. The frequent power outage early warning system based on time series analysis according to claim 4 is characterized in that: In the power outage prediction module, the process of constructing a power outage prediction model input layer based on the long short-term memory network algorithm to evaluate the power outage probability includes: A long short-term memory network model is constructed, which includes an input layer, a hidden layer and an output layer. The input layer receives voltage data, current data and circuit breaker opening and closing state data sequence from the real-time monitoring module after preprocessing and feature extraction as time step data of the input power outage prediction model.

6. A frequent power outage early warning system based on time series analysis according to claim 5, characterized in that: In the power outage prediction module, the process of constructing a hidden layer of a power outage prediction model based on a long short-term memory network algorithm to evaluate the power outage probability includes: The hidden layer captures the long-term dependencies in the time series data received by the input layer through the gating mechanism of its internal long short-term memory unit, and the gating mechanism includes a forget gate, an input gate, and an output gate; The forget gate determines the voltage mutation, current mutation and abnormal opening and closing state information of the circuit breaker to be discarded from the state of the long short-term memory unit, and calculates a candidate value between 0 and 1 through the activation function to indicate the degree of information retention and discarding; The input gate calculates a new candidate value and determines the degree of information update through an activation function, wherein the information update includes adding voltage mutations marked as potential risk of damage to electrical equipment, current mutations with potential risk of overload and short circuit, and abnormal circuit switch states as risk factors into the power outage prediction analysis; Combined with the results of the forget gate and the input gate, the cell state storing long-term information is updated. The updated cell state analyzes the changing trends of voltage, current, and circuit breaker status to identify the pattern that causes power outages. The output gate calculates the risk score of the current time step based on the updated cell state and passes it to the long short-term memory unit of the next time step. If risk factors appear, the output gate outputs a high risk score and issues a power outage warning.

7. A frequent power outage warning system based on time series analysis according to claim 6, characterized in that: In the power outage prediction module, the process of constructing a power outage prediction model output layer based on the long short-term memory network algorithm to evaluate the power outage probability includes: The output layer receives the risk score from the hidden layer. The fully connected layer inside the output layer connects the risk score output by the hidden layer and the output layer neurons to form a linear combination, integrating the voltage, current and circuit breaker status information from different time steps. The Sigmoid activation function is used to perform a nonlinear transformation on the result of the linear combination and output a power outage probability value between 0 and 1.

8. The frequent power outage early warning system based on time series analysis according to claim 7 is characterized in that: In the power outage analysis module, the process of setting the risk score threshold based on the prediction results of the long short-term memory network model includes: The power outage analysis module receives the power outage probability value generated by the output layer, sets the risk score threshold to evaluate the power outage probability, and sends a primary warning message when the risk score reaches 0.3 to remind the operator to pay attention to the changing trend of relevant parameters. When the risk score reaches 0.6, an intermediate warning message is sent to suggest the operator to arrange equipment inspection and adjust the load distribution. When the risk score reaches 0.8, an advanced warning message is sent to start the backup power supply and arrange emergency maintenance. When the risk score is close to 1.0, an extreme warning message is sent to start the emergency plan to ensure the safe operation of key power equipment.

9. A frequent power outage warning system based on time series analysis according to claim 8, characterized in that: In the power outage analysis module, the process of generating a power outage warning report based on the risk score warning level includes: The power outage probability value from the long short-term memory network model is integrated, and the historical data of voltage, current and circuit breaker opening and closing status are combined to analyze the cause of the power outage. When voltage mutation, current mutation and frequent circuit breaker tripping occur at the same time, the power outage analysis module analyzes the correlation of risk factors and generates a power outage warning report according to the risk assessment level. The power outage warning report includes the output of power outage probability and power outage warning level, lists the main risk factors leading to power outages, and proposes specific measures to prevent power outages.

10. A frequent power outage early warning system based on time series analysis according to claim 9, characterized in that: In the power outage analysis module, the process of performing real-time communication and sending power outage warning information to the user in the form of text messages includes: The power outage analysis module uses the MQTT protocol for real-time communication, integrates and formats warning information, calls the SMS gateway API interface to send warning information to the user's mobile phone, regularly monitors SMS sending performance, collects user feedback, and optimizes the communication mechanism according to actual needs.

Images