A method and system for early abnormal diagnosis of LED based on spectral power distribution

Abstract

The application provides an LED early abnormality diagnosis method and system based on spectral power distribution; wherein the method comprises: acquiring first SPD data of a first LED lamp group; performing fitting processing according to the first SPD data and a fitting model to obtain optimal characteristic values corresponding to the fitting model; performing dimension reduction processing on the optimal characteristic values to obtain first target characteristic values, and performing clustering processing on the first target characteristic values to obtain an abnormality judgment threshold; acquiring second SPD data of a second LED lamp group, obtaining second target characteristic values according to the second SPD data, and performing abnormality diagnosis on the second LED lamp group according to the second target characteristic values and the abnormality judgment threshold. The application realizes the diagnosis of the early failure of the LED by analyzing the SPD and using the principal component analysis and KNN dimension reduction method.

Description

Technical Field

[0001] This invention relates to the field of LED testing technology, and more specifically, to a method, system, electronic device, and computer storage medium for early LED anomaly diagnosis based on spectral power distribution. Background Technology

[0002] As the fourth generation of lighting source, LEDs have many advantages such as energy saving, environmental protection, long lifespan, and high reliability, and have been widely used in display screen backlighting, night scene lighting, and other fields. Lifespan prediction and fault diagnosis are important aspects of LED testing. Existing methods mainly employ data-driven approaches and machine learning-based methods to perform LED diagnosis and testing.

[0003] 1) Data-driven approach:

[0004] Chang et al. proposed an anomaly detection technique called similarity-based metric testing for identifying anomalies in historical databases of both healthy and unhealthy data. This research extracts features from spectral power distribution (SPD) using LED peak analysis, reduces the dimensionality of the features using principal component analysis, and uses the k-nearest neighbor (KNN) algorithm to divide the principal component dataset into groups. The distance from the centroid of each cluster to each test point is evaluated, and anomalies are detected when the distance exceeds a threshold. Ultimately, anomalies were detected in less than 1200 hours, reducing LED qualification testing time from the industry standard of 6000 hours to 1200 hours.

[0005] Fan et al. proposed a method to detect anomalies in white LEDs by calculating the Mahalanobis distance between indirect LED performance data (e.g., lead temperature, input drive current, and forward voltage). In this method, early anomalies of the LED are characterized by calculating the Mahalanobis distance between the parameters and defined as a real-time health indicator to reflect LED degradation. The results show that the anomaly detection method can provide early warnings of anomalies at approximately 45% of the lifetime of all evaluated test LEDs before they actually fail.

[0006] Yung et al. proposed a Lévy stochastic process method that combines gamma stochastic processes and Poisson distributions to describe the degradation of small, random, and irregular variations in white LEDs. The reliability function is obtained by using different jump distribution functions. The final results show that the method can obtain more reliability information, but the Lévy method is not as good as the gamma method in characterizing small, random, and irregular variations in degradation.

[0007] In 2015, Huang used an improved Wiener process to simulate the degradation process of medium-power white LEDs, considering dynamic and random variations as well as the nonlinear degradation behavior of the LED device, and also taking into account the degradation caused by color drift during its lifetime. He presented the cumulative failure distribution corresponding to different combinations of lumen maintenance and color drift. The results showed that the joint failure distribution of LED devices can be simplified by modeling the degradation process by simply treating lumen maintenance and color drift as two independent variables.

[0008] Zhai proposed a stochastic Wiener process model based on the accelerated failure time principle and established statistical inference based on maximum likelihood estimation. The model was extended to the Constant Stress Accelerated Degradation Test (CSADT) and validated using an LED accelerated aging test dataset.

[0009] Based on gear performance degradation data, Chen Zhijun et al. concluded that a lifetime prediction and reliability assessment method based on the stochastic Wiener process can more accurately describe the lifetime and reliability of gears, especially when the randomness of product performance degradation needs to be considered. This shows that the stochastic Wiener method can also be applied to LED lighting system research, treating degradation as a process conforming to the stochastic Wiener method for failure analysis.

[0010] 2) Based on machine learning methods:

[0011] Machine learning can be broadly categorized into supervised learning (predictive modeling), semi-supervised learning, and unsupervised learning (descriptive modeling) based on the amount and type of human supervision required.

[0012] Chang et al. proposed a regression model for RVM, which reduced the LED testing time from 6000h to 210h, greatly improving the efficiency of LED degradation monitoring.

[0013] Cao et al. proposed an ANN-based method to predict the luminous flux of LED array modules. They verified the accuracy of this method under various operating conditions, showing that the combined ANN method achieved a luminous flux prediction accuracy of up to 98.5%, demonstrating the model's advantages. Fan et al. used a backpropagation neural network (BP-NN) to predict the spectral characteristic parameters of full-spectrum white LED packages and improved the prediction results by integrating a genetic algorithm.

[0014] Liu et al. employed two ANN algorithms to simplify and improve the accuracy of lifetime prediction for LED lighting systems. ANNs are known methods for effectively and efficiently modeling complex nonlinear systems, allowing for the generalization and tuning of solutions from limited datasets. Depending on the mathematical operations and required parameter settings, neural network structures can be of different types, including feedforward neural networks (FFNNs), backpropagation neural networks (BPNNs), radial basis function neural networks (RBNNs), recurrent neural networks (RNNs), and self-organizing maps (SOMs).

[0015] However, existing research on LED lifetime prediction and fault diagnosis focuses primarily on lumen degradation, neglecting color transformation. Color degradation is a crucial performance parameter for LEDs in applications such as museums, supermarkets, and shopping malls. Therefore, current LED testing research is somewhat one-sided and cannot meet the performance testing requirements of applications like those mentioned above. Summary of the Invention

[0016] In order to at least solve the technical problems existing in the background art, the present invention provides a method, system, electronic device and computer storage medium for early LED anomaly diagnosis based on spectral power distribution.

[0017] The first aspect of the present invention provides a method for early anomaly diagnosis of LEDs based on spectral power distribution, comprising the following steps:

[0018] Obtain the first SPD data of the first LED light group;

[0019] The first SPD data and the fitting model are used for fitting processing to obtain the optimal feature value corresponding to the fitting model;

[0020] The optimal feature value is subjected to dimensionality reduction to obtain a first target feature value, and the target feature value is subjected to clustering to obtain an anomaly detection threshold.

[0021] The second SPD data of the second LED light group is obtained, the second target feature value is obtained based on the second SPD data, and the second LED light group is diagnosed for anomalies based on the second target feature value and the anomaly determination threshold.

[0022] Further, the fitting process based on the first SPD data and the fitting model to obtain the optimal feature value corresponding to the fitting model includes:

[0023] The first SPD data is input into the fitting model for curve fitting. When the degree of fitting meets the preset conditions, the optimal feature value corresponding to the fitting model is extracted.

[0024] Furthermore, the fitting model is an asymmetric bi-S model, as follows:

[0025]

[0026] In the formula, A1 and A2 represent amplitudes, and ω 31 ω 32 Indicates low side energy variance, ω 21 ω 22 This represents the variance of the high-side energy.

[0027] Furthermore, the optimal eigenvalues ​​are amplitudes A1 and A2, and low side energy variance ω. 31 ω 32 High side energy variance ω 21 ω 22 Peak wavelengths xc1 and xc2, full width at half maximum (FWHM) ω1 and ω2, wavelength x.

[0028] Further, the dimensionality reduction processing of the optimal feature value to obtain the first target feature value includes:

[0029] The optimal eigenvalue is reduced to two principal components, namely the first target eigenvalue, by using principal component analysis.

[0030] Further, the step of clustering the first target feature values ​​to obtain an anomaly detection threshold includes:

[0031] The K-means algorithm is used to cluster the first target feature value into m clusters, and the centroid and radius of each cluster are calculated.

[0032] The standard deviation of each cluster is calculated based on the centroid and the radius, and the anomaly detection threshold is determined based on the standard deviation.

[0033] Furthermore, the anomaly detection threshold T i for:

[0034]

[0035] In the formula, σ is the standard deviation, which describes the offset of the test point from the center; D is the scaling factor. In this invention, the scaling factor D is 2 because the optimal eigenvalue is two-dimensional after dimensionality reduction.

[0036] Further, the step of performing anomaly diagnosis on the second LED light group based on the second target feature value and the anomaly determination threshold includes:

[0037] Calculate the distance D between each second target sub-feature value in the second target feature value and the centroid of the corresponding cluster. i ;

[0038] If the distance Di Greater than the anomaly determination threshold T i If the second target feature value is determined to be an anomaly, then the corresponding LED light is determined to have malfunctioned at the corresponding sampling point.

[0039] A second aspect of the present invention provides an early anomaly diagnosis system for LEDs based on spectral power distribution, comprising an acquisition module, a processing module, and a storage module; the processing module is connected to the acquisition module and the storage module.

[0040] The storage module is used to store executable computer program code;

[0041] The acquisition module is used to acquire the first SPD data and the second SPD data, and transmit them to the processing module;

[0042] The processing module is configured to execute the method described in the preceding one by invoking the executable computer program code in the storage module.

[0043] A third aspect of the present invention provides an electronic device comprising: a memory storing executable program code; a processor coupled to the memory; the processor invoking the executable program code stored in the memory to perform the method as described in any of the preceding claims.

[0044] A fourth aspect of the invention provides a computer storage medium storing a computer program that, when executed by a processor, performs the method described in any of the preceding claims.

[0045] To achieve early fault detection of lumen decay and color shift in white LEDs, this invention proposes the aforementioned fault diagnosis method by analyzing SPDs and utilizing principal component analysis and KNN dimensionality reduction. This invention realizes a method for predicting the remaining useful life of phosphor-converted white LEDs (pc-WLEDs) using photometric and chromaticity parameters, enabling earlier detection of early LED anomalies. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart illustrating an early LED anomaly diagnosis method based on spectral power distribution disclosed in an embodiment of the present invention.

[0048] Figure 2 This is a typical SPD illustration for PC-WLED;

[0049] Figure 3 This is a comparison chart of the fitting curves output by the three fitting models;

[0050] Figure 4 This is a schematic diagram of the principal components of SPD training data using 720 data points;

[0051] Figure 5 This is a schematic diagram of CSS trained using SPD data with 720 data points;

[0052] Figure 6 This is a schematic diagram of the KNN-means clustering results using SPD training data with 720 data points;

[0053] Figure 7 This is a schematic diagram of the distance measurement of LED_16 when the clustering result is two clusters in the test experiment of this invention;

[0054] Figure 8 This is a schematic diagram of the distance measurement of LED_16 when the clustering result is three clusters in the test experiment of this invention;

[0055] Figure 9 This is a schematic diagram of the distance measurement of LED_16 when the clustering result is four clusters in the test experiment of this invention;

[0056] Figure 10 This is a schematic diagram of the abnormal detection results (false alarm rate, false alarm rate and abnormal detection rate) in the test of this invention;

[0057] Figure 11 This is a schematic diagram of the structure of an LED early anomaly diagnosis system based on spectral power distribution disclosed in an embodiment of the present invention;

[0058] Figure 12 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0060] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the application. The singular forms “a,” “said,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0061] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0062] It should be understood that although the terms first, second, third, etc., may be used to describe ... in the embodiments of this application, these ... should not be limited to these terms. These terms are only used to distinguish .... For example, without departing from the scope of the embodiments of this application, first ... can also be referred to as second ..., and similarly, second ... can also be referred to as first ....

[0063] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”

[0064] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a product or system comprising a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a product or system. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the product or system that includes said element.

[0065] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0066] Example 1:

[0067] Please see Figure 1 , Figure 1 This is a schematic flowchart of an early LED anomaly diagnosis method based on spectral power distribution disclosed in an embodiment of the present invention. Figure 1 As shown in the figure, an early LED anomaly diagnosis method based on spectral power distribution according to an embodiment of the present invention includes the following steps:

[0068] Obtain the first SPD data of the first LED light group;

[0069] The first SPD data and the fitting model are used for fitting processing to obtain the optimal feature value corresponding to the fitting model;

[0070] The optimal feature value is subjected to dimensionality reduction to obtain a first target feature value, and the first target feature value is subjected to clustering to obtain an anomaly detection threshold.

[0071] The second SPD data of the second LED light group is obtained, the second target feature value is obtained based on the second SPD data, and the second LED light group is diagnosed for anomalies based on the second target feature value and the anomaly determination threshold.

[0072] In this embodiment of the invention, two sets of LED light groups are provided, namely a first LED light group and a second LED light group. The first LED light group can have 10 LEDs and the second LED light group can have 6 LEDs. The SPD data of the first LED light group is used to train the diagnostic model, and then the trained diagnostic model is used to diagnose anomalies in the SPD data of the second LED light group.

[0073] In practice, Avago's 3W high-power pc-WLEDs (ASMT-JN31-NTV01) were used, with 16 test samples driven by a 200 mA DC current under constant aging temperature (Ta = 90°C). SPD data were measured using a spectrometer, and a total of 72 cycles of SPD data were collected. The aging test was stopped when the lumen maintenance rate decreased to approximately 70%.

[0074] Furthermore, the SPD data for each LED in the first SPD data is represented as follows:

[0075]

[0076] In the formula, SPD LED SPD represents the spectrum of a single LED light. i This represents the i-th SPD spectrum extracted from the LED light, and n represents the total number of SPDs.

[0077] In this embodiment of the invention, most commercially available white LEDs use a chip-and-phosphor light emission method. The chip emits short-wavelength light to excite the phosphor to produce long-wavelength visible light. Finally, the remaining light from the chip and the emitted light from the phosphor recombine to form white light. The pc-LED used in this invention can adopt a blue chip + yellow phosphor light emission method. Furthermore, by adjusting the concentration of the phosphor, the intensity ratio of blue and yellow light can be indirectly controlled, thus obtaining white light of different color temperatures. The spectrum of a phosphor-converted white LED typically has multiple peaks: one peak is located in the short-wavelength region (380-495nm), representing the blue light emitted by the LED chip, called the emission spectrum; the other peaks are located in the long-wavelength region (495-745nm), representing the yellow-green light converted from the phosphor, called the conversion spectrum, such as... Figure 2 As shown.

[0078] As a hybrid light source, pc-LED is formed by the superposition of two overlapping single-peak spectra.

[0079] Further, the fitting process based on the first SPD data and the fitting model to obtain the optimal feature value corresponding to the fitting model includes:

[0080] The first SPD data is input into the fitting model for curve fitting. When the degree of fitting meets the preset conditions, the optimal feature value corresponding to the fitting model is extracted.

[0081] In this embodiment of the invention, a corresponding fitting model is selected to fit the first SPD data collected and extracted above. When the degree of fitting meets the preset conditions, the corresponding optimal feature value can be extracted from the fitting model.

[0082] The goodness of fit can be judged by the coefficient of determination:

[0083]

[0084] The goodness of fit is determined by the degree to which the independent variable explains the dependent variable; the closer the value is to 100%, the better the goodness of fit.

[0085] Furthermore, the fitting model is a biGaussian model, as follows:

[0086]

[0087] In the formula, x represents wavelength, xc1 and xc2 represent peak wavelength, and ω1 and ω2 represent full width at half maximum (FWHM).

[0088] Furthermore, the fitting model is a double Lorentz model, as follows:

[0089]

[0090] In the formula, A1 and A2 represent the spectral areas.

[0091] Furthermore, the fitting model is an asymmetric bi-S model (Asym2sig), as follows:

[0092]

[0093] In the formula, A1 and A2 represent amplitudes, and ω 31 ω 32 Indicates low side energy variance, ω 21 ω 22 This represents the variance of the high-side energy.

[0094] In this embodiment of the invention, reference is made to Figure 3 As shown in Table 1, the three curves output by the three fitting models above all have a high degree of fit, at least 95%.

[0095] Table 1. SPD model fit results for the three functions

[0096]

[0097] The double Lorentz function exhibited some errors in fitting the peak and trough values. Specifically, when the wavelength was less than 430 nm, the actual and fitted values ​​deviated somewhat. At wavelengths less than 430 nm, the luminous efficiency was almost zero, so its impact on luminous flux was relatively small. However, its impact was more significant when the wavelength was greater than 430 nm.

[0098] R through the above three curves 2 In comparison, the asymmetric bi-S function achieved the highest fitting efficiency, reaching 99.9%. Therefore, the solution of this invention preferentially uses the asymmetric bi-S function for fitting and feature extraction.

[0099] Furthermore, the optimal eigenvalues ​​are amplitudes A1 and A2, and low side energy variance ω. 31 ω 32 High side energy variance ω 21 ω 22 Peak wavelengths xc1 and xc2, full width at half maximum (FWHM) ω1 and ω2, wavelength x.

[0100] In this embodiment of the invention, when an asymmetric double-S model is used, the corresponding optimal eigenvalues ​​are the 11 parameters involved in the asymmetric double-S model.

[0101] Further, the dimensionality reduction processing of the optimal feature value to obtain the first target feature value includes:

[0102] The optimal eigenvalue is reduced to two principal components, namely the first target eigenvalue, by using principal component analysis.

[0103] In this embodiment of the invention, Principal Component Analysis (PCA), a commonly used linear dimensionality reduction method, maps high-dimensional data to a low-dimensional space through a linear projection, where the data has the greatest information content in the projected dimension. Therefore, PCA uses fewer data dimensions while retaining more features of the original data points. Thus, this invention detects fault points by extracting 11 feature values ​​and transforming them into principal components as initial values.

[0104] The training data, i.e., the first SPD data, is used to evaluate the variance matrix of PCA. The original training set is projected onto selected feature vectors, resulting in a new dimension after dimensionality reduction. Figure 4 In the process, the 11 features extracted by SPD were reduced to two principal components, where the X-axis represents the dimension of the principal component and the Y-axis represents the variance of the corresponding principal component. Points with slight fluctuations were removed, and the two principal components were retained.

[0105] Further, the step of clustering the first target feature values ​​to obtain an anomaly detection threshold includes:

[0106] The K-means algorithm is used to cluster the first target feature value into m clusters, and the centroid and radius of each cluster are calculated.

[0107] The standard deviation of each cluster is calculated based on the centroid and the radius, and the anomaly detection threshold is determined based on the standard deviation.

[0108] In this embodiment of the invention, K-means belongs to unsupervised learning in machine learning. It utilizes unlabeled data to learn the distribution of data and the relationships between data points. In the K-means algorithm, "k" represents k clusters, and "means" indicates that after dividing the data into new clusters, the centroid of each cluster is recalculated using the averaging method to determine the new cluster centers. The cluster sum is obtained by summing the within-cluster sums of all clusters in the dataset, as follows:

[0109]

[0110] In the formula, (x1, x2, x3...x n Given n observations, the K-means clustering method divides the n observations into m clusters, S(S1, S2, S3…S…). m ).

[0111] After dividing the training set into four clusters using K-means, similarity detection can be used to identify fault points. When the training data is divided into m clusters, the number of clusters S is... i The center of mass is μ iThe distance from each data point to the centroid is calculated using Euclidean distance. The distance from each test point to the centroid (D) i To evaluate the distance from each test point to its centroid (μ) i The distance is evaluated. The average radius R(S) is... i ) is the distance from all points in the cluster to the centroid μ. i The average radius is a measure of the density of points in a cluster.

[0112] The centroid and radius are calculated using the following formula:

[0113]

[0114]

[0115] Furthermore, the anomaly detection threshold T i for:

[0116]

[0117] In the formula, σ is the standard deviation corresponding to each cluster, which describes the offset between the test point and the cluster center; D is the scaling factor. In this invention, the scaling factor D is 2 because the optimal feature value is two-dimensional after dimensionality reduction.

[0118] In this embodiment of the invention, ideally, the distribution of these clusters should be circular. Figure 6 In this model, each set approximates an ellipse. Fault point diagnosis depends on the accuracy of the threshold, which should fall between the major and minor axes; thresholds that are too large or too small are detrimental to fault point diagnosis. A threshold that is too large increases the false negative rate, while a threshold that is too small increases the fault detection rate.

[0119] Further, the step of performing anomaly diagnosis on the second LED light group based on the second target feature value and the anomaly determination threshold includes:

[0120] Calculate the distance D between each second target sub-feature value in the second target feature value and the centroid of the corresponding cluster. i ;

[0121] If the distance D i Greater than the anomaly determination threshold T i If the second target feature value is determined to be an anomaly, then the corresponding LED light is determined to have malfunctioned at the corresponding sampling point.

[0122] Example of an experiment:

[0123] Ten datasets, totaling 720 data points, were used as the training set to check LED_16. No alarms were detected when the training set consisted of two and three clusters, as shown below. Figure 7and 8 As shown.

[0124] When the training set consisted of four clusters, fault points were identified, and anomalies were monitored. When the distance to the centroid D... j When T is greater than the threshold j At 789.6 hours, a fault was detected and an alarm was triggered, such as... Figure 9 As shown.

[0125] Next, the training data was divided into 2, 4, 6, 8, 10, and 12 groups, corresponding to 144, 288, 432, 576, 720, and 864 data points respectively, to verify the anomaly detection accuracy for the remaining 14, 12, 10, 8, 6, and 4 LEDs. Error detection was broadly divided into two groups: false alarms and false negatives. False alarms refer to detecting anomalies too early, i.e., judging an LED as faulty under normal conditions; false negatives refer to failing to detect anomalies at the end of the LED's lifespan (luminous flux drops to 70% or color shift reaches 0.007).

[0126] The results are as follows Figure 10 As shown, both false negatives and false positives decrease with increasing training data. When the data reaches 576 and 720 respectively, both false positives and false negatives reach 0, with no such errors occurring. Meanwhile, anomaly detection shows an upward trend. Furthermore, after the training data reaches 720, no false negatives or false positives occur.

[0127] Example 2:

[0128] Please see Figure 7 , Figure 7 This invention also discloses a schematic diagram of an early LED anomaly diagnosis system based on spectral power distribution. (See attached diagram.) Figure 7 As shown, an LED early anomaly diagnosis system based on spectral power distribution according to an embodiment of the present invention includes an acquisition module 101, a processing module 102, and a storage module 103; the processing module 102 is connected to the acquisition module 101 and the storage module 103.

[0129] The storage module 103 is used to store executable computer program code;

[0130] The acquisition module 101 is used to acquire the first SPD data and the second SPD data, and transmit them to the processing module 102;

[0131] The processing module 102 is configured to execute the method described in the preceding one by calling the executable computer program code in the storage module 103.

[0132] The specific functions of the LED early anomaly diagnosis system based on spectral power distribution in this embodiment are the same as those in the above embodiments. Since the system in this embodiment adopts all the technical solutions of the above embodiments, it has at least all the beneficial effects brought about by the technical solutions of the above embodiments, which will not be repeated here.

[0133] Example 3:

[0134] Please see Figure 8 , Figure 8 This invention discloses an electronic device comprising: a memory storing executable program code; a processor coupled to the memory; the processor calling the executable program code stored in the memory to execute the method described in the foregoing embodiments.

[0135] Example 4:

[0136] This invention also discloses a computer storage medium storing a computer program, which is executed by a processor to perform the methods described in the foregoing embodiments.

[0137] The apparatus / system according to embodiments of this disclosure may include a processor, memory for storing and executing program data, permanent memory such as a disk drive, a communication port for processing communication with external devices, and a user interface device, etc. The method is implemented as a software module or may be stored on a computer-readable recording medium as computer-readable code or program instructions executable by a processor. Examples of computer-readable recording media may include magnetic storage media (e.g., read-only memory (ROM), random access memory (RAM), floppy disk, hard disk, etc.), optical reading media (e.g., CD-ROM, DVD, etc.). The computer-readable recording medium may be distributed across computer systems connected to a network, and the computer-readable code may be stored and executed in a distributed manner. The medium may be computer-readable, stored in memory, and executed by a processor.

[0138] Embodiments of this disclosure can be designated as functional block components and various processing operations. Functional blocks can be implemented as various numbers of hardware and / or software components that perform specific functions. For example, embodiments of this disclosure can implement direct circuit components, such as memories, processing circuits, logic circuits, lookup tables, etc., that can perform various functions under the control of one or more microprocessors or other control devices. Components of this disclosure can be implemented by software programming or software components. Similarly, embodiments of this disclosure can include various algorithms implemented by combinations of data structures, procedures, routines, or other programming components, and can be implemented by programming or scripting languages ​​(such as C, C++, Java, assembler, etc.). Functional aspects can be implemented by algorithms executed by one or more processors. Furthermore, embodiments of this disclosure can implement related techniques for electronic environment setup, signal processing, and / or data processing. Terms such as “mechanism,” “element,” “unit,” etc., can be used broadly and are not limited to mechanical and physical components. These terms can refer to a series of software routines associated with processors, etc.

[0139] Specific embodiments are described in this disclosure as examples, and the scope of the embodiments is not limited thereto.

[0140] While embodiments of this disclosure have been described, those skilled in the art will understand that various changes in form and detail may be made therein without departing from the spirit and scope of this disclosure as defined by the appended claims. Therefore, the above embodiments of this disclosure should be interpreted as exemplary and are not limiting in any way. For example, each component described as a single unit may be executed in a distributed manner, and similarly, components described as distributed may be executed in a combined manner.

[0141] All examples or example terms (e.g., etc.) used in the embodiments of this disclosure are for the purpose of describing embodiments of this disclosure and are not intended to limit the scope of embodiments of this disclosure.

[0142] Furthermore, unless otherwise explicitly stated, expressions such as “necessary” or “important” associated with certain components do not necessarily indicate that the components are absolutely necessary.

[0143] Those skilled in the art will understand that embodiments of this disclosure may be implemented in modified forms without departing from the spirit and scope of this disclosure.

[0144] Because this disclosure allows for various changes to the embodiments thereof, it is not limited to the specific embodiments described herein, and it will be understood that all changes, equivalents, and alternatives that do not depart from the spirit and scope of this disclosure are included herein. Therefore, the embodiments of this disclosure described herein should be understood as illustrative in all respects and should not be construed as limiting.

[0145] Furthermore, terms such as "unit" and "module" refer to a unit that can be implemented as hardware or software or a combination of hardware and software to process at least one function or operation. "Unit" and "module" can be stored in a storage medium to be addressed and can be implemented as a program that can be executed by a processor. For example, "unit" and "module" can refer to components such as software components, object-oriented software components, class components, and task components, and can include processes, functions, attributes, procedures, subroutines, program code segments, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, or variables.

[0146] In this disclosure, the statement "A may include one of a1, a2, and a3" can broadly indicate that examples that can be included in element A include a1, a2, or a3. This statement should not be construed as limiting the meaning of examples included in element A to a1, a2, and a3. Therefore, as examples included in element A, elements other than a1, a2, and a3 should not be interpreted as excluding elements. Furthermore, this statement indicates that element A may include a1, a2, or a3. This statement does not imply that the elements included in element A must be selected from a specific set of features. That is, this statement should not be construed restrictively as indicating that a1, a2, or a3 must be selected from a set that includes a1, a2, and a3 to be included in element A.

[0147] Furthermore, in this disclosure, the expression "at least one of a1, a2 and / or a3" means one of "a1", "a2", "a3", "a1 and a2", "a1 and a3", "a2 and a3", and "a1, a2 and a3". Therefore, it should be noted that unless explicitly described as "at least one of a1, at least one of a2, and at least one of a3", the expression "at least one of a1, a2 and / or a3" should not be interpreted as "at least one of a1", "at least one of a2", and "at least one of a3".

Claims

1. A method for early anomaly diagnosis of LEDs based on spectral power distribution, characterized in that, Includes the following steps: Obtain the first SPD data of the first LED light group; The first SPD data and the fitting model are used for fitting processing to obtain the optimal feature value corresponding to the fitting model; The optimal feature value is subjected to dimensionality reduction to obtain a first target feature value, and the first target feature value is subjected to clustering to obtain an anomaly detection threshold. Acquire the second SPD data of the second LED light group, obtain the second target feature value based on the second SPD data, and perform anomaly diagnosis on the second LED light group based on the second target feature value and the anomaly determination threshold. The step of clustering the first target feature value to obtain the anomaly detection threshold includes: The K-means algorithm is used to cluster the first target feature value into m clusters, and the centroid and radius of each cluster are calculated. The standard deviation of each cluster is calculated based on the centroid and the radius, and the anomaly detection threshold is determined based on the standard deviation. The anomaly detection threshold for: ； In the formula, σ is the standard deviation, which describes the offset of the test point from the center; d is the scaling factor, which is set to 2 because the optimal eigenvalue is two-dimensional after dimensionality reduction. The step of diagnosing anomalies in the second LED light group based on the second target feature value and the anomaly determination threshold includes: Calculate the distance between each second target sub-feature value in the second target feature value and the centroid of the corresponding cluster. ; If the distance Greater than the anomaly detection threshold If the second target sub-feature value is determined to be an anomaly, that is, the corresponding LED light is determined to have malfunctioned at the corresponding sampling point time. The fitting model is an asymmetric bi-S model, as follows: ； In the formula, , Indicates amplitude. , Indicates low side energy variance. , This represents the variance of the high-side energy.

2. The method for early LED anomaly diagnosis based on spectral power distribution according to claim 1, characterized in that: The step of performing fitting processing based on the first SPD data and the fitting model to obtain the optimal feature value corresponding to the fitting model includes: The first SPD data is input into the fitting model for curve fitting. When the degree of fitting meets the preset conditions, the optimal feature value corresponding to the fitting model is extracted.

3. The method for early LED anomaly diagnosis based on spectral power distribution according to claim 1, characterized in that: The step of performing dimensionality reduction on the optimal feature value to obtain the first target feature value includes: The optimal eigenvalue is reduced to two principal components, namely the first target eigenvalue, by using principal component analysis.

4. An LED early anomaly diagnosis system based on spectral power distribution, comprising an acquisition module, a processing module, and a storage module; the processing module is connected to the acquisition module and the storage module; The storage module is used to store executable computer program code; The acquisition module is used to acquire the first SPD data and the second SPD data, and transmit them to the processing module; Its features are: The processing module is configured to execute the method as described in any one of claims 1-3 by calling the executable computer program code in the storage module.

5. A computer storage medium storing a computer program, characterized in that: The computer program is executed by the processor to perform the method as described in any one of claims 1-3.

Images