Prescription control method based on continuous age eye transmittance spectrum and scene adaptation

Abstract

The application discloses a formula pre-storage control method based on continuous age eye transmission spectrum and scene adaptation, which comprises the following steps: S1, constructing an eye medium spectral transmittance function covering integer age points of 1-100 years, and establishing a continuous age eye transmission spectrum database; S2, establishing a three-level mapping model, and defining a four-fold parallel constraint screening condition; S3, adopting a hierarchical generation strategy to pre-generate a candidate spectrum set offline for a multi-channel LED light source; S4, extracting a scene spectrum formula; S5, pre-storing all independent formulas as a structured key-value pair database with age, color temperature and scene preference as three index items; and S6, according to the age, color temperature and scene preference slide bar values input by a user, calling a nearest neighbor formula in real time and driving a multi-channel LED to output a target spectrum. The application realizes continuous adaptation of precise photobiological safety protection and scene rhythm lighting in all age groups, and has fast response speed and strong multi-channel light source compatibility.

Description

Technical Field

[0001] This invention belongs to the field of healthy intelligent lighting technology. Specifically, it discloses a formula pre-storage control method based on continuous age eye transmission spectrum and scene adaptation. Background Technology

[0002] In recent years, the field of healthy lighting has focused on spectral modulation technology, aiming to balance visual efficacy with non-visual biological effects by adjusting the spectral power distribution of light sources. With the development of multi-channel mixing technology for LED light sources, personalized spectral optimization solutions for different age groups have begun to be explored. However, existing technologies still face significant technical bottlenecks in areas such as age modeling accuracy, scene adaptability, and safety performance coordination.

[0003] (a) There are defects in age dimension modeling: Current methods generally rely on discrete age-grouping models (such as the child / old age dichotomy) and have not yet constructed a continuous function library of lens transmittance covering ages 1-100. This discontinuous modeling approach makes it impossible to accurately quantify the nonlinear decay characteristics of blue light transmittance caused by lens development and aging. Although some studies have addressed age-related changes in lens transmittance, none have systematically modeled it as an input parameter for the spectral design of lighting sources.

[0004] (ii) Insufficient scene adaptability: While mainstream technologies can adapt to circadian rhythms by adjusting color temperature or illuminance, they fail to effectively resolve the conflicting circadian rhythm factor (CAF) requirements between work and leisure scenarios—work scenarios require high CAF values ​​to enhance alertness, while leisure scenarios require low CAF values ​​to promote physiological relaxation. Existing systems suffer from a lack of compatibility with elderly users when achieving dual-objective extreme value optimization.

[0005] (iii) Security and performance are difficult to coordinate: Existing solutions generally employ fixed blue light suppression spectra, failing to dynamically balance the coupling relationship between age-related decreases in blue light hazard (BLH) sensitivity and CAF response decay. Furthermore, they lack an age-corrected BLH-CAF two-factor optimization model and have not established a BLH safety threshold based on the highest-risk group across all age groups (infants and young children). To reduce BLH, existing technologies often employ high-energy-consumption, low-color-temperature solutions or sacrifice color rendering index (CRI < 80), leading to risks of deteriorated color rendering index and color temperature deviations from the standard range for older users.

[0006] Based on the problems existing in the prior art, this invention aims to establish a continuous quantitative model of the change of blue light transmittance of the lens with age, so as to achieve precise control of the age-differentiated blue light hazard risk; while maintaining a high color rendering index (CRI≥80) and a standard color temperature range (2700K-6500K), it achieves independent extreme value optimization of CAF in both work and leisure scenarios; and establishes a dynamic BLH safety fallback mechanism based on the highest risk group in all age groups, breaking through the industry bottleneck of the traditional solution that "reducing BLH inevitably sacrifices color rendering". Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a pre-stored formula control method based on continuous age-related eye transmission spectra and scene adaptation.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A pre-stored formula control method based on continuous age-related eye transmission spectrum and scene adaptation includes the following steps: S1. Construct a basic database of continuous age-related ocular transmission spectra: Establish an ocular medium spectral transmittance function covering 100 integer age points from 1 to 100 years old; S2. Establish a three-level mapping model and define the screening conditions for four parallel constraints: The three-level mapping model is as follows: the user's age is the first-level mapping input, the user's preferred color temperature is the second-level mapping input, the user's scene preference is the third-level mapping input, and the mapping output is the target spectral formula, that is, the radiative flux ratio of each channel of the multi-channel LED light source. The scenario preference is a continuous variable within [0%, 100%], which is discretized by a preset step size and then mapped to the CAF target value; The selection criteria for the four parallel constraints are: age-adaptive blue light safety constraint BLH(A) ≤ BLH(A,D65), and BLH(A,D65) <= BLH(1,D65) = 7.18 × 10⁻⁶. -4 W / lm (according to) Figure 4 The calculation results show that the maximum value of BLH(A,D65) was obtained at age 1, with a value of 7.18 × 10⁻⁶. -4 W / lm); Color rendering index (CRI) ≥ 80; Color temperature range (CCT) ∈ [2700K, 6500K]; Multi-channel light source color deviation constraints comply with the CCT tolerance and color deviation requirements specified in ANSI C78.377-2008 standard; S3. Generate and screen candidate spectrum sets: For LED light sources with four or more channels, a genetic optimization algorithm is used to pre-generate candidate spectrum sets offline. The genetic optimization algorithm uses the radiant flux ratio of each channel as chromosome genes, and the sum of the ratios of each channel is 1. The fitness function is composed of a weighted sub-objective of color temperature matching degree and a sub-objective of channel balance. The genetic operators include fitness ratio selection operator, arithmetic mean crossover operator and Gaussian perturbation mutation operator. A fixed generation termination strategy is adopted, the population size is 100, and the maximum number of iterations is 80 generations. For the candidate spectra output by genetic optimization, color temperature tolerance filtering and the four parallel constraint screening conditions are sequentially performed. After screening, the spectra are sorted by CAF value and sampled at equal intervals according to scene preference, and the optical parameters of each candidate spectral formulation are output. S4. Extracting scenario-based spectral formulas: For the candidate spectra that have passed the screening, for each preset integer age point, each preset color temperature and each scene preference value after discretization, extract the optimal spectral formula that meets all constraints according to the screening conditions in step S2. The preset integer age points are all integer age points within the range of 1 to 100 years old; the preset color temperature is the color temperature value divided according to the preset color temperature step size within the value range [2700K, 6500K]. S5. Establish an age-scene spectral pre-stored database: The optimal spectral formula extracted in step S4 is pre-stored as a complete "age-color temperature-scene" three-level mapping pre-stored database according to a unified key-value pair format. S6. Real-time retrieval and output of spectral formulas: From the pre-stored database in step S5, query and retrieve matching spectral formulas using age as the first index, color temperature as the second index, and scene preference as the third index.

[0009] The method for constructing the transmittance function τ(λ,A) in step S1 is as follows: Based on CIE standard observer spectral data and combined with the empirical formula for the change of lens optical density with age, a two-dimensional parameterized model of wavelength 380-780nm and age integers 1-100 years is established, and the transmittance value at each wavelength at each age point is output.

[0010] The formula for calculating the CAF value is: ; Let λ represent the spectral power distribution of the LED light source, C(λ) represent the spectrum of the circadian rhythm stimulation, and V(λ) represent the photopic spectral luminous efficiency function.

[0011] In step S5, each formula record in the three-level mapping pre-stored database is indexed using age as the primary key, color temperature as the secondary key, and scene preference as the tertiary key. The record content corresponding to the key value includes the percentage of radiant flux ratio for each LED channel and the corresponding CRI, CCT, CAF, BLH, and D. uv Various optical parameters.

[0012] In step S6, the matching spectral formula is queried and called. PWM dimming is used as the calling method. The peak current of each channel is fixed at the nominal operating point. The radiant flux ratio is controlled by independently adjusting the PWM duty cycle of each channel. Alternatively, PAM is used as an auxiliary dimming method. The output power of each LED channel is independently controlled by adjusting the PWM duty cycle or PAM amplitude of each channel so that its radiant flux ratio matches the formula setting value. The calculation model for the drive current of each channel is as follows: ; The current is the drive current of the ch-th channel, in amperes (A). It represents the actual current value that needs to be output to the LED of this channel, which is achieved by the driver chip adjusting the PWM duty cycle or PAM amplitude. This represents the maximum radiant power of a single LED light source in this channel, expressed in watts (W). The radiative flux ratio of the ch-th channel in the formulation is dimensionless and expressed as a percentage. denoted as the forward voltage drop of the LED in channel ch, in volts (V).

[0013] Compared with the prior art, the present invention has the following advantages: 1. Breakthrough in safety performance across all age groups: A continuous age transmission spectrum function library for eyes aged 1-100 years has been established to replace the traditional discrete grouping model; a dynamic BLH safety safety bottom line threshold is set based on the BLH received by each age group under D65 light irradiation to ensure photobiological safety across all age groups; the lens aging characteristics are adapted in real time through the τ(λ,A) function to avoid color rendering degradation in elderly users.

[0014] 2. Precise Contextualized Rhythm Adaptation: Automatically outputs appropriate CAF spectra for each age group based on user scenario preferences. For users with work-related preferences, outputs spectra with higher CAF values ​​at each age group to enhance alertness; for users with leisure-related preferences, outputs spectra with lower CAF values ​​to promote physiological relaxation. This achieves precise dual-mode matching for 100 independent age groups, meeting the rhythmic health needs of different scenarios.

[0015] 3. Visual performance guarantee: The color rendering index (CRI) is ≥80 for all age groups, and the color temperature strictly complies with the ANSI C78.377-2008 standard (2700K-6500K), overcoming the defect of traditional low BLH solutions that sacrifice color rendering.

[0016] 4. Multi-architecture compatibility and efficient implementation: Supports various LED channel architectures, including four-channel (RGBW), five-channel (RGBYC), and higher-channel options. For four-channel and above (such as RGBW, RGBYC, five-channel, six-channel, and other multi-primary-color combinations), a genetic optimization algorithm is used to pre-generate candidate spectrum sets offline, and offline indexing of spectral recipes enables real-time retrieval. The driver circuit is compatible with PWM / PAM dual-mode dimming, facilitating industrial applications. Attached Figure Description

[0017] Figure 1 The spectral transmittance curves of ocular media for different age groups were calculated based on the CIE 203:2012 standard. Figure 2 The spectral power distribution diagram of the CIE standard light source D65; Figure 3 The curves show the relative sensitivity functions of visual and non-visual spectra according to CIE standards. Figure 4 This is a graph showing the change of BLH(D65,A) with age. Figure 5 This is the main control flowchart of the overall method of the present invention. Detailed Implementation

[0018] See attached document Figures 1-5 .

[0019] This invention belongs to the field of intelligent health lighting technology. Specifically, it discloses a pre-stored control method for formulas based on continuous age-related eye transmission spectrum and scene adaptation. In this embodiment, the invention systematically solves the problems existing in the prior art through a three-in-one spectral control mechanism of "continuous age modeling—continuous scene target adaptation—safety performance coordination." This is achieved through the following steps: S1. Construct a continuous age-based database of ocular transmission spectra: Establish an ocular media spectral transmittance function τ(λ,A) covering 100 integer age points from 1 to 100 years old. Based on CIE standard human eye spectral data, establish an age-parameterized model to quantify the nonlinear decay characteristics of transmittance in the blue light band (400-500nm) with age.

[0020] The method for constructing the eye transmittance function τ(λ,A) is as follows: based on CIE 1931-2 。 Using standard observer spectral data and an empirical formula for the change of lens optical density with age, a two-dimensional parameterized model of wavelength λ (380-780nm) and age A (1-100 years) is established, and the transmittance value at each wavelength is output for each age point.

[0021] Specifically, the optical density D(λ,A) of the eye medium was calculated using the model van de Kraats & van Norren (2007), JOSA A 24(7):1842-1857: ;in, It represents the baseline optical density shift of all ocular media, with a constant value of 0.06 that does not change with age or wavelength, and serves as the basis for the background absorption term of the total optical density of the ocular media; , represents the Rayleigh scattering light density in the ocular medium contributed by the cornea, aqueous humor, lens and vitreous body. Its value changes with age and wavelength, where A is age. It reflects the additional attenuation of short-wavelength blue light in the ocular medium due to Rayleigh scattering. The scattering intensity increases sharply as the wavelength decreases. , which represents the optical density caused by tryptophan absorption in the ocular media, originates from protein components in all ocular media such as the cornea, aqueous humor, lens, and vitreous humor.

[0022] , which represents the light absorption density of 3-hydroxykynurenine glucoside (3HKG) in the lens.

[0023] , represents the light absorption density of the first type of age-related pigment in the lens, which varies with age and wavelength.

[0024] , which represents the light absorption density of the second type of age-related pigment in the lens, and this value varies with age and wavelength.

[0025] Total optical density of ocular media The sum of the above six items, the spectral transmittance τ(λ,A) of the ocular medium, is calculated from the total optical density: , which represents the proportion of light of different wavelengths transmitted through the media of the human eye at age A.

[0026] Figure 1 The spectral transmittance curves of ocular media for different age groups are calculated based on the CIE 203:2012 standard. The colorimetric data are based on the CIE 1931 standard colorimetric observer (2). 。(Field of view). The horizontal axis of the figure represents wavelength λ, covering the range of 380nm to 780nm (the visible light band); the vertical axis represents transmittance (in % or relative value), indicating the proportion of light of different wavelengths transmitted through the lens and other media of the human eye. The figure exemplarily plots transmittance curves for 11 typical age points: 1 year, 10 years, 20 years, 30 years, 40 years, 50 years, 60 years, 70 years, 80 years, 90 years, and 100 years old. Each curve corresponds to the ocular media transmittance function τ(λ,A) for an integer age point. From... Figure 1 As can be seen, the transmittance curve of the ocular media generally shows a downward trend with age, especially in the blue light band (400-500nm), where the attenuation is most significant. This reflects the physiological characteristics of the lens optical density increasing with age and the nonlinear decay of blue light transmittance. This figure is a visualization of the basic database of continuous age-related ocular transmittance spectra constructed in step S1.

[0027] S2. Establish a three-level mapping model and define the screening conditions for four parallel constraints: The three-level mapping model is as follows: the user's age is the first-level mapping input, the user's preferred color temperature is the second-level mapping input, the user's scene preference is the third-level mapping input, and the mapping output is the target spectral formula, that is, the radiative flux ratio of each channel of the multi-channel LED light source. The screening criteria for the four parallel constraints are: age-adaptive blue light safety constraint BLH(A)≤BLH(A,D65); color rendering index CRI≥80; color temperature range CCT∈[2700K,6500K]; and multi-channel light source color deviation constraint follows the CCT tolerance and color deviation requirements specified in the ANSIC78.377-2008 standard. The scenario preference is a continuous variable within the range [0%, 100%]. After discretization with a preset step size, it is mapped to a CAF target value. This invention sets differentiated optimization targets based on user scenario preferences: Users set a scene preference percentage value CAFPCt via a slider. This is a continuously adjustable value between 0% and 100% that the user inputs from the interactive interface, used to represent their subjective preference for the "workability" or "leisureability" of the current lighting scene.

[0028] The first scenario preference target: extreme working scenario, CAFPCt=100%, the system outputs the spectral formula with the highest CAF value at this age and color temperature, in order to improve the user's alertness and cognitive performance; Second scenario preference target: extreme leisure scenario, CAFPCt=0%, the system should output the spectral formula with the lowest CAF value at this age and color temperature to promote the user's physiological relaxation and sleep preparation; Intermediate scene preference (CAFPCt=5%,10%,...,95%): Achieve smooth transition of CAF through linear mapping.

[0029] Meanwhile, this invention sets four parallel constraint screening conditions to ensure both security and visual performance: ① Safety Constraints: This invention sets a uniform upper limit for blue light hazard safety across all age groups. This upper limit uses a standard light source D65 (color temperature 6500K±200K) as a reference light source and selects age-adaptive ocular media transmission characteristics for calculation to obtain the blue light hazard value BLH(D65,A). Since the transmittance of the lens to blue light varies with age, the BLH received by each age group under D65 light exposure is used as the current age-safe threshold.

[0030] Figure 2 This is a spectral power distribution diagram of the CIE standard light source D65. The horizontal axis represents wavelength (380-780nm), and the vertical axis represents the relative spectral power distribution of the D65 light source, which essentially represents the relative intensity or relative energy of light. The spectral power distribution curve of the D65 light source covers the entire visible light band, and the energy distribution is relatively continuous and uniform. Blue band (approximately 400-500nm): The curve reaches a significant peak near 460nm, indicating that the D65 light source contains a rich blue light component. This is consistent with the high energy of blue light in real sunlight, therefore, D65 is used as a benchmark light source for measuring blue light hazard (BLH).

[0031] Figure 3 The figure shows the relative sensitivity function curves of CIE standard visual and non-visual spectra. It compares four important CIE standard spectral response functions on the same coordinate system. (a), (b), (c), and (d) are comparisons of the photopic vision V(λ), scotopic vision V'(λ), circadian rhythm stimulation spectrum C(λ), and blue light hazard B(λ), respectively. The photopic vision function V(λ) serves as a normalized benchmark for brightness perception; the circadian rhythm stimulation spectrum C(λ) quantifies non-visual biological effects; the blue light hazard weighted function B(λ) evaluates the risk of retinal photochemical damage; and the scotopic vision function V'(λ) provides an extended reference. The peak wavelengths of the four functions, from shortest to longest, are B(λ) at 437 nm, C(λ) at 464 nm, V'(λ) at 507 nm, and V(λ) at 555 nm, corresponding precisely to the relationship from photochemical damage and circadian rhythm regulation to visual perception, forming the "safety-health-vision" three-in-one evaluation framework of this invention.

[0032] Figure 4 This is a graph showing the change of BLH(D65,A) with age. The horizontal axis represents age (years), and the vertical axis represents the blue light hazard value (BLH) (W / lm). At age 1, BLH(D65,1) = 7.18 × 10⁻⁶.-4 This is the highest value across all age groups, and this invention uses this value as the unified safety margin for all age groups. The figure also shows, with a horizontal dashed line, the BLH(3000K,1) = 2.55 × 10⁻⁶ for age 1. -4 As a reference for low color temperature light sources.

[0033] This invention constructs a dual-safety quantification model. First, to establish a unified and scientific upper limit for blue light hazard safety across all age groups, it is necessary to use the group with the highest risk of retinal photochemical damage as the benchmark. Given that the lens of one-year-old infants has the highest blue light transmittance (e.g., ... Figure 1 As shown in the figure, it is most sensitive to the harmful effects of blue light, therefore, this invention selects 1 year old as the anchor age for the safety threshold. Secondly, combining existing theories and... Figure 2 As is known, the natural light environment represented by the standard light source D65 is recognized as the typical lighting conditions that humans have adapted to through long-term evolution, and using it as a benchmark has physiological rationality and universality.

[0034] Therefore, this invention sets the blue light hazard value of a 1-year-old child exposed to a D65 standard light source as a rigid safety upper limit applicable to all age groups. This benchmark value is calculated using formula (1): , (1); The spectral power distribution of the CIE standard light source D65; Let A be the transmissivity function of the lens in a human aged A. A weighting function for the hazards of blue light; The spectral luminous efficiency function for photopic vision; Maximum spectral luminous efficacy (683 lm / W); integration range of 380-780 nm (wavelength) ), covering the highest risk across all age groups; Since each function is discretely distributed, the integral can be written as a summation. ; because Since P, B, and V are all dimensionless functions, the dimension of BLH is W / lm. After calculation, it can be seen that The values ​​for each age, when age A=1, the specific value calculated by this formula is the global security threshold of 7.18×10⁻⁶ established in this invention. -4 .

[0035] (2); Therefore, the age-adaptive blue light safety constraint is: BLH(A) ≤ BLH(A, D65), and BLH(A) < 7.18 × 10⁻⁶. -4 ; ② Color rendering index constraint: CRI≥80, to ensure that the color rendering performance of the light source meets the needs of daily lighting; ③ Color temperature range constraint: CCT∈[2700K,6500K], strictly following the requirements of the American National Standards Institute (ANSI) 78.377-2008 standard for color chromaticity specifications of solid-state lighting products.

[0036] ④ Color deviation constraint: The color deviation of four-channel and above LED light sources shall comply with the ANSI C78.377-2008 standard for color temperature tolerance and color deviation specifications, as shown in Table 1.

[0037] Table 1. ANSI C78.377-2008 Standard Color Temperature Tolerance and Color Deviation (Duv) Tolerance Comparison Table: ; Table 1 provides the fixed tolerance parameters for the eight nominal color temperature points in the ANSI C78.377-2008 standard. However, given that step S4 of this invention involves fine-tuning the spectrum, the traversed color temperature range covers a large number of non-nominal color temperature points (i.e., color temperature values ​​from 2800K to 6400K in 100K steps, excluding the aforementioned eight nominal points). To ensure that these flexible color temperature points also meet the color deviation constraints, this embodiment calculates their corresponding tolerances using the following formula, based on the provisions of the flexible color temperature clause in the ANSI C78.377-2008 standard: ; For flexible color temperatures, the values ​​are integer multiples of 100K (e.g., 2700K, 2800K, ..., 6400K), but do not include the eight values ​​listed above; simultaneously, the CCT tolerance of the flexible color temperature point... That is, as long as the measured color temperature falls within the range of ; Within the range of ±ΔT, it is determined to meet the color temperature deviation constraint of the ANSI standard.

[0038] S3. Generate and filter candidate spectral sets.

[0039] A four-channel (RGBW) architecture, or five-channel (RGBYC) or more channels, is employed. Candidate spectra are sequentially screened using the four-fold parallel constraint screening conditions described in S2, eliminating those that do not meet the conditions. A genetic optimization algorithm is used to pre-generate a candidate spectrum set offline. The genetic optimization algorithm uses the radiant flux ratio of each channel as a chromosome gene, with the sum of the ratios for each channel being 1. The fitness function is composed of a weighted sub-objective of color temperature matching and a sub-objective of channel balance. Genetic operators include a fitness ratio selection operator, an arithmetic mean crossover operator, and a Gaussian perturbation mutation operator. A fixed generation termination strategy is adopted, with a population size of 100 and a maximum iteration generation of 80. For the candidate spectra output by genetic optimization, color temperature tolerance filtering and the screening of the four parallel constraints are performed in sequence. After screening, the spectra are sorted by CAF value and sampled at equal intervals according to scene preference, and the optical parameters of each candidate spectral formulation are output. This embodiment uses a four-channel RGBW light source as an example, and the specific implementation of the genetic optimization algorithm is described in detail below: S301, Fundamental Spectrum Modeling and Chromosome Encoding: Define the intensity weights of the four independent channels R, G, B, and W as follows: , , , As four genes of a chromosome in the genetic algorithm, the base spectrum of each channel is modeled as a normalized Gaussian function (the maximum value of the spectral power distribution is 1). The parameters of the four channels are as follows: The candidate spectrum set is generated according to the following sub-steps: R channel: center wavelength 609nm, full width at half maximum 20nm, denoted as R(609,20); G channel: center wavelength 536nm, full width at half maximum 30nm, denoted as G(536,30); Channel B: Center wavelength 460nm, full width at half maximum (FWHM) 20nm, denoted as B(460, 20); W channel: Bimodal Gaussian model, composed of two Gaussian peaks at 450nm (25nm half width at half maximum) and 560nm (110nm half width at half maximum) with a weighted average of 0.6:0.4, denoted as W(450&560, 25&110); Each chromosome =[ , , , ], Satisfying simplex constraints: + + + =1, ≥0; This constraint ensures that the sum of the radiative flux ratios of each channel is 100%, which physically corresponds to a complete spectral power distribution. The search space of the genetic algorithm is all possible combinations of ratios within this four-dimensional simplex.

[0040] S302, Fitness Function and Genetic Operator Configuration: S3021. Fitness Function Configuration: The fitness function is used to evaluate the quality of each candidate spectrum (i.e., each group of channel weight ratios), serving as the basis for "survival of the fittest" in genetic iteration. The fitness function in this step is composed of two sub-objectives: CCT matching degree and channel balance, as detailed below.

[0041] CCT matching sub-objective: The fitness for each candidate spectrum is configured with two objectives: (3) In formula (3), The CCT matching score is dimensionless; CCT is the correlated color temperature (K) of the current candidate spectrum. Let CCT be the target correlated color temperature, K; 15 is the penalty intensity coefficient, dimensionless; the physical meaning of formula (3) is: when CCT = When, fraction =1, reaching the optimal value; when CCT deviates... As time went on, the score gradually decreased; Channel balance sub-objective: , The channel balance score is calculated when the weights of the four channels are completely equal. =1, the channel utilization is most balanced; when the weight of a certain channel approaches 0... When the value approaches 0, the actual effective number of channels degenerates, and the candidate spectrum is given a lower evaluation. The purpose of this sub-objective is to prevent the genetic algorithm from over-relying on a single channel, which would lead to the loss of the advantages of the multi-channel light source architecture, and to ensure the redundancy and flexibility of the mixing scheme.

[0042] Overall fitness The calculation is as follows: ; S3022, Genetic Operator Configuration: Genetic operators drive the evolution of the population from the current generation to the next generation, including three core operations: selection, crossover, and mutation. Selection operator: Fitness-based selection is used, where the probability of each individual being selected is directly proportional to its fitness; individuals with higher fitness have a greater probability of being selected as parents for reproduction. To avoid probability invalidation due to negative fitness, the fitness values ​​are first shifted to a non-negative interval. (4); In formula (4), where, Let be the probability that the i-th individual is selected as the parent. Let be the initial fitness value of the i-th individual. The minimum fitness value in the current population, ε = 1 × 10 -6 Let N be a constant to prevent overflow, and N=100 be the population size. After this shifting and normalization process, the selection probability of each individual is positive and the sum is 1. Individuals with higher fitness have a greater selection probability. Each time, two parent individuals are independently selected for subsequent crossover operations.

[0043] Elite retention: The eight individuals with the highest fitness in each generation (E=8) are directly copied to the next generation without participating in crossover or mutation. This strategy guarantees the monotonic non-degeneracy of GA, ensuring that the current best solution does not decrease in subsequent iterations and that the best solution is not lost due to random crossover or mutation.

[0044] Crossover operator: Arithmetic mean crossover is used, and the offspring chromosome is the element-wise arithmetic mean of the corresponding genes of the two parent chromosomes. Mutation operator: with probability p m =0.15 Trigger mutation operation for each offspring individual, randomly select a channel RGBW, apply Gaussian perturbation N(0,0,1), if the mutated value is negative, truncate it to 0; renormalize the weights of the mutated four channels so that their sum is 1.

[0045] Population and Termination Criteria: Population size N=100, maximum number of iterations G=80. A fixed-generation termination strategy is adopted, meaning optimization terminates after reaching the maximum number of iterations G=80, without using convergence criteria (such as fitness change rate below a threshold). This strategy ensures that each (age, color temperature) combination achieves the same search depth, avoiding inconsistencies caused by premature convergence. The random seed is fixed at 42 to ensure that the algorithm results are fully reproducible under different operating environments.

[0046] S303, Post-processing Filtering and Candidate Spectrum Output: After the genetic algorithm completes 80 generations of iterations, all candidate spectra are extracted from the final population, and five-fold hard constraint filtering is performed sequentially.

[0047] CCT tolerance filtering: Color temperature tolerance is screened according to the ANSI C78.377-2008 standard. Specifically, there are two cases: Nominal rated color temperature: For the eight nominal color temperature points listed in Table 1 (2700K, 3000K, 3500K, 4000K, 4500K, 5000K, 5700K, 6500K), the CCT tolerance range given in Table 1 of this standard is directly used for judgment.

[0048] Flexible color temperature: For non-rated color temperature points (i.e., other color temperature values ​​in the range of 2700K to 6500K with a step size of 100K, such as 2800K, 2900K, 3100K, etc.), the CCT tolerance formula for flexible color temperature points is adopted. ; Then, perform four constraint checks: filter according to the four constraint conditions in step S2; The candidate spectra after the above five-fold filtering are sorted according to their CAF values, and are sampled at intervals within 21 scene preferences (CAFPct=0%, 5%, 10%, ..., 100%) to output the complete optical parameters of each formulation: S4. Extracting scenario-based spectral formulations: For the candidate spectra that have passed the screening, for each preset integer age point, each preset color temperature and each scene preference value after discretization, extract the optimal spectral formulation that meets all constraints according to the screening conditions in step S2. The preset integer age points are all integer age points within the range of 1 to 100 years old; the preset color temperature is a color temperature value divided according to a preset color temperature step size within the range of 2700K to 6500K. For each age A∈{1,2,...,100}, within the color temperature range CCT∈[2700K,6500K] (step size 100K, 39 groups in total) and scene preference 0-100% (step size 5%, 21 groups in total), the optimal spectrum is extracted according to the screening rules and CAF calculation formula described in S2: (5); In formula (5), Let denot be the spectral power distribution of the LED light source, C(λ) be the spectrum of the circadian rhythm stimulation, and V(λ) be the photopic spectral luminous efficiency function.

[0049] Specifically: For the i-th scenario preference =(i-1)×5%, i=1,2,...,21, and select the optimal spectrum according to the CAF target value for each age-color temperature combination; Extreme working scenarios: Select the spectral formulation with the highest CAF; For extreme leisure scenarios: Select the spectral formulation with the lowest CAF; Intermediate scenario preference: Select the CAF value closest to the corresponding Spectral formulations of linear mapping values. Table 2 below shows examples of output scenario-based formulations.

[0050] Table 2. Example of output scenario-based recipes: ; S5. Establish an age-scene spectrum pre-stored database.

[0051] All 100 age points (A=1,2,…,100), 39 groups of feasible color temperatures (2600-6500K) and 21 CAFs were multiplied sequentially to screen approximately 81,900 independent spectral formulations and pre-store them as a structured database.

[0052] The database uses a key-value pair format for storage, with age as the first-level index, color temperature as the second-level index, and scene preference as the third-level index. Each formula record contains the radiant flux ratio, color temperature, color deviation, CAF, and BLH of each LED channel. It is pre-stored as a complete three-level mapping database of "age-color temperature-scene". The sample data is displayed in JSON structure.

[0053] Taking the formula record of a 30-year-old as an example: The primary key is the age value, which serves as the primary index. The secondary key is the color temperature value (CCT), which serves as a secondary index; The third-level key is a scene preference mapping index (0-100%, the closer to 100%, the more inclined to work, the closer to 0, the more inclined to rest), which contains the radiation flux ratio of each channel (R:45%, G:39%, B:16%). The index format is: age, color temperature preference, scene preference; This data structure contains 100 independent configurations (age A = 1 to 100), each including color temperature and scene preference. This key-value pair structure has the advantages of high query efficiency and strong scalability (new channel fields or new scene types can be flexibly added), making it easy for the system to quickly locate and call the target recipe based on the user's input age and scene.

[0054] S6. Real-time calling and output of spectral formulations.

[0055] Based on the user's input of age A', color temperature, and scene preference, the system performs a query and matching operation from a pre-stored database, retrieving the corresponding set of spectral formulations. Specifically, age A' is used as the primary search key, color temperature as the secondary search key, and scene preference as the tertiary index key. It quickly locates and retrieves the radiant flux ratio data (e.g., RGBW four-channel ratio percentage) for the corresponding scene at that age.

[0056] Subsequently, the drive control module generates the corresponding dimming drive signal according to the called spectral formula. The drive circuit can use a dedicated LED driver chip (such as TLC5973) that supports dual-mode dimming of PWM (pulse width modulation) and PAM (pulse amplitude modulation). By adjusting the PWM duty cycle or PAM amplitude of each channel, the output power of each LED channel can be independently controlled so that its radiant flux ratio accurately matches the formula setting value.

[0057] The calculation model for the drive current of each channel is as follows: (6); The driving current (in amperes, A) for the ch-th channel is the actual current value that needs to be output to the LED of that channel, which is achieved by the driver chip adjusting the PWM duty cycle or PAM amplitude.

[0058] This refers to the maximum radiant power (in watts, W) of a single LED light source in this channel. This parameter is determined by the datasheet of the selected LED device. For example, when selecting an LED chip with a rated power of 1W, =1W. This value is an inherent parameter of the device and does not change with the dimming state.

[0059] This represents the radiative flux ratio (dimensionless, expressed as a percentage) of the ch-th channel in the formulation. This value is derived from the spectral formulations pre-stored in the database of step S5. For example, if the red light ratio is 34% in the operating mode, then... =34%, this ratio represents the relative radiative weight of this channel in the full spectrum, and the sum of the ratios of each channel is 100%.

[0060] This represents the forward voltage drop (in volts, V) of the LED in channel ch. This parameter is determined by the datasheet of the selected LED device, and the forward voltage drop varies for different color channels. For example, the forward voltage drop of a red LED is Vch. R The forward voltage drop V of a green LED is approximately 2.0-2.2V. G The forward voltage drop V of a blue LED is approximately 3.0-3.4V. B It is approximately 3.0-3.4V. This value is an inherent parameter of the device and is approximately constant within the operating current range.

[0061] The following example, using a 70-year-old user selecting a working mode, details the complete process of real-time retrieval and output of the spectral formula in step S6 of this invention. This embodiment, as a specific implementation of step S6, together with the preceding steps S1-S5, constitutes the complete technical solution of this invention.

[0062] Step S601, User Input and Parameter Validation: Users input structured parameters through smart terminals (such as mobile apps, touch panels, or voice control devices). The data received by the system is in the following format: {Age: 70, Color Temperature: 3200K, Scene Preference: 21}.

[0063] The input parameters are defined as follows: age A is an integer from 1 to 100 (100 integer age points); color temperature is an integer with a step size of 100 (2600-6500K); and CAF is an integer from 0 to 21, representing 0%, 5%, 10%, etc. The system validates the input parameters: if the age is not an integer or exceeds the range of 1-100, it prompts the user to re-enter the information; if the current color temperature does not meet the four constraints, exceeds the range, or is not entered with a step size of 100K, it maps to the nearest color temperature, selects a color temperature that exceeds the maximum or minimum, or rounds to the appropriate color temperature.

[0064] Step S602, Pre-stored database query and recipe retrieval: The system uses age A as the primary search key, color temperature as the secondary search key, and scene preference as the tertiary index key to perform a query and match in the pre-stored database established in step S5. The pre-stored database is stored in key-value pair format. After a successful query, the system retrieves the recipe (taking RGBW four-channel as an example). If the user inputs 70 years old, 5000K, and 30% scene preference, then Φ is retrieved. R =40%, Φ G =20%, Φ B =19%, Φ W =20%. Table 3 shows typical spectral formulation parameters for some age points in the pre-stored database of this invention, covering typical nodes for all age groups and dual-scene modes.

[0065] Table 3. Typical spectral formulation parameters for some age points in the pre-stored database. ; It should be noted that Table 3 only lists sample data for some age points. The actual pre-stored database contains all 100 age points (A=1,2,...,100), each corresponding to an independent formula for work mode and leisure mode, totaling tens of thousands of sets. As age increases, the proportion of blue light in work mode is gradually adjusted to compensate for the decrease in blue light transmittance caused by lens aging; leisure mode, on the other hand, generally favors a low blue light ratio to promote relaxation.

[0066] Step S603, Drive signal generation and LED output: The system transmits the requested spectral formula to the drive control module, which then executes the spectral output through a multi-channel LED driver chip (such as the Texas Instruments TLC5973, which supports PWM and PAM dual-mode dimming).

[0067] Current Calculation: The driving current for each channel is calculated based on the formula ratio and LED device specifications. Using Formula 6 as a model, and taking this embodiment with an age of 70 years, a color temperature of 6000K, and a scene preference of 30% as an example, the driving current for the red light channel is: VR is the forward voltage drop of the red LED (typically around 2.0-2.2V), I R This represents the required drive current for the red light channel. Similarly, calculate the drive current I for the green and blue light channels. G I B I W PWM dimming command: The driver chip sets the PWM duty cycle for each channel based on the calculated current value. SetPWM(R_channel, dutycycle=33%); SetPWM(G_channel, dutycycle=20%); SetPWM(B_channel, dutycycle=24%); SetPWM(W_channel, dutycycle=24%); The driver chip uses a high-frequency PWM signal (typically >1kHz to avoid flicker) to control the output power of different LED channels, with the radiant flux ratio of each channel precisely matched to the set formula value. The output optical power of each channel is uniformly superimposed in space by a mixing optical device (such as a diffuser or mixing cavity), ultimately outputting the target spectrum P(λ). , in, , , , These are the normalized spectral power distributions for the red, green, blue, and white channels, respectively.

[0068] Output spectrum verification: The output spectrum was measured and verified to meet all constraints. BLH(70) = 2.8 × 10 -4 W / lm <BLH(D65,70)=3.4×10 -4 W / lm, which meets the safety fallback threshold set in step S2 based on the highest-risk groups of different ages; CRI=83>80, CCT=5200K∈[2700K, 6500K], which conforms to the ANSI C78.377 standard; CAF(70) is consistent with the optimization objective of maximizing CAF in the working mode in step S2.

[0069] Figure 5 This is the main control flowchart of the overall method of the present invention, covering the entire process from offline spectral library generation to online real-time invocation.

[0070] A genetic optimization algorithm was used to optimize the radiant flux of multi-channel (RGBW, RGBYC, etc.) LED light sources: 100×80 generations, with CCT bias + channel balance as fitness, through selection-crossover-mutation iterations. Subsequently, age-adaptive blue light safety constraints BLH(A)≤BLH(A,D65), color rendering index CRI≥80, color temperature range CCT∈[2700K,6500K] were applied, and unqualified spectra were removed. Then, CAF(A) was calculated for each age A∈{1,2,...,100}, and the extreme values ​​of CAF and nearest neighbor formulas corresponding to each scene preference were extracted. Finally, approximately 70,000 formulas (100 age points × 30 + color temperature × ≤20 CAFPCt) were stored in a pre-stored database.

[0071] The subsequent process is divided into three stages from top to bottom: User input layer: The user inputs age A', color temperature CCT, and scene preference CAFPCt parameters on the smart terminal. The system receives and verifies these parameters.

[0072] Data processing layer: The system uses the verified age A' as the first-level index, color temperature CCT as the second-level index, and scene preference CAFPCt as the third-level index to perform fast key-value pair queries in the pre-stored database. It can find and call the corresponding spectral formula (including the radiative flux ratio data of each LED channel) in milliseconds.

[0073] Drive execution layer: This layer generates drive signals and outputs the spectrum, producing corresponding dimming drive signals based on the called spectral formula. PWM (Pulse Width Modulation) is used as the primary dimming method (PAM as an auxiliary method). By adjusting the PWM duty cycle of each channel, the output power of each LED channel is independently controlled, ensuring its radiant flux ratio precisely matches the formula setting. The drive current for each channel is calculated based on the formula ratio and LED device specifications (forward voltage drop, maximum radiant power). The multi-channel output optical power is uniformly superimposed through a mixing optics device, ultimately outputting the target spectrum, which is verified to meet all safety and visual constraints.

[0074] This flowchart corresponds to the entire implementation process of steps S1 to S6, reflecting the complete technical closed loop from offline optimization to online control.

[0075] The following is an example of RGBW implementation: I. Channel Definitions: R (609nm, 20nm FWHM), G (536nm, 30nm), B (460nm, 20nm), W (450+560nm); II. Proportioning Space: w R +w G +w B +w W=1, each channel ≥1%, total C(99,3)=156,849 species. Genetic optimization conditions: the population converges to 3900 candidates after 100×80 generations of optimization, and about 630 species are retained per age after quadruple constraints.

[0076] III. Screening Criteria: BLH(A)≤BLH(A,D65); CRI≥80; CCT and Duv are constrained according to ANSI C78.377-2008 standard. IV. CCT 5000K range (According to ANSI standard, reasonable CCT range: 5000±409K) formulation example (take level 5 out of 21 CAF): Table 4 shows the spectral formulation example.

[0077] Table 4. Examples of some CAF graded formulations at the CCT 5000K level: ; From 12% CAF in leisure mode (CAFPCt=0%) to 28% CAF in work mode (CAFPCt=100%), the blue light ratio increases monotonically. CAF values ​​are positively correlated with blue light content; work mode requires high CAF to enhance alertness. W decreases from 27% in leisure mode to 14% in work mode. The white light channel acts to "dilute" blue light; reducing the blue light proportion lowers CAF and promotes relaxation. The red and green channels are relatively stable, with the R channel fluctuating slightly between 36% and 40%, and the G channel fluctuating between 18% and 21%, primarily regulating between the B and W channels. All formulations show lower BLH than age-matched BLH (A, D65).

[0078] In summary, this invention provides a method for pre-storing spectral formulas and controlling LEDs based on a continuous age eye transmission spectrum model and scene adaptation. Through a three-in-one systematic architecture of "continuous age modeling - scene target decoupling - safety performance synergy", it fundamentally solves the long-standing problem of synergistic optimization among age adaptability, scene-based rhythm regulation and photobiological safety in existing healthy lighting technologies.

[0079] This invention first establishes a continuous function library τ(λ,A) of ocular media spectral transmittance covering 100 integer age points from 1 to 100 years old. This upgrades the traditional discretized age-grouping model into a continuous and analytical physiological quantification model, providing a scientific basis for the gradient-based precise control of blue light hazard risk across all age groups. Based on this, a blue light hazard value BLH(A, D65) ≤ BLH(1, D65) = 7.18 × 10⁻⁶ years is constructed using a standard light source D65 and corrected for each age. -4W / lm serves as a dynamic safety margin for all age groups. Combined with a color rendering index (CRI) ≥ 80, rigid visual constraints from ANSI C78.377-2008 regarding correlated color temperatures (CCT) ∈ [2700K, 6500K], and Duv, this creates a selection criterion that ensures both safety and visual quality. Furthermore, a multi-channel LED radiant flux ratio combination is obtained through an ergonomic / genetic optimization algorithm. Spectral formulations are extracted for each age group, and all 81,900+ independent formulations are pre-stored in a structured key-value pair database. This enables precise matching and millisecond-level real-time retrieval of spectral formulations across three dimensions: age, color temperature, and scene preference.

[0080] The core breakthrough of this invention compared to existing technologies lies in: replacing discrete grouping approximation with a continuous age-based physiological model; replacing fixed threshold settings with dynamic safety fallback thresholds; replacing single-target adjustment with dual-scenario independent extreme value optimization; and replacing the inefficient real-time online iterative calculation scheme with a combination of "pre-stored-recall" offline optimization and online control. Ultimately, a globally optimal balance is achieved among suppressing blue light hazards, adapting to circadian rhythm requirements, and ensuring visual color rendering performance.

[0081] The method of this invention has clear steps and an scalable architecture. It supports multi-channel LED light source configurations (RGB, RGBW, RGBYC) and PWM / PAM dual-mode drive control, and has good industrial compatibility and ease of implementation.

[0082] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A pre-stored formula control method based on continuous age-related eye transmission spectrum and scene adaptation, characterized in that, Includes the following steps: S1. Construct a basic database of continuous age-related ocular transmission spectra: Establish an ocular medium spectral transmittance function covering 100 integer age points from 1 to 100 years old; S2. Establish a three-level mapping model and define the screening conditions for four parallel constraints: The three-level mapping model is as follows: the user's age is the first-level mapping input, the user's preferred color temperature is the second-level mapping input, the user's scene preference is the third-level mapping input, and the mapping output is the target spectral formula, that is, the radiative flux ratio of each channel of the multi-channel LED light source. The scenario preference is a continuous variable within [0%, 100%], which is discretized by a preset step size and then mapped to the CAF target value; The selection criteria for the four parallel constraints are: age-adaptive blue light safety constraint BLH(A) ≤ BLH(A,D65), and BLH(A,D65) <= BLH(1,D65) = 7.18 × 10⁻⁶. -4 W / lm (Based on the calculation results shown in Figure 4, the maximum value of BLH(A,D65) was obtained at age 1, with a value of = 7.18 × 10 -4 W / lm); Color rendering index (CRI) ≥ 80; Color temperature range (CCT) ∈ [2700K, 6500K]; Multi-channel light source color deviation constraints comply with the CCT tolerance and color deviation requirements specified in ANSI C78.377-2008 standard; S3. Generate and screen candidate spectrum sets: For LED light sources with four or more channels, a genetic optimization algorithm is used to pre-generate candidate spectrum sets offline. The genetic optimization algorithm uses the radiant flux ratio of each channel as chromosome genes, and the sum of the ratios of each channel is 1. The fitness function is composed of a weighted sub-objective of color temperature matching degree and a sub-objective of channel balance. The genetic operators include fitness ratio selection operator, arithmetic mean crossover operator and Gaussian perturbation mutation operator. A fixed generation termination strategy is adopted, the population size is 100, and the maximum number of iterations is 80 generations. For the candidate spectra output by genetic optimization, color temperature tolerance filtering and the four parallel constraint screening conditions are sequentially performed. After screening, the spectra are sorted by CAF value and sampled at equal intervals according to scene preference, and the optical parameters of each candidate spectral formulation are output. S4. Extracting scenario-based spectral formulas: For the candidate spectra that have passed the screening, for each preset integer age point, each preset color temperature and each scene preference value after discretization, extract the optimal spectral formula that meets all constraints according to the screening conditions in step S2. The preset integer age points are all integer age points within the range of 1 to 100 years old; the preset color temperature is the color temperature value divided according to the preset color temperature step size within the value range [2700K, 6500K]. S5. Establish an age-scene spectral pre-stored database: The optimal spectral formula extracted in step S4 is pre-stored as a complete "age-color temperature-scene" three-level mapping pre-stored database according to a unified key-value pair format. S6. Real-time retrieval and output of spectral formulas: From the pre-stored database in step S5, query and retrieve matching spectral formulas using age as the first index, color temperature as the second index, and scene preference as the third index.

2. The formula pre-storage control method based on continuous age eye transmission spectrum and scene adaptation according to claim 1, characterized in that, The method for constructing the transmittance function τ(λ,A) in step S1 is as follows: Based on CIE standard observer spectral data and combined with the empirical formula for the change of lens optical density with age, a two-dimensional parameterized model of wavelength 380-780nm and age integers 1-100 years is established, and the transmittance value at each wavelength at each age point is output.

3. The formula pre-storage control method based on continuous age eye transmission spectrum and scene adaptation according to claim 2, characterized in that: The formula for calculating the CAF value is: ； Let λ represent the spectral power distribution of the LED light source, C(λ) represent the spectrum of the circadian rhythm stimulation, and V(λ) represent the photopic spectral luminous efficiency function.

4. The formula pre-storage control method based on continuous age eye transmission spectrum and scene adaptation according to claim 1, characterized in that: In step S5, each formula record in the three-level mapping pre-stored database is indexed using age as the primary key, color temperature as the secondary key, and scene preference as the tertiary key. The record content corresponding to the key value includes the percentage of radiant flux ratio for each LED channel and the corresponding CRI, CCT, CAF, BLH, and D. uv Various optical parameters.

5. The formula pre-storage control method based on continuous age eye transmission spectrum and scene adaptation according to claim 1, characterized in that: In step S6, the matching spectral formula is queried and called. PWM dimming is used as the calling method. The peak current of each channel is fixed at the nominal operating point. The radiant flux ratio is controlled by independently adjusting the PWM duty cycle of each channel. Alternatively, PAM is used as an auxiliary dimming method. The output power of each LED channel is independently controlled by adjusting the PWM duty cycle or PAM amplitude of each channel so that its radiant flux ratio matches the formula setting value. The calculation model for the drive current of each channel is as follows: ; The current is the drive current of the ch-th channel, in amperes (A). It represents the actual current value that needs to be output to the LED of this channel, which is achieved by the driver chip adjusting the PWM duty cycle or PAM amplitude. This represents the maximum radiant power of a single LED light source in this channel, expressed in watts (W). The radiative flux ratio of the ch-th channel in the formulation is dimensionless and expressed as a percentage. denoted as the forward voltage drop of the LED in channel ch, in volts (V).

Images