LED Module Breakpoint Detection and Location Method and Device

Abstract

The present application relates to the technical field of LED modules, and discloses a method and device for detecting and locating breakpoints of an LED module. The method includes: collecting voltage and current data of series-parallel nodes of the LED module to obtain an electrical characteristic matrix and a node topology map; performing iterative calculations on the electrical characteristic matrix, processing voltage differences and current change amounts between adjacent nodes to obtain a breakpoint feature vector; inputting the breakpoint feature vector and the node topology map into a multi-task neural network for multi-layer convolution operations to obtain an electrical fault distribution map; performing feature segmentation and depth feature extraction on the electrical fault distribution map to obtain fault region feature data; establishing an electric field distribution model for the fault region feature data, and calculating electric field singular points by the least squares method to obtain breakpoint spatial coordinate data, thereby improving the accuracy of LED module breakpoint detection and achieving precise positioning of LED module breakpoints.

Description

Technical Field

[0001] This application relates to the technical field of LED modules, and particularly to a method and device for detecting and locating breakpoints of LED modules. Background Art

[0002] Due to the complexity of the series-parallel structure of LED modules, breakpoint failures are likely to occur during long-term use. Such failures not only affect the display effect but may also cause the entire module to malfunction. Traditional methods for detecting breakpoints in LED modules mainly rely on manual visual inspection or simple voltage detection, which are inefficient and inaccurate. In recent years, although some researchers have proposed detection methods based on image processing and electrical characteristic analysis, these methods can often only identify obvious faults and lack effective recognition capabilities for potential breakpoint hazards and early fault characteristics.

[0003] The existing LED module breakpoint detection technologies face three main problems: the series-parallel structure of LED modules is complex, and traditional single voltage or current detection methods are difficult to accurately locate the breakpoint position; the electrical characteristics of breakpoint failures are time-varying and non-linear, and simple feature extraction methods cannot effectively capture fault characteristics; existing detection algorithms generally lack the ability to fuse and analyze multi-source information and are difficult to achieve high-precision fault location. Summary of the Invention

[0004] This application provides a method and device for detecting and locating breakpoints of LED modules, thereby improving the accuracy of LED module breakpoint detection and achieving precise positioning of LED module breakpoints.

[0005] In the first aspect of this application, a method for detecting and locating breakpoints of LED modules is provided. The method for detecting and locating breakpoints of LED modules includes:

[0006] Collect voltage and current data of the series-parallel nodes of the LED module to obtain an electrical characteristic matrix and a node topology map;

[0007] Perform iterative calculations on the electrical characteristic matrix, process the voltage difference and current change amount between adjacent nodes, and obtain a breakpoint feature vector;

[0008] Input the breakpoint feature vector and the node topology map into a multi-task neural network for multi-layer convolution operations to obtain an electrical fault distribution map;

[0009] Perform feature segmentation and deep feature extraction on the electrical fault distribution map to obtain fault region feature data;

[0010] Establish an electric field distribution model for the fault region feature data, calculate the electric field singular points by the least squares method, and obtain the breakpoint spatial coordinate data.

[0011] The second aspect of the present application provides an LED module breakpoint detection and positioning device, and the LED module breakpoint detection and positioning device includes:

[0012] An acquisition module, configured to collect voltage and current data of the series-parallel nodes of the LED module to obtain an electrical characteristic matrix and a node topology map;

[0013] A calculation module, configured to perform iterative calculations on the electrical characteristic matrix, process the voltage difference and current change amount between adjacent nodes, and obtain a breakpoint feature vector;

[0014] An operation module, configured to input the breakpoint feature vector and the node topology map into a multi-task neural network for multi-layer convolution operations to obtain an electrical fault distribution map;

[0015] An extraction module, configured to perform feature segmentation and deep feature extraction on the electrical fault distribution map to obtain fault area feature data;

[0016] A building module, configured to build an electric field distribution model for the fault area feature data, and calculate the electric field singular point by the least squares method to obtain breakpoint spatial coordinate data.

[0017] The third aspect of the present application provides an electronic device, including: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor invokes the instructions in the memory to cause the electronic device to execute the above-mentioned LED module breakpoint detection and positioning method.

[0018] The fourth aspect of the present application provides a computer-readable storage medium, in which instructions are stored, and when the instructions run on a computer, the computer is caused to execute the above-mentioned LED module breakpoint detection and positioning method.

[0019] Compared with the prior art, the present application has the following beneficial effects: By establishing a multi-task neural network and a deep feature extraction model, and combining with a voltage domain iterative learning strategy, the accuracy of breakpoint detection is significantly improved. By adopting an electric field distribution model and a least squares optimization algorithm, precise positioning of the breakpoint position is achieved, and the spatial resolution of fault location is enhanced. By combining electrical feature analysis with deep learning methods, the limitations of traditional single detection methods are broken through, and rapid identification of breakpoint faults in complex series-parallel structures is realized. By introducing adaptive feature extraction and multi-level data fusion technologies, the resistance of the system to environmental interference is enhanced, and the detection results are more stable and reliable. Based on the data acquisition scheme of distributed nodes, combined with an efficient feature processing algorithm, real-time monitoring of large-scale LED modules is realized. By applying a depthwise separable convolution module and an attention mechanism, the computational complexity of the algorithm is reduced while ensuring the detection performance. By using the electric field curvature distribution analysis method, the ability of the system to identify early fault signs is improved, and potential breakpoint fault risks can be warned in advance. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0021] The structures, ratios, sizes, etc. shown in the drawings of this specification are only used to cooperate with the content disclosed in the specification for those who are familiar with this technology to understand and read, and are not used to limit the limited conditions that the present invention can be implemented. Therefore, they do not have a substantial technical meaning. Any modification of the structure, change of the proportional relationship, or adjustment of the size, without affecting the effects that the present invention can produce and the purposes that can be achieved, should still fall within the scope that the technical content disclosed by the present invention can cover.

[0022] Figure 1 It is a schematic flow chart of the method for detecting and positioning breakpoints of an LED module provided by an embodiment of the present invention;

[0023] Figure 2 It is a schematic block diagram of the structure of the device for detecting and positioning breakpoints of an LED module provided by an embodiment of the present invention;

[0024] Figure 3 It is a schematic block diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0026] The flowchart shown in the accompanying drawings is only an example, and does not necessarily include all the content and operations / steps, nor does it necessarily need to be executed in the described order. For example, some operations / steps can also be decomposed, combined or partially merged, so the actual execution order may change according to the actual situation.

[0027] It should also be understood that the terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. As used in the specification of this application and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0028] It should be further understood that the term "and / or" used in the specification of this application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. Please refer to Figure 1 , an embodiment of the LED module breakpoint detection and positioning method in the embodiments of this application includes:

[0029] Step 100, collect voltage and current data of the series-parallel nodes of the LED module to obtain an electrical characteristic matrix and a node topology diagram;

[0030] It can be understood that the execution subject of this application can be an LED module breakpoint detection and positioning device, or a terminal or a server. Specifically, it is not limited here. In the embodiments of this application, the server is taken as the execution subject for illustration.

[0031] Specifically, a coordinate system is established for each series-parallel node in the LED module, and spatial mapping calculations are performed to form a node coordinate data set. By analyzing the physical structure of the module and combining with the spatial geometric mapping rules, the position of each node in the actual space is represented by a unified coordinate system. Voltage and current are collected and synchronously sampled and converted for each node in the node coordinate data set, and the collected analog signals are converted into digital signals to form original electrical data, reflecting the real-time electrical characteristics of the LED module at each node. Denoising processing is performed on the original electrical data. Wavelet decomposition operation is used to perform multi-scale decomposition on the original signal, extract the high-frequency components, and perform threshold processing on the high-frequency coefficients to effectively filter out the noise and obtain the denoised signal data. The denoised signal data is divided into multiple time periods for segmented processing, and the characteristic parameters of each segment of data are extracted to generate segmented characteristic data. Zero-mean normalization processing and variance normalization calculation are performed on the voltage and current values in the segmented characteristic data, so that the data distribution has a unified mean and standard deviation, and the calculation deviation caused by the dimension difference or absolute value change is eliminated. The spatially associated operation is performed on the standardized electrical characteristics and the node coordinate data set. Using geometric mapping and node matching technologies, the characteristic mapping transformation is completed to generate a node characteristic association table. The node characteristic association table describes the relationship between the coordinate information of each node in the spatial position and its corresponding standardized electrical characteristics in the form of a matrix. Based on the node characteristic association table and combined with the series-parallel topology structure of the LED module, connection degree analysis and adjacency matrix calculation are performed. By analyzing the physical connection relationship between nodes, a node topology graph reflecting the internal network structure of the module is established. The node topology graph shows the connection paths, relationship strengths, and distribution rules of electrical characteristics in the network of the nodes in the module in the way of graph theory. The node characteristic association table is reconstructed into a characteristic structure with M×N dimensions, and the node characteristics and topology information are organized in a unified format through data rearrangement operation to obtain an electrical characteristic matrix.

[0032] Step 200: Perform iterative calculations on the electrical characteristic matrix to process the voltage difference and current change amount between adjacent nodes, and obtain a breakpoint characteristic vector;

[0033] Specifically, based on the voltage data in the electrical characteristic matrix, by calculating the voltage ratios of adjacent nodes one by one, a node voltage ratio sequence is formed to reflect the relative change relationship of voltages between nodes. Perform time series analysis on the node voltage ratio sequence, extract the dynamic characteristics of voltage changes from the time dimension, generate a voltage change rate matrix, and quantify the voltage change speed at different time points. At the same time, based on the current data in the electrical characteristic matrix, establish a current time series model, and extract the current change amount data by performing a first-order difference operation on the time series. The first-order difference can effectively eliminate the trend interference in the absolute value of the current and prominently reflect abnormal characteristics such as mutations and jumps, making the current change amount data a key indicator for characterizing breakpoint characteristics. Perform Fourier transform on the voltage change rate matrix, extract its frequency domain information, obtain voltage periodic fluctuation data, and reveal the hidden laws in the voltage signal, such as periodic anomalies caused by breakpoints. Use adaptive threshold analysis for the current change amount data, automatically adjust the threshold according to the signal distribution characteristics, and extract the current mutation point data, so as to accurately capture the abnormal behavior of the current at the breakpoint. Based on the extracted voltage periodic fluctuation data, construct a voltage domain iterative learning model, and continuously optimize the voltage feature weights through multiple weight update operations. The weight update process is carried out iteratively, and each round of update is adjusted based on the previous feature extraction results, so that the final voltage feature weights can accurately reflect the voltage abnormal characteristics caused by breakpoints. Perform feature cross-operation on the voltage feature weights and the current mutation point data, fuse the information of two different electrical characteristics, and generate fusion feature data. By introducing the cooperative relationship between multi-dimensional features through cross-operation, the fusion feature data not only contains single voltage or current features, but also captures the interaction characteristics between the two. Perform feature dimensionality reduction and normalization processing on the fusion feature data. The purpose of dimensionality reduction is to remove redundant features, reduce the computational complexity, and at the same time retain the important breakpoint-related features in the data. The normalization processing standardizes the data to zero mean and unit variance, eliminates the influence caused by the dimensional difference between features, and improves the consistency of the data in subsequent calculations. The fusion feature data after dimensionality reduction and normalization processing is the breakpoint feature vector.

[0034] Step 300: Input the breakpoint feature vector and the node topology map into a multi-task neural network for multi-layer convolution operations to obtain an electrical fault distribution map;

[0035] It should be noted that the breakpoint feature vector and the node topology graph are processed through feature mapping to construct the input data of the multi-task neural network. The input data is effectively organized in the design into a format suitable for the multi-task neural network to ensure that the spatial relationship and electrical characteristics between features are retained. The architecture of the multi-task neural network includes a shared feature extraction network and two parallel task branch networks, which are used for node feature learning and electrical boundary feature learning respectively. In the feature extraction stage, the shared feature extraction network generates shared feature data by processing the input data. The network consists of 4 convolutional layers, each convolutional layer using a 3×3 convolutional kernel, combined with the ReLU activation function to ensure the full expression of non-linear features. To stabilize the training process and accelerate convergence, a BatchNorm layer is introduced after each convolutional layer for batch normalization processing, and at the same time, part of the neurons are randomly masked through the Dropout layer to avoid overfitting problems. The shared feature extraction network can extract multi-dimensional electrical features and generate shared feature data with high expression ability. The shared feature data is respectively input into the first task branch network and the second task branch network. In the first task branch network, node feature learning is carried out, and its structure consists of 3 fully connected layers. Each layer uses the LeakyReLU activation function to enhance the gradient sensitivity to the negative value region through the characteristic of a small negative slope. Weight regularization technology is added to the network to limit the scale of parameters to improve the generalization ability and generate the node feature mapping result. In the second task branch network, the focus is on processing electrical boundary feature learning. The network contains 2 transposed convolutional layers for enhancing the resolution of the feature map and uses the Tanh activation function to output smooth boundary features. At the same time, an attention layer is introduced into the network to enhance the expression ability of key features by calculating the importance weights of each feature point and generate the boundary feature mapping result. The node feature mapping result and the boundary feature mapping result are integrated through the feature fusion step. The attention mechanism is used to dynamically adjust the weights of the two types of features to obtain more representative weighted feature data. The weighted feature data is input into the decoder network for feature decoding. The decoder network contains 3 deconvolutional layers, each layer using a 4×4 convolutional kernel and the ReLU activation function to gradually restore the original spatial resolution of the feature map while maintaining high feature expression ability. The output result of the decoder network is the feature distribution data, which contains the spatial distribution information of electrical faults in the LED module. To optimize the feature expression, data reconstruction and residual connection operations are performed on the feature distribution data. The key information in the initial features is retained through the residual connection, and at the same time, the expressiveness of the global features is enhanced. After the feature fusion is completed, the global feature data is dimensionally reduced through a 1×1 convolutional layer to generate a feature representation with low-dimensional and high expression ability. Through the node-level feature classification step, the dimensionally reduced global feature data is parsed into an electrical fault distribution map. This distribution map shows the fault location and features of the LED module in the form of spatial and topological information.

[0036] Step 400: Perform feature segmentation and deep feature extraction on the electrical fault distribution map to obtain fault area feature data;

[0037] Specifically, calculate the voltage gradient and perform statistical analysis on the electrical fault distribution map. During the calculation process, by quantifying the voltage change rate and spatial gradient distribution characteristics, determine the optimal segmentation threshold. This threshold is obtained through multiple iterations of optimization and statistical analysis, and can distinguish the fault area and the normal area to the greatest extent. Based on the optimal segmentation threshold, convert the electrical fault distribution map into a binary feature matrix to obtain the marked fault area data. Perform the extraction operation of the electrical boundary on the marked fault area data. Process the data of the fault area through the voltage gradient operator, including calculating the amplitude change and change direction of the voltage, to extract the key feature data of the electrical boundary. The voltage amplitude reflects the intensity of the electrical signal, and the change direction provides the correlation between the geometric information of the boundary and the electrical characteristics. These data together constitute the electrical boundary feature data. The gradient operator calculation can capture the local details of the electrical changes in space, ensuring the accuracy and integrity of the boundary extraction. Input the extracted electrical boundary feature data into the deep feature network, which is designed to include 5 stacked depthwise separable convolution modules. This module realizes feature extraction and dimensionality reduction through efficient convolution operations. Each module is composed of a 3×3 depth convolution and a 1×1 point convolution. Among them, the depth convolution is responsible for extracting spatial features, and the point convolution completes the channel fusion of features, so as to obtain a highly expressive feature description vector at a relatively low computational cost. Enhance the features of the feature description vector through the self-attention mechanism. The self-attention mechanism dynamically calculates the weight distribution of the feature vector, improves the expression ability of key features, and at the same time suppresses redundant or irrelevant information to generate enhanced feature data. Perform feature fusion on the enhanced feature data and the original fault area data. The feature fusion process combines the global electrical characteristics and local deep features of the fault area to form a comprehensive expression form and generate fused feature data. Reconstruct the electrical features of the fused feature data. Through the reconstruction operation, restore the data of the fault area with high fidelity to generate the reconstructed area data. During the reconstruction process, combine the original features and deep features to ensure that the data contains both local high-precision information and global electrical characteristics. After completing the data reconstruction, perform feature optimization and data fitting on the reconstructed area data to remove existing noise or redundant features and improve the continuity and accuracy of the data. The obtained fault area feature data not only has a high-quality description of electrical characteristics, but also contains complete spatial and boundary information.

[0038] Step 500: Establish an electric field distribution model for the fault area feature data, and calculate the electric field singular points by the least squares method to obtain the break point spatial coordinate data.

[0039] Specifically, the fault area characteristic data is divided into potential grids to generate a potential sampling point matrix in the form of uniformly distributed grids. By spatially discretizing the electrical characteristics of the fault area, it is ensured that the potential distribution of the sampling points can comprehensively cover the entire fault area. The Gauss-Seidel iterative calculation is performed on the potential sampling point matrix to obtain discrete potential data in a rapidly converging manner, and these data locally reflect the distribution law of the potential in the fault area. According to the second-order central difference format, an electric field potential energy equation is established for the discrete potential data, and the electric field intensity distribution matrix is calculated through finite difference operations. By accurately solving the potential gradient, the electric field intensity value of each grid point is generated, which can reflect the change trend and spatial distribution characteristics of the electric field. After obtaining the electric field intensity distribution, each grid point in the distribution matrix is substituted into the Poisson equation, and at the same time, the boundary conditions of the fault area are combined for constraint operations to generate the electric field Poisson distribution data. The Poisson equation can describe the relationship between the potential and the charge distribution, and the physical consistency and accuracy of the distribution data are enhanced through the constraints of the boundary conditions. A least-squares fitting objective function is established for the electric field Poisson distribution data, with the optimization goal of minimizing the error between the electric field intensity and the model prediction value, and the electric field fitting coefficient matrix is optimized through the gradient descent algorithm. The optimized fitting coefficient matrix can completely describe the local characteristics and overall distribution trend of the electric field. The electric field fitting coefficient matrix is input into the Laplace operator model for second-order partial derivative calculation to generate an electric field curvature distribution map. The electric field curvature distribution map can reveal the direction and rate of the electric field change and highlight the abnormal points in the electric field distribution. The singular value decomposition is performed on the electric field curvature distribution map, the main eigenvectors are identified through eigenvalue sorting, and the key characteristics of the electric field distribution are extracted. Using the extreme point detection method, the local extreme points in the main eigenvector are identified to generate a set of singular point candidates, and these candidate points are concentrated in the area of abnormal electric field changes to preliminarily locate the potential position of the break point. Spatial aggregation analysis is performed on the set of singular point candidates. Through the clustering algorithm, the singular points that are close in space are aggregated to obtain the clustering center points, and these center points are used as the candidate positions of the break point, significantly reducing the redundancy of the candidate points. Coordinate transformation and accuracy correction are performed on the clustering center points to eliminate the deviation introduced by coordinate system transformation or calculation errors. The corrected clustering center points accurately represent the space of the break point in the LED module.

[0040] The Poisson distribution data of the electric field is expanded by the N - order Legendre polynomial. This polynomial is characterized by its orthogonality and can efficiently decompose the complex components in the Poisson distribution of the electric field. Through the orthogonal expansion of the Poisson distribution data, the expansion coefficients of each order of the basis function are calculated, obtaining a set of basis function expansion coefficients that describe the projections of the Poisson distribution on each basis function. Based on the basis function expansion coefficients, a least - squares fitting objective function is constructed to optimize the electric field fitting coefficient matrix. In this process, to avoid overfitting and improve the generalization ability of the model, the L2 - norm regularization constraint is introduced into the objective function, forming a loss function with a penalty term. The design of the loss function aims to minimize the fitting error of the electric field distribution while restricting the parameter scale through the penalty term. The variables of the loss function with the penalty term are differentiated, and the gradient direction is calculated through forward - difference and backward - difference operations respectively to obtain the complete gradient direction vector. After the gradient direction vector is determined, the optimization step size is dynamically adjusted by setting an adaptive learning rate to ensure the balance between the convergence speed and accuracy. The model is iteratively optimized using the backpropagation algorithm, continuously fitting and updating the Poisson distribution data of the electric field to obtain the optimized electric field fitting coefficient matrix. This matrix reflects the global trend and local details of the electric field distribution. The electric field fitting coefficient matrix is input into the two - dimensional Laplace operator model, and the second - order partial derivatives are calculated for it through the finite - difference method to generate a second - derivative tensor. The high - precision characteristics of the finite - difference method ensure the reliability of the tensor calculation, and the second - derivative tensor describes the local change rate and bending characteristics of the electric field. The principal curvature of the second - derivative tensor is calculated. According to the formulas of Gaussian curvature and mean curvature, the data in the second - derivative tensor are processed respectively to extract the key curvature feature data. Gaussian curvature reflects the overall bending degree of the surface at each point, while mean curvature reveals the smoothness and trend of the local change of the electric field. The curvature feature data are reconstructed according to the electric field distribution, and the reconstructed data are smoothed and spatially extended through bicubic interpolation to generate the final electric field curvature distribution map. The high - order smoothing characteristics of bicubic interpolation ensure the spatial continuity and accuracy of the curvature distribution map, and the data at each point in the map clearly show the change characteristics of the electric field at the corresponding position.

[0041] In the embodiments of the present application, by establishing a multi-task neural network and a deep feature extraction model, combined with a voltage domain iterative learning strategy, the accuracy of breakpoint detection is significantly improved. The electric field distribution model and the least squares optimization algorithm are used to achieve precise positioning of the breakpoint position and improve the spatial resolution of fault location. The combination of electrical feature analysis and deep learning methods breaks through the limitations of traditional single detection methods and achieves rapid identification of breakpoint faults in complex series-parallel structures. By introducing adaptive feature extraction and multi-level data fusion technology, the system's resistance to environmental interference is enhanced, making the detection results more stable and reliable. A data acquisition solution based on distributed nodes, combined with an efficient feature processing algorithm, realizes real-time monitoring of large-scale LED modules. By applying a deep separable convolution module and an attention mechanism, the computational complexity of the algorithm is reduced while ensuring detection performance. The electric field curvature distribution analysis method is used to improve the system's ability to identify early fault signs and can provide early warning of potential breakpoint failure risks.

[0042] In a specific embodiment, the process of executing step 100 may specifically include the following steps:

[0043] Establish a coordinate system and perform spatial mapping calculations for each series-parallel node in the LED module to obtain a node coordinate data set;

[0044] Perform voltage and current acquisition and synchronous sampling conversion on each node in the node coordinate data set to obtain raw electrical data;

[0045] Perform wavelet decomposition and high-frequency coefficient threshold processing on the original electrical data to obtain the noise-reduced signal data;

[0046] The noise-reduced signal data is segmented to obtain segmented feature data, and the voltage and current values in the segmented feature data are subjected to zero-mean normalization and variance normalization calculation to obtain standardized electrical features;

[0047] Perform spatial correlation operation and feature mapping transformation on the standardized electrical characteristics and the node coordinate data set to obtain a node feature correlation table;

[0048] Based on the series-parallel topology of the LED module, the node feature association table is subjected to connectivity analysis and adjacency matrix calculation to obtain a node topology graph.

[0049] The node feature association table is reconstructed into an M×N dimensional feature structure, and the electrical feature matrix is obtained through data rearrangement operation.

[0050] Specifically, a coordinate system is established for the series - parallel nodes of the LED module, and spatial mapping calculations are performed to obtain a complete node coordinate dataset. By analyzing the physical structure of the LED module, the geometric position of each node in the module is determined and mapped into a three - dimensional coordinate system. Assuming that the physical position of the node is represented by where respectively represent the positions of the th node on the three coordinate axes, then the spatial mapping result of the entire node is expressed as where is the total number of nodes, and represents the node coordinate dataset. The voltage and current of each node in the node coordinate dataset are collected. At the same time, the analog signal is converted into a digital signal through synchronous sampling conversion technology to obtain the original electrical data. Assuming that the collected voltage and current signals are and respectively, where represents time, represents the node number, and the original electrical data of each node is expressed as where is the electrical characteristic signal set of the node . Wavelet decomposition operation and threshold processing of high - frequency coefficients are performed on the original electrical data to obtain the denoised signal data. Wavelet decomposition effectively separates different frequency components by projecting the signal onto a series of time - frequency bases. For the signal , the formula for wavelet decomposition is:

[0051] ;

[0052] where is the wavelet decomposition coefficient, is the th wavelet basis function of the rd layer, is the decomposition layer number, and is the number of coefficients in each layer. High - frequency noise is filtered out through the threshold , and its processing is:

[0053] ;

[0054] The processed generates the denoised signal data. After segmenting the denoised signal data, segmented feature data is obtained. Assuming the segmentation time window is , then the data of each segment of the signal is expressed as . Zero - mean normalization and variance normalization calculations are respectively performed on the voltage and current values in each segment of the signal to ensure the consistency and comparability of data between different nodes. The formula for normalization processing is:

[0055] ;

[0056] wherein and are the mean and standard deviation of the voltage respectively, and are the mean and standard deviation of the current respectively. The standardized data and constitute the standardized electrical features. Perform a spatial association operation on the standardized electrical features and the node coordinate dataset, combine the geometric positions of the nodes, and generate a node feature association table through feature mapping transformation. The feature mapping formula is:

[0057] ;

[0058] wherein is the mapping function, represents the complete feature data of node . Based on the generated node feature association table, combine the series-parallel topological structure of the LED module, analyze the node connectivity, and construct a node topology graph by calculating the adjacency matrix . The definition of the adjacency matrix is:

[0059] ;

[0060] The node topology graph reflects the physical connection relationship of the nodes in the LED module. Reconstruct the node feature association table into a -dimensional feature structure, wherein is the feature dimension, is the number of nodes. Through data rearrangement operation, make it adapt to the input format of subsequent analysis and generate the final electrical feature matrix. The rearrangement formula is:

[0061] ;

[0062] wherein is the electrical feature matrix, which contains the spatial and electrical features of all nodes.

[0063] In a specific embodiment, the process of executing step 200 may specifically include the following steps:

[0064] Calculate the voltage ratio of adjacent nodes according to the voltage data in the electrical feature matrix to obtain a node voltage ratio sequence, and perform a time series analysis on the node voltage ratio sequence to obtain a voltage change rate matrix;

[0065] Establish a current time series model according to the current data in the electrical feature matrix, and obtain the current change data through first-order difference operation;

[0066] The voltage change rate matrix is Fourier transformed to obtain voltage periodic fluctuation data, and the current change data is subjected to adaptive threshold analysis to obtain current mutation point data;

[0067] A voltage domain iterative learning model is constructed based on voltage periodic fluctuation data, and voltage feature weights are obtained through weight update operations.

[0068] The current mutation point data and the voltage feature weight are subjected to feature cross operation to obtain fused feature data, which is then subjected to feature dimension reduction and normalization to obtain the breakpoint feature vector.

[0069] Specifically, the voltage ratio of adjacent nodes is calculated based on the voltage data in the electrical characteristic matrix. Assume that the voltage data in the electrical characteristic matrix is ,in Indicates the The voltage value of each node, is the total number of nodes. The voltage ratio of adjacent nodes is expressed as:

[0070] ;

[0071] in is a node and The first in the voltage ratio sequence Item. The voltage ratios of all adjacent nodes are calculated using the above formula to obtain the node voltage ratio sequence. The time series analysis of the ratio sequence is carried out to analyze its changing pattern over time. Time series analysis is achieved by calculating the rate of change of the ratio over time. Assuming that The ratio sequence when , then the voltage change rate matrix Expressed as:

[0072] ;

[0073] in is a node In time The voltage change rate, is the time interval. The calculated voltage change rate matrix reflects the dynamic change of the voltage ratio of adjacent nodes. At the same time, according to the current data in the electrical characteristic matrix , establish a current time series model for each node. The current time series is expressed as ,in Indicates time, Represents the node number. The current change is calculated by performing the first-order difference operation on the time series:

[0074] ;

[0075] Among them represents the node at time of the current change. The first-order difference can highlight dynamic features such as current mutations. Fourier transform the voltage change rate matrix to extract its frequency-domain features, especially the periodic fluctuation information. The Fourier transform formula is:

[0076] ;

[0077] Among them represents the Fourier coefficient at frequency , is the length of the time series, is the imaginary unit. The result of the Fourier transform is the voltage periodic fluctuation data , reflecting the main periodic components of the voltage ratio change. At the same time, perform adaptive threshold analysis on the current change data to automatically screen out the mutation point data. The core of the adaptive threshold analysis lies in dynamically adjusting the threshold to adapt it to the signal distribution. For the current change data , the recognition rule for mutation points is:

[0078] Mutation point: , ;

[0079] Among them and are the mean and standard deviation of the current change respectively, is the control parameter. In this way, effectively extract the current mutation point data J. Based on the extracted voltage periodic fluctuation data , construct a voltage-domain iterative learning model. This model optimizes the voltage feature weight through a weight update mechanism, and the objective function is to minimize the weight error:

[0080] ;

[0081] Among them is the target value, is the input feature, is the weight. Update the weight through gradient descent:

[0082] ; [[ID=...]]

[0083] Among them is the learning rate, represents the number of iterations. After multiple iterations, obtain the optimized voltage feature weight . The current mutation point data and voltage characteristic weight Perform feature cross operation to generate fusion feature data The operation rules of feature crossover are:

[0084] ;

[0085] in is the first Elements. Fusion feature data By integrating voltage and current information, the breakpoint characteristics are fully reflected. The fused feature data is subjected to dimensionality reduction and normalization. Dimensionality reduction is achieved through principal component analysis, whose goal is to retain the direction with the largest data variance, and normalization is achieved through zero-mean standardization:

[0086] ;

[0087] in and are the feature mean and standard deviation respectively. The normalized data The breakpoint feature vector , used for subsequent breakpoint detection and positioning.

[0088] In a specific embodiment, the process of executing step 300 may specifically include the following steps:

[0089] The breakpoint feature vectors and node topology graph are processed through feature mapping to construct the input data of the multi-task neural network, which includes a shared feature extraction network and two parallel task branch networks;

[0090] The shared feature extraction network is used to extract features from the input data to obtain shared feature data. The shared feature extraction network consists of four convolutional layers. Each convolutional layer uses a 3×3 convolution kernel and a ReLU activation function. BatchNorm and Dropout layers are added between each layer.

[0091] The shared feature data is input into the first task branch network for node feature learning to obtain the node feature mapping result. The first task branch network contains 3 fully connected layers, each layer uses the LeakyReLU activation function and weight regularization;

[0092] The shared feature data is input into the second task branch network for electrical boundary feature learning to obtain the boundary feature mapping result. The second task branch network contains 2 transposed convolutional layers and 1 attention layer, and each layer uses the Tanh activation function.

[0093] Feature fusion is performed on the node feature mapping result and the boundary feature mapping result, and the feature weights are calculated through the attention mechanism to obtain weighted feature data;

[0094] The weighted feature data is input into the decoder network for feature decoding to obtain feature distribution data. The decoder network contains 3 transposed convolutional layers, and each layer uses a 4×4 convolutional kernel and the ReLU activation function;

[0095] Data reconstruction and residual connection operations are performed on the feature distribution data to obtain global feature data, and the global feature data is dimensionally reduced through a 1×1 convolutional layer. After node-level feature classification, an electrical fault distribution map is obtained.

[0096] Specifically, feature mapping processing is performed on the breakpoint feature vector and the node topology graph to construct the input data of the multi-task neural network. Assume that the breakpoint feature vector is represented by the matrix where is the number of nodes, is the feature dimension; the node topology graph is represented by the adjacency matrix where represents that there is a connection between nodes and and represents no connection. Through the feature mapping function , the breakpoint feature is combined with the topological structure to generate the input tensor where is the number of channels after mapping. The input data is input into the shared feature extraction network, which contains 4 convolutional layers, and each convolutional layer uses a convolutional kernel of size where is the convolutional layer number, is the number of channels of the previous layer, and respectively represent the number of channels of the previous layer and the current layer. The convolution operation formula is:

[0097] ;

[0098] where is the activation output of the th layer, is the bias term. A BatchNorm layer is added after each convolutional layer to normalize the activation values to improve the training stability. At the same time, some neurons are randomly masked through the Dropout layer to prevent overfitting. After passing through the shared feature extraction network, the shared feature data is obtained, where and are the size and number of channels after feature dimensionality reduction. The shared feature data They are respectively input into two task branch networks. In the first task branch network, node feature learning is carried out. This network contains 3 fully connected layers, and the output of each layer passes through the formula:

[0099] ;

[0100] where and are the weight and bias matrices respectively, is the input feature of the th layer. The LeakyReLU activation function introduces a small negative slope to keep the negative gradient non-zero, and its formula is:

[0101] ;

[0102] To prevent overfitting, the weight matrix introduces regularization constraints. In the second task branch network, electrical boundary feature learning is carried out. This network contains 2 transposed convolutional layers and 1 attention layer. The purpose of the transposed convolution is to upsample the features and restore the original spatial resolution, and its calculation formula is:

[0103] ;

[0104] where is the output feature map, and the Tanh activation function ensures that the output value is in the range of [-1, 1]. The attention layer enhances the features by learning the feature weights , and its formula is:

[0105] ;

[0106] where represents element-wise multiplication. The node feature mapping result and the boundary feature mapping result are fused. The fusion weight is calculated through the attention mechanism, and the weighting formula is:

[0107] ;

[0108] where and are the node feature and the boundary feature respectively. The fused feature data is input into the decoder network for feature decoding. The decoder network contains 3 deconvolutional layers, and each layer passes through the formula:

[0109] ;

[0110] where is the The decoded output of the layer. The decoded feature data is reconstructed into global feature data, and the original information of the features is maintained through residual connections. Through a convolutional layer to reduce the dimension of the global feature data, and output node-level feature data, and generate an electrical fault distribution map through a classification operation , where is the number of categories.

[0111] In a specific embodiment, the process of executing step 400 may specifically include the following steps:

[0112] Perform voltage gradient calculation and statistical analysis on the electrical fault distribution map to obtain the optimal segmentation threshold, and convert the electrical fault distribution map into a binary feature matrix according to the optimal segmentation threshold to obtain the labeled fault area data;

[0113] Extract the electrical boundary of the labeled fault area data, calculate the voltage amplitude and change direction through the voltage gradient operator to obtain the electrical boundary feature data;

[0114] Input the electrical boundary feature data into a deep feature network to obtain a feature description vector. The deep feature network includes 5 stacked depthwise separable convolution modules, and each depthwise separable convolution module uses a combination of 3×3 depth convolution and 1×1 point convolution;

[0115] Enhance the features of the feature description vector through the self-attention mechanism to obtain enhanced feature data, and fuse the enhanced feature data with the original fault area data to obtain fused feature data;

[0116] Reconstruct the electrical features of the fused feature data to obtain the reconstructed area data, and perform feature optimization and data fitting on the reconstructed area data to obtain the fault area feature data.

[0117] Specifically, perform voltage gradient calculation and statistical analysis on the electrical fault distribution map to determine the optimal segmentation threshold. The gradient calculation uses the Sobel operator and to calculate the horizontal and vertical voltage gradients respectively, and its formula is:

[0118] ;

[0119] ;

[0120] where and respectively represent the voltage gradient components at point , and is the Sobel kernel matrix, which is used to detect changes in the horizontal and vertical directions. The comprehensive gradient amplitude is:

[0121] ;

[0122] Through statistical analysis The histogram distribution of , select the segmentation point that makes the gradient change most obvious as the optimal segmentation threshold , which is solved using Otsu's algorithm:

[0123] ;

[0124] in is the threshold The corresponding inter-class variance. Using the optimal segmentation threshold , the electrical fault distribution map Convert to a binary feature matrix:

[0125] ;

[0126] Among them, the binary matrix Mark the location of the fault area. Perform electrical boundary extraction and use the voltage gradient operator to calculate the voltage amplitude and change direction at the fault area boundary. Calculated by the gradient, and the direction of change Expressed as:

[0127] ;

[0128] The obtained boundary feature data consists of two parts: amplitude and direction, which respectively represent the strength and direction characteristics of the boundary position. The electrical boundary feature data is input into the deep feature network for high-dimensional feature extraction. The deep feature network consists of 5 layers of depth-separable convolution modules, each module includes a Depthwise convolution and a Point convolution. Depth convolution is responsible for extracting local features in the spatial dimension, and its calculation formula is:

[0129] ;

[0130] in is the depthwise convolution output, are input features, is the depth convolution kernel. Point convolution is used for inter-channel feature fusion, the formula is:

[0131] ;

[0132] in is the weight of the point convolution kernel. After 5 layers of stacking, the deep feature network outputs a high-dimensional feature description vector . For the feature description vector , the self-attention mechanism is applied for feature enhancement. The self-attention mechanism calculates the similarity weights of each feature with other features to enhance the expression ability of key features, and its formula is:

[0133] ;

[0134] where and are the query and key representations of the feature vector. The enhanced feature data is:

[0135] ;

[0136] where is the eigenvalue representation. The enhanced feature data is fused with the original fault area data , and the fusion formula is:

[0137] ;

[0138] where and are the fusion weights. The fused feature data is reconstructed for electrical features. The feature data is upsampled through a transposed convolution layer to restore the original resolution, and its formula is:

[0139] ;

[0140] where is the upsampling factor, is the reconstructed feature. The reconstructed area data is optimized for features and data fitting, and the least squares fitting formula is used:

[0141] ;

[0142] where is the target data, and by optimizing the loss function , the optimized fault area feature data is obtained.

[0143] In a specific embodiment, the process of executing step 500 may specifically include the following steps:

[0144] The fault area feature data is divided into a potential grid to obtain a potential sampling point matrix, and the Gauss-Seidel iteration calculation is performed on the potential sampling point matrix to obtain discrete potential data;

[0145] Construct the electric field potential energy equation for the discrete potential data according to the second-order central difference format, and obtain the electric field strength distribution matrix through finite difference operations;

[0146] Substitute each grid point in the electric field strength distribution matrix into the Poisson equation, and obtain the electric field Poisson distribution data through boundary condition constrained operations;

[0147] Establish a least squares fitting objective function for the electric field Poisson distribution data, obtain the electric field fitting coefficient matrix through gradient descent optimization, and input the electric field fitting coefficient matrix into the Laplace operator model for second-order partial derivative calculation to obtain the electric field curvature distribution map;

[0148] Perform singular value decomposition on the electric field curvature distribution map, obtain the main eigenvector through eigenvalue sorting, and perform extreme point detection on the main eigenvector to obtain the candidate set of singular points;

[0149] Spatially aggregate the candidate set of singular points through cluster analysis to obtain the cluster center points, and perform coordinate transformation and accuracy correction on the cluster center points to obtain the break point spatial coordinate data.

[0150] Specifically, divide the fault area feature data into a potential sampling point matrix according to a uniform grid , where and are the number of rows and columns of the grid. The initial potential value of each grid point is obtained by interpolation or mapping from . Perform Gauss-Seidel iterative calculations on the potential sampling point matrix to solve for the distribution of the potential in the equilibrium state. The update formula for Gauss-Seidel iteration is:

[0151] ;

[0152] where represents the potential value of the point at the th iteration, and the iteration continues until the convergence condition is satisfied, is the set error threshold. After the iteration is completed, construct the electric field potential energy equation for the discrete potential data using the second-order central difference format. The and components of the electric field strength are calculated respectively as:

[0153] ;

[0154] where and are the grid spacings. Combine the calculated electric field components into the electric field strength distribution matrix , each of its elements has a size of:

[0155] ;

[0156] Substitute each grid point in the electric field intensity distribution matrix into the Poisson equation , calculate the numerical solution of the Poisson equation through finite differences, and adjust the electric field Poisson distribution data in combination with boundary condition constraints . The boundary conditions are set to fixed potential values (Dirichlet boundary conditions) or fixed electric field intensities (Neumann boundary conditions), and the appropriate constraint form is selected according to the actual scenario. For the electric field Poisson distribution data establish a least squares fitting objective function, and the goal is to fit the optimal electric field distribution parameters. The objective function is defined as:

[0157] ;

[0158] Update the fitting parameter matrix through gradient descent optimization :

[0159] ;

[0160] where is the learning rate, is the fitting coefficient matrix. After fitting, input the electric field fitting coefficient matrix into the Laplace operator model for second-order partial derivative calculation to generate an electric field curvature distribution map. The calculation formula for curvature is:

[0161] ;

[0162] where is the curvature value, which reflects the local change rate of the electric field distribution. After obtaining the electric field curvature distribution map, perform eigenanalysis on it through singular value decomposition. The decomposition form of the curvature matrix is:

[0163] ;

[0164] where is a diagonal matrix, and its diagonal elements are singular values, representing the main feature intensities of the electric field distribution. By sorting the singular values, extract the main eigenvector , and then perform extreme point detection on . The detection conditions for extreme points are:

[0165] ;

[0166] The detected extreme points form a candidate set of singular points. Cluster analysis is performed on the candidate set of singular points, and the candidate points in space are aggregated through algorithms such as K-Means or DBSCAN to obtain the cluster center points. Let the set of singular points be , and the clustering objective is to minimize the within-cluster distance:

[0167] ;

[0168] where is the th cluster, and is its center point. After clustering, coordinate transformation is performed on each center point , converting it from the grid coordinate system to the actual space coordinate system, and accuracy correction is performed in combination with the error model to obtain the spatial coordinates

[0169] In a specific embodiment, the process of performing steps to establish a least-squares fitting objective function for the electric field Poisson distribution data, obtaining the electric field fitting coefficient matrix through gradient descent optimization, and inputting the electric field fitting coefficient matrix into the Laplace operator model for second-order partial derivative calculation to obtain the electric field curvature distribution map may specifically include the following steps:

[0170] Perform basis function expansion on the electric field Poisson distribution data through the Nth-order Legendre polynomial, and calculate the basis function expansion coefficients according to orthogonality;

[0171] Construct a least-squares fitting objective function based on the basis function expansion coefficients, obtain a loss function with a penalty term through L2-norm regularization constraint, and perform variable differentiation on the loss function with a penalty term to obtain the gradient direction vector through forward difference and backward difference operations;

[0172] Set an adaptive learning rate according to the gradient direction vector, perform iterative operations through the backpropagation algorithm to obtain the electric field fitting coefficient matrix, and input the electric field fitting coefficient matrix into the two-dimensional Laplace operator for finite difference to obtain the second-order derivative tensor;

[0173] Perform principal curvature calculation on the second-order derivative tensor, obtain the curvature characteristic data through the Gaussian curvature and mean curvature formulas, and reconstruct the curvature characteristic data according to the electric field distribution, and obtain the electric field curvature distribution map through bicubic interpolation.

[0174] Specifically, perform basis function expansion on the electric field Poisson distribution data through the Nth-order Legendre polynomial. The Legendre polynomial is a set of polynomials orthogonal on the interval [-1,1], used to represent the decomposition of complex functions. Assume that the electric field Poisson distribution data is , where is the two-dimensional coordinate, and the recurrence relation of the Legendre polynomial is:

[0175] ;

[0176] Expand into the basis function form of Legendre polynomials, expressed as:

[0177] ;

[0178] where are the coefficients of the basis function expansion, representing the projection of the Poisson distribution on the Legendre basis. Using the orthogonality of Legendre polynomials, the calculation formula for its expansion coefficients is:

[0179] ;

[0180] The above formula is implemented through numerical integration, taking as the core parameter of the basis function expansion. Based on the calculated basis function expansion coefficients , construct the least squares fitting objective function to minimize the reconstruction error of the electric field Poisson distribution . The objective function is defined as:

[0181] ;

[0182] To prevent overfitting, introduce norm regularization constraint, so that a penalty term is added to the loss function, which is modified to:

[0183] ;

[0184] where is the regularization coefficient, used to balance the fitting error and the complexity of the coefficients. Take the derivative of the loss function with variables to obtain the gradient direction of each coefficient. Assume represents the gradient of , then there is:

[0185] ;

[0186] The gradient calculation is realized through the finite difference operations of forward difference and backward difference to ensure the accuracy of the numerical gradient. Set the adaptive learning rate according to the gradient direction to optimize and update the coefficient matrix :

[0187] ;

[0188] The learning rate is dynamically adjusted to ensure iterative convergence. Through multiple backpropagation optimizations, finally obtain the electric field fitting coefficient matrix The coefficient matrix obtained by fitting is input into the two-dimensional Laplace operator model, and the second-order partial derivatives are calculated using the finite difference method to generate a second derivative tensor . The definition of the two-dimensional Laplace operator is:

[0189] ;

[0190] After discretization, the finite difference calculation of the second derivative is:

[0191] ;

[0192] ;

[0193] The calculation result is stored as a second derivative tensor , and each element of it represents the second-order change of the local electric field distribution. The principal curvature of the second derivative tensor is calculated, and the curvature characteristic data is extracted using the Gaussian curvature and mean curvature formulas. The Gaussian curvature and the mean curvature formulas are:

[0194] ;

[0195] The geometric distribution of the electric field is reconstructed through the curvature characteristic data and . To improve the smoothness of the reconstruction, the bicubic interpolation method is used to interpolate the curvature characteristic data, and the low-resolution curvature data is upsampled to high resolution to generate an electric field curvature distribution map.

[0196] This embodiment further includes: constructing an error compensation model based on the breakpoint spatial coordinate data, and optimizing the parameters through a deep reinforcement learning algorithm to obtain a corrected detection result. Specifically: performing time series analysis and spatial distribution statistics on the breakpoint spatial coordinate data to obtain coordinate error feature data; inputting the coordinate error feature data into a deep reinforcement learning network, and performing feature mapping through a state encoder to obtain a state vector; constructing a Q-learning network based on double time difference to evaluate the action value of the state vector and obtain compensation strategy parameters; injecting exploration noise into the compensation strategy parameters, generating a compensation action sequence through an ε-greedy strategy to obtain a candidate compensation scheme; inputting the candidate compensation scheme into an experience replay buffer pool, screening experience data through priority sampling to obtain a training sample set; performing target network update on the training sample set, calculating the time difference target through a soft update strategy to obtain a network optimization target; constructing a policy gradient loss function based on the network optimization target, calculating the compensation gradient through a double Q network architecture to obtain a parameter update vector; inputting the parameter update vector into an adaptive moment estimation optimizer, and obtaining compensation model parameters through gradient normalization operation; performing stability evaluation on the compensation model parameters, analyzing the convergence domain through a Lyapunov function to obtain a parameter feasible domain; applying the optimal compensation parameters within the parameter feasible domain to the original coordinate data, and obtaining a corrected detection result through coordinate transformation operation.

[0197] The method for detecting and locating breakpoints of the LED module in the embodiment of the present application has been described above. Next, the device 10 for detecting and locating breakpoints of the LED module in the embodiment of the present application will be described. Please refer to Figure 2 , an embodiment of the device 10 for detecting and locating breakpoints of the LED module in the embodiment of the present application includes:

[0198] An acquisition module 11, configured to collect voltage and current data of the series-parallel nodes of the LED module to obtain an electrical characteristic matrix and a node topology diagram;

[0199] A calculation module 12, configured to perform iterative calculations on the electrical characteristic matrix, process the voltage difference and current change amount between adjacent nodes, and obtain a breakpoint feature vector;

[0200] An operation module 13, configured to input the breakpoint feature vector and the node topology diagram into a multi-task neural network for multi-layer convolution operations to obtain an electrical fault distribution map;

[0201] An extraction module 14, configured to perform feature segmentation and deep feature extraction on the electrical fault distribution map to obtain fault area feature data;

[0202] A establishment module 15, configured to establish an electric field distribution model for the fault area feature data, calculate the electric field singular point through the least square method, and obtain the breakpoint spatial coordinate data.

[0203] Through the collaborative cooperation of the above-mentioned various components, by establishing a multi-task neural network and a deep feature extraction model, and combining the voltage-domain iterative learning strategy, the accuracy of breakpoint detection has been significantly improved. By adopting the electric field distribution model and the least squares optimization algorithm, the precise positioning of the breakpoint location has been achieved, and the spatial resolution of fault location has been enhanced. By combining electrical feature analysis with deep learning methods, the limitations of traditional single detection methods have been broken through, and the rapid identification of breakpoint faults in complex series-parallel structures has been realized. By introducing adaptive feature extraction and multi-level data fusion technologies, the resistance of the system to environmental interference has been enhanced, making the detection results more stable and reliable. Based on the data acquisition scheme of distributed nodes, combined with an efficient feature processing algorithm, the real-time monitoring of large-scale LED modules has been realized. Through the application of the depthwise separable convolution module and the attention mechanism, the computational complexity of the algorithm has been reduced, while ensuring the detection performance. By using the electric field curvature distribution analysis method, the ability of the system to identify early fault signs has been improved, and the potential risk of breakpoint faults can be warned in advance.

[0204] Please refer to Figure 3 , Figure 3 FIG. 7 is a schematic block diagram of the structure of the electronic device 300 provided by the embodiment of the present application. The electronic device 300 includes a processor 301 and a memory 302. The processor 301 and the memory 302 are connected through a device bus 303. Among them, the memory 302 may include a non-volatile storage medium and an internal memory.

[0205] The non-volatile storage medium can store a computer program. The computer program includes program instructions. When the program instructions are executed by the processor 301, the processor 301 can be enabled to execute any of the above LED module breakpoint detection and positioning methods.

[0206] The processor 301 is used to provide computing and control capabilities to support the operation of the entire electronic device 300.

[0207] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 301, the processor 301 can be enabled to execute any of the above LED module breakpoint detection and positioning methods.

[0208] Those skilled in the art can understand that Figure 3 the structure shown in FIG. 7 is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the electronic device 300 involved in the solution of the present application. The specific electronic device 300 may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0209] It should be understood that the processor 301 can be a central processing unit (CPU), and the processor 301 can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0210] It should be noted that those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the above-described electronic device 300 can refer to the corresponding process of the foregoing LED module breakpoint detection and positioning method, which will not be elaborated herein.

[0211] The embodiment of the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by one or more processors, the one or more processors are caused to implement the LED module breakpoint detection and positioning method provided by the embodiment of the present application.

[0212] Among them, the computer-readable storage medium can be an internal storage unit of the foregoing embodiment of the electronic device 300, such as the hard disk or memory of the electronic device 300. The computer-readable storage medium can also be an external storage device of the electronic device 300, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped with the electronic device 300.

[0213] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments, which will not be elaborated herein.

[0214] If 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 this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing an electronic 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 this application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0215] As described above, the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit them; although this application 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 recorded 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 this application.

Claims

1. A method for detecting and locating breakpoints of an LED module, characterized in that, The method includes: Collecting voltage and current data of the series - parallel nodes of the LED module to obtain an electrical characteristic matrix and a node topology graph; Performing iterative calculations on the electrical characteristic matrix, processing the voltage difference and current change amount between adjacent nodes to obtain a breakpoint feature vector; Inputting the breakpoint feature vector and the node topology graph into a multi - task neural network for multi - layer convolution operations to obtain an electrical fault distribution graph; Performing feature segmentation and deep feature extraction on the electrical fault distribution graph to obtain fault area feature data; Establishing an electric field distribution model for the fault area feature data, calculating electric field singular points by the least - squares method to obtain breakpoint spatial coordinate data; specifically including: dividing the fault area feature data into potential grids to obtain a potential sampling point matrix, and performing Gauss - Seidel iterative calculations on the potential sampling point matrix to obtain discrete potential data; constructing an electric field potential energy equation for the discrete potential data according to the second - order central difference format, and obtaining an electric field strength distribution matrix through finite - difference operations; substituting each grid point in the electric field strength distribution matrix into the Poisson equation, and obtaining electric field Poisson distribution data through boundary - condition constraint operations; establishing a least - squares fitting objective function for the electric field Poisson distribution data, obtaining an electric field fitting coefficient matrix through gradient descent optimization, and inputting the electric field fitting coefficient matrix into a Laplace operator model for second - order partial derivative calculations to obtain an electric field curvature distribution graph; performing singular value decomposition on the electric field curvature distribution graph, obtaining the main eigenvector through eigenvalue sorting, and performing extreme - point detection on the main eigenvector to obtain a singular - point candidate set; spatially aggregating the singular - point candidate set through clustering analysis to obtain a clustering center point, and performing coordinate transformation and accuracy correction on the clustering center point to obtain breakpoint spatial coordinate data.

2. The method for detecting and positioning breakpoints of the LED module according to claim 1, wherein, The collecting voltage and current data of the series - parallel nodes of the LED module to obtain an electrical characteristic matrix and a node topology graph includes: Establishing a coordinate system and performing spatial mapping calculations for each series - parallel node in the LED module to obtain a node coordinate data set; Collecting voltage and current and performing synchronous sampling conversion for each node in the node coordinate data set to obtain raw electrical data; Performing wavelet decomposition operations and high - frequency coefficient threshold processing on the raw electrical data to obtain denoised signal data; Performing segmentation processing on the denoised signal data to obtain segmented feature data, and performing zero - mean normalization processing and variance normalization calculation on the voltage and current values in the segmented feature data to obtain normalized electrical characteristics; Performing spatial correlation operations and feature mapping transformations on the normalized electrical characteristics and the node coordinate data set to obtain a node feature correlation table; Performing connectivity analysis and adjacency matrix calculations on the node feature correlation table based on the series - parallel topology structure of the LED module to obtain a node topology graph; Reconstructing the node feature correlation table into an M×N - dimensional feature structure, and obtaining an electrical characteristic matrix through data rearrangement operations.

3. The LED module breakpoint detection and positioning method according to claim 2, wherein Performing iterative calculations on the electrical feature matrix, processing the voltage difference and current change amount between adjacent nodes, and obtaining a breakpoint feature vector, including: Calculating the voltage ratio of adjacent nodes according to the voltage data in the electrical feature matrix to obtain a node voltage ratio sequence, and performing time series analysis on the node voltage ratio sequence to obtain a voltage change rate matrix; Establishing a current time series model according to the current data in the electrical feature matrix, and obtaining current change amount data through first-order difference operation; Performing Fourier transform on the voltage change rate matrix to obtain voltage periodic fluctuation data, and performing adaptive threshold analysis on the current change amount data to obtain current mutation point data; Constructing a voltage domain iterative learning model based on the voltage periodic fluctuation data, and obtaining voltage feature weights through weight update operation; Performing feature cross operation on the current mutation point data and the voltage feature weights to obtain fusion feature data, and performing feature dimensionality reduction and normalization processing on the fusion feature data to obtain a breakpoint feature vector.

4. The LED module breakpoint detection and positioning method according to claim 3, characterized in that Inputting the breakpoint feature vector and the node topology map into a multi-task neural network for multi-layer convolution operation to obtain an electrical fault distribution map, including: Constructing the input data of the multi-task neural network through feature mapping processing of the breakpoint feature vector and the node topology map. The multi-task neural network includes a shared feature extraction network and two parallel task branch networks; Performing feature extraction on the input data through the shared feature extraction network to obtain shared feature data. The shared feature extraction network contains 4 convolutional layers, each convolutional layer uses a 3×3 convolutional kernel and a ReLU activation function, and a BatchNorm layer and a Dropout layer are added between each layer; Inputting the shared feature data into the first task branch network for node feature learning to obtain a node feature mapping result. The first task branch network contains 3 fully connected layers, and each layer uses a LeakyReLU activation function and weight regularization; Inputting the shared feature data into the second task branch network for electrical boundary feature learning to obtain a boundary feature mapping result. The second task branch network contains 2 transposed convolutional layers and 1 attention layer, and each layer uses a Tanh activation function; Performing feature fusion on the node feature mapping result and the boundary feature mapping result, and calculating feature weights through an attention mechanism to obtain weighted feature data; Inputting the weighted feature data into a decoder network for feature decoding to obtain feature distribution data. The decoder network contains 3 deconvolutional layers, and each layer uses a 4×4 convolutional kernel and a ReLU activation function; Performing data reconstruction and residual connection operation on the feature distribution data to obtain global feature data, and performing dimensionality reduction processing on the global feature data through a 1×1 convolutional layer, and obtaining an electrical fault distribution map through node-level feature classification.

5. The method for detecting and locating breakpoints of the LED module according to claim 4, characterized in that Performing feature segmentation and depth feature extraction on the electrical fault distribution map to obtain fault region feature data, including: Perform voltage gradient calculation and statistical analysis on the electrical fault distribution map to obtain the optimal segmentation threshold, and convert the electrical fault distribution map into a binary feature matrix according to the optimal segmentation threshold to obtain the labeled fault area data; Extract the electrical boundary from the labeled fault area data, calculate the voltage amplitude and change direction through the voltage gradient operator to obtain the electrical boundary feature data; Input the electrical boundary feature data into a deep feature network to obtain a feature description vector. The deep feature network includes 5 stacked depthwise separable convolution modules, and each depthwise separable convolution module uses a combination of 3×3 depth convolution and 1×1 point convolution; Perform feature enhancement on the feature description vector through the self-attention mechanism to obtain enhanced feature data, and fuse the enhanced feature data with the original fault area data to obtain fused feature data; Reconstruct the electrical features of the fused feature data to obtain the reconstructed area data, and perform feature optimization and data fitting on the reconstructed area data to obtain the fault area feature data.

6. The LED module breakpoint detection and positioning method according to claim 1, characterized in that Establish a least squares fitting objective function for the electric field Poisson distribution data, obtain the electric field fitting coefficient matrix through gradient descent optimization, and input the electric field fitting coefficient matrix into the Laplace operator model for second-order partial derivative calculation to obtain the electric field curvature distribution map, including: Perform basis function expansion on the electric field Poisson distribution data through the N-order Legendre polynomial, and calculate the basis function expansion coefficients according to orthogonality; Construct a least squares fitting objective function based on the basis function expansion coefficients, obtain a loss function with a penalty term through L2 norm regularization constraint, and perform variable differentiation on the loss function with a penalty term. Obtain the gradient direction vector through forward difference and backward difference operations; Set an adaptive learning rate according to the gradient direction vector, perform iterative operations through the backpropagation algorithm to obtain the electric field fitting coefficient matrix, and input the electric field fitting coefficient matrix into a two-dimensional Laplace operator for finite difference to obtain a second-order derivative tensor; Perform principal curvature calculation on the second-order derivative tensor, obtain curvature feature data through the Gaussian curvature and mean curvature formulas, and reconstruct the curvature feature data according to the electric field distribution. Obtain the electric field curvature distribution map through bicubic interpolation.

7. An LED module breakpoint detection and positioning device, characterized in that, For performing the LED module breakpoint detection and positioning method according to any one of claims 1-6, the device includes: An acquisition module for collecting voltage and current data of the series and parallel nodes of the LED module to obtain an electrical feature matrix and a node topology map; A calculation module for performing iterative calculation on the electrical feature matrix, processing the voltage difference and current change amount between adjacent nodes to obtain a breakpoint feature vector; An operation module for inputting the breakpoint feature vector and the node topology map into a multi-task neural network for multi-layer convolution operation to obtain an electrical fault distribution map; An extraction module for performing feature segmentation and deep feature extraction on the electrical fault distribution map to obtain fault area feature data; A building module is configured to establish an electric field distribution model for the fault area feature data, calculate electric field singular points by the least squares method, and obtain breakpoint spatial coordinate data.

8. An electronic device, characterized in that, The electronic device includes: a memory and at least one processor, and instructions are stored in the memory; The at least one processor invokes the instructions in the memory to cause the electronic device to execute the LED module breakpoint detection and positioning method according to any one of claims 1-6.

9. A computer-readable storage medium having instructions stored thereon, characterized in that, When the instructions are executed by the processor, the LED module breakpoint detection and positioning method according to any one of claims 1-6 is implemented.

Images