5G communication base station power operation monitoring system and method

Abstract

The invention relates to the field of power operation monitoring, and particularly discloses a 5G communication base station power operation monitoring system and method, and the method comprises the steps: firstly obtaining a power operation equipment image collected by a camera, power operation equipment temperature values collected by a temperature sensor at a plurality of preset time points, and power system parameter data collected by the sensor; the method comprises the following steps of: performing feature extraction and correlation analysis on the three by using a deep learning technology, and finally obtaining a classification result through a classifier so as to judge whether to send out a 5G communication base station power operation abnormity early warning or not, so that power operation abnormity can be found in time, the reliability and the maintenance efficiency of the 5G base station are improved, and the downtime and the maintenance cost are reduced.

Description

Technical Field

[0001] This application relates to the field of power operation monitoring, and more specifically, to a 5G communication base station power operation monitoring system and method. Background Art

[0002] With the wide application and development of 5G communication technology, as a key facility to support 5G network services, the operation stability and reliability of 5G base stations have become particularly important. 5G base stations are widely deployed globally and undertake the important task of processing high-bandwidth and low-latency data transmission. Due to its complex power system and equipment configuration, the power operation status of the base station directly affects the network performance and service quality. Therefore, an effective power operation monitoring and fault warning system is crucial for ensuring the normal operation of 5G base stations.

[0003] Traditional power monitoring methods mainly rely on static data, such as real-time monitoring by cameras. Since the fault characteristics of power equipment are usually complex, it is difficult to comprehensively evaluate the operation health status of the equipment based on single monitoring data. Moreover, 5G base station equipment is usually deployed in relatively scattered geographical locations, which increases the difficulty of on-site maintenance and fault troubleshooting. If there are problems in the power system, traditional maintenance methods may lead to longer downtime, thus affecting the continuity and quality of network services.

[0004] Therefore, a 5G communication base station power operation monitoring system and method are desired. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. Embodiments of this application provide a 5G communication base station power operation monitoring system and method, which first obtain the power operation equipment images collected by cameras, the power operation equipment temperature values at multiple predetermined time points collected by temperature sensors, and the power system parameter data collected by sensors, then use deep learning technology to perform feature extraction and correlation analysis on the three, and finally obtain a classification result through a classifier to determine whether to issue a 5G communication base station power operation anomaly warning, so as to timely detect power operation anomalies, improve the reliability and maintenance efficiency of 5G base stations, and reduce downtime and maintenance costs.

[0006] According to one aspect of this application, a 5G communication base station power operation monitoring system is provided, which includes:

[0007] A 5G communication base station power operation data acquisition module, configured to obtain the power operation equipment images collected by cameras, the power operation equipment temperature values at multiple predetermined time points collected by temperature sensors, and the power system parameter data collected by sensors;

[0008] A 5G communication base station power operation data processing module is used to extract a multi-modal information management feature vector of power operation equipment and a semantic feature vector associated with power system parameter data from the images of power operation equipment collected by the camera, the temperature values of power operation equipment at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor.

[0009] A power operation analysis module is used to determine whether to issue a 5G communication base station power operation anomaly warning based on the multi-modal information management feature vector of power operation equipment and the semantic feature vector associated with power system parameter data.

[0010] According to another aspect of the present application, a 5G communication base station power operation monitoring method is provided, which includes:

[0011] Obtain the images of power operation equipment collected by the camera, the temperature values of power operation equipment at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor;

[0012] Extract a multi-modal information management feature vector of power operation equipment and a semantic feature vector associated with power system parameter data from the images of power operation equipment collected by the camera, the temperature values of power operation equipment at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor;

[0013] Based on the multi-modal information management feature vector of power operation equipment and the semantic feature vector associated with power system parameter data, determine whether to issue a 5G communication base station power operation anomaly warning.

[0014] Compared with the prior art, a 5G communication base station power operation monitoring system and method provided by the present application first obtain the images of power operation equipment collected by the camera, the temperature values of power operation equipment at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor, then use deep learning technology to perform feature extraction and correlation analysis on the three, and finally obtain a classification result through a classifier to determine whether to issue a 5G communication base station power operation anomaly warning, so as to timely detect power operation anomalies, improve the reliability and maintenance efficiency of 5G base stations, and reduce downtime and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the drawings, the above and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application, and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram of a 5G communication base station power operation monitoring system according to an embodiment of the present application.

[0017] Figure 2 It is a block diagram of a 5G communication base station power operation data processing module in a 5G communication base station power operation monitoring system according to an embodiment of the present application.

[0018] Figure 3 It is a block diagram of a power operation equipment temperature feature extraction unit in a 5G communication base station power operation monitoring system according to an embodiment of the present application.

[0019] Figure 4 It is a block diagram of a power operation analysis module in a 5G communication base station power operation monitoring system according to an embodiment of the present application.

[0020] Figure 5 It is a flowchart of a 5G communication base station power operation monitoring method according to an embodiment of the present application. Detailed implementation manners

[0021] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0022] Figure 1 It is a block diagram of a 5G communication base station power operation monitoring system according to an embodiment of the present application. As Figure 1 shown, a 5G communication base station power operation monitoring system 100 according to an embodiment of the present application includes: a 5G communication base station power operation data acquisition module 110, configured to acquire power operation equipment images collected by a camera, power operation equipment temperature values at multiple predetermined time points collected by a temperature sensor, and power system parameter data collected by a sensor; a 5G communication base station power operation data processing module 120, configured to extract a power operation equipment multimodal information management feature vector and a power system parameter data associated semantic feature vector from the power operation equipment images collected by the camera, the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor; a power operation analysis module 130, configured to determine whether to issue a 5G communication base station power operation anomaly warning based on the power operation equipment multimodal information management feature vector and the power system parameter data associated semantic feature vector.

[0023] In the above 5G communication base station power operation monitoring system 100, the 5G communication base station power operation data acquisition module 110 is used to acquire the power operation equipment images collected by the camera, the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor. It should be understood that with the rapid development of 5G technology, 5G base stations have become the core components of network services, and ensuring their stability and reliability is crucial. 5G base stations are widely distributed globally and are responsible for processing high-bandwidth and low-latency data transmission. The complex power systems and equipment configurations of the base stations make the power status have a direct impact on network performance and service quality. Therefore, establishing an effective power monitoring and fault warning system is necessary to ensure the normal operation of 5G base stations. Traditional power monitoring mostly relies on static data, such as real-time monitoring by cameras. This method is difficult to comprehensively evaluate the operation health status of equipment because the characteristics of power equipment failures are usually complex. In addition, since 5G base stations are usually distributed in different geographical locations, on-site maintenance and fault troubleshooting have become more difficult. Traditional maintenance methods may result in longer downtime, thus affecting the continuity and quality of network services. Therefore, in the technical solution of this application, the power operation equipment images collected by the camera, the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor are acquired, and combined with deep learning technology, by judging whether it is necessary to issue an early warning of abnormal power operation of the 5G communication base station, problems in power operation are timely identified, thereby improving the reliability and maintenance efficiency of 5G base stations, shortening the downtime and reducing the maintenance cost.

[0024] Specifically, the power operation equipment images collected by the camera can provide real-time visual data, which is crucial for detecting the physical state and appearance changes of the equipment. Through image analysis, it can be monitored whether there are obvious physical damages, overheating signs or other visible abnormalities on the equipment. The temperature values of the power equipment collected by the temperature sensor provide dynamic data on the internal temperature of the equipment. The power system parameter data collected by the sensor provides the operation data of the power system, including key parameters such as current, voltage, and power. Combining these three types of data sources, the 5G base station power operation monitoring system can comprehensively consider the visual information, temperature changes and system parameters of the equipment, and establish a comprehensive multi-modal information management feature vector. Such comprehensive data processing ability enables the system to more accurately judge the health status of the equipment, issue early warnings in a timely manner, thereby reducing the risk of failures and improving the operation stability of the base station and the quality of network services.

[0025] In the above 5G communication base station power operation monitoring system 100, the 5G communication base station power operation data processing module 120 is used to extract the power operation equipment multimodal information management feature vector and the power system parameter data associated semantic feature vector from the power operation equipment images collected by the camera, the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor. It should be understood that extracting and fusing these feature vectors not only improves the comprehensive understanding of the operation state of power equipment, but also enhances the accuracy of fault prediction and early warning. Through the combination of multimodal data, the system can achieve more comprehensive equipment health monitoring, improve the operation stability of 5G base stations, and ensure the continuous reliability of network services.

[0026] Figure 2 It is a block diagram of the 5G communication base station power operation data processing module in the 5G communication base station power operation monitoring system according to the embodiment of the present application. As Figure 2 shown, in a specific embodiment of the present application, the 5G communication base station power operation data processing module 120 includes: a power operation equipment image feature extraction unit 121, which is used to extract features from the power operation equipment images collected by the camera to obtain a power operation equipment tracking association feature vector; a power operation equipment temperature feature extraction unit 122, which is used to extract features from the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor to obtain a power operation equipment temperature feature vector; a power operation equipment multimodal feature association unit 123, which is used to associate the power operation equipment tracking association feature vector and the power operation equipment temperature feature vector to obtain the power operation equipment multimodal information management feature vector; a power system parameter data feature extraction unit 124, which is used to extract features from the power system parameter data collected by the sensor to obtain the power system parameter data associated semantic feature vector.

[0027] It should be understood that the core of image feature extraction lies in extracting useful information from the original image, and this information can include features such as the shape, color, texture, and position of the equipment. By processing image data through a deep learning model (such as a convolutional neural network), the system can identify and extract key features, such as the identification of equipment, damage marks, or overheating phenomena. These features are integrated into a feature vector, which is called the power operation equipment tracking association feature vector. The extracted feature vector is combined with other data sources (such as temperature sensors and system parameter data) to improve the accuracy of anomaly detection.

[0028] Furthermore, the core of temperature data feature extraction lies in extracting meaningful features from time series data, which can reflect the temperature change trend, abnormal patterns, and operating status of the device. Specifically, by analyzing temperature data at multiple time points, a temperature feature vector can be generated, which helps analyze the temperature change trend of the device. For example, by calculating statistical features such as the average value, standard deviation, and temperature fluctuation amplitude of the temperature data, the system can identify whether the device is operating within the normal range.

[0029] Furthermore, considering that individual image features and temperature features each provide partial information about the device. Image features mainly reflect the physical state and appearance changes of the device, while temperature features reveal the thermal state and operating load of the device. By correlating these two types of features, a comprehensive state assessment of the device can be obtained. For example, image analysis may show signs of damage on the outside of the device, while temperature data may indicate overheating. Such a comprehensive analysis can more accurately determine whether there is a risk of device failure.

[0030] In particular, power system parameter data usually includes key electrical indicators such as current, voltage, power, and frequency. This data often exists in the form of time series and contains a large amount of numerical information. The purpose of feature extraction is to extract key features from this raw data that help understand the system state, generate a power system parameter data associated semantic feature vector, and the feature vector of power system parameter data can reveal the operating state and performance of the system.

[0031] In a specific embodiment of the present application, the power operation device image feature extraction unit 121 includes: passing the power operation device image collected by the camera through a power operation device image target detection network model to obtain multiple power operation device tracking regions of interest; arranging the multiple power operation device tracking regions of interest into a power operation device tracking input tensor; passing the power operation device tracking input tensor through a power operation device tracking three-dimensional convolutional neural network to obtain the power operation device tracking associated feature vector.

[0032] It should be understood that the object detection of power operation equipment images is a process of automatically identifying and locating objects in images using a deep learning model. By processing the images captured by the camera, the model can detect the regions of interest of different power equipment in the images. The object detection network can automatically identify the specific locations and categories of power equipment, reducing the need for manual intervention. By detecting the regions of interest in the images, the system can accurately locate the equipment and extract relevant information such as equipment type, location, and status. This automated identification process improves the monitoring efficiency and accuracy. The detected regions of interest provide key equipment location information and status data, which can be combined with other monitoring data (such as temperature, vibration, etc.) for comprehensive analysis. Through in-depth analysis of this data, the system can better understand the operating conditions of the equipment, support maintenance decisions, and optimize management. Specifically, the object detection formula of the power operation equipment image object detection network model is used to process the power operation equipment images collected by the camera to obtain the multiple regions of interest for power operation equipment tracking;

[0033] Among them, the object detection formula is:

[0034] ROI = clsψ det , B, Regrψ det , B

[0035] Among them, ψ det is the power operation equipment image collected by the camera, B is the anchor box, ROI is the multiple regions of interest for power operation equipment tracking, clsψ det , B represents classification, and Regrψ det , B represents regression.

[0036] Furthermore, the input tensor for power operation equipment tracking is a multi-dimensional array that contains data of multiple regions of interest (ROIs) extracted from the image. Arranging the multiple regions of interest as a tensor provides a standardized data format, making it more convenient for various computer vision algorithms and models to process and analyze.

[0037] Furthermore, traditional two-dimensional convolutional neural networks (2DCNNs) are mainly used to process static images, while power operation equipment tracking tasks usually require processing dynamic scenes, such as the movement and changes of equipment. Three-dimensional convolutional neural networks (3DCNNs) can perform convolutional operations in both spatial and temporal dimensions, thereby capturing the dynamic features and time-series changes of equipment. This method is more suitable for processing video streams or time-series data than a pure 2D CNN, which helps to more accurately identify and track changes in equipment. 3D CNN can extract features at different levels from the original input tensor through multiple layers of convolution and pooling operations. The initial layer of convolution can extract low-level features such as edges and textures, while deeper convolution layers can extract more complex high-level features such as the shape, position changes, and motion patterns of equipment. This hierarchical feature extraction can help the model to more comprehensively understand and represent the features of power equipment, thereby improving the tracking accuracy. Specifically, each layer of the three-dimensional convolutional neural network for power operation equipment tracking performs the following operations on the input data during the forward pass of the layer: the convolutional units of each layer of the three-dimensional convolutional neural network for power operation equipment tracking perform convolution processing based on a convolution kernel on the input data to obtain a convolutional feature map; the pooling units of each layer of the three-dimensional convolutional neural network for power operation equipment tracking perform pooling processing along the channel dimension on the convolutional feature map to obtain a pooled feature map; and, the activation units of each layer of the three-dimensional convolutional neural network for power operation equipment tracking perform non-linear activation on the feature values at each position in the pooled feature map to obtain an activation feature map; wherein, the output of the last layer of the three-dimensional convolutional neural network for power operation equipment tracking is the distance feature matrix, and the input data of the three-dimensional convolutional neural network for power operation equipment tracking is the input tensor for power operation equipment tracking.

[0038] Figure 3 Block diagram of the power operation equipment temperature feature extraction unit in the 5G communication base station power operation monitoring system according to an embodiment of the present application. As Figure 3 shown, in a specific embodiment of the present application, the power operation equipment temperature feature extraction unit 122 includes: a power operation equipment temperature value arrangement sub-unit 1221, configured to arrange the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor according to the time dimension to obtain a power operation equipment temperature input vector; a power operation equipment temperature correlation matrix calculation sub-unit 1222, configured to multiply the power operation equipment temperature input vector by its transpose to obtain a power operation equipment temperature correlation matrix; and a power operation equipment temperature convolutional encoding sub-unit 1223, configured to pass the power operation equipment temperature correlation matrix through a power operation equipment temperature feature encoder based on a convolutional neural network model to obtain the power operation equipment temperature feature vector.

[0039] It should be understood that the temperature data of power equipment often exhibits time - series characteristics. The operating temperature of the equipment changes over time, reflecting information such as the equipment's load, operating status, and environmental conditions. Arranging the temperature values in the time dimension can preserve the sequential information in the time series, enabling subsequent analysis to consider the temperature change trend and periodicity. This dynamic change over time is crucial for predicting equipment failures, maintenance requirements, and performance changes.

[0040] Furthermore, the temperature input vector contains temperature values recorded at different time points. By multiplying this vector by its transpose, the similarity between temperature data can be calculated. This calculation generates a matrix, where each element represents the correlation between temperature values at different time points. This helps to understand how the temperatures of the equipment at different time points affect each other, thereby revealing the temperature change pattern. There may be a certain time - dependence between the temperatures of power equipment at different time points. By generating the temperature correlation matrix, this time - dependence can be effectively captured.

[0041] Even further, the temperature correlation matrix is a two - dimensional matrix that contains the similarity of temperature data between different time points. Directly extracting features from this matrix can be very complex. A Convolutional Neural Network (CNN) can automatically extract local features and patterns in the matrix through its convolutional layers, such as periodic fluctuations, temperature anomalies, etc. These deep - level features are crucial for understanding the operating status of the equipment and detecting potential problems. The temperature correlation matrix is usually high - dimensional, and its dimension is proportional to the number of time points. The CNN can effectively process these high - dimensional data through convolutional operations and pooling layers. The convolutional layer extracts useful features by applying filters (convolution kernels), and the pooling layer helps to reduce the dimension of the feature map, thereby reducing the computational complexity and retaining important information. Specifically, each layer of the temperature feature encoder of the power operating equipment based on the convolutional neural network model performs convolutional processing, mean pooling processing based on the local feature matrix, and non - linear activation processing on the input data respectively during the forward pass of the layer to output the temperature feature vector of the power operating equipment by the last layer of the temperature feature encoder of the power operating equipment based on the convolutional neural network model, where the input of the temperature feature encoder of the power operating equipment based on the convolutional neural network model is the temperature correlation matrix of the power operating equipment.

[0042] In a specific embodiment of the present application, the power operation equipment temperature correlation matrix calculation sub-unit includes: introducing the Log4j library for building the tool Gradle to add Log4j dependencies; creating a Log4j configuration file for configuring the log level, output format, and log file location; initializing Log4j to ensure the normal operation of the log system; defining the class TemperatureMatrixCalculator, writing code to implement the function of calculating the power operation equipment temperature correlation matrix, and inserting Log4j log records in key steps for monitoring and debugging during runtime; packaging the written code and deploying it to the production environment, and ensuring that the Log4j configuration file is deployed together with the application for the normal operation of the log recording function; running the application in the production environment, and monitoring and debugging by viewing the log files generated by Log4j.

[0043] In this way, the Log4j mechanism is used to deploy and implement multiplying the power operation equipment temperature input vector by its transpose to obtain the power operation equipment temperature correlation matrix. Among them, the log recording function of Log4j can help monitor and debug this function in the production environment to ensure the stability and reliability of the system.

[0044] Among them, some example codes are as follows.

[0045]

[0046]

[0047] Among them, some codes of the og4j configuration file example (log4j.properties) are as follows.

[0048]

[0049]

[0050] It should be understood that through the Log4j logging system, it is ensured that the function can be monitored and debugged in real time in the production environment, thereby improving the stability and reliability of the system. Among them, Log4j is configured through the log4j.properties file, the logging level is set to INFO, and console and file outputs are configured. Console output is used for real-time monitoring, and file output is used for long-term recording and analysis. When the application starts, use PropertyConfigurator.configure("log4j.properties") to load the Log4j configuration file to ensure the normal operation of the logging system. Specifically, define the TemperatureMatrixCalculator class, which contains the calculateTemperatureMatrix method for calculating the temperature correlation matrix. This method receives a temperature input vector, calculates the product of each element through a double loop, and stores the result in a two-dimensional array. More specifically, in the calculateTemperatureMatrix method, Log4j is used to record the start and end of the calculation logs, as well as the specific values of each calculation, for easy monitoring and debugging. Among them, in the main method, configure Log4j, initialize the logger, define an example temperature input vector, and call the calculateTemperatureMatrix method to calculate the temperature correlation matrix. Finally, print the calculation result and record the log.

[0051] In this way, by using Log4j to record the logs of key steps, it is ensured that the function of "multiplying the temperature input vector of the power operation equipment by its transpose to obtain the temperature correlation matrix of the power operation equipment" can be monitored in real time in the production environment. Through the console output, the calculation process and results can be viewed in real time, and problems can be discovered and solved in a timely manner. Through the file output, the logs are recorded in the temperature_matrix.log file, which is convenient for long-term recording and analysis. This is very useful for system maintenance and troubleshooting. By analyzing the log file, the root cause of the problem can be traced and the system performance can be optimized. Inserting Log4j logging in key steps facilitates debugging at runtime. By viewing the log file, the specific values of each calculation step can be understood in detail, which helps developers quickly locate and fix problems. Through the logging function of Log4j, the stability and reliability of the system are ensured. In the production environment, abnormal situations are discovered and handled in a timely manner to avoid system crashes or data loss, and the availability of the system and user satisfaction are improved.

[0052] By using the Log4j mechanism to deploy and implement the function of "multiplying the input vector of the power operation equipment temperature by its transpose to obtain the power operation equipment temperature correlation matrix", we have achieved multiple effects such as real-time monitoring, long-term recording, debugging support, and improved system stability, providing strong technical support for the power operation monitoring of 5G communication base stations.

[0053] In a specific embodiment of the present application, the power system parameter data feature extraction unit 124 includes: obtaining multiple power system parameter data text feature vectors by passing the power system parameter data collected by the sensor through the power system parameter data text context semantic encoder; splicing the multiple power system parameter data text feature vectors into a power system parameter data associated semantic feature vector.

[0054] It should be understood that the data collected by the sensor may contain a large amount of high-dimensional information. By converting this data into feature vectors, the text context semantic encoder can effectively reduce the dimension of the data while retaining key information. This conversion not only reduces the computational burden but also improves the efficiency of data processing and analysis. The text context semantic encoder can extract the implicit features and patterns in the data by converting the power system parameter data into feature vectors. These feature vectors have high expressive power and can better describe the operating state and potential problems of the system. Specifically, perform word segmentation on the power system parameter data collected by the sensor to obtain a power system parameter word sequence; use the embedding layer of the power system parameter data text context semantic encoder to map each power system parameter word in the power system parameter word sequence into a word embedding vector to obtain a sequence of power system parameter word embedding vectors; use the Transformer-based Bert model of the power system parameter data text context semantic encoder to perform global context semantic encoding on the sequence of power system parameter word embedding vectors to obtain multiple power system parameter data text feature vectors.

[0055] Furthermore, each power system parameter data text feature vector usually only reflects the features of a specific aspect, such as a certain dimension of voltage, current, or power. By splicing these feature vectors, information from different dimensions can be fused into a unified feature representation. This fusion enables the final associated semantic feature vector to comprehensively consider the multi-faceted features of the system parameters, thus providing a more comprehensive description of the system state.

[0056] In the above 5G communication base station power operation monitoring system 100, the power operation analysis module 130 is used to determine whether to issue an early warning of abnormal power operation of the 5G communication base station based on the multi-modal information management feature vector of the power operation equipment and the associated semantic feature vector of the power system parameter data. It should be understood that the multi-modal information management feature vector of the power operation equipment usually includes sensor data, equipment status information, historical operation records, etc. The fusion of these information can provide a comprehensive understanding of the equipment operation status. The associated semantic feature vector of the power system parameter data can reveal the complex relationships and interactions between parameters through context correlation analysis of different power system parameter information. By combining multi-modal information and associated semantic features, the system can provide more accurate abnormal detection results and more reliable data support for decision-makers. This helps to formulate targeted maintenance strategies, optimize equipment management, and ensure the stable operation of the 5G communication base station.

[0057] Figure 4 It is a block diagram of the power operation analysis module in the 5G communication base station power operation monitoring system according to an embodiment of the present application. As Figure 4 shown, in a specific embodiment of the present application, the power operation analysis module 130 includes: a communication base station power operation feature fusion unit 131, which is used to fuse the multi-modal information management feature vector of the power operation equipment and the associated semantic feature vector of the power system parameter data to obtain a communication base station power operation abnormal judgment feature vector; a communication base station power operation feature optimization unit 132, which is used to perform prior response-based feature interaction mode reconstruction adaptation on the communication base station power operation abnormal judgment feature vector to obtain an optimized communication base station power operation abnormal judgment feature vector; a power operation abnormal warning judgment unit 133, which is used to pass the optimized communication base station power operation abnormal judgment feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether to issue an early warning of abnormal power operation of the 5G communication base station.

[0058] It should be understood that the power system of the communication base station needs to be monitored in real time to ensure its stable operation. Fusing different feature vectors provides more comprehensive status information, enabling the system to quickly detect abnormalities in the real-time data stream and issue early warnings. This efficient real-time monitoring ability is the key to ensuring the continuous and stable operation of the communication base station. By comprehensively analyzing the specific status of the equipment and the overall operation of the system, more accurate maintenance strategies and fault handling solutions can be formulated, improving the overall management efficiency.

[0059] In particular, the technical solution of the present application involves multiple data sources, including power equipment images, temperature data, and power system parameters, etc. The feature dimensions of these data are inconsistent. Considering that image data usually has a relatively high dimension and may contain thousands of pixel information, while the parameter data of the power system usually has a relatively low dimension (such as temperature, current, voltage, etc.). Since these data sources themselves come from different physical domains (such as image features and sensor data), their distributions and structures in the feature space may be inconsistent. The mismatch of the feature space structure may lead to insufficient representation of data in the model, thereby affecting the accuracy of task objectives (such as anomaly warning). For example, the image features and temperature data of power equipment may have different contribution weights for anomaly judgment. If the feature space is not reasonably optimized, it may lead to the neglect or misjudgment of some important features by the model, thereby affecting the accuracy of the classifier. Therefore, in the technical solution of the present application, the feature interaction pattern of the feature vector for judging the power operation anomaly of the communication base station is reconstructed and adapted based on the prior response to obtain an optimized feature vector for judging the power operation anomaly of the communication base station.

[0060] Among them, reconstructing and adapting the feature interaction pattern of the feature vector for judging the power operation anomaly of the communication base station based on the prior response to obtain an optimized feature vector for judging the power operation anomaly of the communication base station includes: extracting the parametric prior basis matrix of the power operation of the communication base station; performing core prior information feature extraction on the parametric prior basis matrix of the power operation of the communication base station to obtain a set of core prior information feature spectral basis element embedding vectors of the power operation model of the communication base station; constructing a core prior information bilinear association matrix of the power operation of the communication base station between the feature vector for judging the power operation anomaly of the communication base station and each core prior information feature spectral basis element embedding vector in the set of core prior information feature spectral basis element embedding vectors of the power operation model of the communication base station to obtain a set of core prior information bilinear association matrices of the power operation model of the communication base station; calculating the prior information response manifold alignment significance factor of the core prior information bilinear association matrix of the power operation of the communication base station in the set of core prior information bilinear association matrices of the power operation model of the communication base station to obtain a set of prior information response manifold alignment significance factors of the power operation of the communication base station; based on the set of prior information response manifold alignment significance factors of the power operation of the communication base station, performing low-rank constraint fusion on the set of core prior information bilinear association matrices of the power operation model of the communication base station to obtain a prior information response projection coding matrix of the power operation model of the communication base station; mapping the feature vector for judging the power operation anomaly of the communication base station to the feature space of the prior information response projection coding matrix of the power operation model of the communication base station to obtain the optimized feature vector for judging the power operation anomaly of the communication base station.

[0061] Among them, reconstructing and adapting the feature interaction pattern of the communication base station power operation anomaly judgment eigenvector based on prior response to obtain an optimized communication base station power operation anomaly judgment eigenvector includes:

[0062] First, extract the parametric prior basis matrix of the communication base station power operation. It should be understood that extracting the parametric prior basis matrix of the communication base station power operation is not just an information acquisition process, but actually a key operation to structure and computabilize prior knowledge. The principle is to condense domain expertise, model design concepts, or macroscopic laws inverted from data into the mathematical form of the parametric prior basis matrix of the communication base station power operation for the effective utilization of subsequent algorithms.

[0063] Then, perform core prior information feature extraction on the parametric prior basis matrix of the communication base station power operation to obtain a set of core prior information feature spectral basis element embedding vectors of the communication base station power operation model, which is expressed by the core prior information feature extraction formula as:

[0064]

[0065] where U represents the set of core prior information feature spectral basis element embedding vectors of the communication base station power operation model, v1, v2, v m represent the first, second, and mth core prior information feature spectral basis element embedding vectors of the communication base station power operation model respectively, T represents the transpose operation, CoreExtraction represents core prior information feature extraction, M p represents the parametric prior basis matrix of the communication base station power operation, Λ represents the diagonal matrix, and λ1, λ m represent the values at the first and mth positions on the diagonal of the diagonal matrix respectively. It should be understood that the parametric prior basis matrix of the communication base station power operation may contain redundant or high-dimensional information, and direct application may lead to excessive computational burden and interference of non-critical information in the optimization process. Therefore, the principle of core prior information feature extraction is to denoise and refine prior knowledge, just like the filtering process in signal processing, retaining the main components and filtering out the redundancy.

[0066] Next, construct a core prior information bilinear correlation matrix of the communication base station power operation model between the communication base station power operation anomaly judgment eigenvector and each core prior information feature spectral basis element embedding vector in the set of core prior information feature spectral basis element embedding vectors of the communication base station power operation model to obtain a set of core prior information bilinear correlation matrices of the communication base station power operation model, which is expressed by the bilinear correlation formula as:

[0067]

[0068] Among them, x o represents the eigenvector for judging the abnormal power operation of the communication base station, l i x o represents the linear transformation of x o After the linear transformation, the eigenvector has the same characteristic scale as the corresponding core prior information eigen-spectrum basis element embedding vector of the communication base station power operation model, v i represents the i-th core prior information eigen-spectrum basis element embedding vector of the communication base station power operation model, represents matrix multiplication, L represents the length of the core prior information eigen-spectrum basis element embedding vector of the communication base station power operation model, MR i represents the i-th core prior information bilinear correlation matrix of the communication base station power operation model. It should be understood that the core of this step lies in constructing the interaction relationship between the eigenvector for judging the abnormal power operation of the communication base station and the refined prior knowledge. Specifically, through the implicit space mapping, the non-linear response mode of the eigenvector for judging the abnormal power operation of the communication base station to the prior knowledge in different aspects is learned. Essentially, the core prior information bilinear correlation matrix of the communication base station power operation model is a re-encoding of the eigenvector for judging the abnormal power operation of the communication base station from the perspective of prior knowledge, integrating the interpretation and processing of knowledge. Its function exceeds information association, realizing the directional enhancement of feature representation and the extraction of multi-perspective feature information.

[0069] Immediately afterwards, calculate the communication base station power operation prior information response manifold alignment significance factor of each communication base station power operation model core prior information bilinear correlation matrix in the set of the communication base station power operation model core prior information bilinear correlation matrices to obtain a set of communication base station power operation prior information response manifold alignment significance factors, which is expressed by the manifold alignment significance formula as:

[0070] S i = |MR i | F

[0071] Among them, |·| F represents the F-norm of the matrix, S iDenote the significance factor of the manifold alignment of the prior information response of the power operation of the i-th communication base station. It should be understood that the principle of this step is to refine the key information and streamline the feature representation of the bilinear correlation matrix of the core prior information of the power operation model of each communication base station. Just like generating an information summary, the most representative summary or feature vector that can represent the core information is extracted. The significance factor of the manifold alignment of the prior information response of the power operation of the communication base station should be representative and discriminative. It contains the ideas of feature selection and feature aggregation, selects the most informative parts for retention, and aggregates them into a concise manifold alignment significance factor to achieve deeper compression and refinement. The function of the significance factor of the manifold alignment of the prior information response of the power operation of the communication base station is not only information compression, but more importantly, to improve the efficiency and robustness of the subsequent fusion process, reduce the data dimension, and reduce the computational burden, especially in the case of high-dimensional matrices.

[0072] Subsequently, based on the set of the significance factors of the manifold alignment of the prior information response of the power operation of the communication base station, perform low-rank constrained fusion on the set of the bilinear correlation matrices of the core prior information of the power operation model of the communication base station to obtain the prior information response projection coding matrix of the power operation model of the communication base station, which is expressed by the low-rank constrained fusion formula as:

[0073]

[0074] Among them, softmax represents the normalized exponential function, and P represents the prior information response projection coding matrix of the power operation model of the communication base station. It should be understood that the core principle of the low-rank constrained fusion lies in emphasizing the adaptive and selective prior information integration strategy. Sparsity reflects that not all prior information responses are equally important. The fusion should be selective, focusing on more important responses, weakening or ignoring unimportant responses, improving the feature selection ability and generalization ability, and avoiding overfitting. Dynamics means that the weights or methods of the low-rank constrained fusion are not fixed, but are adaptively adjusted according to the input data or the model state, improving the flexibility and adaptability of the model. The essence of performing low-rank constrained fusion on the set of the bilinear correlation matrices of the core prior information of the power operation model of the communication base station is to optimally combine the response information from different prior knowledge perspectives to form the prior information response projection coding matrix of the power operation model of the communication base station that comprehensively reflects the model prior response.

[0075] Finally, map the power operation anomaly judgment feature vector of the communication base station to the feature space of the prior information response projection coding matrix of the power operation model of the communication base station to obtain the optimized power operation anomaly judgment feature vector of the communication base station, which is expressed by the mapping formula as:

[0076]

[0077] where, x optRepresents the optimized communication base station power operation anomaly judgment feature vector. It should be understood that finally, by mapping the communication base station power operation anomaly judgment feature vector to the feature space of the prior information response projection coding matrix of the communication base station power operation model, the optimized communication base station power operation anomaly judgment feature vector is obtained, and the entire adaptation process is completed. The principle of this step is to utilize the feature space defined by the prior information response projection coding matrix of the communication base station power operation model and project the communication base station power operation anomaly judgment feature vector into this space. Substantially, the prior information response projection coding matrix of the communication base station power operation model acts as a transformation matrix to perform linear or non-linear mapping transformation on the communication base station power operation anomaly judgment feature vector, enabling it to be embedded in the feature space integrating model prior information. The optimized communication base station power operation anomaly judgment feature vector better conforms to the constraints and guidance of the model prior, and realizes boundary adaptation on the feature manifold. Compared with the communication base station power operation anomaly judgment feature vector, the optimized communication base station power operation anomaly judgment feature vector is usually significantly improved in terms of expression ability, discriminability, and generalization, providing a better feature representation for subsequent machine learning tasks.

[0078] Furthermore, the optimized communication base station power operation anomaly judgment feature vector usually consists of features in multiple dimensions, including real-time data of devices, historical operation records, environmental parameters, etc. The fusion of this multi-dimensional information can provide a comprehensive understanding of the base station operation status. However, the complexity of these data makes manual analysis and interpretation difficult. As an automated data processing tool, a classifier can effectively identify abnormal patterns from these complex feature vectors. In practical applications, when the optimized communication base station power operation anomaly judgment feature vector is input into the classifier, the classifier can make predictions based on the knowledge learned during its training process to determine whether the current state belongs to the abnormal category. The classifier can use algorithms to perform complex pattern recognition and data analysis to identify potential abnormal situations, thereby reducing the risks of false alarms and missed alarms. By training and optimizing the classifier, its high stability and reliability can be ensured in different operating environments and situations.

[0079] In summary, the embodiment of this application first obtains the power operation device images collected by the camera, the power operation device temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor, then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally obtains the classification result through the classifier to determine whether to issue an early warning for the power operation anomaly of the 5G communication base station, so as to timely discover the power operation anomaly, improve the reliability and maintenance efficiency of the 5G base station, and reduce the downtime and maintenance costs.

[0080] As described above, the 5G communication base station power operation monitoring system 100 according to the embodiments of the present application can be implemented in various terminal devices. In one example, the 5G communication base station power operation monitoring system 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the 5G communication base station power operation monitoring system 100 can be a software module in the operating system of the terminal device, or can be an application program developed for the terminal device; of course, the 5G communication base station power operation monitoring system 100 can also be one of the many hardware modules of the terminal device.

[0081] Alternatively, in another example, the 5G communication base station power operation monitoring system 100 and the terminal device can also be separate devices, and the 5G communication base station power operation monitoring system 100 can be connected to the terminal device through a wired and / or wireless network, and transmit and interact information according to a predefined data format.

[0082] Figure 5 FIG. is a flowchart of a 5G communication base station power operation monitoring method according to an embodiment of the present application. As Figure 5 shown, the 5G communication base station power operation monitoring method according to the embodiments of the present application includes: S110, obtaining power operation device images collected by a camera, power operation device temperature values at multiple predetermined time points collected by a temperature sensor, and power system parameter data collected by a sensor; S120, extracting a power operation device multimodal information management feature vector and a power system parameter data associated semantic feature vector from the power operation device images collected by the camera, the power operation device temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor; S130, based on the power operation device multimodal information management feature vector and the power system parameter data associated semantic feature vector, determining whether to issue a 5G communication base station power operation anomaly warning.

[0083] Here, those skilled in the art can understand that the specific operations of each step in the above 5G communication base station power operation monitoring method have been described in detail in the above description of Figures 1 to 4 the 5G communication base station power operation monitoring system, and therefore, the repeated description thereof will be omitted.

Claims

1. A 5G communication base station power operation monitoring system, characterized in that, Including: A 5G communication base station power operation data acquisition module, which is used to acquire power operation equipment images collected by a camera, power operation equipment temperature values at multiple predetermined time points collected by a temperature sensor, and power system parameter data collected by a sensor; A 5G communication base station power operation data processing module, which is used to extract a power operation equipment multimodal information management feature vector and a power system parameter data associated semantic feature vector from the power operation equipment images collected by the camera, the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor; A power operation analysis module, which is used to determine whether to issue a 5G communication base station power operation anomaly warning based on the power operation equipment multimodal information management feature vector and the power system parameter data associated semantic feature vector.

2. The 5G communication base station power operation monitoring system according to claim 1, characterized in that, The 5G communication base station power operation data processing module includes: A power operation equipment image feature extraction unit, which is used to extract features from the power operation equipment images collected by the camera to obtain a power operation equipment tracking association feature vector; A power operation equipment temperature feature extraction unit, which is used to extract features from the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor to obtain a power operation equipment temperature feature vector; A power operation equipment multimodal feature association unit, which is used to associate the power operation equipment tracking association feature vector and the power operation equipment temperature feature vector to obtain the power operation equipment multimodal information management feature vector; A power system parameter data feature extraction unit, which is used to extract features from the power system parameter data collected by the sensor to obtain the power system parameter data associated semantic feature vector.

3. The 5G communication base station power operation monitoring system according to claim 2, characterized in that, The power operation equipment image feature extraction unit includes: Passing the power operation equipment images collected by the camera through a power operation equipment image target detection network model to obtain multiple power operation equipment tracking regions of interest; Arranging the multiple power operation equipment tracking regions of interest into a power operation equipment tracking input tensor; Passing the power operation equipment tracking input tensor through a power operation equipment tracking three-dimensional convolutional neural network to obtain the power operation equipment tracking association feature vector.

4. The 5G communication base station power operation monitoring system according to claim 3, wherein, The power operation equipment temperature feature extraction unit includes: A power operation equipment temperature value arrangement sub-unit, which is used to arrange the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor according to the time dimension to obtain a power operation equipment temperature input vector; A power operation equipment temperature association matrix calculation sub-unit, which is used to multiply the power operation equipment temperature input vector by its transpose to obtain a power operation equipment temperature association matrix; A power operation equipment temperature convolutional encoding sub-unit, which is used to pass the power operation equipment temperature association matrix through a power operation equipment temperature feature encoder based on a convolutional neural network model to obtain the power operation equipment temperature feature vector.

5. The 5G communication base station power operation monitoring system according to claim 4, wherein The power operation equipment temperature association matrix calculation sub-unit includes: Introducing the Log4j library, which is used to build the tool Gradle to add the Log4j dependency; Create a Log4j configuration file to configure the log level, output format, and log file location; Initialize Log4j to ensure that the logging system can work properly; Define the class TemperatureMatrixCalculator, write code to implement the function of calculating the temperature correlation matrix of power operation equipment, and insert Log4j log records in key steps for monitoring and debugging during runtime; Package the written code and deploy it to the production environment, and ensure that the Log4j configuration file is deployed together with the application for the normal operation of the logging function; Run the application in the production environment, and monitor and debug by viewing the log file generated by Log4j.

6. The 5G communication base station power operation monitoring system according to claim 5, characterized in that, The power system parameter data feature extraction unit includes: Pass the power system parameter data collected by the sensor through the power system parameter data text context semantic encoder to obtain multiple power system parameter data text feature vectors; Concatenate the multiple power system parameter data text feature vectors into a power system parameter data associated semantic feature vector.

7. The 5G communication base station power operation monitoring system according to claim 6, wherein, The power operation analysis module includes: A communication base station power operation feature fusion unit for fusing the power operation equipment multimodal information management feature vector and the power system parameter data associated semantic feature vector to obtain a communication base station power operation anomaly judgment feature vector; A communication base station power operation feature optimization unit for performing a priori response-based feature interaction mode reconstruction adaptation on the communication base station power operation anomaly judgment feature vector to obtain an optimized communication base station power operation anomaly judgment feature vector; A power operation anomaly warning judgment unit for passing the optimized communication base station power operation anomaly judgment feature vector through a classifier to obtain a classification result, and the classification result is used to judge whether to issue a 5G communication base station power operation anomaly warning.

8. The 5G communication base station power operation monitoring system according to claim 7, characterized in that, The communication base station power operation feature optimization unit includes: Extract the communication base station power operation parameterized prior basis matrix; Perform core prior information feature extraction on the communication base station power operation parameterized prior basis matrix to obtain a set of communication base station power operation model core prior information feature spectrum basis element embedding vectors; Construct a communication base station power operation model core prior information bilinear association matrix between the communication base station power operation anomaly judgment feature vector and each communication base station power operation model core prior information feature spectrum basis element embedding vector in the set of communication base station power operation model core prior information feature spectrum basis element embedding vectors to obtain a set of communication base station power operation model core prior information bilinear association matrices; Calculate the communication base station power operation prior information response manifold alignment significance factor of each communication base station power operation model core prior information bilinear association matrix in the set of communication base station power operation model core prior information bilinear association matrices to obtain a set of communication base station power operation prior information response manifold alignment significance factors; Based on the set of significant degree factors for manifold alignment of the prior information response of the communication base station power operation, perform low-rank constraint fusion on the set of bilinear correlation matrices of the core prior information of the communication base station power operation model to obtain the prior information response projection coding matrix of the communication base station power operation model; Map the communication base station power operation anomaly judgment feature vector to the feature space of the prior information response projection coding matrix of the communication base station power operation model to obtain the optimized communication base station power operation anomaly judgment feature vector.

9. A power operation monitoring method for a 5G communication base station, characterized in that Including: Obtain the power operation equipment images collected by the camera, the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor; Extract the power operation equipment multimodal information management feature vector and the power system parameter data associated semantic feature vector from the power operation equipment images collected by the camera, the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor; Based on the power operation equipment multimodal information management feature vector and the power system parameter data associated semantic feature vector, determine whether to issue an early warning for the 5G communication base station power operation anomaly.

10. The 5G communication base station power operation monitoring method according to claim 9, characterized in that, Extracting the power operation equipment multimodal information management feature vector and the power system parameter data associated semantic feature vector from the power operation equipment images collected by the camera, the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor, and the power system parameter data collected by the sensor includes: Perform feature extraction on the power operation equipment images collected by the camera to obtain the power operation equipment tracking association feature vector; Perform feature extraction on the power operation equipment temperature values at multiple predetermined time points collected by the temperature sensor to obtain the power operation equipment temperature feature vector; Associate the power operation equipment tracking association feature vector and the power operation equipment temperature feature vector to obtain the power operation equipment multimodal information management feature vector; Perform feature extraction on the power system parameter data collected by the sensor to obtain the power system parameter data associated semantic feature vector.

Images