Remote monitoring system for mobile substation

Abstract

The invention provides a remote monitoring system for a mobile substation, which relates to the technical field of power systems, and comprises a data acquisition module used for acquiring operation data of a plurality of substation devices in the mobile substation in real time; the edge calculation module is connected with the data acquisition module and is used for analyzing and processing the acquired data in real time through a distributed calculation method; and a high-speed low-delay communication module. The remote monitoring system for the mobile substation has remarkable advantages in the aspects of real-time data acquisition and time sequence characteristic analysis through a deep learning algorithm. Through deep analysis of equipment operation data, the system can accurately extract space and time characteristics, thereby realizing high-precision fault prediction. According to the deep learning method combining time sequence analysis and multi-dimensional data fusion, potential faults of the equipment can be effectively identified, and the problem that a traditional monitoring mode is slow in response to sudden faults is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a mobile substation remote monitoring system. Background Art

[0002] With the continuous development of power systems, mobile substations, as a crucial temporary power facility, are increasingly playing a vital role in various scenarios, including emergency power supply, load regulation, and post-disaster recovery. Mobile substations, characterized by high flexibility, short construction cycles, and rapid commissioning, have led to their widespread use in remote areas and during emergencies. To ensure their safe and stable operation, the industry has developed monitoring systems based on automation, communications, and information technologies that enable status monitoring, data collection, and equipment operation analysis for mobile substations. These systems leverage IoT technology and intelligent sensors for real-time monitoring, providing personnel with information on substation equipment status, load conditions, and fault alarms, ensuring efficient operation. Furthermore, some existing mobile substation monitoring systems feature remote control, enabling operation and maintenance personnel to access and operate them remotely via the internet or dedicated networks. This allows managers to monitor equipment status and make necessary adjustments without having to be physically present on-site. These technological advancements have significantly improved the operational efficiency of mobile substations, providing technical support for the stability and security of power systems.

[0003] Although the existing mobile substation remote monitoring system can meet basic monitoring needs to a certain extent, in complex environments, the monitoring system's data transmission speed is slow and the data delay is large, resulting in the system's inability to quickly respond to equipment failures or abnormal conditions, increasing the risk of equipment damage and accidents. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a mobile substation remote monitoring system. The technical problem to be solved by this invention is: how to monitor substation equipment in real time through multi-dimensional data fusion and time series analysis, accurately predict faults and issue early warnings, and optimize equipment maintenance and fault response.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A mobile substation remote monitoring system, comprising: Data acquisition module, used to collect real-time operating data of multiple substation equipment in the mobile substation; An edge computing module, connected to the data acquisition module, performs real-time analysis and processing of the collected data through a distributed computing method; High-speed, low-latency communication module, using adaptive communication technology, combined with 5G and low-power wide area network technology; Intelligent fault prediction and early warning module, which performs deep learning and big data analysis on real-time collected data based on deep learning algorithms, and establishes a time series analysis and multi-dimensional data fusion model; Remote monitoring terminal, used to display real-time operating data, historical fault trends, health status and fault warning information of substation equipment.

[0006] Preferably, the data acquisition module includes a distributed intelligent sensor network, in which each sensor has data preprocessing capabilities and can perform preliminary analysis, compression and filtering on the original collected data on site to reduce bandwidth consumption and delay in data transmission. The edge computing module includes multiple computing nodes, which can perform multi-level fault detection, anomaly identification, trend analysis and real-time status evaluation based on the collected real-time data.

[0007] Preferably, the remote monitoring terminal and the intelligent fault prediction and early warning module are connected, and two-way data interaction is carried out with the edge computing module through the high-speed and low-latency communication module, supporting remote operation and adjustment to realize active maintenance and fault response of substation equipment.

[0008] Preferably, the distributed intelligent sensor network includes temperature sensors, current sensors, voltage sensors and vibration sensors, which are used to monitor the temperature, current, voltage and vibration data of key substation equipment in the substation in real time.

[0009] Preferably, the computing nodes are respectively deployed near different substation equipment, and the real-time analysis and processing include preprocessing, fault detection and preliminary analysis, thereby reducing the frequency of data transmission to the central server and improving the real-time performance of fault prediction and response.

[0010] Preferably, the time series analysis and multidimensional data fusion model extracts spatial and temporal features from multidimensional data collected in real time, and the specific steps are as follows: S1. The data of different substation devices are mapped to a unified scale range, and spatial features are extracted through a convolution operation. The convolution operation uses the following formula to extract local features: , in, For the convolution output, representing the features extracted by the convolution kernel, For the The weights of the convolution kernels represent the parameters of the convolution filter used for feature extraction. The first Sample values, representing the sample value of each time step in the data stream, is the bias term, which is used to adjust the output of the model. is the length of the convolution kernel, which indicates the size of the data window covered by the convolution kernel, and represents an adaptive convolution kernel size scheme, by adjusting , to achieve feature extraction of different lengths; S2. The spatial features extracted by the convolution operation are input into an LSTM network for time series data modeling, effectively capturing the temporal dependencies of substation equipment status. An LSTM activation function is proposed. This innovative activation function, through two weighted adjustments, can better integrate historical and current status information, improving the accuracy of fault prediction. S3. The time series features generated by LSTM are integrated with big data from multiple sensors to form a high-dimensional feature space. The multidimensional data fusion formula is used to perform a weighted combination of the sensor data to generate the final substation equipment health status feature vector: , in, is the fused substation equipment health status feature vector, For the sensors at time step The output, and is the weighting coefficient of each sensor, This formula adjusts the rate of change to give less weight to sensor data that fluctuates violently, thereby enhancing the system's fault prediction capability in a stable state.

[0011] Preferably, the LSTM activation function is defined as: , in, is the current time step The output of represents the activation value of the current time step, and are two different weight matrices, controlling the current input and the previous hidden state The weighted contribution of and is a bias term, which is used to adjust the output of each activation function. is the hidden state at the last moment, indicating the state of the substation equipment at the last moment. It is the input at the current moment, indicating the current operating data of the substation equipment.

[0012] Preferably, the health status and fault warning information of the substation equipment include fault type, fault location and fault level.

[0013] The present invention provides a mobile substation remote monitoring system, which has the following beneficial effects:

[0014] This mobile substation remote monitoring system, leveraging deep learning algorithms, offers significant advantages in real-time data collection and time-series feature analysis. Through in-depth analysis of equipment operating data, the system accurately extracts spatial and temporal features, enabling highly accurate fault prediction. This deep learning approach, which combines time-series analysis with multidimensional data fusion, effectively identifies potential equipment failures, avoiding the slow response to sudden failures associated with traditional monitoring methods. Unlike traditional systems that rely on simple threshold settings, this solution dynamically adjusts the prediction model using historical and real-time monitoring data, making fault prediction more intelligent and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of a process for implementing the invention; Figure 2 It is a structural diagram for realizing the invention. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0017] Example 1 like Figure 1-2 As shown, an embodiment of the present invention provides a mobile substation remote monitoring system, including a data acquisition module for real-time acquisition of operating data of multiple substation equipment in a mobile substation, the data acquisition module including a distributed intelligent sensor network, wherein each sensor has data preprocessing capability and can perform preliminary analysis, compression and filtering on the original collected data on site to reduce bandwidth consumption and delay in data transmission, the edge computing module includes multiple computing nodes, which can perform multi-level fault detection, anomaly identification, trend analysis and real-time status evaluation based on the collected real-time data, the distributed intelligent sensor network includes temperature sensors, current sensors, voltage sensors and vibration sensors, which are used to respectively monitor the temperature, current, voltage and vibration data of key substation equipment in the substation in real time, the computing nodes are respectively deployed near different substation equipment, and the real-time analysis and processing include preprocessing, fault detection and preliminary analysis, thereby reducing the frequency of data transmission to the central server and improving the real-time performance of fault prediction and response.

[0018] The edge computing module is connected to the data acquisition module and performs real-time analysis and processing of the collected data through distributed computing methods.

[0019] High-speed, low-latency communication module uses adaptive communication technology combined with 5G and low-power wide area network technology.

[0020] The intelligent fault prediction and warning module performs deep learning and big data analysis on real-time collected data based on deep learning algorithms, and establishes a time series analysis and multi-dimensional data fusion model. The time series analysis and multi-dimensional data fusion model extracts spatial and temporal features from the multi-dimensional data collected in real time. The specific steps are as follows: S1. Data from different substation equipment is mapped to a unified scale range, and spatial features are extracted through convolution operations. The convolution operation uses the following formula to extract local features: , in, For the convolution output, representing the features extracted by the convolution kernel, For the The weights of the convolution kernels represent the parameters of the convolution filter used for feature extraction. The first Sample values, representing the sample value of each time step in the data stream, is the bias term, which is used to adjust the output of the model. is the length of the convolution kernel, which indicates the size of the data window covered by the convolution kernel, and represents an adaptive convolution kernel size scheme, by adjusting , to achieve feature extraction of different lengths.

[0021] S2. The spatial features extracted by the convolution operation are input into the LSTM network for time series data modeling, effectively capturing the temporal dependencies of substation equipment status. An LSTM activation function is proposed. This innovative activation function, through two weighted adjustments, can better integrate historical and current status information, improving the accuracy of fault prediction. The LSTM activation function is defined as: , in, is the current time step The output of represents the activation value of the current time step, and are two different weight matrices, controlling the current input and the previous hidden state The weighted contribution of and is a bias term, which is used to adjust the output of each activation function. is the hidden state at the last moment, indicating the state of the substation equipment at the last moment. It is the input at the current moment, indicating the current operating data of the substation equipment.

[0022] S3. The time series features generated by LSTM are integrated with big data from multiple sensors to form a high-dimensional feature space. The multidimensional data fusion formula is used to perform a weighted combination of the sensor data to generate the final substation equipment health status feature vector: , in, is the fused substation equipment health status feature vector, For the sensors at time step The output, and is the weighting coefficient of each sensor, This formula adjusts the rate of change to give less weight to sensor data that fluctuates violently, thereby enhancing the system's fault prediction capability in a stable state.

[0023] The remote monitoring terminal is used to display the real-time operating data, historical fault trends, health status and fault warning information of substation equipment. The remote monitoring terminal is connected to the intelligent fault prediction and warning module, and conducts two-way data interaction with the edge computing module through a high-speed, low-latency communication module. It supports remote operation and adjustment to achieve proactive maintenance and fault response of substation equipment. The health status and fault warning information of substation equipment include fault type, fault location and fault level.

[0024] Example 2 This example describes an intelligent fault prediction and early warning system based on deep learning algorithms, applied to the monitoring and maintenance of substation equipment. By applying deep learning and big data analysis to real-time data collected by substation equipment, the system can predict equipment failures in real time and display relevant fault warning information, historical data trends, and equipment health status through a remote monitoring terminal.

[0025] S1. Data preprocessing and spatial feature extraction In this embodiment, the data acquisition module collects real-time operating data of equipment in the substation through different sensors. The following is part of the actual data collected:

[0026] Device A (temperature sensor): [22.5, 23.0, 23.5, 24.0, 24.5] Device B (voltage sensor): [220, 225, 230, 235, 240] Device C (current sensor): [15, 16, 17, 16.5, 15.5] Represents the real-time sampling values ​​of devices A, B, and C in 5 time steps respectively. In order to unify the data scale, after standardization processing, the following normalized data is obtained:

[0027] Device A normalization: Device B normalization: Device C normalized: [0.0, 0.25, 0.5, 0.375, 0.125] The normalized data is input into the convolutional neural network for spatial feature extraction. Assuming the size of the convolution kernel is , the convolution operation is performed by the following formula:

[0028] , in, is the weight of the convolution kernel, is the sample value of the input data, This convolution operation allows the system to extract local features from the sensor data, such as temperature rise and voltage fluctuation, which in turn provide input for the timing analysis model.

[0029] S2. Time Series Feature Modeling and LSTM Processing The spatial features extracted by the convolution operation are input into the LSTM network for time series modeling. The innovative LSTM activation function, through two weighted adjustments, can better integrate historical data with current input, improving the accuracy of fault prediction. The LSTM activation function is as follows:

[0030] , The LSTM activation function uses two weighted adjustments to combine the current input (such as the device's real-time voltage, current, and other data) with the historical state (such as the device state at the previous moment), enabling more accurate identification of the temporal dependencies of device states.

[0031] S3. Multidimensional data fusion and health status feature extraction The time series features generated by LSTM will be fused with big data from multiple sensors to form a high-dimensional feature space. In this embodiment, we use the following multi-dimensional data fusion formula:

[0032] , in, is the fused substation equipment health status feature vector, For the sensors at time step The output of The data collected by the sensors, is the weighting coefficient, is the weight coefficient, is the gradient (rate of change) of the data.

[0033] This fusion formula reduces the impact of wildly fluctuating sensor data and enhances fault prediction capabilities under stable conditions. For example, current sensor data may fluctuate rapidly, while temperature sensor data fluctuates more steadily. The system automatically adjusts the weights to prioritize temperature trends, minimizing the impact of current fluctuations on health status assessment.

[0034] Remote monitoring terminal and data interaction: The system uses a high-speed, low-latency communication module to exchange data with a remote monitoring terminal. The remote monitoring terminal displays real-time operating data, historical fault trends, health status, and fault warning information for substation equipment. The following is a demonstration of the remote monitoring terminal's functions:

[0035] Real-time data display: The real-time temperature data of device A is displayed graphically, showing the current temperature value and historical temperature changes.

[0036] The real-time voltage data of device B is displayed, and the voltage fluctuation is updated in real time in the chart.

[0037] Historical failure trends: Displays the fault records and trend charts for device A and device B over the past week, helping O&M personnel identify device failure modes.

[0038] The system automatically generates trend charts to analyze the long-term health status of each device.

[0039] Health status and fault warning: Device A's current health status is "Normal." Device B's temperature is about to exceed the threshold. The system alerts the operator with a yellow alarm icon.

[0040] The system issues an early warning indicating that device B may be failing and provides recommended maintenance measures.

[0041] For example, the temperature data of device A is as follows: Real-time data: [22.5, 23.0, 23.5, 24.0, 24.5] Historical fault trends: The system displays the temperature change trend of device A over the past week. It finds that the temperature of device A has gradually increased over the past few days, exceeding the normal threshold. The system predicts that device A may have an overheating fault.

[0042] This implementation demonstrates how an intelligent fault prediction and warning system can monitor the operating status of substation equipment in real time through technologies such as deep learning, time series analysis, and multidimensional data fusion. The system efficiently extracts spatial and temporal characteristics of equipment and generates equipment health status feature vectors, predicting potential faults and displaying equipment health status and fault warning information via a remote monitoring terminal.

[0043] Data accuracy: By combining deep learning with time series analysis, the system can accurately predict equipment failures and reduce the occurrence of failures.

[0044] System response speed: Using adaptive weighting and convolution operations, the system can capture changes in device status in real time and improve response speed.

[0045] Prediction effect: Fault warning information can be issued in advance, helping operation and maintenance personnel take maintenance measures before equipment failure occurs, thus avoiding equipment downtime.

[0046] The system has high real-time and accuracy, and can effectively improve the operating efficiency of substation equipment, reduce fault downtime, and enhance the initiative and intelligence level of equipment maintenance.

[0047] Comparative Example This embodiment is compared with the intelligent fault prediction and early warning system in Example 1 to demonstrate the differences between traditional substation monitoring systems and the intelligent fault prediction and early warning system of the present invention, particularly in terms of data processing, fault prediction, and response efficiency. This comparison clarifies the advantages of the present invention in substation equipment monitoring.

[0048] Traditional substation monitoring system Traditional substation monitoring systems rely primarily on manual inspections and threshold-based alarm mechanisms to monitor equipment status. These systems regularly collect equipment data through sensors (such as temperature and voltage) and transmit it to a central server for storage and processing. Fault detection is typically based on fixed, pre-set thresholds. For example, if a device's voltage exceeds a set value, the system will issue an alarm.

[0049] Data collection and processing methods: Sensors in traditional systems mainly rely on manual installation and settings, and equipment operation data is transmitted to the central server through standardized interfaces.

[0050] Data analysis relies solely on pre-set thresholds and rules. For example, a fault alarm is issued when the current value of a device exceeds 500A.

[0051] When processing data from multiple devices and multiple sensors, the system can only aggregate and analyze data at the center, resulting in slow processing speed and high data latency.

[0052] Fault prediction and response: Traditional monitoring systems can only handle equipment failures after they occur, and usually rely on manual inspections and manual operations to determine equipment failures.

[0053] Fault prediction and warning rely on fixed rules, cannot be adjusted dynamically, and can only detect significant fault conditions.

[0054] Fault response time is long, and it is impossible to accurately predict and handle equipment failures before they occur.

[0055] Summary of advantages and disadvantages: Advantages: The system is simple and easy to use, low cost, and suitable for small-scale substations.

[0056] Disadvantages: Lack of intelligent analysis, inaccurate fault prediction, reliance on manual intervention, and long response time.

[0057] Intelligent fault prediction and early warning system of the present invention In an embodiment of the present invention, the intelligent fault prediction and warning system processes the real-time collected equipment data through multi-dimensional data fusion and deep learning algorithms, and can predict equipment failures before they occur and issue warnings in advance.

[0058] Data collection and processing methods: The data acquisition module includes a distributed intelligent sensor network. The sensors include temperature sensors, current sensors, voltage sensors and vibration sensors, which can monitor the operating data of substation equipment in real time.

[0059] Each sensor has data pre-processing capabilities and can perform preliminary analysis, compression and filtering of raw data on site, reducing bandwidth consumption and delays in data transmission.

[0060] Data is analyzed in real time through edge computing modules. Edge computing nodes are deployed near different equipment in the substation to perform preliminary fault detection, trend analysis and real-time status assessment, thereby reducing the frequency of data transmission to the central server.

[0061] Fault prediction and response: The intelligent fault prediction and warning module is based on deep learning algorithms (such as convolutional neural networks (CNN) and long short-term memory (LSTM) networks). Through time series analysis and multi-dimensional data fusion models, it combines multi-sensor data to generate equipment health status feature vectors, thereby achieving accurate prediction of equipment failures.

[0062] The system can dynamically adjust the prediction model, identify potential faults in advance through a multi-level fault warning mechanism, and provide warning signals of different levels.

[0063] The system can display the equipment health status, historical fault trends and fault warning information in real time through the remote monitoring terminal, and supports remote operation and adjustment to ensure timely response and maintenance measures.

[0064] Summary of advantages and disadvantages: Advantages: The system enables accurate fault prediction and early warning, improving equipment reliability. The data processing process uses distributed computing and deep learning methods, reducing data latency and improving prediction accuracy.

[0065] Disadvantages: System deployment and maintenance are complex, and the initial investment is high, but it can significantly reduce long-term operation and maintenance costs.

[0066] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A mobile substation remote monitoring system, characterized in that: include: Data acquisition module, used to collect real-time operating data of multiple substation equipment in the mobile substation; An edge computing module, connected to the data acquisition module, performs real-time analysis and processing of the collected data through a distributed computing method; High-speed, low-latency communication module, using adaptive communication technology, combined with 5G and low-power wide area network technology; Intelligent fault prediction and early warning module, which performs deep learning and big data analysis on real-time collected data based on deep learning algorithms, and establishes a time series analysis and multi-dimensional data fusion model; Remote monitoring terminal, used to display real-time operating data, historical fault trends, health status and fault warning information of substation equipment.

2. A mobile substation remote monitoring system according to claim 1, characterized in that: The data acquisition module includes a distributed intelligent sensor network, and the edge computing module includes multiple computing nodes.

3. A mobile substation remote monitoring system according to claim 1, characterized in that: The remote monitoring terminal and the intelligent fault prediction and early warning module are connected, and two-way data interaction is performed with the edge computing module through the high-speed and low-latency communication module.

4. A mobile substation remote monitoring system according to claim 2, characterized in that: The distributed intelligent sensor network includes a temperature sensor, a current sensor, a voltage sensor and a vibration sensor.

5. A mobile substation remote monitoring system according to claim 3, characterized in that: The computing nodes are respectively deployed near different substation equipment, and the real-time analysis and processing include pre-processing, fault detection and preliminary analysis.

6. A mobile substation remote monitoring system according to claim 5, characterized in that: The time series analysis and multidimensional data fusion model extracts spatial and temporal features from multidimensional data collected in real time. The specific steps are as follows: S1. Mapping the data of different substation equipment to a unified scale range and extracting spatial features through convolution operation; S2. The spatial features extracted by the convolution operation are input into the LSTM network for time series data modeling, and an LSTM activation function is proposed; S3. The time series features generated by LSTM are integrated with big data from multiple sensors to form a high-dimensional feature space. The multidimensional data fusion formula is used to perform a weighted combination of the sensor data to generate the final substation equipment health status feature vector: ， in, is the fused substation equipment health status feature vector, For the sensors at time step The output, and is the weighting coefficient of each sensor, is the gradient change of sensor data, which indicates the speed of feature change.

7. A mobile substation remote monitoring system according to claim 6, characterized in that: The LSTM activation function is defined as: ， in, is the current time step The output of represents the activation value of the current time step, and are two different weight matrices, controlling the current input and the previous hidden state The weighted contribution of and is a bias term, which is used to adjust the output of each activation function. is the hidden state at the last moment, indicating the state of the substation equipment at the last moment. It is the input at the current moment, indicating the current operating data of the substation equipment.

8. The mobile substation remote monitoring system according to claim 1, characterized in that: The health status and fault warning information of the substation equipment include fault type, fault location and fault level.

Images