Method for detecting content of formaldehyde volatile matters in baby toys based on characteristic wavelength

Abstract

The invention discloses a method for detecting the content of formaldehyde volatile matters in infant toys based on characteristic wavelengths, relates to the technical field of spectrum detection, and aims to solve the problem of poor quantitative detection effect of formaldehyde in toys. The method comprises the following steps: acquiring a three-dimensional surface image and material type information, extracting interference characteristics such as reflection, curvature, pattern density and the like, generating an interference evaluation factor and a path weight, constructing a detection path planning graph and selecting a detection sub-region; slightly heating under a preset temperature control curve and synchronizing with spectrum acquisition, forming a temperature and response intensity data pair, extracting dynamic characteristics such as slope, curvature and inflection point, matching with a formaldehyde characteristic response model, outputting a content value and a risk level according to a concentration mapping relation, and generating a detection result label in combination with a spatial position; and meanwhile, reliability measurement and self-adaptive control are constructed, so that complex surface misjudgment is reduced, and the repeatability and consistency of low-concentration detection are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of spectral detection, more particularly, the present application relates to a method for detecting the content of formaldehyde volatile in baby toys based on characteristic wavelength. BACKGROUND

[0002] In the practical application of safety detection of baby consumer products, formaldehyde, as a typical volatile harmful substance, its release behavior is influenced by material type, surface coating structure and environmental conditions. The existing spectral detection technology is mostly based on the spectral response characteristics of flat single material samples, simplifies the formaldehyde release process as a quantitative absorption response in a stable state, ignores the spatial inconsistency of volatile process caused by complex structure surface, multi-material partition and coating interface in baby toys, and most methods are based on single time point sampling or fixed temperature scanning, without considering the dynamic response curve characteristics of volatile process under temperature excitation and the interaction between material, heat and gas multi-physical processes.

[0003] In the prior art, since the detection method does not couple the interference mechanism of toy surface microstructure and the dynamic induction process of formaldehyde release process, there is lack of response trend modeling means between temperature excitation, characteristic wavelength response and concentration estimation, resulting in poor stability of formaldehyde signal recognition in multi-material splicing or complex geometric area, low concentration response is easily shielded, quantitative estimation deviation is large, causing inaccurate volatile content judgment, deviation of detection results from actual use scene, and reducing the reference value of detection results in quality control and risk management. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the following scheme is provided to solve the problem of poor formaldehyde quantitative detection effect of toys in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0006] A method for detecting the content of formaldehyde volatile in baby toys based on characteristic wavelength, comprising the following steps:

[0007] Based on the characteristic absorption band of formaldehyde in the infrared or Raman spectral region, a formaldehyde characteristic response model is established;

[0008] Collecting three-dimensional surface images and material type information of baby toys, identifying material partition and surface interference area in toy structure, constructing detection path planning graph, and selecting detection sub-area;

[0009] Applying micro-warming operation to the detection sub-area, inducing formaldehyde release behavior through controlled temperature rising process, and simultaneously collecting spectral response signals in the characteristic wavelength interval to form a dynamic spectral sequence varying with time;

[0010] Extract the response change of the target absorption peak position in the dynamic spectrum sequence, construct the dynamic trend curve between temperature and response intensity, and match and compare with the formaldehyde characteristic response model;

[0011] According to the comparison result, the content value of formaldehyde volatile matter is determined, and the detection result label containing the concentration value and the region identifier is output combined with the spatial position of the detection sub-region.

[0012] Further, the formaldehyde characteristic response model content includes target absorption peak position, dynamic response curve characteristics, and mapping relationship corresponding to concentration.

[0013] Further, based on the characteristic absorption band of formaldehyde in the infrared or Raman spectrum interval, the formaldehyde characteristic response model is established, and the specific steps include:

[0014] Collecting spectral response data of formaldehyde standard release samples at different concentrations in the infrared or Raman spectrum interval;

[0015] Extracting the curve characteristics of the target absorption peak position and the response intensity changing with the concentration;

[0016] Constructing the mapping relationship between concentration and response intensity, and generating the formaldehyde characteristic response model for comparison and determination combined with the response trend under the condition of temperature change.

[0017] Further, the three-dimensional surface image and material type information of the baby toy are collected, the material partition and surface interference area in the toy structure are identified, the detection path planning graph is constructed, and the detection sub-region is selected, and the specific steps include:

[0018] Collecting the external three-dimensional surface image of the baby toy, and combining the label information or material database to obtain the material type data corresponding to the surface;

[0019] Identifying different material partitions in the toy structure, extracting the optical interference characteristics of each material partition, including the degree of reflection, surface curvature and color pattern density;

[0020] According to the material partition and interference evaluation factor, the detection path planning graph is constructed, the detection surface is divided into detectable area, limited detection area and priority detection area, and the path score weight is given to each area;

[0021] According to the path score weight, the priority detection area is selected as the detection sub-region for performing formaldehyde volatile content detection.

[0022] Further, a trace heating operation is applied to the detection sub-region, the formaldehyde release behavior is induced through the controlled temperature rising process, and the spectral response signal in the characteristic wavelength interval is collected synchronously to form a dynamic spectrum sequence changing with time, including the following steps:

[0023] Set the starting temperature, target temperature, heating rate and maximum residence time to generate a preset temperature control curve; use a miniature directional heating device to implement local heating on the detection sub-region, so that the surface temperature changes along the preset temperature control curve, and record the temperature time sequence through a temperature sensor;

[0024] Align the temperature control controller with the clock of the spectral acquisition device, set the sampling time interval and integration time, and turn on the synchronous acquisition control;

[0025] During the temperature control induction phase, continuously acquire spectral response data around the characteristic wavelength interval at the set time interval, and record the time stamp and detection sub-region identifier of each acquisition;

[0026] Acquire baseline spectra before heating, reference channel or blank area spectra during heating, and perform background subtraction and stray light correction on the spectra acquired during the heating phase;

[0027] Pair the processed spectral data with the corresponding temperature values and spatial positions in chronological order to construct a dynamic spectral sequence of the detection sub-region.

[0028] Further, extract the response changes of the target absorption peak in the dynamic spectral sequence, construct the dynamic trend curve between temperature and response intensity, and match and compare with the formaldehyde characteristic response model, including the following steps:

[0029] Perform baseline correction and noise suppression on the dynamic spectral sequence to obtain a purified spectral sequence;

[0030] Define the target absorption band within the purified spectral sequence, and use peak search and peak shape tracking to locate the target absorption peak center position and half-peak width at each time point;

[0031] Calculate the peak height, peak area, peak center shift and half-peak width at each time point to generate a time and response index table;

[0032] Pair according to the time stamp and temperature time sequence to construct a temperature and response intensity data pair;

[0033] Perform trend fitting on the temperature and response intensity data pair to obtain a dynamic trend curve and output the slope, curvature and inflection point position characteristics;

[0034] Match the dynamic trend curve characteristics with the standard characteristics in the formaldehyde characteristic response model, calculate the similarity score or generate a matching determination label;

[0035] When the similarity score reaches the preset threshold or the matching determination label is established, confirm the existence of the enhanced characteristics of the formaldehyde target absorption peak, and bind the matching result with the detection sub-region identifier.

[0036] Further, a dynamic trend curve is obtained by performing trend fitting on the temperature and response intensity data pairs, and the slope, curvature, and inflection point position characteristics are output, including the following steps:

[0037] Anomaly value elimination and smoothing processing are performed on the temperature and response intensity data pairs to obtain preprocessed data;

[0038] The preprocessed data is sorted by temperature from low to high and divided into monotonic intervals;

[0039] The fitting method is selected according to the error constraint, and the model parameters are determined, and a dynamic trend curve is constructed by using piecewise polynomial fitting or spline fitting;

[0040] The slope characteristics are calculated based on the dynamic trend curve, and the maximum slope, average slope, and slope change amplitude are extracted;

[0041] The curvature characteristics are calculated based on the dynamic trend curve, and the maximum curvature, average curvature, and curvature change interval are extracted;

[0042] The inflection point is marked at the position where the curvature is close to zero and the slope change sign changes, and the corresponding temperature and response intensity are output;

[0043] The feature vector composed of the slope characteristics, curvature characteristics, and inflection point position is bound with the detection sub-region identifier.

[0044] Further, according to the comparison result, the content value of formaldehyde volatile matter is determined, and combined with the spatial position of the detection sub-region, the detection result label containing the concentration value and the region identifier is output, including the following steps:

[0045] Based on the matching comparison result, the target response index for quantitative calculation is determined, including the peak height, peak area, and peak center offset corrected intensity;

[0046] Temperature influence correction and instrument drift correction are performed on the target response index to obtain standardized response values; the concentration mapping relationship in the formaldehyde characteristic response model is called to perform fitting calculation on the standardized response values, and the initial content value is output;

[0047] Consistency test and uncertainty evaluation are performed on the initial content value, and abnormal data are eliminated to generate a revised content value;

[0048] The revised content value is bound with the spatial position parameter of the detection sub-region to form a detection result label containing the concentration value, region identifier, judgment level, and time information;

[0049] The detection result label is written into the detection report and archived for quality traceability and review call.

[0050] The technical effect and advantages of the method for detecting the formaldehyde volatile content in infant toys based on characteristic wavelengths are as follows:

[0051] The present application realizes dynamic perception and differential identification of the formaldehyde volatile behavior on the complex surface of infant toys by constructing a spectrum detection mechanism based on temperature control induction and characteristic wavelength dynamic response; by collecting and standardizing three-dimensional surface images and material type information, extracting three types of optical interference features including reflectivity, surface curvature and color pattern density, generating interference evaluation factors and detection path score weights, constructing a detection path planning graph and selecting a detection sub-region, implementing micro-warming on the detection sub-region under a preset temperature control curve, strictly time-synchronously collecting spectra in the characteristic wavelength range, forming temperature and response intensity data pairs, extracting dynamic trend features such as slope, curvature and inflection point, and matching and comparing with the formaldehyde characteristic response model, obtaining the content value and risk level based on the concentration mapping relationship, and outputting the detection result label combined with the spatial position of the detection sub-region, the present application realizes non-destructive, regional and traceable quantitative detection.

[0052] After the detection is executed, the stability and consistency of the dynamic trend curve are analyzed, the detection reliability is measured in combination with the abnormal value elimination record, nominal temperature, reference intensity, sampling time interval and integral time, and a stable execution or parameter adjustment signal is generated according to the measurement result, so as to establish a detection closed loop and adaptive control based on dynamic response; compared with static single-point measurement and destructive chemical method, the present application significantly reduces the misjudgment probability in the scene of multiple materials and complex curved surfaces, improves the low-concentration detection capability, result repeatability and cross-scene consistency, shortens the detection period, enhances the linkage with quality traceability and rectification decision, and improves the accuracy of formaldehyde volatile content detection based on characteristic wavelengths. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 The flowchart of the method for detecting the formaldehyde volatile content in infant toys based on characteristic wavelengths is shown. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0055] To achieve the above-mentioned purpose, Figure 1 The structural diagram of the method for detecting the formaldehyde volatile content in infant toys based on characteristic wavelengths is shown, which specifically includes the following steps.

[0056] A formaldehyde characteristic response model is established based on the characteristic absorption band of formaldehyde in the infrared or Raman spectral range.

[0057] A three-dimensional surface image and material type information of the infant toy are collected, material partitions and surface interference areas in the toy structure are identified, a detection path planning graph is constructed, and a detection sub-area is selected;

[0058] A trace heating operation is applied to the detection sub-area, the formaldehyde release behavior is induced through a controlled temperature rise process, and the spectral response signals in the characteristic wavelength range are collected synchronously to form a dynamic spectral sequence varying with time;

[0059] The response change of the target absorption peak in the dynamic spectral sequence is extracted, a dynamic trend curve between temperature and response intensity is constructed, and a comparison is made with the formaldehyde characteristic response model;

[0060] The content value of the formaldehyde volatile matter is determined according to the comparison result, and the detection result label containing the concentration value and the area identifier is outputted in combination with the spatial position of the detection sub-area.

[0061] Step 1, a formaldehyde characteristic response model is established based on the characteristic absorption band of formaldehyde in the infrared or Raman spectral range, and the specific implementation is as follows:

[0062] This example is illustrated in the infrared mode, and the Raman mode can be replaced according to the same process;

[0063] First, obtain formaldehyde standard release samples of different concentration levels, use a stable release source and clean carrier gas to generate formaldehyde gas, pass the carrier gas with controlled mass flow through the stable release source to form a continuous gas flow, combine different flow rates to correspond to several concentration levels, and each concentration level is verified by a calibrated reference analysis device before formal collection, only to confirm that it is within the expected level range, and set a unique concentration level identifier for the level.

[0064] Introduce the standard gas into an infrared gas absorption cell with a known optical path length, set uniform temperature change conditions, including initial temperature, heating rate and holding time, align the temperature control module and the time base of the spectrum collection device to realize time synchronization; collect the blank carrier gas background spectrum as the baseline before entering the heating stage, then continuously collect spectral data at fixed time intervals under each concentration level, covering the heating stage and the holding stage, record the time stamp, temperature value, concentration level identifier and collection method identifier each time, if there is light shielding or condensation, pause the collection and replace the absorption cell under the same conditions to ensure data validity.

[0065] The obtained spectral time series of each concentration level is first subjected to background subtraction and baseline correction, then peak searching and peak shape tracking are performed in the known candidate characteristic wavelength range, and peaks that can be detected under most concentration levels, show time enhancement in the warming-up stage, and have reproducible stability in the holding stage are selected as target absorption peaks; for each target absorption peak, the response characteristics varying with concentration are extracted, including the response intensity corresponding to the peak center position, the peak area, the response intensity increment per unit time in the warming-up stage, the time required to reach half the maximum response intensity, and the response intensity stability in the holding stage, wherein the response intensity increment is obtained by comparing the response intensity difference between adjacent collection times and the time difference between adjacent collection times;

[0066] Based on the above extraction results, a mapping relationship between concentration and response intensity is constructed: each concentration level is identified with its corresponding target response indicator value to form an anchor pair, the anchor pairs are sorted from low to high according to the concentration level, and the overall monotonic change of the target response indicator with the concentration level is checked; when there is individual reverse change, retest once under the same conditions and make one-time correction according to the retest results, the correction is only made under the premise of ensuring overall monotonicity and the correction reason is recorded;

[0067] The mapping relationship is used in the form of lookup table and piecewise interpolation: when the target response indicator value detected is between two adjacent anchor pairs, the concentration estimate between the two end concentration values is given according to the relative position of the target response indicator value between the two end anchor indicator values; when it is lower than the lowest anchor or higher than the highest anchor, no extrapolation is performed, and a determination is made that it is lower than the lower limit of measurement or higher than the upper limit of measurement.

[0068] The dynamic behavior under temperature change conditions is incorporated into the formaldehyde characteristic response model, and the time series of each concentration level under the same temperature change conditions are further subjected to feature extraction to form dynamic response curve characteristics, including the response intensity change rate in the warming-up stage, the time required to reach half the maximum response intensity, the inflection point temperature corresponding to the response intensity change, and the response intensity stability in the holding stage. These dynamic response curve characteristics are stored in the dynamic feature sub-library together with the target absorption peak, and the temperature change conditions, optical path length, integration time and reference background version are recorded in the formaldehyde characteristic response model metadata to ensure consistency with the modeling conditions when subsequent detection is called;

[0069] The obtained formaldehyde characteristic response model consists of three parts: first, a peak position information sub-library that saves the target absorption peak position and its repeatable detection conditions; second, a dynamic characteristic sub-library that saves the dynamic response curve characteristics under temperature change conditions and their acquisition method description; and third, a concentration mapping sub-library that saves anchor pairs sorted by concentration level and the calling rules of table lookup and piecewise interpolation; during on-site detection, first match the consistent model version according to the temperature change conditions, then extract the target response index at the target absorption peak position consistent with the model definition, obtain the concentration estimate through table lookup and piecewise interpolation, and verify the consistency with the dynamic response curve characteristics, and only when the dynamic characteristics are consistent with the formaldehyde characteristic response model entries, the concentration estimate is confirmed to be valid, thereby realizing the complete process of quantitative detection based on the target absorption peak, combined with the dynamic response curve characteristics and through the mapping relationship corresponding to the concentration.

[0070] Step 2, collect three-dimensional surface images and material type information of baby toys, identify material partition and surface interference area in toy structure, construct detection path planning graph, and select detection sub-area, the specific implementation is:

[0071] First, perform three-dimensional surface acquisition and material type information collection on baby toys, obtain external three-dimensional surface images using structured light scanning or multi-view imaging, and obtain surface grids as a spatial carrier; combine product label information or material database to check the material type of each surface grid, so that each surface grid has corresponding material type data, forming a three-dimensional surface model with material type data. To ensure data reproducibility, set uniform lighting and fixed camera parameters before scanning, and record lighting method, relative position of camera and toy, resolution and reconstruction error range in model metadata; for areas with occlusions, supplement by changing the shooting angle until the three-dimensional surface is continuous and corresponds to the material type data one by one.

[0072] Subsequently, material partition identification and surface interference feature extraction are performed on the complete three-dimensional surface model. The material partition identification divides the surface into several material partitions according to the material type data, and each partition corresponds to a material type. Three optical interference features are extracted in each material partition, namely, the degree of reflection, surface curvature, and color pattern density. The degree of reflection is determined by evaluating the proportion of highlight areas and the brightness fluctuation of the partition image under uniform lighting. The brightness fluctuation is described by the consistency of the brightness change of adjacent small areas, and is not represented by a formula. It only requires recording whether the brightness change is stable and whether there is local strong reflection. The surface curvature is determined by comparing the spatial orientation difference of adjacent facets in the partition. The curvature is described by whether the angle difference between the normal vectors of adjacent facets is concentrated or has a sudden change. The more concentrated the angle difference, the smaller the curvature. The color pattern density is determined by counting the number of color or texture boundaries per unit area on the texture map of the partition. The more dense the boundaries, the higher the color pattern density. To facilitate comprehensive evaluation, the three interference features are mapped to a unified scale according to field experience specifications to form a set of interference evaluation factors, where a higher value indicates stronger interference. The mapping process records the reference plate image and lighting conditions used in the model metadata, ensuring consistency in different batches of evaluations.

[0073] After obtaining the material partition and interference evaluation factors, a detection path planning graph is constructed and regional classification is completed. The three-dimensional surface model is projected onto a set of inspectable viewing angles. In the reachable surface range, each candidate detection point is evaluated based on the material partition and interference evaluation factors. When any one of the degree of reflection, surface curvature, or color pattern density of a candidate point reaches an unsuitable detection condition, the point is classified into a restricted detection area. When all three interferences are in an easy-to-detect condition and the material partition of the point is representative, the point is classified into a priority detection area. Other candidate points that do not meet the priority conditions and do not trigger the restricted conditions are classified into a detectable area.

[0074] To achieve quantifiable path sequencing, path scoring weights are set for candidate points. On the basis of uniform scoring, plus or minus points are given according to the interference strength of the degree of reflection, surface curvature, and color pattern density. The weaker the interference, the more points are added, and the stronger the interference, the more points are subtracted. To avoid excessive concentration of candidate points in the same material partition, candidate points that are too close to each other in the same material partition are given a weight reduction, and candidate points with more uniform spatial distribution are preferentially retained. Candidate points that have occlusion risks or have too difficult incident angles are given a weight reduction or are removed. Specific methods include "strong reflection directly reduced to restricted, high curvature priority weight reduction, texture density weight reduction, same partition too close weight reduction, and insufficient visibility removal, etc. The final output detection path planning graph is a list of candidate detection points arranged from high to low according to path scoring weights. On the three-dimensional surface, the three types of areas are marked as layers: detectable area, restricted detection area, and priority detection area.

[0075] Finally, representative priority detection areas are selected from the detection path planning graph as detection sub-regions for performing formaldehyde volatile content detection. Three principles are followed in the selection: first, the path score weight of the priority detection area should be in the top; second, the distribution between different material partitions should be balanced to ensure that each material type has a representative detection sub-region; third, sufficient spatial distance should be maintained between adjacent detection sub-regions to avoid mutual influence due to local heat conduction or optical interference. For each selected detection sub-region, the spatial position parameters (position in the three-dimensional surface model coordinates), material partition number, visibility and occlusion state, recommended light incidence angle range and reference focusing distance are recorded, a unique region identifier is generated, and the above information is archived with the detection path planning Figure 1 graph.

[0076] If the field lighting or camera parameters change, the interference evaluation factor is calibrated once by quickly reviewing the reference board image and a small number of candidate point photos; when the calibration result does not change the top-ranked candidate points, the established detection sub-regions are used, otherwise the ranking is updated and the affected detection sub-regions are replaced according to the same rules, ensuring consistency and traceability of different batches of detection.

[0077] Example scenario: the target toy is a baby stroller, the label and material database show that the body is ABS plastic (bright surface, partially covered with colored stickers), the handle is beech wood surface varnish (semi-matte), and the safety strap is cotton fabric (dark color);

[0078] First step (collect three-dimensional surface image and material type information): use structured light scanning to obtain the external three-dimensional surface image and generate a continuous surface mesh; associate the material type data in the product label with the mesh surface by surface piece to form a three-dimensional surface model containing material type data, and replace the angle for supplementary sampling of the chassis edge not covered due to occlusion until each visible area has corresponding material type data. Record the lighting method, camera and toy relative position, resolution and reconstruction error range for easy reproduction;

[0079] Step (identify material partitions and extract optical interference features): divide the surface into three "material partitions" according to the material type data: ABS plastic area, beech wood varnish area, and cotton fabric area. Extract three optical interference features for each partition: reflectivity, under uniform lighting, the high-light area on the ABS body cover and stickers is obvious, judged as high reflectivity; the beech handle is semi-matte, judged as medium reflectivity; the cotton fabric has no mirror high-light, judged as low reflectivity. Surface curvature, the edge of the wheel and the adjacent surface of the body edge change greatly, with high curvature; the wood handle cylinder segment has medium curvature; the central flat area of the fabric strap has low curvature. Color pattern density, the ABS sticker area has dense boundaries and high density; the wood handle texture is uniform and low density; the fabric is dark color fabric with few pattern boundaries and low density;

[0080] The three results are mapped to a unified scale of interference evaluation factor, with higher values indicating stronger interference. The reference plate images and lighting conditions used are written into the metadata to ensure comparability between different batches.

[0081] Step 3: Generate a set of candidate detection points under the reachable viewing angle, and assign a path score weight to each candidate point based on material partition and interference evaluation factor: the lower the degree of reflection, the smaller the curvature, and the lower the color pattern density, the higher the weight. Reduce the weight of candidate points that are too close within the same material partition, and prioritize uniform distribution. Accordingly, the region is divided into priority detection areas (e.g., the flat area of the beech handle middle section, the paper-free matte area on the inside of the car body bottom plate, and the central flat area of the fabric strap), detectable areas (e.g., the curved end of the beech handle and the fabric edge area), and limited detection areas (e.g., the high-brightness reflective area of the ABS cover, the sharp curvature edge of the wheel, and the dense color paper area). Finally, three detection sub-areas are selected from the priority detection area: A1 (flat area of beech handle middle section, material partition = beech varnish, moderate degree of reflection, low curvature, and low color pattern density), A2 (central flat area of fabric strap, material partition = cotton fabric, low degree of reflection, low curvature, and low color pattern density), and A3 (matte area on the inside of the ABS bottom plate without paper, material partition = ABS plastic, moderate to low degree of reflection, low curvature, and low color pattern density). A unique region identifier is generated for each detection sub-area, and the spatial position parameters (position in the three-dimensional surface model coordinates), material partition number, visibility and occlusion state, recommended light incidence angle range, and reference focusing distance are recorded. The region identifiers of the three areas and the detection path planning Figure 1 are archived. If the on-site lighting or camera parameters change, a one-time calibration of the interference evaluation factor is performed by quickly reviewing a few reference plate images and a small number of candidate point photos. If the ranking does not change after calibration, the original detection sub-area is used, and if the change affects the top ranking, the detection sub-area is updated according to the same rules to ensure consistency and traceability between different batches.

[0082] Step 3: Apply a small amount of warming operation to the detection sub-area, induce formaldehyde release behavior through a controlled temperature rise process, and simultaneously collect spectral response signals in the characteristic wavelength range to form a dynamic spectral sequence that changes over time. The specific implementation is as follows:

[0083] A detection sub-region S1 is selected, the ambient temperature is read as the initial temperature, the target temperature, the heating rate and the maximum residence time are set, the preset temperature control curve is generated and written into the temperature control controller. The preset temperature control curve is composed of temperature set points arranged in time sequence, the temperature control controller gradually outputs power according to the set points to approach the target temperature, a miniature directional heating device is used to implement local heating on S1, the heating device is fixed at a position with a stable included angle with the surface normal, so as to ensure that the heating area covers S1 and does not block the light path for spectrum acquisition. A temperature sensor is used to record the temperature time sequence. The temperature sensor can be a miniature contact sensor attached near S1, or a small spot infrared temperature sensor aligned with S1. Regardless of which type is used, the sampling interval of the sensor is kept constant and consistent with the preset temperature control curve, forming a continuous and traceable temperature time sequence.

[0084] Before the temperature rise is performed, the clock alignment of the temperature control controller and the spectrum acquisition device is completed, and the sampling time interval and the integration time are set. The sampling time interval is used to determine the time interval between two consecutive spectrum acquisitions, and the integration time is used to determine the duration of the detector to accumulate photons during each acquisition. Once set, they remain unchanged in this sequence acquisition to avoid temperature and spectrum synchronization caused by changes in sampling rhythm. After the synchronous acquisition control is started, the spectrum acquisition device continuously acquires spectrum response data around the characteristic wavelength interval at the sampling time interval. Each acquisition records the time stamp and the detection sub-region identifier, and also records the temperature value output by the current temperature sensor, so that each spectrum and the corresponding temperature form a one-to-one correspondence. To ensure that the characteristic wavelength interval covers the target absorption peak, the wavelength range and grating settings consistent with the formaldehyde characteristic response model are loaded through the device configuration interface before acquisition, and the light path parameters for this acquisition are locked.

[0085] The baseline spectrum is acquired before the temperature rise, which is used as the reference for subsequent background subtraction. During the temperature rise stage, in addition to continuous acquisition of S1, the reference channel or the blank area spectrum is also acquired synchronously. The reference channel can be an internal reference light path without passing through the sample path, and the blank area can be a neighboring area with the same material but not heated at a certain interval from S1. Both of them are used to measure the ambient light and system drift. For each spectrum obtained during the temperature rise stage, first, background subtraction is performed: the baseline spectrum closest in time to the spectrum is used as the reference, and the points are subtracted to eliminate the influence of static background. Then, stray light correction is performed: the average intensity of the discrete reference band not containing the target absorption peak is calculated, and the average intensity is used as the stray component to subtract from the full spectrum. If the reference channel or the blank area shows a step change in ambient light, an additional baseline acquisition is performed before and after the time point, and the background subtraction and stray light correction are repeated accordingly to ensure consistency in processing.

[0086] After background subtraction and stray light correction, the processed spectral data is paired with corresponding temperature value and spatial position in time sequence to construct the dynamic spectral sequence of detection sub-region S1. The pairing rule is: taking the time stamp as the primary key, the spectrum and temperature value of the same time stamp or the nearest adjacent time stamp are stored together; if the spectrum or temperature record at a certain time is missing, it is marked as missing and continues to be collected at the next time, without interpolation to avoid introducing uncertainty. The dynamic spectral sequence also saves metadata such as sampling time interval, integration time, preset temperature control curve version number, reference channel or blank area identification, heating device power change record, etc., which are used for subsequent response change extraction at the target absorption peak position, construction of dynamic trend curve between temperature and response intensity, and matching comparison with the formaldehyde characteristic response model.

[0087] Step 4: Extract the response change of the target absorption peak in the dynamic spectral sequence, construct the dynamic trend curve between temperature and response intensity, and match and compare with the formaldehyde characteristic response model. The specific implementation is:

[0088] Firstly, baseline correction and noise suppression are performed on the dynamic spectral sequence to obtain a purified spectral sequence; baseline correction uses the baseline spectrum collected at the beginning of the sequence or before heating as a reference, and each processed spectrum is subtracted from the closest baseline spectrum in time to eliminate the static background; if the reference channel or blank area shows a step change in ambient light during the collection process, an additional baseline collection is added before and after the step, and the closest baseline spectrum at that time is used for correction;

[0089] Noise suppression uses a combination of sliding average of multiple adjacent spectra in time and local smoothing within the waveband: first estimate the random fluctuation level in the flat area outside the target absorption waveband, then perform mild smoothing within the full waveband without changing the peak shape to avoid weakening the real peak shape details;

[0090] Then, the target absorption waveband is defined in the purified spectral sequence (the waveband is consistent with the peak position range in the formaldehyde characteristic response model), and peak search and peak shape tracking are performed on the spectrum at each time point: first locate the local maximum response position as the initial value of the peak center position, then find the position reaching half the peak height on both sides of the initial value to determine the half-peak width; if the peak center position of adjacent time points deviates too much but the peak shape has good continuity, then the peak center position of the previous time is used as the preferred reference to ensure the continuity of the peak shape in time.

[0091] The peak height, peak area, peak center offset and half-peak width are calculated for each time point, and these indicators are recorded in the time and response indicator table together with the corresponding timestamp; the peak area is approximated by integrating the spectrum between the boundary points on both sides of the peak center position, and the specific method is to accumulate the rectangular area of adjacent sampling points from low to high in wavelength or Raman shift in this interval and make a compromise by linear extrapolation of the boundary; the peak center offset is obtained by comparing the difference between the peak center position at the current time and the peak center position at the starting time of the sequence.

[0092] Then, the time and response indicator table is paired with the temperature time sequence one by one to build temperature and response intensity data pairs. The response intensity is principle peak height; if the peak height is unstable due to local reflection or microstructure in this material, the peak area is used as an alternative response intensity, and the alternative relationship is clearly marked in the record;

[0093] Time matching takes timestamp as the primary key: when the spectrum collection time and temperature recording time are not completely consistent, the temperature record with the smallest time difference is selected to pair with the spectrum; if there is a missing, it is recorded according to the missing flag and does not use interpolation; after obtaining the temperature and response intensity data pairs, trend fitting is performed to obtain the dynamic trend curve and output the slope, curvature and inflection point position characteristics;

[0094] Before trend fitting, outlier rejection and smoothing are performed: if the response intensity change direction of a point is obviously opposite to its two adjacent points and the change amplitude is much higher than the adjacent change amplitude, the point is marked as an outlier and does not participate in fitting; the remaining data is lightly smoothed to keep the peak value and change inflection point not to be smoothed. The preprocessed data is sorted by temperature from low to high and divided into monotonic intervals: when the response intensity of adjacent points generally rises with temperature, it belongs to the same monotonic increasing interval, and when it appears obvious slowing down or platform, a new interval is started. Select fitting method and determine model parameters according to error constraint: if the residual direction is basically consistent in each interval and the residual amplitude changes slowly with temperature, use segmented polynomial fitting to ensure local approximation in the interval; if the residual direction frequently alternates in the interval, indicating that there are detailed bends, use spline fitting to preserve the natural transition of the curve;

[0095] After the dynamic trend curve is established, the slope characteristics, curvature characteristics and inflection point position characteristics are calculated, the steps are:

[0096] The slope feature is obtained by calculating the local change rate of the response intensity increment to temperature increment between adjacent temperature points, and extracting the maximum slope, average slope and slope change amplitude on the whole curve. The curvature feature is measured by comparing the difference between adjacent local slopes to measure the bending degree of the curve, and the maximum curvature, average curvature and curvature change interval are extracted in the whole curve range. The inflection point position is determined by finding the transition point where the slope changes from increasing to decreasing, or from positive to near zero and then to negative, and the corresponding temperature and response intensity are output.

[0097] Finally, the dynamic trend curve features are matched and compared with the standard features in the formaldehyde feature response model, the similarity score is calculated or the matching judgment label is generated, and when the judgment condition is met, the enhanced features of the target absorption peak are confirmed and bound with the detection sub-region identification;

[0098] The matching comparison is carried out in the following order:

[0099] First, compare the trend direction and stage features, which require a general upward trend in the warming stage and a stage where the upward trend turns into a plateau or a slow upward trend. Second, compare the relative relationship between the slope and the curvature, which requires a larger slope in the low temperature stage, a smaller slope in the high temperature stage, and a significant peak in the curvature near the turning point. Third, compare the inflection point position, which requires the temperature corresponding to the inflection point to be consistent with the temperature range recorded in the model entry. The similarity score is calculated by scoring each feature entry that meets the criteria and adding them up, including consistent trend direction, consistent slope range, curvature peak value, inflection point temperature falling within the range, and consistent peak center shift direction with the model. When the cumulative score reaches the passing threshold set by the model, or when the key entries (such as consistent trend direction and consistent inflection point temperature) are met simultaneously, a matching judgment label is generated as true. Once the matching result is confirmed, it is immediately bound with the detection sub-region identification, and the metadata is written, including the model version used, the target absorption waveband range, the trend fitting method, the abnormal value elimination record and the data timestamp. If the similarity score does not reach the threshold, the sequence is retained and the re-sampling strategy is prompted (such as extending the maximum dwell time or adjusting the integral time appropriately), but no quantitative results are output, ensuring that the data used for quantitative estimation is fully consistent, reproducible and auditable with the formaldehyde feature response model in dynamic behavior.

[0100] Step 5, according to the comparison result, determine the content value of formaldehyde volatiles, and output the detection result label containing the concentration value and the region identification combined with the spatial position of the detection sub-region, the specific implementation is:

[0101] After completing the matching comparison with the formaldehyde feature response model, the target response index for quantitative calculation is first determined, and the specific process is as follows:

[0102] For the same detection sub-region, the peak height is selected as the target response index for the dynamic spectral sequence; when the peak height is significantly affected by surface reflection or microstructure and fluctuates obviously, the peak area is used as the target response index; when the peak center position shows an observable shift during the temperature rise and affects the reading of the peak height or peak area, the intensity after correction of the peak center shift is used as the target response index;

[0103] The intensity after correction of the peak center shift is obtained by the following method: in the purified spectrum at this time point, the target absorption peak recorded by the model is taken as the reference, and two sampling points close to the reference position are selected, and the intensity at the reference position is determined according to the relative intensity of the two points and the distance from the reference position, so as to avoid directly using the original intensity value after displacement;

[0104] Then, temperature influence correction and instrument drift correction are performed on the target response index to obtain the standardized response value. The temperature influence correction is based on the temperature change condition recorded in the model metadata: if there is a difference between the temperature at a certain time point in the field detection and the nominal temperature used to establish the concentration mapping relationship of the formaldehyde characteristic response model, then the dynamic characteristics near the curve at the same time point (such as the response intensity change rate at this temperature segment) is used to move along the temperature axis on the curve to the nominal temperature position, and the expected intensity at this position is used as the intensity after temperature correction;

[0105] When the nominal temperature is located at the junction of two segments, the stability of the two sides is compared respectively, and the side with smoother change is selected as the correction basis.

[0106] The instrument drift correction is performed by collecting reference intensity at the beginning and end of detection through the reference channel or blank area, comparing the difference between the two, and gradually modifying the target response index of the main channel in a uniform time distribution manner: when the expected drift is a slow change, the difference is distributed to each time point according to time; when the reference channel shows a step change at a certain time, the main channel is segmented and corrected using the respective reference intensity before and after the step, respectively. After the two types of correction, the standardized response value is obtained, and the nominal temperature, reference intensity and segmentation time are recorded to form a traceable correction chain.

[0107] The standardized response value is sent into the concentration mapping relationship in the formaldehyde characteristic response model to obtain an initial content value. Specifically, in the anchor point pair table of the formaldehyde characteristic response model, the upper and lower adjacent anchor point pairs closest to the standardized response value are found, and the relative position of the standardized response value between the two target response index values is observed. When closer to the lower anchor point pair, the initial content value is closer to the concentration value at the lower end. When closer to the upper anchor point pair, the initial content value is closer to the concentration value at the upper end. When exactly in the middle of the two ends, the initial content value is taken as the middle position of the concentration values at the two ends. If the standardized response value is lower than the lowest anchor point pair or higher than the highest anchor point pair, no extrapolation is performed, and a determination of being lower than the lower limit of measurement or higher than the upper limit of measurement is given.

[0108] The three initial content values obtained by the three paths of peak height, peak area and peak position center offset corrected intensity can be calculated in parallel at the same moment. Consistency check is performed: when the three values fall within the same concentration interval, the middle result of the three is taken as the initial content value at that moment; when the three values are dispersed, the one with the largest difference from the other two (considered abnormal) is first removed, and the average of the remaining two is taken as the initial content value, and the reason for removing the abnormal value is recorded in the data record;

[0109] After completing the consistency check, a series of initial content values obtained by the detection sub-region in the entire temperature rising and holding stage are subjected to uncertainty evaluation: the distance between the maximum value and the minimum value in the series of results is calculated, and half of the distance is taken as the uncertainty range of the detection sub-region estimated this time, and the typical moment and the corresponding temperature range producing the range are recorded for quick positioning by the reviewer;

[0110] Based on the sequence results after consistency check, the middle position of the sequence or the position close to the stable platform stage is taken as the representative value of the corrected content value. If the sequence abnormally fluctuates in the platform stage, the median result of the most densely distributed small section in the platform stage is selected as the corrected content value, and the time range of the small section is taken as an explanation.

[0111] Finally, the corrected content value is bound with the spatial position parameters of the detection sub-region to form a detection result label containing the concentration value, region identification, judgment level and time information;

[0112] The spatial position parameter is jointly marked by position, material partition number and visibility state in three-dimensional surface model coordinates to ensure subsequent repeat positioning; the determination level is obtained by comparing the modified content value with the preset safety limit value interval: when the modified content value is lower than the lower limit of the safety limit value and the upper limit of the uncertainty is still lower than the lower limit, it is marked as qualified; when the modified content value is close to the safety limit value, or the upper limit of the uncertainty exceeds the safety limit value, it is marked as warning; when the modified content value is higher than the safety limit value and the lower limit of the uncertainty is still higher than the limit value, it is marked as out of limit. The time information includes detection start and end time and representative time for quantification, and is accompanied by model version, nominal temperature, reference intensity, trend fitting method and abnormal value elimination record and other metadata.

[0113] The generated detection result label is written into the detection report and archived, the detection report contains original record index for quality traceability and review call; when there are multiple detection sub-areas for the same toy, each label is listed according to the area mark, and the overall determination and the priority order of the recommended retest or disposal are given in the report abstract, so as to realize the closed-loop output from quantitative results to spatial position to disposal suggestion.

[0114] The present application realizes dynamic perception and differential identification of volatile formaldehyde on complex surface of infant toys by constructing a spectral detection mechanism based on temperature control induction and characteristic wavelength dynamic response; by collecting and standardizing three-dimensional surface images and material type information, extracting three types of optical interference features of reflectivity, surface curvature and color pattern density, generating interference evaluation factor and detection path score weight, constructing detection path planning graph and selecting detection sub-area, implementing micro-warming on the detection sub-area under the preset temperature control curve, strictly time-synchronized collecting spectral in the characteristic wavelength interval, forming temperature and response intensity data pairs, extracting dynamic trend features such as slope, curvature and inflection point, and matching and comparing with formaldehyde characteristic response model, obtaining content value and risk level based on concentration mapping relationship, combining the spatial position of the detection sub-area to output the detection result label, realizing non-destructive, regional and traceable quantitative detection;

[0115] After detection execution, the stability and consistency of the dynamic trend curve are analyzed, the detection reliability is measured in combination with abnormal value elimination record, nominal temperature, reference intensity, sampling time interval and integration time, and the stable execution or parameter adjustment signal is generated according to the measurement result, so as to establish a detection closed loop and adaptive control based on dynamic response; compared with static single-point measurement and destructive chemical method, the present application significantly reduces the misjudgment probability in the scene of multiple materials and complex curved surface, improves the low-concentration detection capability, result repeatability and cross-scene consistency, shortens the detection period and enhances the linkage with quality traceability and rectification decision, and overall improves the accuracy of formaldehyde volatile content detection based on characteristic wavelength.

[0116] The above formulas are all de-dimensioned to calculate the numerical values, the formulas are obtained by collecting a large amount of data to simulate the most recent real situation, and the preset parameters in the formulas are set by a person skilled in the art according to the actual situation.

[0117] The above embodiments can be realized wholly or partially by software, hardware, firmware or any other combination. When realized by software, the above embodiments can be realized wholly or partially in the form of a computer program product.

[0118] Those skilled in the art can realize that the modules and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solutions. A person skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0119] In addition, each functional module in each embodiment of the present application can be integrated in one processing module, or each module can exist physically alone, or two or more modules can be integrated in one module.

[0120] The above is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0121] Finally, the above is only the preferred embodiment of the present application, and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for detecting the formaldehyde volatile content in infant toys based on characteristic wavelengths, characterized in that: Includes the following steps: Based on the characteristic absorption bands of formaldehyde in the infrared or Raman spectral range, a characteristic response model of formaldehyde is established. Collect 3D surface images and material type information of infant and toddler toys, identify material partitions and surface interference areas in the toy structure, construct a detection path planning map, and select detection sub-regions; A micro-heating operation is applied to the detection sub-region to induce formaldehyde release behavior through a controlled temperature rise process, while spectral response signals in the characteristic wavelength range are collected simultaneously to form a dynamic spectral sequence that changes over time. The response changes of the target absorption peak position are extracted from the dynamic spectral sequence, a dynamic trend curve between temperature and response intensity is constructed, and it is matched and compared with the formaldehyde characteristic response model. Based on the comparison results, the content value of formaldehyde volatiles is determined, and combined with the spatial location of the detection sub-region, a detection result label containing the concentration value and the region identifier is output.

2. The method for detecting formaldehyde volatile content in infant toys based on characteristic wavelengths according to claim 1, characterized in that: The formaldehyde characteristic response model includes the target absorption peak position, dynamic response curve characteristics, and the mapping relationship with concentration.

3. The method for detecting formaldehyde volatile content in infant toys based on characteristic wavelengths according to claim 2, characterized in that: Based on the characteristic absorption bands of formaldehyde in the infrared or Raman spectral range, a characteristic response model for formaldehyde is established. The specific steps include: Collect spectral response data of formaldehyde standard release samples at different concentrations in the infrared or Raman spectral range; Extract the target absorption peak position and the curve characteristics of the response intensity as a function of concentration; A mapping relationship between concentration and response intensity is constructed, and a formaldehyde characteristic response model for comparison and judgment is generated by combining the response trend under temperature change conditions.

4. The method for detecting formaldehyde volatile content in infant toys based on characteristic wavelengths according to claim 3, characterized in that: The process involves acquiring 3D surface images and material type information of infant and toddler toys, identifying material partitions and surface interference areas within the toy structure, constructing a detection path planning map, and selecting detection sub-regions. Specific steps include: Collect external three-dimensional surface images of infant and toddler toys, and combine them with label information or material databases to obtain material type data corresponding to the surface; Identify different material zones in the toy structure and extract the optical interference features of each material zone, including reflectivity, surface curvature, and color pattern density. Based on material zoning and interference evaluation factors, a detection path planning map is constructed, dividing the detection surface into detectable areas, restricted detection areas, and priority detection areas, and assigning path scoring weights to each area. Based on the path scoring weights, priority detection areas are selected as detection sub-regions for performing formaldehyde volatile content detection.

5. The method for detecting formaldehyde volatile content in infant toys based on characteristic wavelengths according to claim 4, characterized in that: A micro-heating operation is applied to the detection sub-region to induce formaldehyde release behavior through a controlled temperature rise process. Simultaneously, spectral response signals in characteristic wavelength ranges are acquired to form a dynamic spectral sequence that changes over time. This includes the following steps: Set the initial temperature, target temperature, heating rate, and maximum dwell time to generate a preset temperature control curve; use a micro directional heating device to locally heat the detection sub-region, so that the surface temperature changes along the preset temperature control curve, and record the temperature time series through a temperature sensor; Align the clock of the temperature controller with the clock of the spectral acquisition device, set the sampling time interval and integration time, and enable synchronous acquisition control; During the temperature-controlled induction phase, spectral response data are continuously acquired at set time intervals around the characteristic wavelength range, and the timestamp and detection sub-region identifier of each acquisition are recorded. Baseline spectra are acquired before heating, and spectra of reference channels or blank regions are acquired during heating. Background subtraction and stray light correction are performed on the spectra acquired during the heating stage. The processed spectral data are paired with corresponding temperature values ​​and spatial locations in chronological order to construct a dynamic spectral sequence for the detection sub-region.

6. The method for detecting formaldehyde volatile content in infant toys based on characteristic wavelengths according to claim 5, characterized in that: The response changes of the target absorption peak position are extracted from the dynamic spectral sequence, a dynamic trend curve between temperature and response intensity is constructed, and it is matched and compared with the formaldehyde characteristic response model, including the following steps: Baseline correction and noise suppression are performed on the dynamic spectral sequence to obtain a purified spectral sequence; The target absorption band is defined within the purified spectral sequence, and peak search and peak shape tracking are used to locate the center position and half-peak width of the target absorption peak at each time point. Calculate the peak height, peak area, peak center offset, and half-peak width at each time point, and generate a table of time and response indicators; Pair the timestamp with the temperature time series to construct temperature and response intensity data pairs; The temperature and response intensity data are fitted to a trend to obtain a dynamic trend curve and output the slope, curvature and inflection point location features. The dynamic trend curve features are matched and compared with the standard features in the formaldehyde feature response model to calculate the similarity score or generate a matching judgment label. When the similarity score reaches the preset threshold or the matching judgment label is established, the enhanced feature of the formaldehyde target absorption peak is confirmed, and the matching result is bound to the detection sub-region identifier.

7. The method for detecting formaldehyde volatile content in infant toys based on characteristic wavelengths according to claim 6, characterized in that: The process of fitting temperature and response intensity data to the execution trend yields a dynamic trend curve and outputs slope, curvature, and inflection point location features, including the following steps: Outlier removal and smoothing were performed on the temperature and response intensity data to obtain preprocessed data. The preprocessed data were sorted from low to high temperature and then divided into monotonic intervals. Based on the error constraints, select the fitting method and determine the model parameters, and construct the dynamic trend curve by piecewise polynomial fitting or spline fitting. Calculate slope characteristics based on dynamic trend curves and extract maximum slope, average slope and slope change range; Curvature features are calculated based on dynamic trend curves, and maximum curvature, average curvature, and curvature variation range are extracted. The inflection point is marked at the position where the curvature is close to zero and the sign of the slope change is altered, and the corresponding temperature and response intensity are output. The feature vector composed of slope features, curvature features, and inflection point positions is bound to the detection sub-region identifier.

8. The method for detecting formaldehyde volatile content in infant toys based on characteristic wavelengths according to claim 6, characterized in that: Based on the comparison results, the formaldehyde volatile content is determined, and combined with the spatial location of the detection sub-region, a detection result label containing the concentration value and region identifier is output, including the following steps: Based on the matching comparison results, the target response index for quantitative calculation is determined. The target response index includes peak height, peak area and intensity after peak position center offset correction. Temperature effect correction and instrument drift correction are applied to the target response index to obtain the standardized response value; The concentration mapping relationship in the formaldehyde characteristic response model is used to fit and calculate the standardized response value, and the initial content value is output. Perform consistency checks and uncertainty assessments on the initial content values, remove outliers, and generate corrected content values. The corrected content value is bound to the spatial location parameters of the detection sub-region to form a detection result label that includes concentration value, region identifier, judgment level and time information; The test result labels are written into the test report and archived for quality traceability and review.

Images