A fault monitoring system for a power adapter

Abstract

The application belongs to the technical field of power management, and discloses a fault monitoring system of a power adapter, which comprises a data collection module, an electrical performance data collection module and an adapter working environment data collection module; a data processing module, which is used for extracting features of the power adapter parameter data, the electrical performance data and the working environment data to obtain an adapter parameter feature dataset, an electrical performance feature dataset and a working environment feature dataset; a fault state detection module, which is used for weighting and fusing the adapter parameter feature dataset, the electrical performance feature dataset and the working environment feature dataset through a weighting model to obtain a comprehensive feature dataset; an adapter fault detection model is obtained by training according to the comprehensive feature dataset; a power adapter fault state evaluation label is obtained by prediction through the adapter fault detection model, so as to determine whether the power adapter is in a fault state; and a sudden fault can be warned in advance.

Description

Technical Field

[0001] This invention relates to the field of power management technology, and more specifically, to a fault monitoring system for a power adapter. Background Technology

[0002] Patent publication number CN117310348A discloses a method and system for real-time monitoring of power adapter faults, comprising: acquiring the output voltage and resistance temperature of the power adapter to obtain a temperature-compensated discrete voltage sequence; obtaining a trend component sequence and a residual component sequence based on the temperature-compensated discrete voltage sequence, and obtaining a trend stability index based on the trend component sequence; obtaining a voltage local nonlinear oscillation index based on the residual component sequence; obtaining an adaptive k-value based on the voltage local nonlinear oscillation index and the trend stability index of the temperature-compensated discrete voltage sequence, obtaining an anomaly factor for each data point in the temperature-compensated discrete voltage sequence based on the adaptive k-value, and completing real-time monitoring of power adapter faults based on the anomaly factors. This invention solves the problem that different voltage fluctuation characteristics during power adapter operation can easily lead to inaccurate fault detection.

[0003] Existing power adapter fault monitoring systems mainly suffer from the following problems:

[0004] Without timestamping and sorting the collected data, the temporal sequence and continuity of the data cannot be guaranteed. Subsequent data processing and analysis may be biased due to data inconsistencies or missing data, leading to a decrease in the accuracy of the results. Furthermore, the failure to use empirical mode decomposition (EMD) methods to decompose complex signals may prevent the full revelation of different frequency components and local variation characteristics in the data. Power adapter operating data may contain multiple frequency components, and simple traditional signal processing methods may not be able to effectively capture these subtle differences, thus affecting the understanding of the data's intrinsic characteristics and leading to the risk of misdiagnosis and failure to accurately identify potential equipment faults.

[0005] Failure to dynamically adjust the integration variable range of the Hilbert transform may lead to transformation errors and instability. Signals with different frequency components may require different integration ranges for accurate transformation. Ignoring this may cause errors in signals with large or small frequency components, thus affecting the results of the Hilbert transform. In particular, when there are large differences in frequency components in the electrical performance data of power adapters, such errors may directly affect the accuracy and stability of fault detection.

[0006] Without constructing a directed graph model of the power adapter circuit and assigning weights to each edge based on different types of faults, it is impossible to accurately reflect the degree of fault risk among electrical components; faults between different components are interconnected, and the lack of a comprehensive network model may lead to the omission of key fault paths, causing fault analysis to remain local and unable to fully identify potential fault points of the adapter.

[0007] Without a mechanism for dynamically adjusting and updating factors, it is impossible to adapt to different fault types and the actual conditions of electrical components in a timely manner; the lack of flexible analysis methods makes it difficult to react quickly when dealing with complex and ever-changing fault scenarios, and may fail to detect sudden faults in time or quickly correct incorrect fault diagnosis paths. Without accurate fault location and dynamic optimization mechanisms, maintenance personnel cannot quickly diagnose problems through accurate fault location, which increases the difficulty and cost of long-term maintenance.

[0008] In view of this, the present invention proposes a fault monitoring system for a power adapter to solve the above problems. Summary of the Invention

[0009] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a fault monitoring system for a power adapter, comprising:

[0010] The data collection module is used to collect power adapter parameter data, electrical performance data, and adapter operating environment data;

[0011] The data processing module is used to extract features from power adapter parameter data, electrical performance data, and operating environment data based on improved empirical mode decomposition combined with Hilbert transform, to obtain adapter parameter feature dataset, electrical performance feature dataset, and operating environment feature dataset.

[0012] The fault state detection module is used to perform weighted fusion of adapter parameter feature dataset, electrical performance feature dataset and working environment feature dataset through a weighted model to obtain a comprehensive feature dataset; train the adapter fault detection model, input the comprehensive feature dataset into the trained adapter fault detection model, predict the probability of power adapter fault state, and determine whether the power adapter is in a fault state.

[0013] The fault type diagnosis module generates an early warning message immediately when the power adapter is in a fault state, and automatically collects fault state data; it trains and obtains an adapter fault diagnosis model based on the fault state data, and predicts the fault type of the power adapter through the adapter fault detection model.

[0014] The fault response module, based on directed graph modeling, combines a weighted Dijkstra algorithm with a dynamically adjusted update factor to locate adapter faults. The power adapter intelligent monitoring terminal automatically triggers the corresponding fault response strategy based on the obtained power adapter fault type and location. The modules are connected to each other via wired and / or wireless means.

[0015] Furthermore, the power adapter parameter data includes the power adapter's input voltage, input current, output voltage, output current, operating time, and temperature; the electrical performance data includes the power adapter's output power, harmonic voltage, harmonic current, and total harmonic distortion; and the adapter's operating environment data includes the temperature, humidity, vibration frequency, and electromagnetic interference level in the adapter's operating environment.

[0016] Furthermore, the method for feature extraction of power adapter parameter data, electrical performance data, and operating environment data based on improved empirical mode decomposition combined with Hilbert transform includes:

[0017] S31. Timestamp the collected power adapter parameter data, electrical performance data, and adapter operating environment data at a daily collection interval, and sort the collected power adapter parameter data, electrical performance data, and adapter operating environment data in ascending order according to the timestamps.

[0018] S32. Integrate the sorted power adapter parameter data, electrical performance data, and adapter operating environment data to obtain a data sequence X. j For each data sequence X j Data variables in Empirical mode decomposition is performed using the empirical mode decomposition formula, decomposing it into v eigenmode function components;

[0019] The empirical mode decomposition formula is: in, Let be the signal value of the i-th variable in the j-th data at time t; K represents the k-th eigenmode function component of the i-th variable signal in the j-th data type. i The number of intrinsic mode function components obtained from the decomposition of the i-th variable signal; k is the index of the intrinsic mode function component decomposed in each signal; t represents the low-frequency signal value remaining during the signal decomposition process; j represents the index of the collected data category; t represents the sampling time of the power adapter parameter data, electrical performance data, and adapter operating environment data; i represents the index of each variable in the collected data category.

[0020] The number K of intrinsic mode function components obtained by decomposing the i-th variable signal using the intrinsic mode function component constraint formula is K. i To impose restrictions, the formula for restricting the intrinsic mode function components is as follows: Wherein, K′ i To limit the number of eigenmode function components; Q j P represents the number of data categories collected. i The number of variable categories in the collected data; a and b are constant factors that adjust the proportions of the intrinsic modal function component constraint formula;

[0021] S33. For each eigenmode function component The complex signal of each component is obtained by performing a Hilbert transform using the Hilbert transform formula.

[0022] The Hilbert transform formula is: in, For each intrinsic mode function component Complex signals; For each eigenmode function component The imaginary part of the signal obtained by performing a Hilbert transform; i′ is the imaginary unit;

[0023] The Where τ is the integral variable used to calculate the Hibbert transform at time t;

[0024] The range of the integral variable τ used to calculate the Hibbert transform at time t is dynamically adjusted and constrained by the integral variable range adjustment formula.

[0025] The formula for adjusting the range of the integral variable is: Wherein, K′ i To limit the number of intrinsic mode function components; Let τ be the range of the integral variable;

[0026] S34, from complex signal The data sequence X is extracted using the instantaneous amplitude calculation formula and the instantaneous frequency calculation formula. j The instantaneous amplitude and instantaneous frequency;

[0027] The formula for calculating the instantaneous amplitude is: in, This is the instantaneous amplitude calculated at time t; For each intrinsic mode function component The square of the imaginary part of the signal obtained by performing the Hilbert transform;

[0028] The formula for calculating the instantaneous frequency is:

[0029] in, The instantaneous frequency calculated at time t; The arctangent of the ratio of the imaginary part to the real part of the signal represents the phase of the complex signal; This indicates taking the derivative with respect to t;

[0030] S35. Constructing a data sequence X using instantaneous amplitude and instantaneous frequency through Hilbert's energy spectrum formula. j Hilbert energy spectrum for each variable;

[0031] The Hilbert energy spectrum formula is: in, Let be the Hilbert energy spectrum of the i-th variable in the j-th data type; The Dirac function represents the instantaneous frequency of the signal. A pulse at a given location; ω is a continuous variable of angular frequency, used to represent and analyze the energy distribution of a signal at different frequency components;

[0032] S36, for each data variable The intrinsic mode function components are constructed with features, and the statistical features of each intrinsic mode function are extracted. The statistical features include the standard deviation of the instantaneous amplitude and the average value of the instantaneous frequency. By calculating the statistical features of each intrinsic mode function component, a feature vector is formed. The feature vectors of each data variable are integrated into the corresponding datasets to form the adapter parameter feature dataset, the electrical performance feature dataset, and the working environment feature dataset, respectively.

[0033] Furthermore, the method for obtaining a comprehensive feature dataset by weighted fusion of the adapter parameter feature dataset, electrical performance feature dataset, and operating environment feature dataset using a weighted model includes:

[0034] Let D1 be the adapter parameter feature dataset, D2 be the electrical performance feature dataset, and D3 be the working environment feature dataset; the weighted model is: Qz = D1·δ1 + D2·δ2 + D3·δ3; where Qz is the comprehensive feature dataset; δ1 is the weight coefficient of the adapter parameter feature dataset; δ2 is the weight coefficient of the electrical performance feature dataset; and δ3 is the weight coefficient of the working environment feature dataset.

[0035] Furthermore, the training method for the adapter fault detection model includes:

[0036] The dataset is divided into training, validation and test sets to build an adapter fault detection model; the sample set is a subset of the dataset, and each sample set includes historical comprehensive feature dataset and corresponding power adapter fault state probability.

[0037] The model's input data is a historical comprehensive feature dataset, and the model's output label is the probability of a power adapter's fault state; the sigmoid function is used as the activation function; the adapter fault detection model is a gradient boosting decision tree model.

[0038] Initialize the adapter fault detection model and set the hyperparameters for the number of trees, learning rate, and tree depth. Train the model using different combinations of hyperparameters through k-fold cross-validation, and fine-tune the initially set hyperparameters using the cross-validation results, selecting the best-performing hyperparameter combination.

[0039] The logarithmic loss function is used to measure the difference between the model's predictions and the true labels; the formula for calculating the logarithmic loss function is as follows:

[0040] Where N is the total number of samples in the adapter fault detection model; y c Let q be the true label of the c-th sample. c Predict the probability that the c-th sample is a positive sample for the model; c is the index of the sample.

[0041] The adapter fault detection model is trained on the training set by training a new decision tree based on the residuals of the current model in each iteration. The model is tuned by adjusting the hyperparameters according to the model performance feedback. The model is retrained using the adjusted hyperparameters. After the training process is completed, the optimal hyperparameters are selected by cross-validation.

[0042] Training stops once the preset model complexity or convergence condition is reached, resulting in the final trained adapter fault detection model. The trained adapter fault detection model is then used to predict the current comprehensive feature dataset to obtain the probability of power adapter fault states.

[0043] Furthermore, the method for determining whether the power adapter is in a faulty state includes:

[0044] If the predicted probability of a power adapter failure is greater than or equal to the preset power adapter failure probability threshold, then the power adapter is determined to be in a failure state.

[0045] If the predicted probability of a power adapter failure is less than the preset power adapter failure probability threshold, then the power adapter is determined to be in a non-faulty state.

[0046] Furthermore, the fault status data includes abnormal current, abnormal voltage, abnormal output power, abnormal temperature, abnormal vibration frequency, abnormal electrical noise, and abnormal harmonic data of the power adapter.

[0047] Furthermore, the training method for the adapter fault diagnosis model includes:

[0048] The dataset is divided into training, validation, and test sets; an adapter fault diagnosis model is constructed, which includes an input layer, a self-attention layer, a feedforward network layer, and an output layer; the input layer of the model is used to input historical fault state data, and the output layer of the model is used to output the power adapter fault type; the softmax function is used as the activation function; the adapter fault diagnosis model is a self-attention mechanism model.

[0049] Multi-class cross-entropy is used as the model's loss function to measure the difference between the model's predicted values ​​and the actual values; the multi-class cross-entropy loss function is: Where L is the average loss of the dataset; N′ is the total number of samples in the adapter fault diagnosis model; D is the number of power adapter fault types; y gd Let p be the true label of the g-th sample for the d-th type; gd Let g be the probability predicted by the model that the g-th sample belongs to the d-th type; g is the index of the sample.

[0050] The adapter fault diagnosis model is trained using the training set, and the model parameters are updated using the backpropagation algorithm to minimize the loss function. The performance of the adapter fault diagnosis model is evaluated by calculating the accuracy metric using the validation set.

[0051] The SGD optimization algorithm was selected as the optimizer. The model was tuned based on the performance feedback from the validation set. The model parameters were adjusted until the performance no longer improved or the preset number of iterations was reached. The performance of the model in the prediction task was evaluated using the test set. The trained adapter fault diagnosis model was used to predict the current fault state data to obtain the power adapter fault type.

[0052] Furthermore, the method for adapter fault location based on directed graph modeling, combined with the weighted Dijkstra algorithm and dynamically adjusted update factors, includes:

[0053] Power adapter fault types include adapter overheating fault, adapter short circuit fault, and output voltage instability fault;

[0054] S91. Model the circuit of the power adapter as a directed graph, which contains V nodes and E edges; the nodes represent the electrical components in the power adapter, the input end of the graph is the starting node, and the output end of the graph is the target node; the electrical components include the input end, converter, filter, sensor and output end of the power adapter; the edges represent the connection relationship between the electrical components.

[0055] S92. Assign a weight to each edge according to different power adapter fault types, representing the degree of adapter fault risk; the edge weight is:

[0056] in, Let be the weight of edge (e,f), representing the fault risk level from node e to node f; α is the weight coefficient corresponding to adapter short-circuit fault; β is the weight coefficient corresponding to output voltage instability fault; γ is the weight coefficient corresponding to adapter overheating fault; |I ef -I nom | is the current current I ef relative to the preset normal current I nom Deviation; |V ef -V nom | is the current voltage V ef Compared to the preset normal voltage V nom Deviation; T ef -T se The current temperature T ef Exceeding the preset safe temperature T se Deviation;

[0057] S93. Starting from the starting node s, find the shortest path from the starting node s to the target node o using Dijkstra's algorithm; define the cost of the shortest path from the starting node s to any node h as d(h), where d(h) is the sum of the minimum failure risk levels from the starting node s to any node h.

[0058] For each node u, the shortest path cost of all connected arbitrary nodes h is updated using an iterative update formula; the iterative update formula is: Where d′(h) is the updated shortest path cost; Let ξ be the weight of edge (u,h); ξ is the update factor for the shortest path cost.

[0059] The update factor of the shortest path cost is dynamically adjusted using an update factor adjustment formula, which is: Where ξ′ is the update factor of the shortest path cost after dynamic adjustment; L is the total number of nodes; and w is the number of iterations.

[0060] S94, The preset fault risk level threshold is: After finding the shortest path from the starting node s to the target node o, if the fault risk level of any node h on the path is greater than a preset fault risk level threshold... This node is then identified as the location of the adapter failure.

[0061] Furthermore, the method by which the intelligent power adapter monitoring terminal automatically triggers the corresponding fault response strategy based on the obtained power adapter fault type and fault location includes:

[0062] After the power adapter intelligent monitoring terminal detects the location of the adapter short circuit fault, it immediately triggers the power-off protection mechanism, cuts off the power input of the power adapter, and sends an alarm to the staff, informing them of the existence of a short circuit fault and its location.

[0063] After the power adapter intelligent monitoring terminal detects the location of the output voltage instability fault, it automatically adjusts the output voltage of the power adapter. If it cannot restore the normal voltage output, it automatically switches to the backup power supply and sends an alarm to the staff, informing them of the output voltage instability fault and its location.

[0064] Once the power adapter intelligent monitoring terminal detects the location of the adapter overheating fault, it immediately activates the cooling system to reduce the adapter temperature. If the cooling measures are ineffective and the temperature continues to rise, it automatically shuts down the power adapter and sends an alarm to the staff, informing them of the adapter overheating fault and its location.

[0065] The technical effects and advantages of the fault monitoring system for a power adapter of the present invention are as follows:

[0066] This invention ensures the temporality and continuity of collected power adapter parameter data, electrical performance data, and adapter operating environment data by timestamping and sorting them by timestamp, providing a solid foundation for subsequent data processing and analysis. It uses empirical mode decomposition (EMD) to decompose complex data signals into several intrinsic mode function (EMF) components, each representing a different frequency component in the data, facilitating a deeper understanding of the data's intrinsic characteristics. By limiting the number of EMF components, overfitting during data decomposition is avoided, while maintaining data simplicity and representativeness. During Hilbert transform, the range of the integral variable is dynamically adjusted using an adjustment formula, ensuring the accuracy and stability of the transform. This helps reduce transform errors caused by differences in frequency components and improves the reliability of the transform results.

[0067] By constructing a directed graph model of the power adapter circuit and assigning weights to each edge based on different types of faults (such as overheating, short circuits, and unstable output voltage), the method can accurately reflect the fault risk level of each electrical component in the adapter. By representing electrical components as nodes and the connections between components as edges, this method can cover all electrical components and connections in the adapter, thus achieving a comprehensive analysis of adapter faults. Using Dijkstra's algorithm to find the shortest path from the starting node to the target node, the method can accurately locate the fault location. Through iterative update formulas and update factor adjustment formulas, this method can dynamically adjust the update factor of the shortest path cost, thereby accelerating the algorithm's running speed and improving the efficiency of fault location. By accurately and quickly locating the fault location of the power adapter, repair time can be significantly shortened and repair costs reduced. Attached Figure Description

[0068] Figure 1 This is a schematic diagram of a fault monitoring system for a power adapter according to the present invention;

[0069] Figure 2 This is a schematic flowchart of a fault monitoring method for a power adapter according to the present invention. Detailed Implementation

[0070] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0071] Example 1

[0072] Please see Figure 1 As shown in the figure, this embodiment of a power adapter fault monitoring system includes:

[0073] The data collection module is used to collect power adapter parameter data, electrical performance data, and adapter operating environment data;

[0074] The data processing module is used to extract features from power adapter parameter data, electrical performance data, and operating environment data based on improved empirical mode decomposition combined with Hilbert transform, to obtain adapter parameter feature dataset, electrical performance feature dataset, and operating environment feature dataset.

[0075] The fault state detection module is used to perform weighted fusion of adapter parameter feature dataset, electrical performance feature dataset and working environment feature dataset through a weighted model to obtain a comprehensive feature dataset; train the adapter fault detection model, input the comprehensive feature dataset into the trained adapter fault detection model, predict the probability of power adapter fault state, and determine whether the power adapter is in a fault state.

[0076] The fault type diagnosis module generates an early warning message immediately when the power adapter is in a fault state, and automatically collects fault state data; it trains and obtains an adapter fault diagnosis model based on the fault state data, and predicts the fault type of the power adapter through the adapter fault detection model.

[0077] The fault response module, based on directed graph modeling, combines a weighted Dijkstra algorithm with a dynamically adjusted update factor to locate adapter faults. The power adapter intelligent monitoring terminal automatically triggers the corresponding fault response strategy based on the obtained power adapter fault type and location. The modules are connected to each other via wired and / or wireless means.

[0078] The power adapter parameter data includes the power adapter's input voltage, input current, output voltage, output current, operating time, and temperature; the electrical performance data includes the power adapter's output power, harmonic voltage, harmonic current, and total harmonic distortion; and the adapter's operating environment data includes the temperature, humidity, vibration frequency, and electromagnetic interference level in the adapter's operating environment.

[0079] Based on improved empirical mode decomposition combined with Hilbert transform, the method for feature extraction of power adapter parameter data, electrical performance data, and operating environment data includes:

[0080] S31. Timestamp the collected power adapter parameter data, electrical performance data, and adapter operating environment data at a daily collection interval, and sort the collected power adapter parameter data, electrical performance data, and adapter operating environment data in ascending order according to the timestamps.

[0081] S32. Integrate the sorted power adapter parameter data, electrical performance data, and adapter operating environment data to obtain a data sequence X. j For each data sequence X j Data variables in Empirical mode decomposition is performed using the empirical mode decomposition formula, decomposing it into v eigenmode function components;

[0082] The empirical mode decomposition formula is: in, Let be the signal value of the i-th variable in the j-th data at time t; K represents the k-th eigenmode function component of the i-th variable signal in the j-th data type. i The number of intrinsic mode function components obtained from the decomposition of the i-th variable signal; k is the index of the intrinsic mode function component decomposed in each signal; t represents the low-frequency signal value remaining during the signal decomposition process; j represents the index of the collected data category; t represents the sampling time of the power adapter parameter data, electrical performance data, and adapter operating environment data; i represents the index of each variable in the collected data category.

[0083] The number K of intrinsic mode function components obtained by decomposing the i-th variable signal using the intrinsic mode function component constraint formula is K. i To impose restrictions, the formula for restricting the intrinsic mode function components is as follows: Wherein, K′ i To limit the number of eigenmode function components; Q j P represents the number of data categories collected. i The number of variable categories in the collected data; a and b are constant factors that adjust the proportions of the intrinsic modal function component constraint formula;

[0084] For example, suppose we have the following data situation: the number of data categories Q collected. j The number of variable types in the first type of data collected is 3, the number of variable types in the second type of data collected is 5, the number of variable types in the third type of data collected is 8, and the number of variable types in the third type of data collected is 3. The constant factors a and b for adjusting the proportion of the intrinsic mode function component constraint formula are 10 and 2, respectively.

[0085] Calculate each K′ i quantity:

[0086] For the first type of data:

[0087] For the second type of data:

[0088] For the third type of data:

[0089] The intrinsic modal function component constraint formula reveals that as the number of collected variable types increases, the number of intrinsic modal function components decreases. Limiting the number of intrinsic modal function components controls complexity. Conversely, when the number of collected variable types is small, a larger number of intrinsic modal function components is allowed, helping to control decomposition complexity and ensuring that the number of intrinsic modal function components for each variable is within a reasonable range.

[0090] S33. For each eigenmode function component The complex signal of each component is obtained by performing a Hilbert transform using the Hilbert transform formula.

[0091] The Hilbert transform formula is: in, For each intrinsic mode function component Complex signals; For each eigenmode function component The imaginary part of the signal obtained by performing a Hilbert transform; i′ is the imaginary unit;

[0092] Where τ is the integral variable used to calculate the Hibbert transform at time t;

[0093] The range of the integral variable τ used to calculate the Hibbert transform at time t is dynamically adjusted and constrained by the integral variable range adjustment formula.

[0094] The formula for adjusting the range of integral variables is: Wherein, K′ i To limit the number of intrinsic mode function components; Let τ be the range of the integral variable;

[0095] For example, assuming the sampling time t is 10, the number of intrinsic mode function components K′ after constraint i It is 4.

[0096] According to the formula: The range of integrals can be calculated:

[0097] That is, when performing the Hilbert transform, the integration range of the integration variable τ is restricted to [8,12] instead of [-∞,∞]; the signal characteristics in this range will have a more direct impact on the Hilbert transform result near time t.

[0098] If the number of intrinsic mode functions K′ is increased i For example, K′ i =9, then the range of integration becomes: When K′ i When the value is increased, the integration range is further reduced to [9,11], which is more concentrated around time t=10, showing stronger instantaneous characteristics. This adjustment mechanism allows the Hilbert transform to dynamically adapt to the component complexity of the signal and capture instantaneous change characteristics more flexibly.

[0099] S34, from complex signal The data sequence X is extracted using the instantaneous amplitude calculation formula and the instantaneous frequency calculation formula. j The instantaneous amplitude and instantaneous frequency;

[0100] The formula for calculating instantaneous amplitude is: in, This is the instantaneous amplitude calculated at time t; For each intrinsic mode function component The square of the imaginary part of the signal obtained by performing the Hilbert transform;

[0101] The formula for calculating instantaneous frequency is:

[0102] in, Let be the instantaneous frequency calculated at time t, and represent the instantaneous rate of change of the signal frequency; The arctangent of the ratio of the imaginary part to the real part of the signal represents the phase of the complex signal; This indicates taking the derivative with respect to t;

[0103] S35. Constructing a data sequence X using instantaneous amplitude and instantaneous frequency through Hilbert's energy spectrum formula. j Hilbert energy spectrum for each variable;

[0104] Hilbert's energy spectrum formula is: in, Let be the Hilbert energy spectrum of the i-th variable in the j-th data type; The Dirac function represents the instantaneous frequency of the signal. A pulse at a given location; ω is a continuous variable of angular frequency, used to represent and analyze the energy distribution of a signal at different frequency components; specifically, ω varies over the entire real number range, while It is a specific point within this continuous range;

[0105] S36, for each data variable The intrinsic mode function components are constructed with features, and the statistical features of each intrinsic mode function are extracted. The statistical features include the standard deviation of the instantaneous amplitude and the average value of the instantaneous frequency. By calculating the statistical features of each intrinsic mode function component, a feature vector is formed. The feature vectors of each data variable are integrated into the corresponding datasets to form the adapter parameter feature dataset, the electrical performance feature dataset, and the working environment feature dataset, respectively.

[0106] Methods for obtaining a comprehensive feature dataset by weighted fusion of adapter parameter feature datasets, electrical performance feature datasets, and operating environment feature datasets using a weighted model include:

[0107] Let D1 be the adapter parameter feature dataset, D2 be the electrical performance feature dataset, and D3 be the working environment feature dataset. The weighted model is: Qz = D1·δ1 + D2·δ2 + D3·δ3; where Qz is the comprehensive feature dataset; δ1 is the weight coefficient of the adapter parameter feature dataset; δ2 is the weight coefficient of the electrical performance feature dataset; and δ3 is the weight coefficient of the working environment feature dataset.

[0108] Training methods for adapter fault detection models include:

[0109] The dataset is divided into training, validation and test sets to build an adapter fault detection model; the sample set is a subset of the dataset, and each sample set includes historical comprehensive feature dataset and corresponding power adapter fault state probability.

[0110] The model's input data is a historical comprehensive feature dataset, and the model's output label is the probability of a power adapter failure state; the sigmoid function is used as the activation function; the adapter failure detection model is a gradient boosting decision tree model.

[0111] Initialize the adapter fault detection model and set the hyperparameters for the number of trees, learning rate, and tree depth. Train the model using different combinations of hyperparameters through k-fold cross-validation, and fine-tune the initially set hyperparameters using the cross-validation results, selecting the best-performing hyperparameter combination.

[0112] The log loss function is used to measure the difference between the model's predictions and the true labels; the formula for calculating the log loss function is:

[0113] Where N is the total number of samples in the adapter fault detection model; y c Let q be the true label of the c-th sample. c Predict the probability that the c-th sample is a positive sample for the model; c is the index of the sample.

[0114] The adapter fault detection model is trained on the training set by training a new decision tree based on the residuals of the current model in each iteration. The model is tuned by adjusting the hyperparameters according to the model performance feedback. The model is retrained using the adjusted hyperparameters. After the training process is completed, the optimal hyperparameters are selected by cross-validation.

[0115] Training stops once the preset model complexity or convergence condition is reached, resulting in the final trained adapter fault detection model. The trained adapter fault detection model is then used to predict the current comprehensive feature dataset to obtain the probability of power adapter fault states.

[0116] Methods for determining whether a power adapter is faulty include:

[0117] If the predicted probability of a power adapter failure is greater than or equal to the preset power adapter failure probability threshold, then the power adapter is determined to be in a failure state.

[0118] If the predicted probability of a power adapter failure is less than the preset power adapter failure probability threshold, then the power adapter is determined to be in a non-faulty state.

[0119] The fault status data includes abnormal current, abnormal voltage, abnormal output power, abnormal temperature, abnormal vibration frequency, abnormal electrical noise, and abnormal harmonic data of the power adapter.

[0120] Training methods for adapter fault diagnosis models include:

[0121] The dataset is divided into training, validation, and test sets. An adapter fault diagnosis model is constructed, which includes an input layer, a self-attention layer, a feedforward network layer, and an output layer. The input layer of the model is used to input historical fault state data, and the output layer of the model is used to output the power adapter fault type. The softmax function is used as the activation function. The adapter fault diagnosis model is a self-attention mechanism model.

[0122] Multi-class cross-entropy is used as the model's loss function to measure the difference between the model's predicted values ​​and the actual values; the multi-class cross-entropy loss function is: Where L is the average loss of the dataset; N′ is the total number of samples in the adapter fault diagnosis model; D is the number of power adapter fault types; y gd Let p be the true label of the g-th sample for the d-th type; gd Let g be the probability predicted by the model that the g-th sample belongs to the d-th type; g is the index of the sample.

[0123] The adapter fault diagnosis model is trained using the training set, and the model parameters are updated using the backpropagation algorithm to minimize the loss function. The performance of the adapter fault diagnosis model is evaluated by calculating the accuracy metric using the validation set.

[0124] The SGD optimization algorithm was selected as the optimizer. The model was tuned based on the performance feedback from the validation set. The model parameters were adjusted until the performance no longer improved or the preset number of iterations was reached. The performance of the model in the prediction task was evaluated using the test set. The trained adapter fault diagnosis model was used to predict the current fault state data to obtain the power adapter fault type.

[0125] Methods for adapter fault location based on directed graph modeling, combined with weighted Dijkstra's algorithm and dynamically adjusted update factors, include:

[0126] Power adapter fault types include adapter overheating fault, adapter short circuit fault, and output voltage instability fault;

[0127] S91. Model the circuit of the power adapter as a directed graph, which contains V nodes and E edges. The nodes represent the electrical components in the power adapter. The input end of the graph is the starting node, and the output end of the graph is the target node. The electrical components include the input end, converter, filter, sensor, and output end of the power adapter. The edges represent the connection relationships between the electrical components.

[0128] S92. Assign a weight to each edge based on different power adapter fault types, representing the degree of adapter fault risk; the edge weights are:

[0129] in, Let be the weight of edge (e,f), representing the fault risk level from node e to node f; α is the weight coefficient corresponding to adapter short-circuit fault; β is the weight coefficient corresponding to output voltage instability fault; γ is the weight coefficient corresponding to adapter overheating fault; |I ef -I nom | is the current current I ef relative to the preset normal current I nom Deviation; |V ef -V nom | is the current voltage V ef Compared to the preset normal voltage V nom Deviation; T ef -T se The current temperature T ef Exceeding the preset safe temperature T se Deviation;

[0130] S93. Starting from the starting node s, find the shortest path from the starting node s to the target node o using Dijkstra's algorithm; define the cost of the shortest path from the starting node s to any node h as d(h), where d(h) is the sum of the minimum failure risk levels from the starting node s to any node h.

[0131] For each node u, update the shortest path cost of all connected nodes h using an iterative update formula; the iterative update formula is: Where d′(h) is the updated shortest path cost; Let ξ be the weight of edge (u,h); ξ is the update factor for the shortest path cost.

[0132] The update factor of the shortest path cost is dynamically adjusted using the update factor adjustment formula, which is: Where ξ′ is the update factor of the shortest path cost after dynamic adjustment; L is the total number of nodes; and w is the number of iterations.

[0133] For example, if the total number of nodes is 5 and the number of iterations is 8, then the update factor ξ′ of the shortest path cost after dynamic adjustment is approximately 0.38.

[0134] S94, The preset fault risk level threshold is: After finding the shortest path from the starting node s to the target node o, if the fault risk level of any node h on the path is greater than a preset fault risk level threshold... This node is then identified as the location of the adapter failure.

[0135] The methods by which a smart power adapter monitoring terminal automatically triggers corresponding fault response strategies based on the obtained power adapter fault type and fault location include:

[0136] After the power adapter intelligent monitoring terminal detects the location of the adapter short circuit fault, it immediately triggers the power-off protection mechanism, cuts off the power input of the power adapter, and sends an alarm to the staff, informing them of the existence of a short circuit fault and its location.

[0137] After the power adapter intelligent monitoring terminal detects the location of the output voltage instability fault, it automatically adjusts the output voltage of the power adapter. If it cannot restore the normal voltage output, it automatically switches to the backup power supply and sends an alarm to the staff, informing them of the output voltage instability fault and its location.

[0138] Once the power adapter intelligent monitoring terminal detects the location of the adapter overheating fault, it immediately activates the cooling system to reduce the adapter temperature. If the cooling measures are ineffective and the temperature continues to rise, it automatically shuts down the power adapter and sends an alarm to the staff, informing them of the adapter overheating fault and its location.

[0139] The preset power adapter failure probability threshold is set by the staff. Different power adapter failure probabilities are collected through the power adapter intelligent monitoring terminal, and the average of multiple power adapter failure probabilities is taken as the preset power adapter failure probability threshold; similarly, the failure risk level threshold is set.

[0140] In this embodiment, by timestamping and sorting the collected power adapter parameter data, electrical performance data, and adapter operating environment data according to the timestamps, the temporal sequence and continuity of the data are ensured, providing a solid foundation for subsequent data processing and analysis. The Empirical Mode Decomposition (EMD) method is used to decompose the complex data signal into several intrinsic mode function (EMF) components, each representing a different frequency component in the data, which helps to understand the intrinsic characteristics of the data more deeply. By limiting the number of EMF components, overfitting problems during the data decomposition process are avoided, while ensuring the simplicity and representativeness of the data. When performing the Hilbert transform, the range of the integral variable is dynamically adjusted and constrained through the integral variable range adjustment formula, ensuring the accuracy and stability of the transform. This helps to reduce transform errors caused by differences in frequency components and improve the reliability of the transform results.

[0141] By constructing a directed graph model of the power adapter circuit and assigning weights to each edge based on different types of faults (such as overheating, short circuits, and unstable output voltage), the method can accurately reflect the fault risk level of each electrical component in the adapter. By representing electrical components as nodes and the connections between components as edges, this method can cover all electrical components and connections in the adapter, thus achieving a comprehensive analysis of adapter faults. Using Dijkstra's algorithm to find the shortest path from the starting node to the target node, the method can accurately locate the fault location. Through iterative update formulas and update factor adjustment formulas, this method can dynamically adjust the update factor of the shortest path cost, thereby accelerating the algorithm's running speed and improving the efficiency of fault location. By accurately and quickly locating the fault location of the power adapter, repair time can be significantly shortened and repair costs reduced.

[0142] Example 2

[0143] Please see Figure 2 As shown, parts not described in detail in this embodiment are described in Embodiment 1. A fault monitoring method for a power adapter is provided, including:

[0144] S1. Collect power adapter parameter data, electrical performance data, and adapter operating environment data;

[0145] S2. Based on the improved experience mode decomposition combined with Hilbert transform, feature extraction is performed on the power adapter parameter data, electrical performance data and working environment data to obtain adapter parameter feature dataset, electrical performance feature dataset and working environment feature dataset.

[0146] S3. The adapter parameter feature dataset, electrical performance feature dataset and working environment feature dataset are weighted and fused using a weighted model to obtain a comprehensive feature dataset. The adapter fault detection model is trained and obtained. The comprehensive feature dataset is input into the trained adapter fault detection model to predict the probability of power adapter fault state and determine whether the power adapter is in a fault state.

[0147] S4. If the power adapter is in a faulty state, the power adapter intelligent monitoring terminal will immediately generate an early warning message and automatically collect fault status data; train and obtain the adapter fault diagnosis model based on the fault status data, and predict the power adapter fault type through the adapter fault detection model.

[0148] S5. Based on directed graph modeling, combined with the weighted Dijkstra algorithm and dynamically adjusted update factors, adapter fault location is performed. The power adapter intelligent monitoring terminal automatically triggers the corresponding fault response strategy according to the obtained power adapter fault type and adapter fault location.

[0149] Since the electronic device described in this embodiment is the electronic device used to implement the fault monitoring system based on a power adapter in the embodiments of this application, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the fault monitoring system for a power adapter described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. Any electronic device used by those skilled in the art to implement the fault monitoring system for a power adapter in the embodiments of this application falls within the scope of protection of this application.

[0150] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0151] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A fault monitoring system for a power adapter, characterized in that, include: The data collection module is used to collect power adapter parameter data, electrical performance data, and adapter operating environment data; The data processing module is used to extract features from power adapter parameter data, electrical performance data, and operating environment data based on improved empirical mode decomposition combined with Hilbert transform, to obtain adapter parameter feature dataset, electrical performance feature dataset, and operating environment feature dataset. The fault state detection module is used to perform weighted fusion of adapter parameter feature dataset, electrical performance feature dataset and working environment feature dataset through a weighted model to obtain a comprehensive feature dataset. Train the adapter fault detection model, input the comprehensive feature dataset into the trained adapter fault detection model, predict the probability of power adapter fault state, and determine whether the power adapter is in a fault state. The fault type diagnosis module will immediately generate an early warning message and automatically collect fault status data if the power adapter is in a faulty state. The adapter fault diagnosis model is trained based on fault status data, and the power adapter fault type is predicted by the adapter fault detection model. The fault response module, based on directed graph modeling, combines the weighted Dijkstra algorithm with a dynamically adjusted update factor to locate adapter faults. The power adapter intelligent monitoring terminal automatically triggers the corresponding fault response strategy based on the obtained power adapter fault type and adapter fault location. The method for adapter fault location based on directed graph modeling, combined with weighted Dijkstra's algorithm and dynamically adjusted update factors, includes: Power adapter fault types include adapter overheating fault, adapter short circuit fault, and output voltage instability fault; S91. Model the circuit of the power adapter as a directed graph, which contains V nodes and E edges; the nodes represent the electrical components in the power adapter, the input end of the graph is the starting node, and the output end of the graph is the target node; the electrical components include the input end, converter, filter, sensor and output end of the power adapter; the edges represent the connection relationship between the electrical components. S92. Assign a weight to each edge according to different power adapter failure types to indicate the degree of adapter failure risk; S93. Starting from the starting node s, find the shortest path from the starting node s to the target node o using Dijkstra's algorithm; define the cost of the shortest path from the starting node s to any node h as d(h), where d(h) is the sum of the minimum failure risk levels from the starting node s to any node h. For each node u, the shortest path cost of all connected arbitrary nodes h is updated using an iterative update formula; the iterative update formula is: Where d′(h) is the updated shortest path cost; Let ξ be the weight of edge (u,h); ξ is the update factor for the shortest path cost. The update factor of the shortest path cost is dynamically adjusted using an update factor adjustment formula, which is: Where ξ′ is the update factor of the shortest path cost after dynamic adjustment; L is the total number of nodes; and w is the number of iterations. S94, The preset fault risk level threshold is: After finding the shortest path from the starting node s to the target node o, if the fault risk level of any node h on the path is greater than a preset fault risk level threshold... This node is then identified as the location of the adapter failure. The modules are connected to each other via wired and / or wireless means.

2. The fault monitoring system for a power adapter according to claim 1, characterized in that, The power adapter parameter data includes the power adapter's input voltage, input current, output voltage, output current, operating time, and temperature. Electrical performance data includes the power adapter's output power, harmonic voltage, harmonic current, and total harmonic distortion. The adapter operating environment data includes the temperature, humidity, vibration frequency, and electromagnetic interference level in the adapter's operating environment.

3. The fault monitoring system for a power adapter according to claim 2, characterized in that, The method for feature extraction of power adapter parameter data, electrical performance data, and operating environment data based on improved empirical mode decomposition combined with Hilbert transform includes: S31. Timestamp the collected power adapter parameter data, electrical performance data, and adapter operating environment data at a daily collection interval, and sort the collected power adapter parameter data, electrical performance data, and adapter operating environment data in ascending order according to the timestamps. S32. Integrate the sorted power adapter parameter data, electrical performance data, and adapter operating environment data to obtain a data sequence X. j For each data sequence X j Data variables in Empirical mode decomposition is performed using the empirical mode decomposition formula, decomposing it into v eigenmode function components; The empirical mode decomposition formula is: in, Let be the signal value of the i-th variable in the j-th data at time t; K represents the k-th eigenmode function component of the i-th variable signal in the j-th data type; i The number of intrinsic mode function components obtained from the decomposition of the i-th variable signal; k is the index of the intrinsic mode function component decomposed in each signal; t represents the low-frequency signal value remaining during the signal decomposition process; j represents the index of the collected data category; t represents the sampling time of the power adapter parameter data, electrical performance data, and adapter operating environment data; i represents the index of each variable in the collected data category. The number K of intrinsic mode function components obtained by decomposing the i-th variable signal using the intrinsic mode function component constraint formula is K. i To impose restrictions, the formula for restricting the intrinsic mode function components is as follows: Among them, K i ′ represents the number of intrinsic mode function components after constraint; Q j P represents the number of data categories collected. i The number of variable categories in the collected data; a and b are constant factors that adjust the proportions of the intrinsic modal function component constraint formula; S33. For each eigenmode function component The complex signal of each component is obtained by performing a Hilbert transform using the Hilbert transform formula. The Hilbert transform formula is: in, For each intrinsic mode function component Complex signals; For each eigenmode function component The imaginary part of the signal obtained by performing a Hilbert transform; i′ is the imaginary unit; The Where τ is the integral variable used to calculate the Hibbert transform at time t; The range of the integral variable τ used to calculate the Hibbert transform at time t is dynamically adjusted and constrained by the integral variable range adjustment formula. The formula for adjusting the range of the integral variable is: Among them, K i ′ represents the number of intrinsic mode function components after restriction; Let τ be the range of the integral variable; S34, from complex signal The data sequence X is extracted using the instantaneous amplitude calculation formula and the instantaneous frequency calculation formula. j The instantaneous amplitude and instantaneous frequency; The formula for calculating the instantaneous amplitude is: in, This is the instantaneous amplitude calculated at time t; For each intrinsic mode function component The square of the imaginary part of the signal obtained by performing the Hilbert transform; The formula for calculating the instantaneous frequency is: in, The instantaneous frequency calculated at time t; The arctangent of the ratio of the imaginary part to the real part of the signal represents the phase of the complex signal; This indicates taking the derivative with respect to t; S35. Constructing a data sequence X using instantaneous amplitude and instantaneous frequency through Hilbert's energy spectrum formula. j Hilbert energy spectrum for each variable; The Hilbert energy spectrum formula is as follows: in, Let be the Hilbert energy spectrum of the i-th variable in the j-th data type; The Dirac function represents the instantaneous frequency of the signal. A pulse at a given location; ω is a continuous variable of angular frequency, used to represent and analyze the energy distribution of a signal at different frequency components; S36, for each data variable The intrinsic mode function components are constructed with features, and the statistical features of each intrinsic mode function are extracted. The statistical features include the standard deviation of the instantaneous amplitude and the average value of the instantaneous frequency. By calculating the statistical features of each intrinsic mode function component, a feature vector is formed. The feature vectors of each data variable are integrated into the corresponding datasets to form the adapter parameter feature dataset, the electrical performance feature dataset, and the working environment feature dataset, respectively.

4. The fault monitoring system for a power adapter according to claim 3, characterized in that, The method for obtaining a comprehensive feature dataset by weighted fusion of adapter parameter feature dataset, electrical performance feature dataset, and operating environment feature dataset using a weighted model includes: Let D1 be the adapter parameter feature dataset, D2 be the electrical performance feature dataset, and D3 be the working environment feature dataset; the weighted model is: Qz = D1·δ1 + D2·δ2 + D3·δ3; where Qz is the comprehensive feature dataset; δ1 is the weight coefficient of the adapter parameter feature dataset; δ2 is the weight coefficient of the electrical performance feature dataset; and δ3 is the weight coefficient of the working environment feature dataset.

5. The fault monitoring system for a power adapter according to claim 4, characterized in that, The training method for the adapter fault detection model includes: The dataset is divided into training, validation and test sets to build an adapter fault detection model; the sample set is a subset of the dataset, and each sample set includes historical comprehensive feature dataset and corresponding power adapter fault state probability. The model's input data is a historical comprehensive feature dataset, and the model's output label is the probability of a power adapter's fault state; the sigmoid function is used as the activation function; the adapter fault detection model is a gradient boosting decision tree model. Initialize the adapter fault detection model and set the hyperparameters for the number of trees, learning rate, and tree depth. Train the model using different combinations of hyperparameters through k-fold cross-validation, and fine-tune the initially set hyperparameters using the cross-validation results, selecting the best-performing hyperparameter combination. The logarithmic loss function is used to measure the difference between the model's predictions and the true labels; the formula for calculating the logarithmic loss function is as follows: Where N is the total number of samples in the adapter fault detection model; y c Let q be the true label of the c-th sample. c Predict the probability that the c-th sample is a positive sample for the model; c is the index of the sample. The adapter fault detection model is trained on the training set by training a new decision tree based on the residuals of the current model in each iteration. The model is tuned by adjusting the hyperparameters according to the model performance feedback. The model is retrained using the adjusted hyperparameters. After the training process is completed, the optimal hyperparameters are selected by cross-validation. Training stops once the preset model complexity or convergence condition is reached, resulting in the final trained adapter fault detection model. The trained adapter fault detection model is then used to predict the current comprehensive feature dataset to obtain the probability of power adapter fault states.

6. The fault monitoring system for a power adapter according to claim 5, characterized in that, The method for determining whether the power adapter is in a faulty state includes: If the predicted probability of a power adapter failure is greater than or equal to the preset power adapter failure probability threshold, then the power adapter is determined to be in a failure state. If the predicted probability of a power adapter failure is less than the preset power adapter failure probability threshold, then the power adapter is determined to be in a non-faulty state.

7. The fault monitoring system for a power adapter according to claim 6, characterized in that, The fault status data includes abnormal current, abnormal voltage, abnormal output power, abnormal temperature, abnormal vibration frequency, abnormal electrical noise, and abnormal harmonic data of the power adapter.

8. The fault monitoring system for a power adapter according to claim 7, characterized in that, The training method for the adapter fault diagnosis model includes: The dataset is divided into training, validation, and test sets; an adapter fault diagnosis model is constructed, which includes an input layer, a self-attention layer, a feedforward network layer, and an output layer; the input layer of the model is used to input historical fault state data, and the output layer of the model is used to output the power adapter fault type; the softmax function is used as the activation function; the adapter fault diagnosis model is a self-attention mechanism model. Multi-class cross-entropy is used as the model's loss function to measure the difference between the model's predicted values ​​and the actual values; the multi-class cross-entropy loss function is: Where L is the average loss of the dataset; N′ is the total number of samples in the adapter fault diagnosis model; D is the number of power adapter fault types; y gd Let p be the true label of the g-th sample for the d-th type; gd Let g be the probability predicted by the model that the g-th sample belongs to the d-th type; g is the index of the sample. The adapter fault diagnosis model is trained using the training set, and the model parameters are updated using the backpropagation algorithm to minimize the loss function. The performance of the adapter fault diagnosis model is evaluated by calculating the accuracy metric using the validation set. The SGD optimization algorithm was selected as the optimizer. The model was tuned based on the performance feedback from the validation set. The model parameters were adjusted until the performance no longer improved or the preset number of iterations was reached. The performance of the model in the prediction task was evaluated using the test set. The trained adapter fault diagnosis model was used to predict the current fault state data to obtain the power adapter fault type.

9. The fault monitoring system for a power adapter according to claim 8, characterized in that, In S92, the edge weights include: in, Let be the weight of edge (e,f), representing the fault risk level from node e to node f; α is the weight coefficient corresponding to adapter short-circuit fault; β is the weight coefficient corresponding to output voltage instability fault; γ is the weight coefficient corresponding to adapter overheating fault; |I ef -I nom | is the current current I ef relative to the preset normal current I nom Deviation; |V ef -V nom | is the current voltage V ef Compared to the preset normal voltage V nom Deviation; T ef -T se The current temperature T ef Exceeding the preset safe temperature T se The deviation.

10. The fault monitoring system for a power adapter according to claim 9, characterized in that, The method by which the intelligent power adapter monitoring terminal automatically triggers a corresponding fault response strategy based on the obtained power adapter fault type and fault location includes: After the power adapter intelligent monitoring terminal detects the location of the adapter short circuit fault, it immediately triggers the power-off protection mechanism, cuts off the power input of the power adapter, and sends an alarm to the staff, informing them of the existence of a short circuit fault and its location. After the power adapter intelligent monitoring terminal detects the location of the output voltage instability fault, it automatically adjusts the output voltage of the power adapter. If it cannot restore the normal voltage output, it automatically switches to the backup power supply and sends an alarm to the staff, informing them of the output voltage instability fault and its location. Once the power adapter intelligent monitoring terminal detects the location of the adapter overheating fault, it immediately activates the cooling system to reduce the adapter temperature. If the cooling measures are ineffective and the temperature continues to rise, it automatically shuts down the power adapter and sends an alarm to the staff, informing them of the adapter overheating fault and its location.

Images