An Aging Test Device and Method for a POE Switch Used for Security Protection

Abstract

The present invention relates to the technical field of switch aging testing, and discloses an aging testing device and method for a POE switch for security protection. The method includes: fixing the POE switch to be tested on a test platform, applying standard loads to the POE power supply ports respectively through a load simulator and obtaining power supply parameters and port states to obtain initial test data; classifying to generate a hierarchical test benchmark matrix; respectively performing a full-load test on the POE ports, a port isolation data flow test, and an auto-negotiation interaction test, and collecting a full-scale test data set in a hierarchical and progressive manner; performing PSE power supply characteristic calculation, MAC address aging characteristic calculation, and port performance characteristic calculation to generate multiple device state characteristic vectors; calculating the device deterioration trend through an aging prediction model set to obtain a device deterioration prediction index. The present invention realizes the correlation analysis of multi-dimensional performance parameters of the device, and enhances the adaptability and prediction accuracy of the testing method in extreme environments.

Description

Technical Field

[0001] The present invention relates to the technical field of switch aging testing, and particularly to an aging testing device and method for a POE switch used for security. Background Art

[0002] With the rapid development and popularization of security systems, as the core device for power supply and network transmission of security equipment, the reliability and stability of POE switches directly affect the operation of the entire security system. Traditional POE switch testing methods mainly focus on the verification of single performance parameters, such as power supply capacity testing or network performance testing, lacking comprehensive performance evaluation of the device in complex application environments.

[0003] Currently, there are problems in the aging testing of POE switches, such as incomplete test coverage, strong independence of test parameters, and insufficient analysis of the correlation of test data. Especially in complex application scenarios such as dual-protocol support of IEEE802.3af / at standards, full-port full-load power supply, and port isolation, it is difficult to accurately evaluate the aging trend of the device and predict the reliability of the device, resulting in unpredictable failures that may occur during the actual operation of the device. In addition, the existing POE switch testing methods lack intelligence and self-adaptability, unable to dynamically adjust the test strategy according to the real-time state of the device, nor to deeply mine and analyze multi-dimensional test data, making it difficult to establish an accurate device life prediction model, affecting the long-term stable operation of the security system. Summary of the Invention

[0004] The present invention provides an aging testing device and method for a POE switch used for security. The present invention realizes the correlation analysis of multi-dimensional performance parameters of the device, enhancing the adaptability and prediction accuracy of the testing method in extreme environments.

[0005] In a first aspect, the present invention provides an aging testing method for a POE switch used for security. The aging testing method for the POE switch used for security includes:[[]]

[0006] Fix the POE switch to be tested on the test platform, and apply standard loads to the POE power supply ports respectively through a load simulator to obtain power supply parameters and port states, obtaining initial test data;

[0007] Classify the initial test data according to POE power supply parameters, port negotiation states, and temperature distribution data to generate a hierarchical test benchmark matrix;

[0008] Based on the hierarchical test benchmark matrix, respectively execute POE port full-load testing, port isolation data stream testing, and auto-negotiation interaction testing through an intelligent PSE controller, and collect a full set of test data in a hierarchical and progressive manner;

[0009] Perform PSE power supply characteristic calculation, MAC address aging characteristic calculation, and port performance characteristic calculation on the full-scale test data set to generate multiple device status feature vectors;

[0010] Input the multiple device status feature vectors into the aging prediction model set to perform device deterioration trend calculation, and obtain device deterioration prediction indicators.

[0011] In a second aspect, the present invention provides a POE switch aging test device for security protection. The POE switch aging test device for security protection includes:

[0012] An acquisition module, configured to fix the POE switch to be tested on a test platform, apply standard loads to the POE power supply ports respectively through a load simulator, and obtain power supply parameters and port statuses to obtain initial test data;

[0013] A classification module, configured to classify the initial test data according to POE power supply parameters, port negotiation statuses, and temperature distribution data to generate a hierarchical test benchmark matrix;

[0014] A test module, configured to perform a POE port full-load test, a port isolation data stream test, and an auto-negotiation interaction test respectively through an intelligent PSE controller based on the hierarchical test benchmark matrix, and collect a full-scale test data set in a hierarchical progressive manner;

[0015] A calculation module, configured to perform PSE power supply characteristic calculation, MAC address aging characteristic calculation, and port performance characteristic calculation on the full-scale test data set to generate multiple device status feature vectors;

[0016] A processing module, configured to input the multiple device status feature vectors into the aging prediction model set to perform device deterioration trend calculation, and obtain device deterioration prediction indicators.

[0017] In a third aspect of the present invention, there is provided a computer device, including: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor calls the instructions in the memory to enable the computer device to execute the above-mentioned POE switch aging test method for security protection.

[0018] In a fourth aspect of the present invention, there is provided a computer-readable storage medium, wherein instructions are stored in the computer-readable storage medium, and when it runs on a computer, it enables the computer to execute the above-mentioned POE switch aging test method for security protection.

[0019] In the technical solution provided by the present invention, through a hierarchical progressive test strategy, a comprehensive test of the power supply capacity, network performance, and environmental adaptability of the POE switch is achieved, improving the test coverage rate; by using an intelligent PSE controller and multi-dimensional data fusion technology, a device state feature vector is established, making the test results more objective and accurate; by introducing an aging prediction model set based on deep learning and combining with the weighted fusion mechanism of the AdaBoost algorithm, the accuracy of predicting the deterioration trend of the device is improved; through a dynamic test parameter adjustment mechanism, the test process can be adaptively optimized according to the device state, reducing the waste of test resources; a complete multi-modal feature sequence analysis system is established to realize the correlation analysis of multi-dimensional performance parameters of the device, providing a reliable basis for device maintenance decision-making; through a hierarchical early warning mechanism and temperature compensation strategy, the adaptability and prediction accuracy of the test method in extreme environments are enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0021] Figure 1 It is a schematic diagram of the steps of the aging test method for the POE switch for security in the embodiments of the present invention;

[0022] Figure 2 It is a schematic diagram of the structure of the aging test device for the POE switch for security in the embodiments of the present invention;

[0023] Figure 3 It is a schematic block diagram of the structure of the computer device in the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The embodiments of the present invention provide an aging test device and method for a POE switch for security. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above drawings are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0025] For ease of understanding, the specific process of the embodiment of the present invention will be described below. Please refer to Figure 1 , an embodiment of the aging test method for the POE switch for security in the embodiment of the present invention includes:

[0026] Step S1: Fix the POE switch to be tested on the test platform, apply standard loads to the POE power supply ports respectively through the load simulator, and obtain the power supply parameters and port states to obtain the initial test data;

[0027] It can be understood that the execution subject of the present invention can be an aging test device for the POE switch for security, or a terminal or a server, and specific limitations are not made here. In the embodiment of the present invention, the server is taken as the execution subject for illustration.

[0028] Specifically, fix the POE switch to be tested on a standard 13-inch rack to ensure the stability of the device during the test and avoid test errors caused by unstable positions. Use a scanner to number and identify the physical ports of the POE switch to be tested to obtain accurate position mapping data of the ports. Configure load simulators compliant with the IEEE802.3af / at standard for each of the 16 POE ports of the switch to be tested. The IEEE802.3af / at standard is a common standard for POE power supply. By simulating the power consumption requirements of different devices for the POE power supply ports, the load simulator ensures that the test conditions conform to the actual usage. After applying appropriate loads to each port through the load simulator, use a digital multimeter to collect the voltage values at the power supply ends of each port to obtain the initial voltage data of the power supply ports. Through the initial voltage data, understand the power supply status of the device under standard loads. Input the initial voltage data into the voltage regulation controller, and scan the output voltage of each POE port within the power range of 0 to 30W through PWM wave modulation. Test the voltage fluctuation of the POE switch ports under different loads to obtain the curve data of port power supply. Through the analysis of the power supply curve data, evaluate the power supply stability and reliability of each port, and then judge the working ability of the port under high-load conditions. Compare the power supply curve data of the ports with thresholds, and use a voltage comparator to set the power supply reference value and voltage deviation value for each POE port. In this way, obtain the power supply reference data of the ports. According to the power supply reference data, use the intelligent PSE controller to perform the Auto MDI / MDI-X automatic flip test on each POE port to obtain the port line pair identification data. MDI / MDI-X automatic flip refers to automatically switching the transmission mode of the network connection port. Through this test, it can be ensured that the POE port can correctly identify the type of connected device and optimize the connection stability of the network. Conduct a cross-matrix analysis on the port line pair identification data. Through Boolean algebraic operations, generate a cross-correlation table between ports, reflecting the state relationships between each port, and obtain the port status matrix. The port status matrix helps to evaluate the performance of each port under different network configurations. To test the port performance, apply a standard load of 15.4W to each POE port in the port status matrix on this basis. After applying the standard load, use temperature sensors to collect the temperature values of the power supply module, PSE controller, and Ethernet chip respectively to obtain the temperature distribution data of the target points. By monitoring the temperature changes of the device under load conditions, evaluate its thermal stability and potential heat dissipation problems. Integrate the initial voltage data of the power supply ports, the port power supply curve data, the port status matrix, and the temperature distribution data of the target points to obtain the initial test data.

[0029] Step S2: Classify the initial test data according to POE power supply parameters, port negotiation status, and temperature distribution data to generate a hierarchical test benchmark matrix;

[0030] Specifically, numerical classification is performed on the POE power supply parameters in the initial test data. The power supply is segmented by the 15.4W threshold of the IEEE802.3af standard and the 30W threshold of the IEEE802.3at standard to obtain interval data for different power supply levels. The interval data helps analyze the power supply performance under different loads and divides the power supply capabilities of devices into different grade intervals, providing a clear classification standard for subsequent tests and evaluations. Based on the interval data of the power supply levels, standardized calculations are performed on the power supply parameters of each POE port to obtain the power supply reference value of the port. Standardized calculations can eliminate the influence caused by hardware differences or different test conditions between different ports, enabling the power supply performance of each port to be compared under the same standard. The power supply reference values are arranged in the order of the port numbers, and a 16×3-dimensional power supply parameter matrix is generated through matrix operations. This matrix contains the power supply data of 16 POE ports under different load conditions, reflecting the stability and power distribution capabilities of each port under different working conditions, and obtaining the POE power supply reference data. Extract the Auto MDI / MDI-X status data in the port status matrix. The Auto MDI / MDI-X automatic flip function is used to automatically identify the network connection type to ensure correct connection between different devices. After extracting the status data of the ports, verify the negotiation capabilities of each port through the port isolation function to ensure that each port can correctly identify and adapt to different types of network devices. Through this step, the negotiation status data of each port is obtained, reflecting the adaptability of the switch in different network environments. Combine and analyze the port negotiation status data with the IEEE802.3X flow control status to evaluate the stability and data processing capabilities of the switch under complex conditions such as high load and flow control. Through the status mapping of these data, generate the negotiation reference parameters for each POE port to obtain the negotiation status reference data. Process the temperature distribution data of the key points regarding temperature in the initial test data. These data include the temperature data of the power module, PSE controller, and Ethernet chip, which directly affect the stability and lifespan of the device. By performing zonal statistics according to the positions of the temperature data, obtain the temperature distribution of each region, and calculate and generate the temperature distribution characteristics according to the temperature gradient, reflecting the thermal stability of the device under different working environments and helping to evaluate the reliability of the device under high or low temperature conditions, obtaining the temperature reference data. Arrange the POE power supply reference data, negotiation status reference data, and temperature reference data in the form of a two-dimensional matrix to form a hierarchical feature matrix. Through linear transformation of these data, unify different types of test data into a standard framework to obtain the test hierarchy data. Reorganize the structure of the test hierarchy data, and integrate the test parameters of the power supply layer, negotiation layer, and temperature layer through hierarchical mapping rules to obtain the hierarchical test reference matrix.

[0031] Step S3: Based on the hierarchical test benchmark matrix, the intelligent PSE controller respectively performs a full-load test on the POE port, a port isolation data stream test, and an auto-negotiation interaction test, and collects a full set of test data in a hierarchical and progressive manner;

[0032] Specifically, the first - layer test is performed based on the power - supply layer parameters in the hierarchical test benchmark matrix. The tests in this stage mainly focus on the load - bearing capacity and continuous power - supply performance of the POE ports. In the first - layer test, a 15.4W rated load is applied to a single POE port and power is supplied continuously for 8 hours to simulate the performance of the device under normal working loads. Through this test, the power - supply data of a single port under the rated load is obtained, and its stability and durability are evaluated. According to the single - port rated - load data, the first - layer progressive test is carried out. In this test, a larger load, the maximum load of 30W, is applied to a single POE port and power is supplied continuously for another 8 hours. The progressive test can evaluate the working ability of the device under higher loads and its response to load fluctuations. The first - layer full - load test is performed based on the single - port maximum - load data. A load with a total power of 240W is applied to all POE ports and power is supplied continuously for 24 hours. The long - term operation ability of the entire switch system under high - load conditions is comprehensively evaluated. Through the test, the full - port full - load test data is obtained to confirm the stability, power - supply efficiency and potential failure risks of the device under full - load conditions. Based on the performance - layer parameters in the hierarchical test benchmark matrix, the second - layer test is entered. In this stage, the network performance of the device is concerned, especially the data - flow isolation ability between ports. Through the port - isolation function, all ports are isolated to prevent signal interference between different ports, and the port - isolation configuration data is obtained. These data can help evaluate whether the data transmission between different ports of the switch is independent and avoid the risks of network congestion or data conflicts. After completing the port - isolation configuration test, the second - layer progressive test is carried out. Through the IEEE802.3X flow - control protocol, the concurrent data - transmission test is performed on all isolated ports to obtain the data - throughput test data, revealing the performance of the switch under high - concurrent transmission conditions, especially key indicators such as data throughput, latency and error rate. According to the temperature - layer parameters in the hierarchical test benchmark matrix, the third - layer test is entered. The working ability and stability of the device in different temperature environments are tested. In a low - temperature environment of - 10°C, 30% of the load of all ports is maintained for 4 hours to obtain the low - temperature test data and evaluate the start - up performance and long - term operation stability of the device under low - temperature conditions. According to the low - temperature test data, the progressive test of the third layer is carried out. In a high - temperature environment of 55°C, the same 30% load of all ports is maintained for 4 hours to obtain the high - temperature test data, evaluating the heat - dissipation performance, thermal stability and potential overheating problems of the device under high - temperature conditions, and helping to predict the long - term durability of the device. The full - port full - load test data, data - throughput test data, low - temperature test data and high - temperature test data are integrated according to the chronological relationship to obtain the full - volume test data set.

[0033] Step S4: Calculate the PSE power - supply characteristic, MAC - address aging characteristic and port - performance characteristic for the full - volume test data set to generate multiple device - status feature vectors;

[0034] Specifically, data normalization is performed on the voltage data in the full-port full-load test data. The voltage data is converted into a standardized form to eliminate numerical differences under different ports or test conditions, enabling consistent comparison of voltage fluctuations at each port. Based on the voltage fluctuation data under the 240W full-load condition, the value of voltage stability is calculated to obtain the power supply fluctuation characteristic data, which reflects the voltage fluctuation range of the POE switch under heavy load, thereby evaluating its power supply stability during long-term use. Analyze the POE port power supply parameters according to the power supply fluctuation characteristic data. By calculating the voltage ripple coefficient and power factor at the 52V power output terminal, the PSE power supply quality data is obtained, which reflects the power supply stability, efficiency, and comprehensive quality of the power supply system of the device under different load conditions. According to the data throughput test data, the MAC address learning rate test is performed on each POE port. By calculating the address update time in the IEEE802.3X full-duplex mode, the address processing rate data of each port is obtained, which reflects the efficiency of the switch in processing MAC address learning and thus affects its network traffic forwarding ability. According to the address processing rate data, statistical operations are performed on the MAC address aging time and failure probability of each POE port to obtain the address aging characteristic data. The MAC address aging time and failure probability are important indicators of the switch performance, directly affecting the address table management ability and data processing efficiency of the switch after long-term operation. Based on the low-temperature test data and high-temperature test data, the slope of the temperature rise curve of the power supply module, PSE controller, and Ethernet chip is calculated to obtain the temperature change characteristic data. The temperature rise curve reflects the thermal response characteristics of the device under different working environments. By calculating the slope of the temperature rise curve, the thermal stability of the device after long-term operation is evaluated. The test data under high-temperature and low-temperature conditions helps to understand the heat dissipation ability of the device and its operating conditions in extreme environments, ensuring the reliability of the device in various temperature environments. Perform a correlation analysis on the PSE power supply quality data and the temperature change characteristic data. By calculating the coupling coefficient between the power supply parameters and the temperature change, the power supply state characteristic vector is obtained, which reflects the degree of association between the power supply quality and the device temperature change, thereby evaluating the impact of temperature fluctuations on the device power supply state. Calculate the performance score of the address aging characteristic data. Based on the data forwarding efficiency and address table entry utilization rate in the port isolation mode, the port performance characteristic vector is obtained to evaluate the data processing ability of each port under different working conditions. The port performance characteristic vector can reflect the stability and data forwarding efficiency of the port after the switch has been running for a long time, especially under high network load. Combine the power supply state characteristic vector and the port performance characteristic vector, and generate multiple device state characteristic vectors according to the classification structure of power supply characteristics, aging characteristics, and performance characteristics.

[0035] Step S5: Input the multiple device status feature vectors into the aging prediction model set to calculate the device deterioration trend and obtain the device deterioration prediction index.

[0036] Specifically, the power supply status feature vector among multiple device status feature vectors is input into the PSE power supply weak classifier. This classifier consists of three main parts: a feature extraction layer, a time series analysis layer, and a status prediction layer. The feature extraction layer adopts a three-layer fully connected network structure. Feature dimensionality reduction is performed successively in each layer, with dimensions of 512, 256, and 128 respectively. This process is processed through the ReLU activation function and the Dropout layer to prevent overfitting and increase the robustness of the model. The time series analysis layer adopts a bidirectional LSTM (Long Short-Term Memory Network) structure. This layer contains 64 hidden units and is used to capture the dynamic features of power supply parameters changing over time. The bidirectional LSTM can handle the forward and backward dependencies in time series data and more accurately analyze the changing trend of the power supply status. The status prediction layer uses a Softmax classifier for classification output to obtain the PSE power supply degradation score, which reflects the degradation degree of the power supply system during the long-term operation of the device. During the processing of the PSE power supply weak classifier, through the dimensionality reduction of the three-layer fully connected network and the time series modeling of the bidirectional LSTM network, the time series features of the power supply status can be extracted. The port performance feature vector among multiple device status feature vectors is input into the port performance weak classifier. The port performance weak classifier consists of a data preprocessing layer, a feature enhancement layer, and a classification output layer. The data preprocessing layer adopts a batch normalization structure, which can standardize the input data to improve the training efficiency and stability of the model. The feature enhancement layer adopts a residual network structure. This structure has a three-layer stacking structure, with each layer containing two 3x3 convolutional layers and a shortcut connection to enhance the feature expression ability, prevent the problem of gradient disappearance, and accelerate training. Each convolutional layer is followed by a LeakyReLU activation function to further improve the non-linear expression ability of the model. Through this structure, the port performance time series features are extracted, and the classification output layer uses the cross-entropy loss function to output the port performance degradation score, measuring the performance decline degree of the switch port after long-term operation, including aspects such as data forwarding efficiency and processing ability. The temperature change feature data is input into the temperature feature weak classifier. This classifier consists of a temperature curve analysis layer, a trend extraction layer, and a result output layer. The temperature curve analysis layer is processed using a one-dimensional convolutional network. The number of convolutional kernels is set to 32, and the size of each convolutional kernel is 3. The feature dimension is compressed through the max pooling layer. This layer mainly extracts effective time series features from the temperature change data to reveal the thermal response of the device under different temperature conditions. The trend extraction layer adopts an attention mechanism. This mechanism can automatically focus on the most important time points in the temperature curve and perform weighted processing on the time series changes of the temperature data through the scaled dot-product structure to obtain the temperature change time series features. The result output layer calculates the temperature degradation score based on the extracted features, reflecting the performance change trend of the device under different temperature environments.The deterioration scores of PSE power supply, port performance, and temperature are weighted and fused through the AdaBoost algorithm. The AdaBoost algorithm is an ensemble learning method that improves the accuracy of the final model by combining the prediction results of multiple weak classifiers. In this application, the AdaBoost algorithm weights and fuses the deterioration scores in three aspects: PSE power supply, port performance, and temperature, to obtain a comprehensive device deterioration prediction index and evaluate the overall health status of the device during long-term operation.

[0037] Perform time series segmentation on the PSE time series characteristics, port performance time series characteristics, and temperature change time series characteristics to better process and analyze these data. Perform the first time series segmentation on the PSE time series characteristics to obtain PSE sequence data; perform the second time series segmentation on the port performance time series characteristics to obtain port sequence data; perform the third time series segmentation on the temperature change time series characteristics to obtain temperature sequence data. Perform unified sampling processing on the sequence data through timestamp alignment. Through the cubic spline interpolation algorithm, generate a multi-modal feature sequence with a unified sampling interval to obtain aligned feature data. Next, perform the first-stage feature fusion. By calculating the coupling coefficients between PSE power supply and temperature and between port performance and temperature, obtain the first fusion weight coefficients, which reflect the mutual influence between power supply, port performance, and temperature changes. Based on the fusion coefficients, perform weighted processing on the deterioration scores of PSE power supply and port performance. This process is implemented by the first-level classifier of the improved AdaBoost algorithm, where the weak classifier group with a temperature compensation factor is further used to optimize the weighting of the scores. The weighted scores form the deterioration index of the first stage, which reflects the overall deterioration trend of the device under the current conditions. Perform the second-stage feature fusion on the first-stage deterioration index and the temperature deterioration score, and calculate the second fusion weight coefficients through a multi-level weighting method with an adaptive temperature threshold. Automatically adjust the weights of different features according to the impact of actual temperature changes on device deterioration to obtain device deterioration prediction data. Based on the weighted calculation of the first-stage deterioration index and the second fusion weight coefficients, obtain the final device deterioration prediction index. To refine the prediction results, use a grading standard to calibrate the degree of device deterioration. Set the prediction results as mild deterioration for [0.3, 0.6), moderate deterioration for [0.6, 0.8), and severe deterioration for [0.8, 1.0] to obtain the device deterioration prediction index.

[0038] In the embodiments of the present invention, through a hierarchical progressive test strategy, a comprehensive test of the power supply capacity, network performance, and environmental adaptability of the POE switch is realized, improving the test coverage rate; by adopting an intelligent PSE controller and multi-dimensional data fusion technology, a device state feature vector is established, making the test results more objective and accurate; introducing an aging prediction model set based on deep learning and combining with the weighted fusion mechanism of the AdaBoost algorithm to improve the accuracy of predicting the degradation trend of the device; through a dynamic test parameter adjustment mechanism, the test process can be adaptively optimized according to the device state, reducing the waste of test resources; establishing a complete multi-modal feature sequence analysis system to realize the correlation analysis of multi-dimensional performance parameters of the device, providing a reliable basis for device maintenance decision-making; through a hierarchical early warning mechanism and temperature compensation strategy, enhancing the adaptability and prediction accuracy of the test method in extreme environments.

[0039] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0040] Fix the POE switch to be tested on a standard 13-inch rack, and use a scanner to number and identify the physical ports of the POE switch to be tested to obtain port position mapping data;

[0041] Configure load simulators compliant with the IEEE802.3af / at standard for 16 POE ports in the port position mapping data respectively, and use a digital multimeter to collect the voltage values at the 52V power output terminal to obtain the initial voltage data of the power supply ports;

[0042] Input the initial voltage data of the power supply ports into a voltage regulation controller, and perform power scanning in the range of 0-30W for the output voltage of each POE port through PWM wave modulation to obtain port power supply curve data;

[0043] Compare the port power supply curve data with thresholds, set the power supply reference value and voltage deviation value for each POE port through a voltage comparator to obtain port power supply reference data, and based on the port power supply reference data, perform an Auto MDI / MDI-X automatic flip test on each POE port through an intelligent PSE controller to obtain port line pair identification data;

[0044] Perform cross-matrix analysis on the port line pair identification data, generate a cross-correlation table between ports through Boolean algebra operations to obtain a port state matrix, and apply a standard load of 15.4W to each POE port in the port state matrix, and use a temperature sensor to collect the temperature values of the power supply module, PSE controller, and Ethernet chip to obtain the target point temperature distribution data;

[0045] Fuse the initial voltage data of the power supply port, the port power supply curve data, the port status matrix, and the target point temperature distribution data to obtain the initial test data.

[0046] Specifically, fix the POE switch to be tested on a standard 13-inch rack. As a standardized test platform, the 13-inch rack can ensure the stability and uniformity of the device during the test. Use a scanner to number and identify the physical ports of the POE switch when fixing the device. Automatically scan each port through the scanner to obtain the number of each port and its physical location on the switch, and generate port location mapping data. According to the 16 POE ports in the port location mapping data, configure load simulators that conform to the IEEE802.3af / at standard respectively. The load simulator simulates the current and power required by the POE port during actual use. By simulating the working states under different load conditions, evaluate the performance of the switch during actual operation. Measure the voltage value at the 52V power output terminal of each POE port with a digital multimeter to obtain the initial voltage data of the power supply port, reflecting the power supply state of the POE switch without load. Input the initial voltage data into the voltage regulation controller. The voltage regulation controller precisely controls the output voltage of each POE port by adopting PWM wave modulation technology. During the test, the controller performs a power scan in the range of 0 to 30W for each POE port to obtain the port power supply curve data. By scanning the voltage response at different powers, understand the power supply capacity and stability of the switch under different load conditions. Compare the port power supply curve data with thresholds. Through a voltage comparator, set the power supply reference value and voltage deviation value for each POE port. The power supply reference value represents the voltage value that the port should maintain under standard working conditions, while the voltage deviation value reflects the deviation between the actual voltage of the port and the standard voltage, which is specifically expressed by the following formula:

[0047] ;

[0048] Wherein, represents the voltage deviation, is the measured actual voltage value, is the standard voltage value. In this way, the power supply stability of the POE port is judged and compared with the standard value to generate port power supply reference data. Based on these power supply reference data, the intelligent PSE controller performs the Auto MDI / MDI-X automatic flip test. MDI / MDI-X refers to the automatic identification function of the network port, which judges the port connectivity and whether data is correctly exchanged. Through the automatic flip test, the reliability of the port connection is ensured, and port wire pair identification data is generated according to the test results. Cross matrix analysis is performed on the port wire pair identification data. Cross matrix analysis is a technology that generates a cross-correlation table between ports through Boolean algebraic operations. By analyzing the mutual relationship between different ports, a port status matrix is generated. This matrix can reflect the working status of each port, such as whether it is in the enabled state and whether there is signal interference. A standard load of 15.4W is applied to each POE port in the port status matrix. This is a load level based on the IEEE802.3af standard, which can simulate the typical load status of the switch during actual use. While applying the load, a temperature sensor is used to monitor the temperature values of the power module, PSE controller, and Ethernet chip in real time. Through the temperature data collected by the temperature sensor, the temperature distribution data of the target points is obtained, reflecting the temperature changes under different working conditions on different components. The acquisition formula for the target point temperature distribution data is expressed as:

[0049] ;

[0050] where, represents the temperature of the power module, represents the applied load power, represents the test time. Through this formula, the changing trend of temperature with load is dynamically monitored to ensure the reliability of the device in a high-temperature environment. The power supply port initial voltage data, port power supply curve data, port status matrix, and target point temperature distribution data are fused. The data from different sources are comprehensively processed to generate a comprehensive initial test data set.

[0051] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0052] Numerically classify the POE power supply parameters in the initial test data, segment them through the 15.4W threshold of the IEEE802.3af standard and the 30W threshold of the IEEE802.3at standard to obtain power supply level interval data;

[0053] Based on the power supply level interval data, perform standardized calculation on the power supply parameters of each POE port to obtain the port power supply reference value, arrange the port power supply reference values in the order of port numbers, and generate a 16×3-dimensional power supply parameter matrix through matrix operation to obtain POE power supply reference data;

[0054] Extract the Auto MDI / MDI-X status data in the port status matrix, verify the negotiation ability of each port through the port isolation function, and obtain the port negotiation status data;

[0055] Combine and analyze the port negotiation status data with the IEEE802.3X flow control status, generate the negotiation reference parameters for each POE port through the status mapping table, and obtain the negotiation status reference data;

[0056] Partition and statistically analyze the key point temperature distribution data according to the positions of the power module, PSE controller, and Ethernet chip, generate the temperature distribution characteristics through temperature gradient calculation, and obtain the temperature reference data;

[0057] Arrange the POE power supply reference data, negotiation status reference data, and temperature reference data in the form of a two-dimensional matrix, generate a hierarchical feature matrix through linear transformation, and obtain the test hierarchy data;

[0058] Restructure the test hierarchy data, integrate the test parameters of the power supply layer, negotiation layer, and temperature layer through the hierarchical mapping rule, and obtain the hierarchical test reference matrix.

[0059] Specifically, numerically classify the POE power supply parameters in the initial test data. Segment by the 15.4W threshold of the IEEE802.3af standard and the 30W threshold of the IEEE802.3at standard. Divide the power supply status at different powers into several grade intervals. Set two classification thresholds according to different power supply standards - 15.4W corresponds to the IEEE802.3af standard, while 30W is the IEEE802.3at standard. The generation process of the power supply grade interval data is expressed by the following formula:

[0060] ;

[0061] Among them, represents the power supply grade interval, which distinguishes the power supply types of POE ports through different power ranges. After classification, the power supply grade interval provides an identifier for each POE port, facilitating standardization and analysis. Based on the power supply grade interval data, perform standardized calculations on the power supply parameters of each POE port to eliminate the differences caused by different voltages or powers between different ports, so that the data of different ports can be comparable. The standardization uses the linear standardization formula:

[0062] ;

[0063] Among them, represents the standardized power supply parameter, is the actually measured voltage or power value, is the average value of the power supply parameters for all ports, is the standard deviation. Through standardization, the deviation of voltage and power data of different ports is eliminated, making the power supply parameters of all POE ports comparable. Arrange the power supply reference values of each POE port in the order of port numbers. Assume that each port has three key parameters (for example, voltage, current, power), and a 16×3-dimensional power supply parameter matrix is generated through matrix operations. Each row represents a POE port, and each column represents a power supply parameter of that port. The matrix form is expressed as:

[0064] ;

[0065] where, represents the th power supply parameter of the th port. Through matrix operations, a POE power supply reference data set that comprehensively reflects the power supply status of POE ports is obtained. Extract the Auto MDI / MDI-X status data in the port status matrix. Auto MDI / MDI-X is the automatic detection function of the Ethernet port, which can automatically judge the network type connected to the port and the way of exchanging data. Verify the negotiation ability of each port through the port isolation function to obtain the port negotiation status data. The port negotiation status reflects the working status of the port when connected to different devices, ensuring the stability and flexibility of network interconnection. The port negotiation status is represented by a binary value. For example, 1 means successful negotiation, and 0 means failed negotiation. These data are converted and analyzed through the status mapping table to generate the negotiation reference parameters of each POE port. For each port, the negotiation status reference data is expressed as:

[0066] };

[0067] where, represents the port negotiation status, reflecting the negotiation ability of the port in actual work. Process the key point temperature distribution data. Temperature distribution is a very critical indicator in the test process, reflecting the thermal performance of the device under different workloads. Obtain the temperature values of the power supply module, PSE controller, and Ethernet chip through temperature sensors, and then perform area division and temperature gradient calculation based on these data to obtain the temperature distribution characteristics. The temperature gradient calculation formula is expressed as:

[0068] ;

[0069] where, represents the temperature gradient, and are the highest temperature and the lowest temperature of the measured area respectively, is the distance between temperature sensors. This calculation helps the system understand the heat distribution generated by different components during operation. Arrange the POE power supply reference data, negotiation status reference data, and temperature reference data in the form of a two-dimensional matrix. The two-dimensional matrix contains data in multiple aspects such as power supply, negotiation, and temperature, and generates a hierarchical feature matrix through linear transformation to obtain test level data. The test level data is restructured through a hierarchical mapping rule. Integrate the test parameters of the power supply layer, negotiation layer, and temperature layer through the hierarchical mapping rule to obtain a hierarchical test reference matrix.

[0070] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0071] Perform the first layer of testing based on the power supply layer parameters in the hierarchical test reference matrix, apply a 15.4W rated load to a single POE port and supply power continuously for 8 hours to obtain single-port rated load data;

[0072] Perform the first layer of progressive testing according to the single-port rated load data, apply a 30W maximum load to a single POE port and supply power continuously for 8 hours to obtain single-port maximum load data;

[0073] Perform the first layer of full load testing based on the single-port maximum load data, apply a total power of 240W to all POE ports and supply power continuously for 24 hours to obtain full-port full load test data;

[0074] Perform the second layer of testing based on the performance layer parameters in the hierarchical test reference matrix, isolate the data flow of all ports through the port isolation function to obtain port isolation configuration data;

[0075] Perform the second layer of progressive testing according to the port isolation configuration data, perform concurrent data transmission on all isolated ports through IEEE802.3X flow control to obtain data throughput test data;

[0076] Perform the third layer of testing based on the temperature layer parameters in the hierarchical test reference matrix, keep all ports running at 30% load in a -10°C temperature environment for 4 hours to obtain low-temperature test data;

[0077] Perform the third layer of progressive testing according to the low-temperature test data, keep all ports running at 30% load in a 55°C temperature environment for 4 hours to obtain high-temperature test data;

[0078] Integrate the full-port full load test data, data throughput test data, low-temperature test data, and high-temperature test data according to the time sequence relationship to obtain a full-volume test data set.

[0079] Specifically, according to the power supply layer parameters in the hierarchical test benchmark matrix, the first layer of testing is performed. Evaluate the performance of a single POE port under rated load to ensure that each port can maintain stable power supply under typical working conditions. Apply a rated load of 15.4W to each port and supply power continuously for 8 hours to simulate the power supply ability of the POE switch to devices under normal use. During this test, record the voltage change, power output, and current fluctuation of each port within 8 hours. The formula is expressed as:

[0080] ;

[0081] Among them, is the rated load power, is the port output voltage, is the port output current. Through this formula, calculate the actual power output of the port when a 15.4W load is applied and monitor its voltage stability. Perform the first-layer progressive test according to the single-port rated load data, increase the load, apply a maximum load of 30W, and keep the power supply for 8 hours. Simulate the working ability of the POE switch under overload conditions and evaluate its power supply stability and performance under the maximum load. During the test, record parameters such as voltage, current, and power, and compare and analyze their fluctuations, especially whether the system can remain within a reasonable range. The power calculation formula is as follows:

[0082] ;

[0083] Among them, is the maximum load power, and are the port output voltage and current respectively. When performing the maximum load test, pay attention to the stability of the system under a long-term load, whether there are abnormal situations such as voltage drop or excessive current. Based on the single-port maximum load data, perform the first-layer full-load test. Apply a total power of 240W to all POE ports and supply power continuously for 24 hours. Simulate the long-term working state of the device under the maximum load to ensure that all ports can operate stably under extreme conditions. Calculate the total power through the following formula:

[0084] ;

[0085] Among them, is the total power of all 16 ports, represents the The power of each port. The test data includes the voltage, current, power of each port, and stability analysis within 24 hours to verify the durability and stability of the POE switch. Perform the second-layer test based on the performance layer parameters in the hierarchical test benchmark matrix and conduct port isolation tests. Use the port isolation function to isolate the data streams of all ports to ensure that the performance of each port can be evaluated independently. This operation can effectively eliminate the influence of mutual interference between ports and ensure the accuracy of data transmission tests. After isolation, record the configuration data of each port and perform the second-layer progressive test based on this. In the second-layer progressive test, perform concurrent data transmission on all isolated ports through IEEE802.3X flow control to test their data throughput. IEEE802.3X flow control can control the transmission rate of data streams to avoid network congestion and ensure that the load and data transmission performance of each port are not interfered. The throughput test involves key indicators such as the packet transmission rate, latency, and packet loss rate of the port. Assume the throughput is , then calculate the throughput through the following formula:

[0086] ;

[0087] where, is the data throughput, is the total number of packets transmitted, is the total test time. By analyzing the data throughput, evaluate the processing capacity and network performance of the port under high load. Perform the third-layer test based on the temperature layer parameters in the hierarchical test benchmark matrix. Evaluate the working stability and reliability of the POE switch in extreme temperature environments. In the low-temperature test, set the temperature to -10°C and keep all ports running at 30% load for 4 hours to simulate the working performance of the device in a cold environment, record the voltage, current, and power of the port, and analyze whether there is a performance degradation caused by low temperature. Calculate the voltage deviation in the low-temperature test data through the following formula:

[0088] ;

[0089] where, is the voltage deviation, and They are the voltage values at the start and end of the test respectively. According to the low-temperature test data, a high-temperature test is carried out. The temperature is raised to 55 °C and all ports are operated at 30% load for 4 hours. The test simulates the working state of the device in a high-temperature environment to verify its heat resistance and long-term stability. During the high-temperature test, parameters such as the voltage, current, power, and temperature change of the ports are recorded, and the performance stability of the device at high temperature is evaluated. The full-port full-load test data, data throughput test data, low-temperature test data, and high-temperature test data are integrated according to the chronological relationship to obtain a complete full-scale test data set. By analyzing these data, it is possible to effectively judge the performance of the POE switch during long-term operation and the possible failure risks. For example, by comparing the voltage change and data throughput change in different temperature environments, the stability of the device in different working environments can be evaluated, the service life of the device can be predicted, and feedback can be provided for subsequent optimized design.

[0090] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0091] Perform data normalization on the voltage data in the full-port full-load test data, calculate the voltage stability value through the voltage fluctuation under the 240W full-load condition, and obtain the power supply fluctuation characteristic data;

[0092] Analyze the POE port power supply parameters according to the power supply fluctuation characteristic data, calculate the voltage ripple coefficient and power factor of the 52V power supply output terminal, and obtain the PSE power supply quality data;

[0093] Based on the data throughput test data, perform a MAC address learning rate test on each POE port, and calculate through the address update time in the IEEE802.3X full-duplex mode to obtain the address processing rate data;

[0094] Based on the address processing rate data, perform statistical operations on the MAC address aging time and failure probability of each POE port to obtain the address aging characteristic data;

[0095] Based on the low-temperature test data and high-temperature test data, calculate the slope of the temperature rise curves of the power supply module, PSE controller, and Ethernet chip to obtain the temperature change characteristic data;

[0096] Conduct a correlation analysis based on the PSE power supply quality data and temperature change characteristic data, calculate the coupling coefficient between the power supply parameters and temperature change, and obtain the power supply state characteristic vector;

[0097] Perform a performance score calculation on the address aging characteristic data, and obtain the port performance characteristic vector through the data forwarding efficiency and address table entry utilization rate in the port isolation mode;

[0098] Combine the power supply status feature vector and the port performance feature vector, and generate multiple device status feature vectors according to the classification structure of power supply features, aging features, and performance features.

[0099] Specifically, perform data normalization on the voltage data in the full-port full-load test data to eliminate the deviation caused by different test environments and ensure that the data can be compared and analyzed on a unified scale. The normalization process is achieved by dividing the original voltage data by the maximum voltage value, and the data is scaled to the interval of [0,1]. The following formula is used for normalization:

[0100] ;

[0101] where, is the original voltage value, and are the minimum and maximum values in the test data respectively, is the normalized voltage value. Based on the normalized voltage data, analyze the voltage fluctuation of the full port to calculate the voltage stability. Voltage fluctuation is an important indicator reflecting the stability of the power supply system and is obtained by calculating the standard deviation of the voltage. The smaller the standard deviation, the smaller the voltage fluctuation and the more stable the power supply system. The voltage fluctuation is calculated by the following formula:

[0102] ;

[0103] where, is the standard deviation of the voltage fluctuation, is the th normalized voltage value of the port, is the average value of the voltages of all ports, is the number of ports. Convert the voltage fluctuation into power supply fluctuation characteristic data to evaluate the voltage stability. Based on the power supply fluctuation characteristic data, analyze the power supply quality of the POE port by calculating the voltage ripple coefficient and the power factor. The voltage ripple coefficient reflects the ratio of the AC signal to the DC signal in the voltage output, and the calculation formula is:

[0104] ;

[0105] where, is the AC component of the voltage, is the DC component of the voltage. The power factor reflects the ratio of the active power to the total power of the power system and is calculated by the following formula:

[0106] ;

[0107] where, is the active power, is the total power. The closer the power factor is to 1, the higher the power supply efficiency and the better the power supply quality. Based on the data throughput test data, the MAC address learning rate test is performed on each POE port. The MAC address learning rate test evaluates its processing ability by recording the speed at which the port learns MAC addresses. In the IEEE802.3X full-duplex mode, each address update takes a certain amount of time. By calculating the number of MAC addresses that can be learned per second, the address processing rate is obtained. Assume the number of addresses learned per second is , then the address learning rate is calculated by the following formula:

[0108] ;

[0109] where is the total number of learned MAC addresses, is the total test time. Through this step, the MAC address learning rate of each port is obtained, and then its data processing ability is evaluated. Based on the address processing rate data, statistical operations are performed on the MAC address aging time and failure probability of each POE port. MAC address aging means that if a MAC address does not communicate within a certain period of time, it will be deleted from the address table. By statistically counting the aging time of each port within a certain period of time and combining the usage of the port, the failure probability is calculated. Assume the MAC address aging time of the port is , then the failure probability is expressed by the following formula:

[0110] ;

[0111] where is the probability of MAC address failure, is the MAC address aging time, is the maximum aging time. Through this data, the stability and long-term operation ability of the port are evaluated. Based on the low-temperature test data and high-temperature test data, the slope of the temperature rise curve of the power supply module, PSE controller, and Ethernet chip is calculated. The temperature rise curve reflects the temperature change of the device during operation, and the slope is a measure of the temperature rise speed. By the magnitude of the slope, the thermal stability of the device in different temperature environments is judged. The calculation formula for the temperature rise slope is:

[0112] ;

[0113] where is the temperature rise slope, is the temperature change amount, is the time variation. By analyzing the temperature rise curves of different components (such as power supply modules, PSE controllers, and Ethernet chips), characteristic data of temperature changes are obtained to evaluate the performance of the device under different temperature conditions. Correlation analysis is performed based on the PSE power supply quality data and the characteristic data of temperature changes. By calculating the coupling coefficient between the power supply parameters and the temperature changes, the relationship between the power supply stability and the device temperature is obtained. The coupling coefficient is calculated by the following formula:

[0114] ;

[0115] wherein, is the coupling coefficient, and respectively represent the data of the power supply parameters and the temperature changes, and are their means. The calculation of the coupling coefficient can reveal the correlation between the power supply quality and the temperature changes. Performance scoring calculation is performed on the address aging characteristic data. The performance of the port is evaluated by the data forwarding efficiency and the address table entry utilization rate in the port isolation mode. The data forwarding efficiency is calculated by the following formula:

[0116] ;

[0117] wherein, is the data forwarding efficiency, is the amount of successfully forwarded data, is the total amount of data. The power supply state feature vector and the port performance feature vector are combined in terms of features. By splicing the feature vectors of different dimensions according to the classification structure of power supply features, aging features, and performance features, multiple device state feature vectors are generated to reflect the performance of the POE switch under different working conditions, providing a basis for subsequent fault prediction and performance optimization. The device state feature vectors are synthesized by means of linear combination or weighted summation.

[0118] In a specific embodiment, the process of executing step S5 may specifically include the following steps:

[0119] Input the power supply state feature vector in the multiple device state feature vectors into the PSE power supply weak classifier. The PSE power supply weak classifier includes a feature extraction layer, a time series analysis layer, and a state prediction layer. The feature extraction layer adopts a three-layer fully connected network structure, the time series analysis layer adopts a bidirectional LSTM structure, and the state prediction layer adopts a Softmax classifier to obtain the PSE power supply degradation score; the three-layer fully connected network in the PSE power supply weak classifier sequentially performs feature dimensionality reduction of 512, 256, and 128 dimensions, processes the output of each layer through the ReLU activation function and the Dropout layer, and the bidirectional LSTM includes 64 hidden units to obtain the PSE time series features;

[0120] Input the port performance feature vector among multiple device status feature vectors into the port performance weak classifier. The port performance weak classifier includes a data preprocessing layer, a feature enhancement layer, and a classification output layer. The data preprocessing layer adopts a batch normalization structure, the feature enhancement layer adopts a residual network structure, and the classification output layer adopts a cross-entropy loss function to obtain the port performance degradation score. The residual network structure of the port performance weak classifier is a three-layer stacked structure, with each layer containing two 3x3 convolutional layers and a shortcut connection, and the feature map is processed through the LeakyReLU activation function to obtain the port performance time series feature;

[0121] Input the temperature change feature data into the temperature feature weak classifier. The temperature feature weak classifier includes a temperature curve analysis layer, a trend extraction layer, and a result output layer. The temperature curve analysis layer adopts a one-dimensional convolutional network, and the trend extraction layer adopts an attention mechanism to obtain the temperature degradation score. The one-dimensional convolutional network of the temperature feature weak classifier is set with 32 convolutional kernels, the size of the convolutional kernel is 3, the feature dimension is compressed through a max pooling layer, and the attention mechanism adopts a scaled dot-product structure to obtain the temperature change time series feature;

[0122] Based on the PSE time series feature, the port performance time series feature, and the temperature change time series feature, the AdaBoost algorithm is used to perform weighted fusion on the PSE power supply degradation score, the port performance degradation score, and the temperature degradation score to obtain the device degradation prediction index.

[0123] Specifically, input the power supply status feature vector among multiple device status feature vectors into the PSE power supply weak classifier. The structure of the PSE power supply weak classifier includes a feature extraction layer, a time series analysis layer, and a status prediction layer. The feature extraction layer adopts a three-layer fully connected network structure to process the input power supply status feature. The output of each layer of the network is subjected to a non-linear transformation through the ReLU activation function, and the Dropout layer is used for regularization to prevent overfitting. In the first layer, the dimension of the fully connected network is set to 512, the second layer is 256, and the third layer is 128. The output of each layer is transformed through the ReLU activation function and processed through the Dropout layer to enhance the generalization ability of the model. Assuming the input power supply status feature is , then the output of the feature extraction layer is expressed as:

[0124] ;

[0125] Among them, is the weight matrix of the fully connected layer, is the bias term. After being processed by the three-layer fully connected network, the extracted power supply feature . The timing analysis layer of the PSE power supply weak classifier adopts a bidirectional LSTM (Long Short-Term Memory) structure. The bidirectional LSTM can capture the timing features in the input sequence from both the forward and reverse directions and understand the dynamic changes in the data. The number of hidden units of the bidirectional LSTM is set to 64. For the input features , the bidirectional LSTM generates a timing feature :

[0126] ;

[0127] Among them, Bi-LSTM represents the bidirectional LSTM network. This timing feature captures the dynamic information of the power supply feature changing over time. In the state prediction layer, a Softmax classifier is used to classify the timing feature and output the PSE power supply degradation score. The Softmax function converts the input timing feature into a class probability distribution. Suppose the output of the classifier is , then:

[0128] ;

[0129] Among them, is the weight matrix of the Softmax layer, is the bias term. The PSE power supply degradation score is obtained, and this score reflects the health degree of the device power supply state. The port performance feature vector in multiple device state feature vectors is input into the port performance weak classifier. The structure of the port performance weak classifier includes a data preprocessing layer, a feature enhancement layer, and a classification output layer. In the data preprocessing layer, batch normalization is used to process the input data to accelerate training and improve the stability of the model. Suppose the input port performance feature is , then the output of the batch normalization layer is:

[0130] ;

[0131] In the feature enhancement layer, a residual network structure is used to improve the representation ability of the model. The residual network alleviates the gradient vanishing problem in the deep network by introducing skip connections. The residual network structure is a three-layer stacked structure, and each layer contains two 3x3 convolutional layers and a short circuit connection. Suppose the output of the feature enhancement layer is , then it is expressed as:

[0132] ;

[0133] Among them, ResNet represents the Residual Network. Through the feature enhancement layer, the enhanced port performance features are obtained. At the classification output layer, the cross-entropy loss function is used to calculate the port performance degradation score. The calculation formula of the cross-entropy loss function for multi-class classification problems is:

[0134] ;

[0135] Among them, is the actual label, is the predicted probability. Through this process, the port performance degradation score is obtained. The temperature change feature data is input into the temperature feature weak classifier. The structure of the temperature feature weak classifier includes a temperature curve analysis layer, a trend extraction layer, and a result output layer. In the temperature curve analysis layer, a one-dimensional convolutional network (1D ConvNet) is used to analyze the temporal features of temperature changes. The one-dimensional convolutional network slides the convolutional kernel to process the time series data and extracts local features from it. Assuming the input is the temperature feature , after the convolution operation, the temperature temporal features are obtained:

[0136] ;

[0137] Among them, the size of the convolutional kernel is 3, and the number of convolutional kernels is set to 32. Dimensionality reduction is performed through the max pooling layer. In the trend extraction layer, the attention mechanism is used to focus on the important change regions in the temperature curve. The attention mechanism dynamically adjusts the attention degree of the model to different inputs through weighted summation. Assuming the feature after the attention mechanism is , it is expressed as:

[0138] ;

[0139] In the result output layer, the temperature change features are converted into the temperature degradation score through the Softmax classifier. Based on the PSE temporal features, the port performance temporal features, and the temperature change temporal features, the AdaBoost algorithm is used to perform weighted fusion on the PSE power supply degradation score, the port performance degradation score, and the temperature degradation score to obtain the device degradation prediction index. The AdaBoost algorithm combines the outputs of multiple classifiers through weighted averaging to obtain a comprehensive prediction result. Assuming the weight of each score is , then the device degradation prediction index is:

[0140] ;

[0141] Through weighted fusion, the device degradation prediction index is obtained, and based on this index, the health status of the device is judged to carry out fault warning and maintenance decision-making.

[0142] In a specific embodiment, the process of executing step S6 may specifically include the following steps:

[0143] Perform the first time series segmentation on the PSE time series characteristics to obtain PSE sequence data; perform the second time series segmentation on the port performance time series characteristics to obtain port sequence data; perform the third time series segmentation on the temperature change time series characteristics to obtain temperature sequence data;

[0144] Align the time stamps of the PSE sequence data, port sequence data, and temperature sequence data, and generate a multi-modal feature sequence with a unified sampling interval through the cubic spline interpolation algorithm to obtain aligned feature data;

[0145] Perform the first-stage feature fusion on the aligned feature data, and obtain the first fusion weight coefficient by calculating the PSE power supply-temperature coupling coefficient and the port performance-temperature coupling coefficient;

[0146] Input the first fusion weight coefficient into the first-level classifier of the improved AdaBoost algorithm, and weight the PSE power supply degradation score and the port performance degradation score through a weak classifier group introducing a temperature compensation factor to obtain the first-stage degradation index;

[0147] Perform the second-stage feature fusion on the first-stage degradation index and the temperature degradation score, and calculate the final weight distribution through a multi-level weighting method introducing an adaptive temperature threshold to obtain the second fusion weight coefficient;

[0148] Perform a weighted calculation on the first-stage degradation index and the second fusion weight coefficient, and obtain the device degradation prediction index by setting the grading standard of [0.3, 0.6) for mild degradation, [0.6, 0.8) for moderate degradation, and [0.8, 1.0] for severe degradation.

[0149] Specifically, perform the first time series segmentation on the PSE time series characteristics, and cut it into multiple time period data sequences according to a fixed time window. For example, assume that the data of the PSE time series characteristics is , and segment it according to the time step to obtain multiple subsequences:

[0150] ;

[0151] Among them, represents the th data sequence after time series segmentation, and is the total number after segmentation. Similarly, the port performance time series characteristics and the temperature change time series characteristics are also respectively subjected to the second time series segmentation and the third time series segmentation to obtain the port sequence data and the temperature sequence data , that is:

[0152] ;

[0153] ;

[0154] The cubic spline interpolation algorithm is adopted to smoothly connect multiple data sequences with inconsistent timestamps through an interpolation function to obtain a unified sampling interval. Suppose there are three time-series data that are not completely aligned and . Find a common time axis for each sequence, that is, adjust the timestamps of these data through cubic spline interpolation so that all data sequences are aligned on a unified time axis to obtain aligned feature data:

[0155] ;

[0156] Among them, SplineInterpolation represents the cubic spline interpolation operation, is the multi-modal feature data after alignment. Perform the first-stage feature fusion on the aligned feature data. Calculate the first fusion weight coefficient by calculating the coupling coefficient between PSE power supply and temperature and the coupling coefficient between port performance and temperature. The coupling coefficient reflects the degree of mutual influence between two different features. Suppose represents the PSE power supply feature, represents the temperature feature, represents the port performance feature. Calculate the coupling coefficients between PSE and temperature, and between port performance and temperature by calculating the Pearson correlation coefficient:

[0157] ;

[0158] ;

[0159] Among them, represents the covariance, , , are the standard deviations of the PSE, port, and temperature features respectively. Through these coupling coefficients, calculate the fusion weight coefficient of the first stage:

[0160] ;

[0161] Input the weight coefficient into the first-stage classifier of the improved AdaBoost algorithm. The AdaBoost algorithm improves the accuracy of the model by weighted combination of multiple weak classifiers. The first-stage classifier weights the PSE power supply degradation score and the port performance degradation score through a weak classifier group introducing a temperature compensation factor to obtain the first-stage degradation index:

[0162] ;

[0163] Among them, and respectively represent the PSE power supply score and the port performance score. The first-stage degradation index is weighted and fused with the temperature degradation score. An adaptive temperature threshold is introduced, and the final weight distribution is calculated through a multi-level weighting method. This means that the change in temperature not only affects the PSE power supply and port performance but also has an impact on the final device degradation prediction result. By calculating the adaptive weight coefficient , the fusion process is optimized:

[0164] ;

[0165] The first-stage degradation index and the second-stage fusion weight coefficient are weighted to obtain the final device degradation prediction index :

[0166] ;

[0167] By setting the classification criteria for device degradation, [0.3, 0.6) is mild degradation, [0.6, 0.8) is moderate degradation, and [0.8, 1.0] is severe degradation. According to the final device degradation prediction index the health state of the device is determined.

[0168] The above described the aging test method for the security POE switch in the embodiment of the present invention. Next, the aging test device for the security POE switch in the embodiment of the present invention will be described. Please refer to Figure 2 One embodiment of the aging test device for the security POE switch in the embodiment of the present invention includes:

[0169] An acquisition module, configured to fix the POE switch to be tested on the test platform, apply standard loads to the POE power supply ports respectively through a load simulator, and acquire the power supply parameters and port states to obtain initial test data;

[0170] A classification module, configured to classify the initial test data according to the POE power supply parameters, port negotiation status, and temperature distribution data to generate a hierarchical test benchmark matrix;

[0171] A test module, configured to perform a full-load test on the POE port, a port isolation data stream test, and an auto-negotiation interaction test respectively through an intelligent PSE controller based on the hierarchical test benchmark matrix, and acquire a full-scale test data set in a hierarchical progressive manner;

[0172] A calculation module, configured to calculate PSE power supply characteristics, MAC address aging characteristics, and port performance characteristics for the full-scale test data set, and generate multiple device status feature vectors;

[0173] A processing module, configured to input multiple device status feature vectors into an aging prediction model set to calculate the device deterioration trend, and obtain a device deterioration prediction index.

[0174] Through the collaborative cooperation of the above-mentioned components and a hierarchical progressive test strategy, a comprehensive test of the power supply capacity, network performance, and environmental adaptability of the POE switch is realized, improving the test coverage rate; by adopting an intelligent PSE controller and multi-dimensional data fusion technology, a device status feature vector is established, making the test results more objective and accurate; by introducing an aging prediction model set based on deep learning and combining the weighted fusion mechanism of the AdaBoost algorithm, the accuracy of device deterioration trend prediction is improved; through a dynamic test parameter adjustment mechanism, the test process can be adaptively optimized according to the device status, reducing the waste of test resources; a complete multi-modal feature sequence analysis system is established to realize the correlation analysis of multi-dimensional performance parameters of the device, providing a reliable basis for device maintenance decisions; through a hierarchical early warning mechanism and a temperature compensation strategy, the adaptability and prediction accuracy of the test method in extreme environments are enhanced.

[0175] Refer to Figure 3 , in an embodiment of the present invention, a computer device is further provided. The computer device may be a server, and its internal structure may be as shown in Figure 3 . The computer device includes a processor, a memory, a display screen, an input device, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0176] Those skilled in the art can understand that Figure 3 the structure shown in

[0177] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0178] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0179] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described system, system and unit can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0180] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0181] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of various embodiments of the present invention.

Claims

1. A method for aging test of a POE switch for security protection, characterized in that, The method includes: Fix the POE switch to be tested on the test platform, apply standard loads to the POE power supply ports respectively through a load simulator, and obtain power supply parameters and port states to obtain initial test data; Classify the initial test data according to POE power supply parameters, port negotiation states, and temperature distribution data to generate a hierarchical test benchmark matrix; Based on the hierarchical test benchmark matrix, perform full-load tests on POE ports, port isolation data stream tests, and auto-negotiation interaction tests respectively through an intelligent PSE controller, and collect a full set of test data in a hierarchical progressive manner. The full set of test data includes full-port full-load test data, data throughput test data, low-temperature test data, and high-temperature test data; Calculate PSE power supply feature, MAC address aging feature, and port performance feature for the full set of test data to generate multiple device state feature vectors; Input the multiple device state feature vectors into an aging prediction model set to calculate the device degradation trend and obtain device degradation prediction indicators.

2. The aging test method of the POE switch for security protection according to claim 1, characterized in that The step of fixing the POE switch to be tested on the test platform, applying standard loads to the POE power supply ports respectively through a load simulator, and obtaining power supply parameters and port states to obtain initial test data includes: Fix the POE switch to be tested on a standard 13-inch rack, and number and identify the physical ports of the POE switch to be tested through a scanner to obtain port position mapping data; Configure load simulators complying with IEEE802.3af / at standards for 16 POE ports in the port position mapping data respectively, and collect the voltage values at the output end of the 52V power supply through a digital multimeter to obtain initial voltage data of the power supply ports; Input the initial voltage data of the power supply ports into a voltage regulation controller, and perform power scanning within the range of 0-30W on the output voltage of each POE port through PWM wave modulation to obtain port power supply curve data; Compare the port power supply curve data with thresholds, set the power supply reference value and voltage deviation value for each POE port through a voltage comparator to obtain port power supply reference data, and based on the port power supply reference data, perform Auto MDI / MDI-X auto-reversal tests on each POE port through an intelligent PSE controller to obtain port line pair identification data; Perform cross-matrix analysis on the port line pair identification data, generate a cross-correlation table between ports through Boolean algebra operations to obtain a port state matrix, apply a 15.4W standard load to each POE port in the port state matrix, and collect the temperature values of the power supply module, PSE controller, and Ethernet chip through a temperature sensor to obtain target point temperature distribution data; Fuse the initial voltage data of the power supply ports, the port power supply curve data, the port state matrix, and the target point temperature distribution data to obtain initial test data.

3. The aging test method of the POE switch for security protection according to claim 2, characterized in that, The step of classifying the initial test data according to POE power supply parameters, port negotiation states, and temperature distribution data to generate a hierarchical test benchmark matrix includes: Numerically classify the POE power supply parameters in the initial test data, segment them by the 15.4W threshold of the IEEE802.3af standard and the 30W threshold of the IEEE802.3at standard to obtain power supply level interval data; Based on the power supply level interval data, perform standardized calculations on the power supply parameters of each POE port to obtain the port power supply reference value, arrange the port power supply reference values in the order of port numbers, and generate a 16×3-dimensional power supply parameter matrix through matrix operations to obtain POE power supply reference data; Extract the Auto MDI / MDI-X status data in the port status matrix, and verify the negotiation capabilities of each port through the port isolation function to obtain port negotiation status data; Combine and analyze the port negotiation status data with the IEEE802.3X flow control status, and generate the negotiation reference parameters for each POE port through the status mapping table to obtain negotiation status reference data; Partition and statistically analyze the target point temperature distribution data according to the positions of the power supply module, PSE controller, and Ethernet chip, and generate temperature distribution characteristics through temperature gradient calculation to obtain temperature reference data; Arrange the POE power supply reference data, the negotiation status reference data, and the temperature reference data in the form of a two-dimensional matrix, and generate a hierarchical feature matrix through linear transformation to obtain test level data; Restructure the test level data, and integrate the test parameters of the power supply layer, negotiation layer, and temperature layer through hierarchical mapping rules to obtain a hierarchical test reference matrix.

4. The aging test method of the POE switch for security protection according to claim 3, characterized in that, Based on the hierarchical test reference matrix, the intelligent PSE controller respectively executes the full-load test of the POE port, the port isolation data stream test, and the auto-negotiation interaction test, and collects the full-scale test data set in a hierarchical progressive manner, including: Execute the first-layer test based on the power supply layer parameters in the hierarchical test reference matrix, apply a 15.4W rated load to a single POE port and supply power continuously for 8 hours to obtain single-port rated load data; Execute the first-layer progressive test according to the single-port rated load data, apply a 30W maximum load to a single POE port and supply power continuously for 8 hours to obtain single-port maximum load data; Execute the first-layer full-load test based on the single-port maximum load data, apply a total power of 240W to all POE ports and supply power continuously for 24 hours to obtain full-port full-load test data; Execute the second-layer test based on the performance layer parameters in the hierarchical test reference matrix, isolate the data stream of all ports through the port isolation function to obtain port isolation configuration data; Execute the second-layer progressive test according to the port isolation configuration data, perform concurrent data transmission on all isolated ports through IEEE802.3X flow control to obtain data throughput test data; Execute the third-layer test based on the temperature layer parameters in the hierarchical test reference matrix, keep all ports running at 30% load in a -10°C temperature environment for 4 hours to obtain low-temperature test data; Perform the third-level progressive test based on the low-temperature test data, and keep all ports running at 30% load for 4 hours in a 55°C temperature environment to obtain high-temperature test data; Integrate the full-port full-load test data, the data throughput test data, the low-temperature test data, and the high-temperature test data according to the timing relationship to obtain a full-scale test data set.

5. The aging test method for the POE switch for security protection according to claim 4, characterized in that, Perform PSE power supply characteristic calculation, MAC address aging characteristic calculation, and port performance characteristic calculation on the full-scale test data set to generate multiple device status feature vectors, including: Perform data normalization processing on the voltage data in the full-port full-load test data, and calculate the voltage stability value through the voltage fluctuation under the 240W full-load condition to obtain power supply fluctuation characteristic data; Analyze the POE port power supply parameters according to the power supply fluctuation characteristic data, and calculate the voltage ripple coefficient and power factor of the 52V power output terminal to obtain PSE power supply quality data; Perform MAC address learning rate test on each POE port based on the data throughput test data, and calculate the address update time in the IEEE802.3X full-duplex mode to obtain address processing rate data; Based on the address processing rate data, perform statistical operations on the MAC address aging time and failure probability of each POE port to obtain address aging characteristic data; Based on the low-temperature test data and the high-temperature test data, calculate the slope of the temperature rise curves of the power supply module, PSE controller, and Ethernet chip to obtain temperature change characteristic data; Perform correlation analysis according to the PSE power supply quality data and the temperature change characteristic data, and calculate the coupling coefficient of the power supply parameters and the temperature change to obtain the power supply state feature vector; Perform performance scoring calculation on the address aging characteristic data, and obtain the port performance feature vector through the data forwarding efficiency and address table entry utilization rate in the port isolation mode; Combine the power supply state feature vector and the port performance feature vector, and generate multiple device status feature vectors according to the classification structure of power supply characteristics, aging characteristics, and performance characteristics.

6. The aging test method for the POE switch used for security protection according to claim 5, wherein, Input the multiple device status feature vectors into the aging prediction model set to calculate the device degradation trend, and obtain the device degradation prediction index, including: Input the power supply state feature vector in the multiple device status feature vectors into the PSE power supply weak classifier. The PSE power supply weak classifier includes a feature extraction layer, a timing analysis layer, and a state prediction layer. The feature extraction layer adopts a three-layer fully connected network structure, the timing analysis layer adopts a bidirectional LSTM structure, and the state prediction layer adopts a Softmax classifier to obtain the PSE power supply degradation score; the three-layer fully connected network in the PSE power supply weak classifier performs feature dimensionality reduction of 512, 256, and 128 dimensions in sequence, and processes the output of each layer through the ReLU activation function and the Dropout layer. The bidirectional LSTM contains 64 hidden units to obtain the PSE timing feature; Input the port performance feature vector among the multiple device status feature vectors into the port performance weak classifier. The port performance weak classifier includes a data preprocessing layer, a feature enhancement layer, and a classification output layer. The data preprocessing layer adopts a batch normalization structure, the feature enhancement layer adopts a residual network structure, and the classification output layer adopts a cross-entropy loss function to obtain a port performance degradation score. The residual network structure of the port performance weak classifier is a three-layer stacked structure, with each layer containing two 3x3 convolutional layers and a short connection, and the feature map is processed through a LeakyReLU activation function to obtain port performance time series features. Input the temperature change feature data into the temperature feature weak classifier. The temperature feature weak classifier includes a temperature curve analysis layer, a trend extraction layer, and a result output layer. The temperature curve analysis layer adopts a one-dimensional convolutional network, and the trend extraction layer adopts an attention mechanism to obtain a temperature degradation score. The one-dimensional convolutional network of the temperature feature weak classifier is set with 32 convolutional kernels, the size of the convolutional kernel is 3, and the feature dimension is compressed through a max-pooling layer. The attention mechanism adopts a scaled dot-product structure to obtain temperature change time series features. Based on the PSE time series features, the port performance time series features, and the temperature change time series features, the AdaBoost algorithm is used to perform weighted fusion on the PSE power supply degradation score, the port performance degradation score, and the temperature degradation score to obtain a device degradation prediction index.

7. The aging test method for the POE switch for security protection according to claim 6, characterized in that, The above-mentioned based on the PSE time series features, the port performance time series features, and the temperature change time series features, using the AdaBoost algorithm to perform weighted fusion on the PSE power supply degradation score, the port performance degradation score, and the temperature degradation score to obtain a device degradation prediction index, includes: Perform the first time series segmentation on the PSE time series features to obtain PSE sequence data; perform the second time series segmentation on the port performance time series features to obtain port sequence data; perform the third time series segmentation on the temperature change time series features to obtain temperature sequence data. Align the time stamps of the PSE sequence data, the port sequence data, and the temperature sequence data, and generate a multi-modal feature sequence with a unified sampling interval through a cubic spline interpolation algorithm to obtain aligned feature data. Perform the first-stage feature fusion on the aligned feature data, and calculate the PSE power supply-temperature coupling coefficient and the port performance-temperature coupling coefficient to obtain the first fusion weight coefficient. Input the first fusion weight coefficient into the first-level classifier of the improved AdaBoost algorithm, and weight the PSE power supply degradation score and the port performance degradation score through a weak classifier group introducing a temperature compensation factor to obtain a first-stage degradation index. Perform the second-stage feature fusion on the first-stage degradation index and the temperature degradation score, and calculate the final weight allocation through a multi-level weighted method introducing an adaptive temperature threshold to obtain the second fusion weight coefficient. Perform weighted calculation on the first-stage deterioration index and the second fusion weight coefficient, and obtain the equipment deterioration prediction index by setting the grading criteria of [0.3, 0.6) for mild deterioration, [0.6, 0.8) for moderate deterioration, and [0.8, 1.0] for severe deterioration.

8. An aging test device for a POE switch used for security, characterized in that, For performing the aging test method of the POE switch for security protection as described in any one of claims 1-7, the aging test device of the POE switch for security protection includes: An acquisition module, configured to fix the POE switch to be tested on a test platform, apply standard loads to the POE power supply ports respectively through a load simulator, and acquire power supply parameters and port states to obtain initial test data; A classification module, configured to classify the initial test data according to POE power supply parameters, port negotiation states, and temperature distribution data to generate a hierarchical test reference matrix; A test module, configured to perform a full-load test of the POE port, a port isolation data stream test, and an auto-negotiation interaction test respectively through an intelligent PSE controller based on the hierarchical test reference matrix, and collect a full-scale test data set in a hierarchical progressive manner, where the full-scale test data set includes full-port full-load test data, data throughput test data, low-temperature test data, and high-temperature test data; A calculation module, configured to perform PSE power supply characteristic calculation, MAC address aging characteristic calculation, and port performance characteristic calculation on the full-scale test data set to generate a plurality of device state characteristic vectors; A processing module, configured to input the plurality of device state characteristic vectors into an aging prediction model set to perform equipment deterioration trend calculation to obtain an equipment deterioration prediction index.

9. A computer device, characterized in that, It includes a memory and a processor, and the memory stores a computer program that can run on the processor. It is characterized in that when the processor executes the computer program, the aging test method of the POE switch for security protection as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, on which a computer program is stored, and when the computer program is run by a processor, the processor is caused to execute the aging test method of the POE switch for security protection as described in any one of claims 1 to 7.

Images