LED light color detection analysis method and system

Abstract

The invention discloses a light color detection analysis method and system for an LED, and relates to the technical field of LED detection, and the system comprises a light color collection module which is used for obtaining spectrum data generated in the lightening process of a target LED; the time sequence modeling module is used for collecting a plurality of spectral data points at fixed time intervals within a continuous preset time period after the target LED starts to emit light, calculating to obtain a dominant wavelength, a color coordinate and a fitting color temperature value, forming a multi-moment light color parameter sequence, and constructing a time sequence model based on the sequence, and outputting the time sequence model and the corresponding original parameter sequence. According to the method, the change curve of the light intensity and the color coordinates is constructed, and a convergence trend analysis mechanism is introduced, so that the LED light-emitting stable state is automatically judged, wrong judgment caused by unstable early data is effectively avoided, and the representativeness and the accuracy of detection data are improved.

Description

Technical Field

[0001] The present invention relates to the field of LED detection technology, and in particular to a method and system for detecting and analyzing the light color of an LED (light emitting diode). Background Art

[0002] As a highly efficient, long-life light-emitting device, LEDs' light color consistency and stability are key factors in quality control. As LEDs are widely used in lighting, displays, automotive lighting, and other fields, different application scenarios require more accurate and representative testing of parameters such as dominant wavelength, CIE color coordinates, and fitted color temperature. Therefore, establishing a standardized and quantifiable light color testing process has become a key industry priority during LED production and shipment.

[0003] A search revealed a Chinese patent (publication number: CN112200200B) that discloses a method for detecting the color of LED lights. This patent involves fixing an image acquisition device, setting an LED light acquisition position, and placing LED lights of various colors in the LED light acquisition position. The image acquisition device then acquires a color digital image of the LED lights, which is then sent to a processor for processing. The color characteristic parameter value ranges of the various colors of LED lights are calculated to form a color characteristic parameter value range database. The LED lights to be detected are placed in the LED light acquisition position and illuminated. The image acquisition device then acquires a color digital image of the LED lights to be detected, which is then sent to a processor for processing. The color characteristic parameter values ​​of the LED lights to be detected are calculated and matched against the color characteristic parameter value range database to determine the color of the LED lights to be detected.

[0004] In practical applications, due to the thermal response process during the initial lighting of LEDs, their light intensity and color coordinates often fluctuate within a short period of time. Furthermore, LED light has certain spatial light distribution angle differences, which can lead to inaccurate color difference determination when sampling at a single angle. Traditional static sampling methods generally fail to consider the temporal evolution trend of light color parameters and lack a spatial fusion mechanism for acquiring information from multiple angles. This makes it difficult to cover LED devices with rapidly changing luminous characteristics or asymmetric structures. Therefore, the present invention proposes a method and system for detecting and analyzing light color in LED light-emitting diodes. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for detecting and analyzing the light color of LED light emitting diodes to solve the problems mentioned in the above background technology.

[0006] The present invention can be implemented through the following technical solutions: a light color detection and analysis system for LED light emitting diodes, the system comprising: a light color acquisition module, a time sequence modeling module, a light color recognition module and a processing module;

[0007] The light color acquisition module is used to obtain the spectral data generated by the target LED during the lighting process. The spectral data includes the main wavelength value, CIE color coordinates (x, y), fitted color temperature value and light intensity value;

[0008] The time series modeling module is used to collect multiple spectral data points at fixed time intervals within a continuous preset time period after the target LED starts to emit light, and calculate the dominant wavelength, color coordinates and fitted color temperature value based on each spectral data point to form a multi-time light color parameter sequence, that is, a light color data set arranged in time;

[0009] The time series modeling module constructs a time series model based on the above light color parameter sequence to describe the trend of light color parameters such as main wavelength, color coordinates, and color temperature changing over time, and outputs the time series model and the corresponding original parameter sequence;

[0010] The light color recognition module is used to receive the time series model and light color parameter sequence output by the time series modeling module;

[0011] Identifying whether the LED has entered a stable lighting state according to a preset judgment standard;

[0012] The processing module is used to receive the light color data of the target LED after the light color recognition module confirms the stable state, and extract the average value of the light color data of all sampling points in the stable state as the input data for classification judgment and project it into the established multi-dimensional light color feature model;

[0013] The multidimensional light color feature model is a set of training samples with labeled light color levels. Each sample contains four indicators: dominant wavelength, color coordinates (x, y), fitted color temperature value, and dynamic change trend direction, forming a four-dimensional feature vector. The dynamic change trend direction is represented by +1, 0, and -1 to indicate rising, stable, or falling, respectively.

[0014] Each vector in the sample library is bound to a light color grade label, which is used for similarity matching of input vectors and light color grade classification.

[0015] The processing module classifies and judges the light color data through the minimum distance matching algorithm or the K nearest neighbor method, and outputs the light color level and color difference evaluation index corresponding to the light color data.

[0016] A further technical improvement of the present invention is that the method for the light color acquisition module to obtain spectral data includes:

[0017] A1. The system provides a constant driving current to the target LED to make it enter the light-emitting state, and obtains the LED light intensity in real time through the light intensity acquisition channel to obtain the light intensity value;

[0018] In this embodiment, the constant driving current is automatically set according to the rated operating current of the LED or selected by looking up a table to ensure that the light emitting process meets the typical light emission characteristics;

[0019] If the light intensity values ​​collected at three consecutive sampling moments are all higher than the set light intensity threshold, and the light intensity change rate is less than the preset threshold, the system confirms that the LED has entered the sampleable state and enters step A2;

[0020] A2. At each sampling moment, the spectrum acquisition module continuously collects no less than three frames of spectrum data, with a frame interval of no more than 5ms;

[0021] And the spectrum acquisition module compares the three frames of spectrum data collected continuously and extracts the main wavelength and CIE color coordinates (x, y);

[0022] If the main wavelength difference between frames is less than 1nm and the color coordinate difference is less than 0.003, the spectrum data at that moment is determined to be valid;

[0023] If it is judged to be an invalid frame, the sampling is delayed and retried until a valid data frame is obtained;

[0024] A3. Standardizing the spectrum data determined to be valid to obtain standard spectrum data, including the following steps:

[0025] The wavelength sampling range is unified to 380nm to 780nm, with a sampling step of 1nm, for a total of 401 wavelength points;

[0026] Perform maximum normalization on the spectral intensity data so that the main peak intensity value is set to 1.0;

[0027] Apply a sliding median filter to remove local spectral outliers and improve data smoothness and anti-interference performance;

[0028] A4. Based on the standard spectrum data processed in step A3, calculate the following light color parameters:

[0029] Dominant wavelength: obtained by projecting spectral data onto the isochromatic lines in the CIE chromaticity diagram;

[0030] CIE color coordinates (x, y): Two-dimensional chromaticity coordinates calculated based on the CIE1931 color space, used to locate the position of light color in the chromaticity diagram;

[0031] Fitted color temperature value: The fitted color temperature obtained based on the minimum distance fitting method between the CIE color coordinates and the blackbody locus;

[0032] The light color acquisition module packages the above light color parameters and their sampling timestamps into structured single-frame light color data for use by subsequent recognition or classification modules. Specifically, the structured single-frame light color data includes: dominant wavelength, CIE color coordinates (x, y), fitted color temperature value, sampling timestamp, sampling frame number, and LED number.

[0033] A further technical improvement of the present invention is that the step of the light color recognition module recognizing whether the LED has entered a stable light emitting state includes:

[0034] Z1, the light color acquisition module collects the light intensity value and light color parameters of the target LED at fixed time intervals, and forms a light intensity change curve and a color coordinate change curve within a preset sampling period;

[0035] Z2, the light color recognition module analyzes the continuous sampling points corresponding to the light intensity change curve and the color coordinate change curve in Z1, and determines whether the following convergence conditions are met:

[0036] Z21, light intensity convergence: the light intensity change rate of multiple consecutive sampling points is less than the set light intensity change rate threshold;

[0037] Z22, color coordinate convergence: the color coordinate change amplitude of multiple consecutive sampling points is less than the set change amplitude threshold;

[0038] When the two conditions Z21 and Z22 are met at the same time, it is determined that the LED has reached a stable lighting state.

[0039] A further technical improvement of the present invention is that the system sets a delayed inspection time based on the time required for multiple LEDs in the same batch to reach a stable lighting state, including:

[0040] At the beginning of batch testing, several LED samples are selected and the luminous stability judgment process is performed separately, and the time required for each LED to reach a stable state is recorded;

[0041] Based on the sample stabilization time, the minimum stabilization time Tmin and the maximum stabilization time Tmax are determined to form the dynamic value range [Tmin, Tmax] of the delay check time;

[0042] The system initially uses Tmin as the delayed inspection time, and as the number of inspected LEDs increases, the actual delayed inspection time is dynamically adjusted from Tmin to Tmax;

[0043] When the number of inspections reaches the set threshold or the delay time approaches Tmax, the current delayed inspection time is fixed as the inspection waiting time for subsequent LEDs in the batch.

[0044] A further technical improvement of the present invention is that when the maximum value Tmax and the minimum value Tmin in the dynamic range interval [Tmin, Tmax] change, the system dynamically adjusts the delay check time, including:

[0045] K1. The system obtains the stable time set through the sample LED and extracts the original minimum stable time and maximum stabilization time , construct the initial delay time interval ;

[0046] K2. When detecting the kth LED, the system uses the following formula to calculate the delay inspection time ;

[0047] ;

[0048] K3. As more LEDs complete stability testing, the system continues to update Tmin and Tmax, which are recorded as 、 , and calculate its relative change percentage:

[0049] ;

[0050] ;

[0051] K4, the system checks the current delay time based on the amplitude of Δmin and Δmax Fine-tune and adjust the strategy as follows:

[0052] ;

[0053] Among them, α and β are adjustable adjustment coefficients (such as 0.3 to 0.5) used to control the correction ratio.

[0054] A further technical improvement of the present invention is that the light color acquisition module further includes a plurality of optical acquisition channels, which are respectively arranged at a plurality of angular positions in the light emitting direction of the target LED;

[0055] At each sampling moment, the system synchronously acquires spectral data from different angles through the multiple channels to form a multi-angle light color data set;

[0056] Multi-angle light color data includes the dominant wavelength value, CIE color coordinates (x, y), fitted color temperature value and light intensity value in multiple directions. Each direction corresponds to a sampling channel and number.

[0057] To eliminate the color deviation caused by different light distribution angles of LEDs, the system assigns weights to the collected values ​​in each direction based on a preset light distribution model or empirical distribution, and adopts the following weighted fusion strategy:

[0058] For any light color parameter P, the fusion value at the sampling time ti is:

[0059] , where j represents the direction number, is the weight, is the light color parameter of the j-th direction at the i-th moment;

[0060] Among them, the weight It can be set based on the target LED light distribution curve, angle response function, or empirical statistical model to make the fusion value closer to the subjective color perception of the human eye;

[0061] The fused light color parameters will serve as the representative input data at the current sampling moment for use in subsequent time series modeling, steady-state judgment, and light color level evaluation modules.

[0062] A further technical improvement of the present invention is that the method for the processing module to classify and judge the light color data includes:

[0063] M1. When the target LED reaches a stable state, the system receives all valid sampling point data in the stable state from the light color recognition module, extracts the light color parameters of each frame, and calculates the average value of all valid frames as the light color feature vector of the target LED;

[0064] M2, the preset multi-dimensional light color feature model consists of m training samples, each sample vector is: , j=1,2,...,m; each sample vector corresponds to a label , indicating its light color level;

[0065] M3, calculate the light color level using minimum Euclidean distance matching or K nearest neighbor classification;

[0066] pass Calculate color difference evaluation index .

[0067] A method for detecting and analyzing the light color of an LED light emitting diode comprises the following steps:

[0068] Step 1: Light color data collection:

[0069] Obtain spectral data generated by the target LED during lighting, including the dominant wavelength, CIE color coordinates (x, y), fitted color temperature, and light intensity.

[0070] Step 2: Time series sampling and modeling:

[0071] After the target LED starts emitting light, multiple spectral data points are collected at fixed time intervals within a continuous preset period of time, and the dominant wavelength, color coordinates, and fitted color temperature values ​​are calculated based on each spectral data point to form a time-arranged light color parameter sequence;

[0072] A time series model is constructed based on the light color parameter sequence to describe the changing trends of the main wavelength, color coordinates and color temperature parameters over time;

[0073] Step 3: Stability judgment:

[0074] Based on the constructed time series model and parameter sequence, determine whether the target LED has reached a stable lighting state, wherein the judgment basis includes a preset light color change amplitude and trend threshold judgment criteria;

[0075] Step 4: Average light color vector extraction:

[0076] After the target LED is identified as reaching a stable state, the light color parameters of all valid sampling points within the stable time period are extracted, and their average value is calculated to form a stable light color feature vector as the input for subsequent classification judgment;

[0077] Step 5: Light color grade judgment and color difference evaluation:

[0078] The average light color feature vector is projected into a pre-built multidimensional light color feature model, and the light color level corresponding to the target LED is determined by the minimum Euclidean distance matching algorithm or the K-nearest neighbor classification method, and the color difference evaluation index between it and the reference sample is output.

[0079] Compared with the prior art, the present invention has the following beneficial effects:

[0080] The present invention constructs a curve of light intensity and color coordinates and introduces a convergence trend analysis mechanism to automatically determine the stable state of LED lighting. This effectively avoids erroneous judgments caused by unstable early data and improves the representativeness and accuracy of the detection data.

[0081] Furthermore, the present invention records the time required for multiple LEDs in the same batch to reach a stable state, constructs a dynamic delay inspection time adjustment mechanism, and gradually corrects the waiting time according to the number of test samples, effectively reducing redundant waiting time, while improving the system's automatic adjustment capability and enhancing the intelligent level of batch detection.

[0082] On the other hand, the present invention introduces a multi-angle sampling and weighted fusion mechanism, combines the main wavelength, color coordinates, color temperature and their changing trends, constructs a four-dimensional light color feature vector, and performs classification and judgment through KNN or minimum distance algorithm, which can effectively deal with the local color difference caused by the lighting distribution angle and improve the credibility and consistency of the light color grade evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] To facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0084] Figure 1 is a system block diagram of the present invention;

[0085] Figure 2 This is a functional diagram of the light color acquisition module in the present invention;

[0086] Figure 3 This is a functional diagram of the timing modeling module in the present invention;

[0087] Figure 4 Schematic diagram of the function of the light color recognition module in the present invention;

[0088] Figure 5 This is a functional diagram of the processing module in the present invention. DETAILED DESCRIPTION

[0089] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0090] Example 1

[0091] See also Figure 1-5 As shown, the present invention provides a light color detection and analysis system for LED light emitting diodes, which includes: a light color acquisition module, a time sequence modeling module, a light color recognition module and a processing module;

[0092] The light color acquisition module is used to obtain the spectral data generated by the target LED during the lighting process. The spectral data includes the main wavelength value, CIE color coordinates (x, y), fitted color temperature value and light intensity value;

[0093] In this embodiment, the light color acquisition module includes at least one optical acquisition channel and a spectrum detection device connected thereto, the spectrum detection device has a wavelength detection range of 380nm to 780nm, and a sampling resolution of no more than 1nm;

[0094] The method for the light color acquisition module to obtain spectral data includes:

[0095] A1. The system provides a constant driving current to the target LED to make it enter the light-emitting state, and obtains the LED light intensity in real time through the light intensity acquisition channel to obtain the light intensity value;

[0096] In this embodiment, the constant driving current is automatically set according to the rated operating current of the LED or selected by looking up a table to ensure that the light emitting process meets the typical light emission characteristics;

[0097] The light intensity acquisition channel is a dedicated signal acquisition path for collecting the light intensity output by the target LED during the lighting process. It includes: an optical receiving component, a photoelectric sensor, and a signal amplification and conversion circuit. It is used to convert the total luminous flux or local light intensity emitted by the target LED into a calculable electrical signal, which is used as the basis for judging the lighting status in subsequent processing.

[0098] If the light intensity values ​​collected at three consecutive sampling moments are all higher than the set light intensity threshold (such as 80% of the rated light intensity), and the light intensity change rate is less than the preset threshold (such as ±5%), the system confirms that the LED has entered the sampling state and enters step A2;

[0099] A2. At each sampling moment, the spectrum acquisition module continuously collects no less than three frames of spectrum data, with a frame interval of no more than 5ms;

[0100] And the spectrum acquisition module compares the three frames of spectrum data collected continuously and extracts the main wavelength and CIE color coordinates (x, y);

[0101] If the main wavelength difference between frames is less than 1nm and the color coordinate difference is less than 0.003, the spectrum data at that moment is determined to be valid;

[0102] If it is determined to be an invalid frame, the sampling is delayed and retried (i.e. the current sampling is invalidated, and a delay period is waited for, and a complete three-frame sampling process is performed again) until a valid data frame is obtained;

[0103] A3. Standardizing the spectrum data determined to be valid to obtain standard spectrum data, including the following steps:

[0104] The wavelength sampling range is unified to 380nm to 780nm, with a sampling step of 1nm, for a total of 401 wavelength points;

[0105] Perform maximum normalization on the spectral intensity data so that the main peak intensity value is set to 1.0;

[0106] Apply a sliding median filter to remove local spectral outliers and improve data smoothness and anti-interference performance;

[0107] A4. Based on the standard spectrum data processed in step A3, calculate the following light color parameters:

[0108] Dominant wavelength: obtained by projecting spectral data onto the isochromatic lines in the CIE chromaticity diagram;

[0109] CIE color coordinates (x, y): Two-dimensional chromaticity coordinates calculated based on the CIE1931 color space, used to locate the position of light color in the chromaticity diagram;

[0110] Fitted color temperature value: The fitted color temperature obtained based on the minimum distance fitting method between the CIE color coordinates and the blackbody locus;

[0111] The light color acquisition module packages the above light color parameters and their sampling timestamps into structured single-frame light color data for use by subsequent recognition or classification modules. Specifically, the structured single-frame light color data includes: dominant wavelength, CIE color coordinates (x, y), fitted color temperature value, sampling timestamp, sampling frame number, and LED number;

[0112] The time series modeling module is used to collect multiple spectral data points at fixed time intervals within a continuous preset period of time after the target LED starts to emit light. Based on each spectral data point, the dominant wavelength, color coordinates, and fitted color temperature values ​​are calculated to form a multi-moment light color parameter sequence, that is, a time-arranged light color data set.

[0113] In this embodiment, the preset time is 1s-10s, the sampling interval is 0.1s-1s, and the number of sampling points is not less than five;

[0114] The time series modeling module constructs a time series model based on the above light color parameter sequence to describe the trend of light color parameters such as main wavelength, color coordinates, and color temperature changing over time, and outputs the time series model and the corresponding original parameter sequence;

[0115] The light color recognition module is used to receive the time series model and light color parameter sequence output by the time series modeling module;

[0116] According to the preset judgment criteria, identify whether the LED has entered a stable lighting state;

[0117] The steps of the light color recognition module identifying whether the LED has entered a stable light emission state include:

[0118] Z1. Collection and real-time monitoring of light intensity / color coordinate change curves:

[0119] The light color acquisition module starts working and collects the light intensity and light color parameters of the target LED in real time;

[0120] Within the preset sampling period, multiple light intensity values ​​and light color parameter sampling point data are continuously acquired at fixed time intervals to form a light intensity change curve and a color coordinate change curve, reflecting the change trend of light output and light color parameters over time during the LED lighting process;

[0121] Specifically, the formation of the light intensity variation curve includes:

[0122] Within the preset sampling period T after the target LED is lit, the system continuously samples at a fixed time interval Δt, and obtains the light intensity value Ii of the current light output each time;

[0123] Arrange all sampling points in time sequence to form a two-dimensional data set: {(ti,Ii)}, where Where, Indicates the sampling time point corresponding to the i-th sampling point; Indicates the sampling start time (i.e. the time of the first sampling point);

[0124] i represents the serial number of the sampling point, usually counting from 0, that is, i=0,1,...,n-1, where n is the number of sampling points;

[0125] In the coordinate system, this dataset uses "time ti" as the horizontal axis and "light intensity value Ii" as the vertical axis. Connecting each point forms a continuous light intensity change curve, reflecting the change trend of the luminous intensity of the LED over time during the lighting process.

[0126] The formation of the color coordinate change curve includes:

[0127] In synchronization with the light intensity acquisition process, each sampling also records the CIE color coordinates (xi, yi) of the target LED at that moment, and the data set of these CIE color coordinates changing over time is represented as: {(ti, xi)} and {(ti, yi)}, with time ti as the horizontal axis and x value or y value as the vertical axis, respectively, which can be formed:

[0128] x-axis change curve (time – color coordinate x);

[0129] y-axis change curve (time-color coordinate y);

[0130] These two curves together reflect the changing trend of the color output of the LED during the lighting process;

[0131] Z2. Convergence trend analysis and stability determination:

[0132] The light color recognition module analyzes the light intensity and color coordinate data of several consecutive sampling points to determine whether their change range meets the stability requirements, including:

[0133] Light intensity convergence: The light intensity change rate of multiple consecutive sampling points is less than the set light intensity change rate threshold;

[0134] Color coordinate convergence: The color coordinate change amplitude of multiple consecutive sampling points is less than the set change amplitude threshold;

[0135] When the above two convergence conditions are met at the same time, the light color recognition module determines that the target LED has reached a stable lighting state and allows the system to execute subsequent steps;

[0136] In this embodiment, the judgment criteria include all of the following conditions:

[0137] (1) The absolute value of the dominant wavelength change at three consecutive sampling points is less than 1 nm;

[0138] (2) The change range of the corresponding color coordinates (x, y) is less than 0.005;

[0139] (3) The absolute value of the change in the fitted color temperature value is no more than 50K;

[0140] When all the above judgment conditions are met, the system determines that the LED has reached a stable lighting state and allows subsequent light color judgment to be performed;

[0141] The processing module is used to receive the light color data of the target LED after the light color recognition module confirms the stable state, and extract the average value of the light color data of all sampling points in the stable state as the input data for classification judgment and project it into the established multi-dimensional light color feature model;

[0142] The multidimensional light color feature model is a set of training samples with labeled light color levels. Each sample contains four indicators: dominant wavelength, color coordinates (x, y), fitted color temperature value, and dynamic change trend direction, forming a four-dimensional feature vector. The dynamic change trend direction is represented by +1, 0, and -1 to indicate rising, stable, or falling, respectively.

[0143] Each vector in the sample library is bound to a light color grade label, which is used for similarity matching of input vectors and light color grade classification.

[0144] The processing module classifies and judges the light color data through the minimum distance matching algorithm, and outputs the light color grade and color difference evaluation index corresponding to the light color data;

[0145] The method for the processing module to classify and judge the light color data includes:

[0146] M1. When the target LED reaches a stable state, the system receives all valid sampling point data in the stable state from the light color recognition module, extracts the light color parameters of each frame (dominant wavelength, CIE color coordinates (x, y), fitted color temperature value, color temperature drift direction), and calculates the average value of all valid frames as the light color feature vector of the target LED;

[0147] That is, after the target color temperature value LED color temperature value enters a stable state, a total of n valid moments of light color parameter data are collected, and the four light color parameters at the i-th moment are recorded as: , i=1,2,...,n;

[0148] Then the light color feature vector (i.e. target vector) in the stable state ) is defined as the mean value vector of these n sampling points:

[0149] ;

[0150] Where, ;

[0151] ;

[0152] ;

[0153] ;

[0154] M2, the preset multi-dimensional light color feature model consists of m training samples, each sample vector is: , j=1,2,...,m; each sample vector corresponds to a label , indicating its light color level;

[0155] M3, using the minimum Euclidean distance matching method for judgment:

[0156] Calculate the Euclidean distance between the target vector and each sample:

[0157] ;

[0158] Take the envoy smallest , and its corresponding label That is the determined light color level of the target LED;

[0159] At the same time, the color difference evaluation index Defined as: ;

[0160] The smaller the value, the closer the target LED light color is to a reference grade sample, indicating that the light color is stable and acceptable.

[0161] like If the color difference exceeds the preset threshold, it will be marked as an abnormal sample or tested again.

[0162] Example 2

[0163] A light color detection and analysis system for LED light emitting diodes, the system comprising: a light color acquisition module, a time sequence modeling module, a light color recognition module and a processing module;

[0164] The light color acquisition module is used to obtain the spectral data generated by the target LED during the lighting process. The spectral data includes the main wavelength value, CIE color coordinates (x, y), fitted color temperature value and light intensity value;

[0165] The time series modeling module is used to collect multiple spectral data points at fixed time intervals within a continuous preset period of time after the target LED starts to emit light. Based on each spectral data point, the dominant wavelength, color coordinates, and fitted color temperature values ​​are calculated to form a multi-moment light color parameter sequence, that is, a time-arranged light color data set.

[0166] The time series modeling module constructs a time series model based on the above light color parameter sequence to describe the trend of light color parameters such as main wavelength, color coordinates, and color temperature changing over time, and outputs the time series model and the corresponding original parameter sequence;

[0167] The light color recognition module is used to receive the time series model and light color parameter sequence output by the time series modeling module;

[0168] According to the preset judgment criteria, identify whether the LED has entered a stable lighting state;

[0169] The processing module is used to receive the light color data of the target LED after the light color recognition module confirms the stable state, and extract the average value of the light color data of all sampling points in the stable state as the input data for classification judgment and project it into the established multi-dimensional light color feature model;

[0170] The processing module classifies and judges the light color data through the minimum distance matching algorithm or the K nearest neighbor method, and outputs the light color level and color difference evaluation index corresponding to the light color data.

[0171] Compared to Example 1, the system in Example 2 sets a delayed inspection time based on the time required for multiple LEDs in the same batch to reach a stable lighting state, including:

[0172] At the beginning of batch testing, several LED samples are selected and the luminous stability judgment process is performed separately, and the time required for each LED to reach a stable state is recorded;

[0173] Based on the sample stabilization time, the minimum stabilization time Tmin and the maximum stabilization time Tmax are determined to form the dynamic value range [Tmin, Tmax] of the delay check time;

[0174] The system initially uses Tmin as the delayed inspection time, and as the number of inspected LEDs increases, the actual delayed inspection time is dynamically adjusted from Tmin to Tmax;

[0175] When the number of inspections reaches the set threshold or the delay time approaches Tmax, the current delayed inspection time is fixed as the inspection waiting time for subsequent LEDs in the batch;

[0176] Specifically, the following steps are included:

[0177] B1. Obtaining sample stability time:

[0178] At the beginning of LED testing, several target LED samples are selected from the current testing batch, and the luminous stability judgment process is performed on each LED. The time required for each LED to reach a stable state is recorded to form a stable time data set:

[0179] ;

[0180] B2. Calculation of stable time range:

[0181] The system extracts the maximum value Tmax and the minimum value Tmin from the above set and constructs the dynamic range interval [Tmin, Tmax] of the delay inspection time;

[0182] B3. Delayed inspection time initialization:

[0183] In the initial stage, the system sets the delayed inspection time DCT = Tmin as the initial delay waiting time for the first few LED inspections;

[0184] B4. Dynamic approximation adjustment of delay time:

[0185] As the number of tests increases, the system dynamically approaches the delayed inspection time DCT from Tmin to Tmax according to the set strategy, for example, Calculate; where, is the delay inspection time of the current k-th LED, k is the number of LEDs that have completed the inspection, and N is the upper limit of the number of LEDs allowed to be adjusted.

[0186] When the maximum value Tmax and the minimum value Tmin in the dynamic range interval [Tmin, Tmax] change, the system dynamically adjusts the delay check time, including:

[0187] K1. The system obtains the stable time set through the sample LED and extracts the original minimum stable time and maximum stabilization time , construct the initial delay time interval ;

[0188] K2. When detecting the kth LED, the system uses the following formula to calculate the delay inspection time ;

[0189] ;

[0190] K3. As more LEDs complete stability testing, the system continues to update Tmin and Tmax, which are recorded as 、 , and calculate its relative change percentage:

[0191] ;

[0192] ;

[0193] K4, the system checks the current delay time based on the amplitude of Δmin and Δmax Fine-tune and adjust the strategy as follows:

[0194] ;

[0195] Among them, α and β are adjustable adjustment coefficients (such as 0.3 to 0.5) used to control the correction ratio;

[0196] Moreover, in this embodiment, when the cumulative number of detected samples reaches a set ratio (such as 30% of the total number of samples), the system will 、 Alternative 、 As a new interval reference to further improve adaptability.

[0197] Example 3

[0198] A light color detection and analysis system for LED light emitting diodes, the system comprising: a light color acquisition module, a time sequence modeling module, a light color recognition module and a processing module;

[0199] The light color acquisition module is used to obtain the spectral data generated by the target LED during the lighting process. The spectral data includes the main wavelength value, CIE color coordinates (x, y), fitted color temperature value and light intensity value;

[0200] The time series modeling module is used to collect multiple spectral data points at fixed time intervals within a continuous preset period of time after the target LED starts to emit light. Based on each spectral data point, the dominant wavelength, color coordinates, and fitted color temperature values ​​are calculated to form a multi-moment light color parameter sequence, that is, a time-arranged light color data set.

[0201] The time series modeling module constructs a time series model based on the above light color parameter sequence to describe the trend of light color parameters such as main wavelength, color coordinates, and color temperature changing over time, and outputs the time series model and the corresponding original parameter sequence;

[0202] The light color recognition module is used to receive the time series model and light color parameter sequence output by the time series modeling module;

[0203] According to the preset judgment criteria, identify whether the LED has entered a stable lighting state;

[0204] The processing module is used to receive the light color data of the target LED after the light color recognition module confirms the stable state, and extract the average value of the light color data of all sampling points in the stable state as the input data for classification judgment and project it into the established multi-dimensional light color feature model;

[0205] The processing module classifies and judges the light color data through the minimum distance matching algorithm or the K nearest neighbor method, and outputs the light color level and color difference evaluation index corresponding to the light color data.

[0206] Compared with Example 1 and Example 2, the light color acquisition module in Example 3 further includes multiple optical acquisition channels, which are respectively set at multiple angular positions in the light emission direction of the target LED;

[0207] At each sampling moment, the system synchronously acquires spectral data from different angles through multiple channels to form a multi-angle light and color data set;

[0208] Multi-angle light color data includes the dominant wavelength value, CIE color coordinates (x, y), fitted color temperature value and light intensity value in multiple directions. Each direction corresponds to a sampling channel and number.

[0209] To eliminate the color deviation caused by different light distribution angles of LEDs, the system assigns weights to the collected values ​​in each direction based on a preset light distribution model or empirical distribution, and adopts the following weighted fusion strategy:

[0210] For any light color parameter P, the fusion value at the sampling time ti is:

[0211] , where j represents the direction number, is the weight, is the light color parameter of the j-th direction at the i-th moment;

[0212] Among them, the weight It can be set based on the target LED light distribution curve, angle response function, or empirical statistical model to make the fusion value closer to the subjective color perception of the human eye;

[0213] The fused light color parameters will serve as the representative input data at the current sampling moment for use in subsequent time series modeling, steady-state judgment, and light color level evaluation modules.

[0214] The method for the processing module to classify and judge the light color data includes:

[0215] M1. When the target LED reaches a stable state, the system receives all valid sampling point data in the stable state from the light color recognition module, extracts the light color parameters of each frame (dominant wavelength, CIE color coordinates (x, y), fitted color temperature value, color temperature drift direction), and calculates the average value of all valid frames as the light color feature vector of the target LED;

[0216] That is, after the target color temperature value LED color temperature value enters a stable state, a total of n valid moments of light color parameter data are collected, and the four light color parameters at the i-th moment are recorded as: , i=1,2,...,n;

[0217] Then the light color feature vector (i.e. target vector) in the stable state ) is defined as the mean value vector of these n sampling points:

[0218] ;

[0219] Where, ;

[0220] ;

[0221] ;

[0222] ;

[0223] M2, the preset multi-dimensional light color feature model consists of m training samples, each sample vector is: , j=1,2,...,m; each sample vector corresponds to a label , indicating its light color level;

[0224] M3, using K-nearest neighbor classification (KNN) for judgment:

[0225] Select The smallest first k samples are recorded as a set , and then vote to decide the final label:

[0226] Where, is the light color grade;

[0227] At the same time, the color difference evaluation index Defined as: ;

[0228] The smaller the value, the closer the target LED light color is to a reference grade sample, indicating that the light color is stable and acceptable.

[0229] like If the color difference exceeds the preset threshold, it will be marked as an abnormal sample or tested again.

[0230] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0231] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A light color detection and analysis system for LED light emitting diodes, characterized in that: include: Light color acquisition module, used to obtain the spectrum data generated by the target LED during the lighting process; The time series modeling module is used to collect multiple spectral data points at fixed time intervals within a continuous preset time period after the target LED starts to emit light, calculate the dominant wavelength, color coordinates and fitted color temperature values, form a multi-time light color parameter sequence, and build a time series model based on the sequence, and output the time series model and the corresponding original parameter sequence; A light color recognition module is configured to receive the time series model and light color parameter sequence output by the time series modeling module, and identify whether the LED has entered a stable light emission state according to a preset judgment standard; The processing module is used to receive the light color data of the target LED after the light color recognition module confirms the stable state, and extract the average value of the light color data of all sampling points in the stable state, project it into the established multi-dimensional light color feature model as input data for classification judgment, and classify and judge the light color data, and output the light color grade and color difference evaluation index corresponding to the light color data.

2. A light color detection and analysis system for LED light emitting diodes according to claim 1, characterized in that: The method for the light color acquisition module to obtain spectral data includes: A1. The system provides a constant driving current to the target LED to make it enter the light-emitting state, and obtains the LED light intensity in real time through the light intensity acquisition channel to obtain the light intensity value; If the light intensity values ​​collected at three consecutive sampling moments are all higher than the set light intensity threshold, and the light intensity change rate is less than the preset threshold, it is confirmed that the LED has entered the sampling state and enters step A2; A2. At each sampling moment, the spectrum acquisition module continuously collects no less than three frames of spectrum data; And the spectrum acquisition module compares the three frames of spectrum data collected continuously and extracts the main wavelength and CIE color coordinates (x, y); If the main wavelength difference between frames is less than 1nm and the color coordinate difference is less than 0.003, the spectrum data at that moment is determined to be valid; If it is judged to be an invalid frame, the sampling is delayed and retried until a valid data frame is obtained; A3. performing a standardization operation on the spectrum data determined to be valid to obtain standard spectrum data; A4. Calculate the light color parameters based on the standard spectrum data processed in step A3. The light color parameters include: dominant wavelength, CIE color coordinates (x, y), and fitted color temperature. The light color acquisition module packages the above light color parameters and their sampling timestamps into structured single-frame light color data for use by subsequent recognition or classification modules.

3. The LED light color detection and analysis system according to claim 1, characterized in that: The step of the light color recognition module recognizing whether the LED has entered a stable light emitting state includes: Z1, the light color acquisition module collects the light intensity value and light color parameters of the target LED at fixed time intervals, and forms a light intensity change curve and a color coordinate change curve within a preset sampling period; Z2, the light color recognition module analyzes the continuous sampling points corresponding to the light intensity change curve and the color coordinate change curve in Z1, and determines whether the following convergence conditions are met: Z21, light intensity convergence: the light intensity change rate of multiple consecutive sampling points is less than the set light intensity change rate threshold; Z22, color coordinate convergence: the color coordinate change amplitude of multiple consecutive sampling points is less than the set change amplitude threshold; When the two conditions Z21 and Z22 are met at the same time, it is determined that the LED has reached a stable lighting state.

4. The LED light color detection and analysis system according to claim 1, characterized in that: The system sets the delayed inspection time based on the time required for multiple LEDs in the same batch to reach a stable lighting state, including: At the beginning of batch testing, several LED samples are selected and the luminous stability judgment process is performed separately, and the time required for each LED to reach a stable state is recorded; Based on the sample stabilization time, the minimum stabilization time Tmin and the maximum stabilization time Tmax are determined to form the dynamic value range [Tmin, Tmax] of the delay inspection time; The system initially uses Tmin as the delayed inspection time, and as the number of inspected LEDs increases, the actual delayed inspection time is dynamically adjusted from Tmin to Tmax; When the number of inspections reaches the set threshold or the delay time approaches Tmax, the current delayed inspection time is fixed as the inspection waiting time for subsequent LEDs in the batch.

5. The LED light color detection and analysis system according to claim 4, characterized in that: When the maximum value Tmax and the minimum value Tmin in the dynamic range interval [Tmin, Tmax] change, the system dynamically adjusts the delay check time, including: K1. The system obtains the stable time set through the sample LED and extracts the original minimum stable time and the original maximum settling time , construct the initial delay time interval ; K2. When detecting the kth LED, the system uses the following formula to calculate the delay inspection time ; ； K3. As more LEDs complete stability testing, the system continues to update Tmin and Tmax, which are recorded as 、 , and calculate its relative change percentage: ； ； K4, the system checks the current delay time based on the amplitude of Δmin and Δmax Make adjustments to obtain the corrected delayed check time : ； Where α and β are adjustable adjustment coefficients.

6. The LED light color detection and analysis system according to claim 1, characterized in that: The light color acquisition module further includes a plurality of optical acquisition channels, which are respectively arranged at a plurality of angular positions in the light emitting direction of the target LED; At each sampling moment, the system synchronously acquires spectral data from different angles through the multiple channels to form a multi-angle light color data set; Multi-angle light color data includes the dominant wavelength value, CIE color coordinates (x, y), fitted color temperature value and light intensity value in multiple directions. Each direction corresponds to a sampling channel and number. The system assigns weights to the collected values ​​in each direction and performs weighted fusion of the light and color parameters in each direction; The fused light color parameters will serve as the representative input data at the current sampling moment for use in subsequent time series modeling, steady-state judgment, and light color level evaluation modules.

7. The LED light color detection and analysis system according to claim 1, characterized in that: The method for the processing module to classify and judge the light color data includes: M1. When the target LED reaches a stable state, the system receives all valid sampling point data in the stable state from the light color recognition module, extracts the light color parameters of each frame, and calculates the average value of all valid frames as the light color feature vector of the target LED; M2, the preset multi-dimensional light color feature model consists of m training samples, each sample vector is: , j=1,2,...,m; each sample vector corresponds to a label , indicating its light color level; M3, calculate the light color level using minimum Euclidean distance matching or K nearest neighbor classification; pass Calculate color difference evaluation index .

8. A method for detecting and analyzing the light color of LED light emitting diodes, characterized in that: The method uses the analysis system according to any one of claims 1 to 7, comprising: Step 1: Collect spectral data of the target LED during lighting, including dominant wavelength, CIE color coordinates (x, y), fitted color temperature, and light intensity. Step 2: Collect multiple spectral data points at fixed time intervals within a continuous preset time period to form a light color parameter sequence, and construct a time series model to describe the change trend of the light color parameters over time; Step 3: Based on the time series model and parameter sequence, determine whether the LED has reached a stable lighting state according to the preset light color change amplitude and trend threshold; Step 4: After identifying the stable state, extract all valid light color parameters in the time period and calculate the average value as the stable light color feature vector; Step 5: Project the average light color vector into the multidimensional light color feature model, determine the light color level through minimum distance matching or K-nearest neighbor classification algorithm, and output the color difference evaluation index.

Images