Method for detecting quality of low odor wire harness material

Abstract

This invention relates to a quality testing method for low-odor wire harness materials, belonging to the field of quality testing technology. The method includes: placing a wire harness sample in an intelligent sealed chamber; dynamically adjusting temperature, humidity, and ventilation rate using the Internet of Things (IoT) to simulate the entire lifecycle of VOCs slow-release processes, generating a test gas that closely resembles real-world operating conditions; capturing ppb-level trace VOCs molecular signals using a chip equipped with specific biological olfactory receptor proteins, forming a high signal-to-noise ratio dataset; performing toxicity verification by matching a sensitive bacterial strain library with an AI-based microbial sensor array, outputting component-toxicity correlation assessment results; and outputting quantitative detection results after multi-channel filtering calibration and operating condition deviation correction. Based on the results, an N-BEATS model is constructed to achieve risk warning, matching remediation solutions, and forming a closed loop of detection-warning-control-verification. This invention clearly defines the concentration and toxicity contribution of trace components, captures synergistic / antagonistic toxicity effects between components, and significantly improves detection accuracy and data reliability.

Description

Technical Field

[0001] This invention belongs to the field of quality testing technology, specifically a quality testing method for low-odor wire harness materials. Background Technology

[0002] Traditional detection methods often rely on manual olfaction or simple instruments, resulting in low accuracy. They can only qualitatively determine the presence or absence of odor, failing to identify trace VOC components at the ppb level, and making it even more difficult to quantify the correlation between concentration and toxicity. They also lack adaptability to various operating conditions, failing to dynamically simulate the entire lifecycle of the product, leading to a disconnect between the tested object and actual service conditions, thus limiting the reference value of the results. Furthermore, they lack closed-loop management capabilities, remaining solely at the detection level. They lack risk warning and intelligent control mechanisms, making it impossible to trace the source of toxicity, assess the synergistic hazards of components, and are prone to human intervention and large errors, failing to meet the stringent quality control requirements of high-end fields for low-odor materials. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, this invention proposes a quality inspection method for low-odor wire harness materials. This invention primarily addresses the problems of traditional inspection methods, which involve significant manual intervention and large errors, making them unsuitable for meeting the stringent quality control requirements of high-end applications for low-odor materials.

[0004] The present invention provides a quality testing method for low-odor wire harness materials, comprising: S1: extracting wire harness material samples into an intelligent sealed chamber, adjusting temperature, humidity and ventilation rate in real time through Internet of Things sensing, simulating the slow release of VOCs throughout the entire life cycle, and generating dynamic test gas that is highly consistent with real working conditions.

[0005] S2: Based on the dynamic capture of VOCs molecular characteristic signals of the gas to be tested, a chip equipped with specific biological olfactory receptor proteins is used to identify trace components at the ppb level. The components are distinguished and their concentrations are correlated through electrochemical signal conversion, forming a high signal-to-noise ratio dataset.

[0006] S3: Based on the high signal-to-noise ratio, AI is used to match the corresponding sensitive strain library, a microbial sensor array is used to perform toxicity verification, the rate of change of fluorescence signal is monitored, the toxicity level is quantified by comparing with the calibration curve, and the component toxicity correlation assessment results are output.

[0007] S4: Based on the component toxicity correlation assessment results, multi-channel signal filtering calibration technology is used to purify the information, and deviations are corrected by combining operating condition parameters. The results are then output as quantitative test results through standard sample comparison calibration.

[0008] S5: Construct an N-BEATS model based on the quantitative detection results, train the model to identify the critical threshold and mutation pattern of VOCs toxicity exceeding the standard, embed an IoT real-time monitoring interface, perform automatic risk level early warning, and obtain early warning model signals.

[0009] S6: Based on the early warning model signal, the system calls upon the preset treatment strategy library to match the control scheme. For highly toxic trace components, the system activates the targeted adsorption module, simultaneously adjusting temperature, humidity, and ventilation rate to suppress the slow release of VOCs, forming a closed loop of detection-early warning-control-verification, and outputting a treatment acceptance report.

[0010] According to the quality inspection method for low-odor wire harness materials provided by the present invention, the specific steps in step S1 for generating a dynamic test gas that highly fits the actual working conditions are as follows: S11: Select a representative wire harness material sample, remove surface contaminants, and cut it into uniform small pieces to ensure that the sample volume is compatible with the sealed chamber, thus obtaining a standardized test sample.

[0011] S12: Standardized test samples are placed in an intelligent sealed chamber. The sealed chamber is connected to an environmental parameter monitoring terminal via an IoT sensor module to collect initial data on temperature, humidity, and ventilation rate in real time, forming an initial parameter set for the operating conditions.

[0012] S13: Based on the initial parameter set of the working condition, and according to the full life cycle working condition curve of the actual service of the wire harness, dynamically adjust the temperature and humidity cycle and ventilation rate to promote the natural slow release of VOCs and odor substances from the sample, and generate mixed release gas.

[0013] S14: Collect the mixed released gas, filter to remove particulate impurities, and generate a dynamic test gas that closely matches the actual working conditions.

[0014] According to the quality inspection method for low-odor wire harness materials provided by the present invention, the specific steps for forming a high signal-to-noise ratio dataset in step S2 are as follows: S21: The dynamic gas to be tested is introduced into the sensor detection chamber, and the constant temperature and humidity environment inside the chamber is controlled to obtain the gas flow status.

[0015] S22: Based on the gas flow state, use a specific bio-olfactory receptor protein chip to capture changes in electrochemical signals induced by VOCs molecule binding, and identify electrochemical signal change data of trace target components at the ppb level.

[0016] S23: Input the electrochemical signal change data into the signal-concentration formula to calculate the different VOCs component types and associate them with the corresponding concentration values.

[0017] S24: Utilize the correlation data between component types and concentrations, employ signal denoising algorithms to further optimize data quality, filter feature information, and form a high signal-to-noise ratio set.

[0018] According to the quality detection method for low-odor wire harness materials provided by the present invention, the specific steps for identifying the electrochemical signal change data of trace target components at the ppb level in step S22 are as follows: Based on a stable gas flow state, a chip activation program is used to activate the receptor proteins on the electrode surface from a dormant state to an active state, exposing all binding sites and outputting an activated chip.

[0019] By activating the active chip, the dynamic gas to be tested flows at a constant speed and continuously contacts the chip surface. VOCs molecules bind specifically to their corresponding receptor proteins, and the original electrochemical signal corresponding to the VOCs component is output.

[0020] The signal fluctuations of the raw electrochemical signal, the peak voltage and response current characterizing the sensor's response sensitivity are collected, and interference is subtracted after baseline calibration to output an effective signal segment containing trace component characteristics at the ppb level.

[0021] The effective signal fragments are compared with standard characteristic signals to identify trace target components at the ppb level and output electrochemical signal change data.

[0022] According to the quality testing method for low-odor wire harness materials provided by the present invention, the specific steps for outputting the component toxicity correlation assessment results in step S3 are as follows: S31: Import the high signal-to-noise ratio set into the trained intelligent matching model. The intelligent matching model selects exclusive strain combinations that are highly sensitive to low concentrations of odor components based on signal purity and strain response specificity, and outputs a strain library list.

[0023] S32: Transfer the strain library list to the microbial sensor array chip for array detection, configure a constant temperature and humidity detection environment, and form a toxicity sensing detection system.

[0024] S33: The volatile gas of the low-odor wire harness material to be tested is introduced into the toxicity sensing detection system. The changes in fluorescence signals produced by each strain in the toxicity sensing detection system after being stimulated by toxicity are monitored. The signal intensity data at different time points are recorded, and the dynamic curve of fluorescence signal change rate is output.

[0025] S34: Compare the dynamic curve of fluorescence signal change rate with the preset toxicity-fluorescence response calibration curve, calculate the toxicity value matched by the signal change rate of each volatile component, and output the preliminary toxicity level judgment result.

[0026] S35: Based on the preliminary toxicity level determination results and the response specificity of different strains in the strain library to specific components, construct a component-toxicity correlation matrix, analyze the toxicity contribution and synergistic effect of each odor component, and output a component toxicity correlation assessment report.

[0027] According to the quality testing method for low-odor wire harness materials provided by the present invention, the specific steps for constructing the component-toxicity correlation matrix in step S35 are as follows: Based on the preliminary toxicity level determination results and the response specific parameters of the strain library, the correspondence between components, target strains, and toxicity values ​​was sorted out, and a three-dimensional raw data comparison table was output.

[0028] Based on the three-dimensional raw data comparison table, the odor components are set as row vectors, and the toxicity level and strain response characteristics are set as column vectors. A matrix framework is built and the data is filled in. The response coefficients are labeled, and the initial draft of the component-toxicity correlation matrix is ​​output.

[0029] Based on the initial draft of the component-toxicity correlation matrix, principal component analysis was used to calculate the toxicity contribution weight of each component, identify synergies between components, and output the correlation matrix.

[0030] According to the quality testing method for low-odor wire harness materials provided by the present invention, the specific steps for outputting quantitative test results in step S4 are as follows: S41: Employs multi-channel signal filtering and calibration technology to separate and purify multi-dimensional detection information of concentration and toxicity in the component toxicity correlation assessment results, and outputs a multi-dimensional detection information set.

[0031] S42: Based on the multi-dimensional detection information set and the temperature, humidity, and pressure parameters recorded in the sealed chamber, the data is calibrated through the deviation correction model, and the corrected detection dataset is output.

[0032] S43: Compare the gradient concentration standard samples with the calibration and correction test dataset to establish a precise mapping relationship between the signal and the true concentration, realize the upgrade from semi-quantitative data to precise quantification, and output calibrated quantitative data.

[0033] S44: Optimize the component-concentration-toxicity correspondence graph based on calibration and quantification data, label the influence weight of operating conditions, and output the quantitative detection results.

[0034] According to the quality inspection method for low-odor wire harness material provided by the present invention, the specific steps for obtaining the early warning model signal in step S5 are as follows: S51: Extract VOCs component concentrations, toxicity values, and corresponding time series data based on the quantitative detection results, perform data cleaning and normalization, and construct a multi-dimensional time series dataset containing operating condition parameters.

[0035] S52: Using multi-dimensional time-series datasets as training samples, build an N-BEATS model architecture, set the number of iterations and learning rate parameters, train the model to discover the critical threshold and mutation patterns of VOCs toxicity exceeding the standard, and output a converged basic model.

[0036] S53: Develop an embedded adapter module for the IoT real-time monitoring interface based on the convergence basic model, complete the protocol docking between the model and sensors and data transmission devices, and output the embedded early warning model.

[0037] S54: Based on the embedded early warning model, set the VOCs toxicity risk level judgment rules, simulate real-time monitoring scenarios to conduct stress tests, optimize the early warning response delay, and output early warning model signals.

[0038] According to the quality inspection method for low-odor wire harness materials provided by the present invention, the specific steps for outputting the convergent basic model in step S52 are as follows: The core architecture of the N-BEATS model, consisting of stacked fully connected layer modules, is constructed. Trend and fluctuation decomposition branches are divided to adapt to the characteristics of VOCs toxicity data and output the initial N-BEATS model.

[0039] Based on the initial N-BEATS model, the number of iterations and learning rate are set, the input and output dimensions of the multi-dimensional time series dataset are bound, and the model to be trained is output.

[0040] Multi-dimensional time-series datasets are input into the model to be trained for iterative training. Weights are optimized through backpropagation to explore the critical threshold and mutation patterns of VOCs toxicity exceeding the standard, and a set of model performance indicators is output.

[0041] Verify whether the loss value of the model performance index set reaches the convergence threshold and whether the pattern recognition accuracy meets the standard. If yes, output the converged basic model.

[0042] According to the quality testing method for low-odor wire harness materials provided by the present invention, the specific steps for outputting the treatment acceptance report in step S6 are as follows: S61: Extract VOCs toxicity level, exceeding components and concentration information from the early warning model signal, match the corresponding control scheme according to risk level-component type, and output the VOCs treatment implementation scheme.

[0043] S62: Based on the VOCs treatment implementation plan, targeted adsorption is initiated for highly toxic trace components, the temperature and humidity of the sealed chamber and the ventilation rate are adjusted to inhibit the continuous slow release of VOCs, and real-time control condition data is output.

[0044] S63: Restart VOCs quantitative detection based on real-time control condition data and linkage detection system to verify the removal effect of control measures on toxic components and output VOCs quantitative detection results.

[0045] S64: Compare the quantitative test results before and after regulation to evaluate the effectiveness of the governance plan. If the standard is not met, feedback is sent to the strategy library to optimize the plan and repeat the regulation. If the standard is met, a complete closed loop is formed and a governance acceptance report is output.

[0046] This invention provides a quality testing method for low-odor wire harness materials. The beneficial effects of this invention are as follows: 1. This invention overcomes the limitations of traditional wire harness material odor detection, which is characterized by semi-quantitative methods and susceptibility to interference, by combining biosensing and multi-dimensional calibration technologies. Utilizing a chip equipped with specific biological olfactory receptor proteins and electrochemical signal conversion, it can accurately identify trace VOC components at the ppb level. Combined with wavelet threshold denoising and principal component analysis, the data signal-to-noise ratio is improved to over 30dB, effectively eliminating interference signals. Through microbial sensor arrays and fluorescence signal change rate analysis, the toxicity level is quantified by comparing with calibration curves. Further multi-channel filtering, operating condition deviation correction, and standard sample calibration achieve an upgrade from semi-quantitative to precise quantification, establishing a precise correspondence between component, concentration, and toxicity. Compared to traditional detection methods that only determine the presence or absence of odor, this method can clearly define the concentration, toxicity level, and contribution of each trace component, and even capture synergistic / antagonistic toxicity effects between components, significantly improving detection accuracy and data reliability.

[0047] 2. This invention provides end-to-end data collection and analysis, from operational simulation and component capture to quantitative detection. It utilizes the N-BEATS model to uncover patterns of toxicity exceedances, embeds an IoT interface for real-time early warning, and proactively mitigates risks. Based on warning signals, it automatically matches treatment plans, initiating targeted adsorption and operational condition regulation. The treatment effect is then verified through a detection system, forming a closed-loop management system. From sample standardization and automatic signal acquisition to model-based early warning and intelligent control, the entire process relies on AI and IoT technologies for automated operation. This avoids human error, enables rapid response to toxicity risks, and significantly improves detection and treatment efficiency.

[0048] 3. This invention is based on the full lifecycle operating conditions of wire harness materials, solving the problem of the disconnect between traditional testing and actual service scenarios. Through a smart sealed chamber, it dynamically simulates real-world operating conditions such as high and low temperature cycles and humidity, generating a highly accurate dynamic test gas. This ensures that the tested object closely approximates the actual slow-release state. The influence weights of operating conditions are clearly labeled, and the optimized spectrum can adapt to the data correction needs of different service environments. Control measures simultaneously reference actual operating condition parameters, making the treatment solution more practical. This condition-oriented design allows the test results to not only accurately reflect the quality of the material itself but also predict its odor and toxicity risks in actual use. It provides comprehensive data support for the R&D optimization, production quality control, and service safety of wire harness materials, meeting the stringent requirements for low-odor materials in high-end fields such as automotive and electronics. Attached Figure Description

[0049] The invention will now be further described with reference to the accompanying drawings.

[0050] Figure 1 This is a flowchart illustrating the steps of a quality testing method for low-odor wire harness materials provided in an embodiment of the present invention.

[0051] Figure 2This is a flowchart of a quality testing method for low-odor wire harness materials provided in an embodiment of the present invention.

[0052] Figure 3 This is a flowchart illustrating the steps involved in forming a high signal-to-noise ratio dataset, as provided in an embodiment of the present invention. Detailed Implementation

[0053] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below according to specific embodiments.

[0054] like Figures 1 to 3 As shown in the figure, an embodiment of the present invention provides a quality testing method for low-odor wire harness materials, the method comprising: S1: First, extract wire harness material samples and place them in an intelligent sealed chamber. Then, collect data in real time through IoT sensors and dynamically adjust high and low temperature cycles, humidity, and ventilation rate to simulate the slow release process of VOCs and odor substances throughout the entire life cycle, generating dynamic test gas that closely matches the actual working conditions.

[0055] S11: Select a representative wire harness material sample, remove surface contaminants, and cut it into uniform small pieces to ensure that the sample volume is compatible with the sealed chamber, thus obtaining a standardized test sample.

[0056] S12: Standardized test samples are placed in an intelligent sealed chamber. The sealed chamber is connected to an environmental parameter monitoring terminal via an IoT sensor module to collect initial data on temperature, humidity, and ventilation rate in real time, forming an initial parameter set for the operating conditions.

[0057] S13: Based on the initial parameter set of the working condition, and according to the full life cycle working condition curve of the actual service of the wire harness, dynamically adjust the temperature and humidity cycle and ventilation rate to promote the natural slow release of VOCs and odor substances from the sample, and generate mixed release gas.

[0058] S14: Collect the mixed released gas, filter to remove particulate impurities, and generate a dynamic test gas that closely matches the actual working conditions.

[0059] S2: Based on the dynamic capture of VOCs molecular characteristic signals of the gas to be tested, the chip is equipped with specially modified biological olfactory receptor proteins to identify trace components at the ppb level. Through intelligent conversion of electrochemical signals and molecular binding efficiency, the types of odor components are distinguished and their concentrations are correlated to form a high signal-to-noise ratio component feature and concentration dataset.

[0060] S21: The dynamic gas to be tested is introduced into the sensor detection chamber, and the constant temperature and humidity environment inside the chamber is controlled to ensure that the gas flows through the chip surface at a uniform speed.

[0061] The dynamic gas to be tested is introduced into the sensor detection chamber through an inert gas guide tube. The precision flow controller at the front end of the tube is turned on, and the inlet pressure is adjusted to 0.1-0.3MPa to stabilize the gas flow rate at 5-10mL / min. At the same time, the flow rate data is fed back in real time through the flow monitoring module at the rear end of the chamber. The deviation is controlled within ±0.2mL / min to ensure that the gas flows through the chip surface at a uniform speed.

[0062] The embedded temperature and humidity control unit inside the chamber is activated, and the temperature is stabilized at 25±0.5℃ using a PID temperature control algorithm. The relative humidity is controlled at 50±2%RH through the coordinated action of the steam generator and dehumidification module. At the same time, the gas mixing device inside the chamber is activated to ensure uniform temperature, humidity, and gas concentration distribution. Deviations are continuously monitored and dynamically corrected to maintain a stable reaction environment.

[0063] S22: Based on the gas flow state under stable reaction conditions, the chip equipped with specific biological olfactory receptor proteins is activated to capture the electrochemical signal changes caused by VOCs molecule binding, and accurately identify trace target components at the ppb level.

[0064] Under stable gas flow conditions, a constant operating voltage is applied to the chip carrying specific bio-olfactory receptor proteins to activate the chip's electrochemical signal acquisition unit. The chip is then preheated for 30 minutes to allow the receptor proteins to reach optimal activity. Simultaneously, standard VOCs gas of known concentration is introduced to calibrate the chip's signal response sensitivity, ensuring that the signal acquisition accuracy meets ppb-level detection requirements.

[0065] After calibration, VOCs molecules in the gas to be tested, which flows at a constant speed, bind specifically to the specific bio-olfactory receptor protein on the chip surface. The binding process causes a change in protein conformation, which in turn leads to a change in the electron transfer efficiency on the chip electrode surface, generating characteristic electrochemical signal fluctuations. These fluctuation signals are captured in real time by a high-sensitivity signal acquisition probe (sampling frequency ≥ 1 kHz) built into the chip.

[0066] Based on a pre-defined feature signal library, the waveforms, amplitudes, and other characteristic parameters of the captured electrochemical signals are compared to eliminate interference signals and preliminarily identify trace VOCs components at the ppb level, providing targeted signal data for subsequent concentration calculations.

[0067] S23: Based on the electrochemical signal change data captured by the chip, the system performs calculations using a preset signal-concentration conversion formula to distinguish different VOCs components and associate them with their corresponding concentration values.

[0068] The captured raw electrochemical signal was initially filtered to extract characteristic signal parameters for each suspected component, including the peak voltage Uᵢ of the signal response, the stable current value Iᵢ, and the signal duration tᵢ. At the same time, the volume V of the gas participating in the reaction in the sensor detection chamber was recorded.

[0069] When a single molecule of a target VOC binds to a sensitive material, it will release a fixed amount of VOCs. The characteristic chemical energy. Within a fixed volume V of the sensor detection chamber, the total number N of target VOCs molecules is directly proportional to their molar concentration C, satisfying... ,in, Let be Avogadro's constant. Based on the above relationship, the total chemical energy released by the reaction of all target VOC molecules in the detection chamber can be expressed as: In the formula, V is the total chemical energy, C is the volume of gas participating in the reaction in the sensor detection chamber, and N is the molar concentration of the target VOCs component. A is Avogadro's constant, and e0 is the characteristic chemical energy released by the binding of a single molecule.

[0070] The electrical energy of the electrochemical signal is determined by the signal current I, the response voltage U, and the signal duration t acquired by the sensor, as expressed by the formula: In the formula, U is the total electrical energy captured by the electrochemical sensor, U is the electrochemical signal response voltage, I is the electrochemical signal response current, and t is the signal stabilization duration.

[0071] The total chemical energy released when VOCs molecules specifically bind to receptor proteins Equal to the total electrical energy captured by the electrochemical sensor .

[0072] The formula for calculating VOCs concentration and electrochemical signal parameters is expressed as follows: In the formula, U is the sensor operating voltage, I is the response current, t is the signal acquisition time, and V is the volume of gas participating in the reaction inside the detection chamber. denoted as Avogadro's constant, e as the elementary charge, and C as the molar concentration of the target VOCs component.

[0073] Based on the differences in characteristic signal parameters corresponding to different components and the calculation results of the formula, the different VOCs components in the gas to be tested are clearly distinguished, a one-to-one correspondence between component type, characteristic signal and concentration value is established, and a preliminary component-concentration correlation data table is generated.

[0074] S24: Utilize the correlation data between component types and concentrations, employ signal denoising algorithms to further optimize data quality, filter out effective feature information, and form a high signal-to-noise ratio component feature and concentration dataset.

[0075] A wavelet threshold denoising algorithm was used to optimize the initial correlation data. By setting a reasonable threshold, invalid signals caused by environmental interference and equipment noise were filtered out, while effective signals that are strongly correlated with the types and concentrations of VOCs were retained, thereby improving the data signal-to-noise ratio to over 30dB.

[0076] Principal component analysis (PCA) was used to extract features from the denoised data, selecting core feature parameters with a contribution rate ≥ 90% and removing redundant information. Simultaneously, data consistency was checked to remove outliers.

[0077] The selected effective feature information is integrated with the corresponding component types and concentration values, and then classified and archived according to component type to form a component feature and concentration dataset with standardized structure, high signal-to-noise ratio, and reliable data, providing data support for subsequent detection result analysis and model optimization.

[0078] S3: Based on the high signal-to-noise ratio component characteristics and concentration dataset, AI automatically matches the corresponding sensitive strain library according to the component type, schedules the microbial sensor array to carry out targeted biotoxicity verification, contacts the test gas with the strain according to the concentration gradient, monitors the fluorescence signal change rate in real time through the fluorescence sensing module, quantifies the comprehensive toxicity level of the VOCs mixture by comparing it with the preset toxicity level calibration curve, cross-validates the bioaccumulation hazard of trace toxic components, and outputs the component-toxicity correlation assessment results.

[0079] S31: To adapt the AI-sensitive strain library with a high signal-to-noise ratio set, the characteristic signal spectrum of the volatiles of the target wire harness material needs to be imported into the trained intelligent matching model. The model uses an algorithm that correlates signal purity with strain response specificity to select exclusive strain combinations that are highly sensitive to low concentrations of odor components and outputs a list of strain libraries.

[0080] The characteristic signal spectrum of volatiles in the target wire harness material is extracted, and the signal is denoised and purified to retain the characteristic peaks of low-concentration odor components, forming a high signal-to-noise ratio signal dataset to provide accurate input for subsequent AI matching.

[0081] The high signal-to-noise ratio signal dataset is imported into the trained intelligent matching model. The model calls the correlation algorithm between signal purity and strain response specificity to establish a mapping relationship between signal features and strain sensitivity.

[0082] Based on the mapping relationship, strain types with high responsiveness to low concentrations of odor components were screened out, strains with high cross-interference rates were removed, and a list of exclusive strain combinations was formed to complete the precise construction of the strain library.

[0083] S32: The sensitive strain library is mounted on the microbial sensor array chip. The strains are uniformly fixed on the array detection unit through microfluidic technology. At the same time, a constant temperature and humidity detection environment is configured to form a toxicity sensing detection system that can respond in real time.

[0084] Based on a list of specific strain combinations, microfluidic chip spotting technology is used to precisely inject different strains into the corresponding detection units of the microbial sensor array, thereby achieving the partitioning and fixation of strains.

[0085] The sensor array with the strain immobilized is sealed and encapsulated, and a constant temperature and humidity detection environment of 25±1℃ and 60±5% is configured to ensure the activity and response stability of the strain, thus forming a pre-detection sensor chip.

[0086] The pre-detection sensor chip was subjected to no-load response testing to verify the signal baseline stability of each detection unit, eliminate environmental interference factors, and finally form a toxicity sensing detection system that can respond to the toxicity of volatile substances in real time.

[0087] S33: Connect the prepared microbial sensor array to the fluorescence signal acquisition device, introduce the volatile gas of the low-odor wire bundle material to be detected, continuously monitor the changes in fluorescence signals generated by each strain unit in the array after being stimulated by toxicity, record the signal intensity data at different time points, and output a complete dynamic curve of fluorescence signal change rate.

[0088] Based on the toxicity sensing detection system, the microbial sensor array within the system is connected to the high-resolution fluorescence signal acquisition device through both hardware and software. On the hardware side, a stable connection is established through the data transmission interface, and the signal acquisition frequency and accuracy parameters are adjusted to ensure that the acquisition device can accurately capture the weak fluorescence signal changes of the strain after being stimulated by toxicity.

[0089] A precise amount of volatile gas from the low-odor wire bundle material to be detected is introduced into the detection chamber of the sensor array using a precision sample introduction device. The amount introduced is precisely calculated based on the chamber volume and the target concentration. Subsequently, a fluorescence signal acquisition program is initiated to track the fluorescence intensity of each strain unit in the array in real time. Signal data is continuously recorded at preset time points to form a raw fluorescence intensity-time dataset for each strain unit.

[0090] Based on the collected fluorescence intensity-time raw dataset, the fluorescence intensity values ​​at adjacent time points are differentially calculated using the following formula: In the formula, The rate of change of fluorescence signal, The fluorescence intensity value at the next time point. This represents the fluorescence intensity value at the previous time point. To represent the time value at the next time point, This is the time value of the previous time point.

[0091] The variation amplitude data of all strain units in the integrated array are visualized and integrated according to the time axis, and the output is a complete dynamic curve of fluorescence signal change rate that can intuitively reflect the relationship between toxic stimulation and signal response.

[0092] S34: Using the dynamic curve of fluorescence signal change rate, and comparing it with the pre-constructed toxicity-fluorescence response calibration curve, the toxicity value matched by the signal change rate of each volatile component is calculated by the data fitting algorithm. Combined with the strain response threshold, the quantitative classification is completed, and the preliminary toxicity level judgment result of each component is output.

[0093] The dynamic curve of fluorescence signal change rate is compared with the pre-constructed toxicity-fluorescence response calibration curve. The calibration curve is a baseline curve of signal change rate-toxicity value obtained by testing standard samples with known toxicity gradients. During the comparison process, the focus is on matching the overlap interval of the two curves. The toxicity value of the calibration curve corresponding to this interval is the initial toxicity reference range of the analyte, which defines the boundary for subsequent accurate calculation.

[0094] Using a toxicity reference range as a constraint, a least squares data fitting algorithm was employed to fit and calculate the effective data points of the dynamic curve of fluorescence signal change rate, obtaining the specific toxicity values ​​corresponding to the signal change rates of each volatile component. A response threshold from the strain library was introduced as a correction parameter to exclude invalid signal data below the strain's response lower limit, ensuring that the calculated toxicity values ​​accurately reflect the toxic intensity of the components.

[0095] Based on the pre-defined toxicity value range classification criteria, the fitted toxicity values ​​are mapped to the corresponding toxicity level ranges. During this process, outlier data caused by fluctuations in strain activity and environmental interference are removed to ensure the accuracy of the results. The toxicity values ​​and corresponding levels of all effective components are integrated, and preliminary toxicity level determination results for each component are output, providing core data support for correlation analysis.

[0096] S35: Toxicity level data of each component, combined with the response specificity of different strains in the strain library to specific components, construct a component-toxicity correlation matrix, analyze the toxicity contribution and synergistic effect of each odor component through statistical methods, and output a component toxicity correlation assessment report.

[0097] The preliminary toxicity level assessment results of each component were integrated, including the toxicity level range, corresponding fluorescence signal change rate, and fitted toxicity value for each odor component in the low-odor wiring harness material. Response-specific parameters of different strains in the strain library to specific components were also included, such as the strain's response threshold, signal response intensity, and cross-reactivity rate. A three-dimensional correspondence table was established, using components as the core link, relating odor components, target-response strains, and toxicity level values.

[0098] A mathematical framework for a component-toxicity correlation matrix was constructed based on a three-dimensional correspondence table of components, strains, and toxicity. The row vectors of the matrix represent all odor components in the low-odor wire harness material, with row labels indicating the component names. The column vectors of the matrix have two dimensions: one is the quantitative index of the toxicity level of each component, and the other is the response characteristic parameter of the corresponding strain, with column labels clearly distinguishing the parameter types. Then, according to the row and column correspondences, the data from the three-dimensional correspondence table were filled into the matrix cells one by one, while simultaneously labeling each cell with the strain response coefficient. After completing the matrix data filling, a component-toxicity correlation matrix was formed that intuitively reflects the relationship between components, toxicity, and strain responses, presenting the basic toxicity data and response traceability information of each component.

[0099] Based on the component-toxicity correlation matrix, a deep statistical analysis was conducted: First, principal component analysis (PCA) was used to reduce the dimensionality of the multi-dimensional data in the matrix, identifying the core components that contribute the most to overall toxicity and eliminating redundant data interference. Then, correlation tests were performed to analyze the toxicity correlation between different odor components, determining whether there is a synergistic effect of toxicity additive or an antagonistic effect of toxicity offsetting. Subsequently, based on the PCA results, the toxicity contribution weight of each component was calculated, clarifying the proportion and priority of each component in overall toxicity. Finally, toxicity level data, strain response specificity analysis, and conclusions on component synergistic / antagonistic effects were integrated to form a component toxicity correlation assessment report that includes data support, analytical process, and recommended conclusions.

[0100] S4: Based on the component-toxicity correlation assessment results, multi-channel signal filtering and calibration technology is used to separate and purify multi-dimensional detection information of components, concentrations, and toxicities. The deviation is corrected by combining the working condition parameters recorded in the sealed chamber. The semi-quantitative data is upgraded to precise quantification through standard sample comparison calibration. The component-concentration-toxicity correspondence spectrum is optimized, and the quantitative detection results with both working condition adaptability and toxicity correlation are output.

[0101] S41: Employing multi-channel signal filtering and calibration technology, the concentration and toxicity multi-dimensional detection information in the component toxicity correlation assessment results are separated and purified. Electrochemical signal characteristic parameters (peak voltage, response current) contained in the component toxicity correlation assessment results are extracted. These parameters directly reflect the sensor's response sensitivity under the current operating conditions, and output a multi-dimensional detection information set.

[0102] Multi-channel signal filtering and calibration technology addresses the coupled concentration and toxicity detection information in the evaluation results by separating channels according to signal frequency characteristics—different component concentration signals and toxicity signals correspond to different frequency bands. Bandpass filtering is used to filter target band signals, while adaptive calibration algorithms eliminate signal interference caused by equipment drift and environmental noise. Finally, the separated and purified concentration values, toxicity levels, and strain response characteristics are structurally integrated to output a multi-dimensional detection information set free of redundancy and interference.

[0103] S42: Based on the multi-dimensional detection information set and the temperature, humidity, and pressure parameters recorded in the sealed chamber, the data is calibrated using a deviation correction model, and a corrected detection dataset is output. The deviation correction model is specifically a nonlinear polynomial regression compensation model. Its input variables include: the original concentration and toxicity values ​​from the multi-dimensional detection information set, and the real-time temperature, humidity, and pressure parameters recorded in the sealed chamber. Its internal algorithm logic is as follows: the model has a pre-calibrated experimentally defined working condition parameter-sensitivity drift mapping table. The model uses a polynomial fitting function (such as a second- or third-order polynomial) to calculate the theoretical drift value of the sensor sensitivity under the current working condition. The drift value is used as a correction coefficient to perform nonlinear compensation on the input original detection data.

[0104] The specific mathematical expression is: In the formula, This is the corrected concentration / toxicity value. These are the raw measured values, where T, RH, and P represent real-time temperature, humidity, and pressure. , , This is the standard operating condition value. This is the temperature sensitivity coefficient, representing the relative change in sensitivity for every 1°C deviation of the temperature from the standard operating condition. Electrochemical sensors typically exhibit a negative coefficient. This is the humidity sensitivity coefficient, representing the relative change in sensitivity for every 1% RH deviation from standard operating conditions. Some sensors experience a decrease in sensitivity under high humidity. The pressure sensitivity coefficient represents the relative change in sensitivity for every 1 kPa deviation of the pressure from the standard operating condition. The gas diffusion rate increases with increasing pressure and is mostly a positive coefficient. The output of this model is the corrected detection dataset that eliminates environmental interference.

[0105] Based on a multi-dimensional detection information set and the temperature, humidity, and pressure parameters recorded in the sealed chamber, the influence of temperature and humidity fluctuations on bacterial activity and sensor response sensitivity is analyzed. This pattern is then transformed into correction coefficients, and the mechanism by which pressure changes affect the diffusion rate of volatile gases is analyzed. These patterns are then embedded into a pre-defined deviation correction model. The multi-dimensional detection information set is substituted into the model to perform nonlinear deviation compensation on concentration and toxicity data, correcting detection errors caused by deviations from standard operating conditions. The output is a corrected detection dataset that eliminates operating condition interference and significantly improves data accuracy.

[0106] S43: Compare the gradient concentration standard samples with the calibration and correction test dataset to establish a precise mapping relationship between the signal and the true concentration, realize the upgrade from semi-quantitative data to precise quantification, and output calibrated quantitative data.

[0107] A standard sample with a known accurate concentration is configured, and the same detection procedure as the test sample is used to acquire the signal response data of the standard sample, establishing a baseline correlation curve between the standard concentration and the detection signal. Subsequently, the corrected detection dataset is compared point-by-point with the baseline correlation curve, and a precise mapping relationship between the test sample signal and the true concentration is established through least squares fitting. This overcomes the limitations of the original semi-quantitative detection, achieving an upgrade from semi-quantitative to precise quantification, and ultimately outputting calibrated quantitative data. The calibrated quantitative data is specifically a structured data object, which includes: data in the form of a two-dimensional data matrix or a JSON format file; fields consisting of component identifiers (chemical names or CAS numbers of VOCs components); precise concentration values ​​based on the quantitative concentration obtained from the standard sample mapping; toxicity quantification values; and uncertainty representing the confidence interval of the measured value.

[0108] S44: Optimize the component-concentration-toxicity correspondence graph based on calibration and quantification data, label the influence weight of operating conditions, and output the quantitative detection results.

[0109] The component-concentration-toxicity correlation map is a structured and visualized correlation map. It uses VOCs standard sample calibration curves, microbial toxicity calibration curves, and equipment operating condition deviation calibration results as its core benchmark framework. The overall architecture is built by combining the response characteristics of bio-olfactory receptor chips, the toxicological response patterns of microbial strains, and the interference patterns of environmental conditions on detection signals. A four-dimensional correlation model of component-concentration-toxicity-operating conditions is adopted, and all correlation logic has been verified through orthogonal experiments and repeatability calibration experiments. The data sources for the component-concentration-toxicity correlation map are divided into two categories: static basic data and dynamic real-time data. Static basic data is the initial framework data of the map, including gradient VOCs standard sample calibration data, microbial strain toxicity calibration data, operating condition deviation correction model coefficient data, and industry VOCs and toxicity limit standards for wire harness materials. Dynamic real-time data is the incremental data of the map, including component concentration data from a single detection, toxicity quantification data, real-time temperature / humidity / pressure operating condition data, historical sample detection data, and post-treatment retest data. The component-concentration-toxicity correlation graph update mechanism employs a three-level automated update rule: First, a real-time incremental update is triggered after each test, synchronously recording the current calibration quantification data and operating condition weights into the graph, replacing the original semi-quantitative information; Second, a full iterative update is performed according to the test batch or fixed cycle, summarizing all valid data within the cycle, recalculating the contribution of component toxicity and the weight of operating condition influence, and optimizing the correlation of each dimension; Third, abnormal data filtering is enabled throughout the process, automatically removing outliers caused by equipment failure, sample contamination, and drastic fluctuations in operating conditions based on standard baselines and calibration thresholds, and automatically issuing equipment calibration reminders when data deviations exceed limits, ensuring the stability and reliability of the graph data.

[0110] The calibrated concentration and toxicity values ​​are added to the spectrum, replacing the original semi-quantitative data. Combined with the deviation correction model parameters, the influence weights of operating condition parameters such as temperature, humidity, and pressure on the test results are calculated and marked on the spectrum, clarifying the correction direction and magnitude for data under different operating conditions. Finally, a new version of the spectrum is formed, combining data accuracy and operating condition adaptability. It integrates the core conclusions of the spectrum with the quantitative detection data of each component, outputting quantitative test results and providing a comprehensive and accurate technical basis for the toxicity assessment of low-odor wire harness materials. The data composition of the quantitative test results includes: the basic detection layer, which includes the name, precise concentration value, and toxicity level of each VOC component; the analysis weight layer, which includes the toxicity contribution weight of each component calculated based on principal component analysis (PCA) and the influence weights of operating condition parameters (temperature and humidity) on the test results; and the spatiotemporal attribute layer, which includes the corresponding time series data and sampling point information. The output is an interactive component-concentration-toxicity correlation graph. The underlying data is stored in a structured table format, containing all the above values. The graph uses different colored blocks to mark the concentration and toxicity contribution of each component, and overlays a correction curve for the influence weight of operating conditions.

[0111] S5: The quantitative test results, which combine working condition adaptability and toxicity correlation, are compared with the multi-field industry standard library that is updated in real time by AI. The results are used to comprehensively evaluate the material's odor compliance, the risk level of toxic components and working condition adaptability. The core sources of odor and toxicity are located in reverse through the component traceability map, and a targeted optimization path and problem traceability report are generated.

[0112] S51: Extract VOCs component concentrations, toxicity values, and corresponding time series data based on the quantitative detection results, perform data cleaning and normalization, and construct a multi-dimensional time series dataset containing operating condition parameters.

[0113] The concentration values ​​and toxicity levels of each VOC component were selected from the quantitative detection results, along with the corresponding signal acquisition timestamps. Operating parameters such as temperature, humidity, and pressure recorded in the sealed chamber were also integrated. Data cleaning was performed to remove outliers and missing values ​​caused by sensor malfunctions or sudden environmental interference. A min-max normalization algorithm was used to map all data to the [0,1] interval to eliminate the impact of dimensional differences on model training. The data was then structured and integrated along the dimensions of time series, component concentration, toxicity value, and operating parameters, outputting a multi-dimensional time-series dataset containing operating parameters.

[0114] S52: Using multi-dimensional time-series datasets as training samples, build an N-BEATS model architecture, set the number of iterations and learning rate parameters, train the model to discover the critical threshold and mutation patterns of VOCs toxicity exceeding the standard, and output a converged basic model.

[0115] An N-BEATS model architecture was constructed, consisting of a network of multiple stacked fully connected layers. Each module includes a trend decomposition branch and a fluctuation decomposition branch. The trend branch captures the long-term trend of VOC toxicity, while the fluctuation branch focuses on the short-term fluctuation characteristics of toxicity mutations. The input dimension matches the number of features in the dataset. Training parameters were then configured, including the number of iterations, initial learning rate, batch size, and MSE as the loss function. A standardized multi-dimensional time-series dataset was input into the model for iterative training, with real-time monitoring of the loss value on the validation set. Model convergence was determined when the loss value stabilized after several consecutive rounds. During training, the model adaptively decomposed the time-series data to uncover the correlation between VOC concentration changes and toxicity exceedances, accurately identifying the critical threshold and mutation patterns of toxicity exceedances, and finally outputting a converged basic model.

[0116] S53: Develop an embedded adapter module for the IoT real-time monitoring interface based on the convergence basic model, complete the protocol docking between the model and sensors and data transmission devices, and output the embedded early warning model.

[0117] To address the embedded deployment requirements of real-time IoT monitoring scenarios, an embedded adaptation module for the model was developed. This module reduces the model's computational power consumption through lightweight model technology, making it compatible with the hardware performance of the sensor terminal. Subsequently, protocol integration between the model and front-end sensors and data transmission devices was completed, establishing a unified data transmission format. This enables seamless input of real-time data collected by sensors into the model, while ensuring that early warning commands output by the model can be quickly transmitted to terminal displays or alarm devices. By integrating the lightweight model and the embedded adaptation module, an embedded early warning model with real-time data access and processing capabilities is output.

[0118] S54: Based on the embedded early warning model, set the VOCs toxicity risk level judgment rules, simulate real-time monitoring scenarios to conduct stress tests, optimize the early warning response delay, and output early warning model signals.

[0119] A multi-level VOCs toxicity risk assessment rule was established: toxicity value < threshold 1 indicates low risk, threshold 1 ≤ toxicity value < threshold 2 indicates medium risk, and toxicity value ≥ threshold 2 indicates high risk. Different risk levels correspond to different early warning triggering mechanisms. Subsequently, a simulated real-time monitoring scenario was built, simulating the continuous transmission of time-series data of different VOCs concentrations from sensors. Stress tests were conducted on the embedded early warning model, focusing on verifying the model's response speed and early warning accuracy under high-concurrency data input. To address the early warning delay issue observed during testing, the model inference process and data transmission link were optimized to reduce response latency. A stable and reliable early warning model signal is output, containing key information such as real-time toxicity values, risk levels, and early warning trigger reasons, achieving automatic early warning of VOCs toxicity risks.

[0120] S6: Based on the early warning model signal, the preset governance strategy library is invoked to match the control scheme. The targeted adsorption module is activated for highly toxic trace components, and the temperature, humidity and ventilation rate are adjusted simultaneously to inhibit the slow release of VOCs, forming a closed loop of detection-early warning-control-verification.

[0121] S61: Extract VOCs toxicity level, exceeding components and concentration information from the early warning model signal, match the corresponding control scheme according to risk level-component type, and output the VOCs treatment implementation scheme.

[0122] Key information is precisely extracted from the warning signals: the toxicity risk level of each VOC component, the specific types and peak concentrations of the exceeding components, and the time points of toxicity mutations, while also correlating historical data of operating parameters from step S42. Subsequently, a pre-defined treatment strategy library is invoked, which contains built-in association rules between risk level, component type, and control parameters. For example, high-risk halogenated hydrocarbon components are matched with activated carbon targeted adsorption strategies, while medium-risk alcohol components are matched with ventilation dilution strategies. The system intelligently matches these strategies according to the principle of prioritizing risk level and adapting to component type, generating a VOCs treatment execution plan that includes parameters such as the activation sequence of the targeted adsorption module, temperature and humidity adjustment thresholds, and ventilation rate gradients.

[0123] S62: Based on the VOCs treatment implementation plan, targeted adsorption is initiated for highly toxic trace components, the temperature and humidity of the sealed chamber and the ventilation rate are adjusted to inhibit the continuous slow release of VOCs, and real-time control condition data is output.

[0124] For the highly toxic trace components identified in the plan, the corresponding targeted adsorption module is activated. This module, equipped with specific adsorption materials, can accurately capture target component molecules, avoiding ineffective adsorption of low-toxicity components. Simultaneously, the operating conditions of the sealed chamber are adjusted according to the parameters set in the plan: temperature and humidity are controlled within the optimal range for the slow release of VOCs from the inhibitory materials, and the ventilation rate is adjusted in a gradient manner to reduce the continuous release of VOCs at the source. During the control process, data such as the operating status of the adsorption module, real-time temperature and humidity values, and ventilation rate are collected in real time by sensors inside the chamber. This data is then structured and integrated to output real-time control condition data, providing a data traceability basis for subsequent effect verification.

[0125] S63: Restart VOCs quantitative detection based on real-time control condition data and linkage detection system to verify the removal effect of control measures on toxic components and output VOCs quantitative detection results.

[0126] Based on the confirmed stable operation of the control equipment using operating data, the system automatically triggers a restart command and performs quantitative VOCs detection within the chamber according to standard procedures: from multi-channel signal filtering and purification, correction of operating parameter deviations, to standard sample calibration, the entire process replicates the original detection logic to ensure the comparability of the test results. During the detection process, key indicators such as changes in the concentration of exceeding standards and the reduction in toxicity values ​​are closely monitored. The adjusted concentration and toxicity data are initially compared with warning thresholds and treatment targets, ultimately outputting the quantitative VOCs detection results.

[0127] S64: Compare the quantitative test results before and after regulation to evaluate the effectiveness of the governance plan. If the standard is not met, feedback is sent to the strategy library to optimize the plan and repeat the regulation. If the standard is met, a complete closed loop is formed and a governance acceptance report is output.

[0128] By comparing the two sets of quantitative detection data before and after regulation, the system calculates the decrease in concentration and change in toxicity level of the exceeding components to assess the effectiveness of the treatment plan. If the toxicity value after regulation is lower than the safety threshold, it is considered compliant. If it is not compliant, the data is automatically fed back to the treatment strategy library, and the algorithm optimizes the adsorption module's runtime, operating condition adjustment parameters, etc., to generate an iterative regulation plan. For compliant scenarios, the system integrates data from the entire process: from initial detection data, early warning signals, regulation plans, operating condition data to final detection results, forming a complete treatment acceptance report. The report clearly marks the achievement of treatment objectives, the trajectory of key parameter adjustments, and the data traceability link, completing the entire closed-loop process.

[0129] In summary, this embodiment provides a quality inspection method for low-odor wire harness materials. By combining biosensing and multi-dimensional calibration technologies, it overcomes the bottlenecks of traditional wire harness material odor detection, which is characterized by semi-quantitative methods and susceptibility to interference. Utilizing a chip equipped with specific biological olfactory receptor proteins, combined with electrochemical signal conversion, it can accurately identify trace VOC components at the ppb level. Combined with wavelet threshold denoising and principal component analysis, the data signal-to-noise ratio is improved to over 30dB, effectively eliminating interference signals. Through microbial sensor arrays and fluorescence signal change rate analysis, the toxicity level is quantified by comparing with calibration curves. Further multi-channel filtering, operating condition deviation correction, and standard sample calibration achieve an upgrade from semi-quantitative to precise quantification, establishing a precise correspondence between component, concentration, and toxicity. Compared to traditional detection methods that can only determine the presence or absence of odor, this method can clearly define the concentration, toxicity level, and contribution of each trace component, and even capture synergistic / antagonistic toxicity effects between components, significantly improving detection accuracy and data reliability.

[0130] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A quality testing method for low-odor wire harness materials, characterized in that, include: S1: Extract wire harness material samples into a smart sealed chamber, and adjust the temperature, humidity and ventilation rate in real time through IoT sensors to simulate the slow release of VOCs throughout the entire life cycle and generate dynamic test gas that is highly consistent with real working conditions. S2: Based on the dynamic gas to be tested, capture the molecular characteristic signal of VOCs, use a chip equipped with a specific biological olfactory receptor protein to identify ppb-level trace components, and distinguish components and correlate concentrations through electrochemical signal conversion to form a high signal-to-noise ratio dataset; S3: Based on the high signal-to-noise ratio, AI is used to match the corresponding sensitive strain library, a microbial sensor array is used to perform toxicity verification, the fluorescence signal change rate is monitored, the toxicity level is quantified by comparing with the calibration curve, and the component toxicity correlation assessment results are output. S4: Based on the component toxicity correlation assessment results, the information is purified using multi-channel signal filtering calibration technology, the deviation is corrected by combining operating parameters, and the quantitative detection results are output by calibration with standard sample comparison. S5: Construct an N-BEATS model based on the quantitative detection results, train the model to identify the critical threshold and mutation pattern of VOCs toxicity exceeding the standard, embed the IoT real-time monitoring interface, perform automatic risk level early warning, and obtain early warning model signal; S6: Based on the warning model signal, call the preset treatment strategy library to match the control scheme; activate the targeted adsorption module for highly toxic trace components, and simultaneously adjust the temperature, humidity and ventilation rate to inhibit the slow release of VOCs, forming a closed loop of detection-early warning-control-verification, and outputting a treatment acceptance report.

2. The quality testing method for a low-odor wire harness material according to claim 1, characterized in that: In step S1, the specific steps for generating a dynamic test gas that highly fits the actual operating conditions are as follows: S11: Select a representative wire harness material sample, remove surface contaminants and cut it into uniform small pieces to ensure that the sample volume is compatible with the sealed chamber and obtain a standardized test sample. S12: Place the standardized test sample into the intelligent sealed chamber. The sealed chamber is connected to the environmental parameter monitoring terminal through the Internet of Things sensor module to collect the initial data of temperature, humidity and ventilation rate in the chamber in real time, forming the initial parameter set of the working condition. S13: Based on the initial parameter set of the working condition, dynamically adjust the temperature and humidity cycle and ventilation rate according to the full life cycle working condition curve of the actual service of the wire harness, so as to promote the natural slow release of VOCs and odor substances from the sample and generate mixed release gas. S14: Collect the mixed released gas, filter to remove particulate impurities, and generate a dynamic test gas that closely matches the actual working conditions.

3. The quality testing method for a low-odor wire harness material according to claim 1, characterized in that: In step S2, the specific steps for forming a high signal-to-noise ratio dataset are as follows: S21: The dynamic gas to be tested is introduced into the sensor detection chamber, and the constant temperature and humidity environment inside the chamber is controlled to obtain the gas flow state; S22: Based on the gas flow state, use a specific bio-olfactory receptor protein chip to capture the electrochemical signal changes triggered by VOCs molecule binding, and identify the electrochemical signal change data of trace target components at the ppb level. S23: Input the electrochemical signal change data into the signal-concentration formula to calculate the different VOCs component types and associate them with the corresponding concentration values; S24: Using the correlation data between the types and concentrations of the components, a signal denoising algorithm is used to further optimize the data quality, filter feature information, and form a high signal-to-noise ratio set.

4. The quality testing method for a low-odor wire harness material according to claim 3, characterized in that: In step S22, the specific steps for identifying electrochemical signal change data of trace target components at the ppb level are as follows: Based on the gas flow state, a chip activation program is used to convert the receptor protein on the electrode surface from a dormant state to an active state, exposing all binding sites and outputting an activated chip. According to the activation chip, the dynamic gas to be tested flows at a constant speed and continuously contacts the chip surface. VOCs molecules bind specifically to the corresponding receptor proteins and output the original electrochemical signal corresponding to the VOCs components. The signal fluctuations, peak voltages, and stable currents of the original electrochemical signals are collected, and interference is subtracted after baseline calibration to output an effective signal segment containing trace component characteristics at the ppb level. The effective signal fragments are compared with standard characteristic signals to identify trace target components at the ppb level and output electrochemical signal change data.

5. The quality testing method for a low-odor wire harness material according to claim 1, characterized in that: In step S3, the specific steps for outputting the component toxicity correlation assessment results are as follows: S31: Import the high signal-to-noise ratio dataset into the trained intelligent matching model. The intelligent matching model selects exclusive strain combinations that are highly sensitive to low concentration odor components based on signal purity and strain response specificity, and outputs a strain library list. S32: Transfer the strain library list to the microbial sensor array chip for array detection, configure a constant temperature and humidity detection environment, and form a toxicity sensing detection system; S33: The volatile gas of the low-odor wire harness material to be tested is introduced into the toxicity sensing detection system, the fluorescence signal changes of each strain in the toxicity sensing detection system after being stimulated by toxicity are monitored, the signal intensity data at different time points are recorded, and the dynamic curve of fluorescence signal change rate is output. S34: Compare the dynamic curve of fluorescence signal change rate with the preset toxicity-fluorescence response calibration curve, calculate the toxicity value matched by the signal change rate of each volatile component, and output the preliminary toxicity level determination result. S35: Based on the preliminary toxicity level determination results and the response specificity of different strains in the strain library to specific components, construct a component-toxicity correlation matrix, analyze the toxicity contribution and synergistic effect of each odor component, and output a component toxicity correlation assessment report.

6. The quality testing method for a low-odor wire harness material according to claim 5, characterized in that: In step S35, the specific steps for constructing the component-toxicity correlation matrix are as follows: Based on the preliminary toxicity level determination results and the response specific parameters of the strain library, the correspondence between components, target strains, and toxicity values ​​is sorted out, and a three-dimensional raw data comparison table is output. Based on the three-dimensional raw data reference table, the odor components are set as row vectors and the toxicity level and strain response characteristics are set as column vectors. A matrix framework is built and data is filled in. The response coefficients are labeled, and a draft of the component-toxicity correlation matrix is ​​output. Based on the initial draft of the component-toxicity correlation matrix, principal component analysis was used to calculate the toxicity contribution weight of each component, the synergies between components were marked, and the correlation matrix was output.

7. The quality testing method for a low-odor wire harness material according to claim 1, characterized in that: In step S4, the specific steps for outputting the quantification detection result are as follows: S41: Using multi-channel signal filtering and calibration technology, the concentration and toxicity multi-dimensional detection information in the component toxicity correlation assessment results are separated and purified to output a multi-dimensional detection information set; S42: Based on the multi-dimensional detection information set and the temperature, humidity, and pressure parameters recorded in the sealed chamber, the data is calibrated using a deviation correction model, and a corrected detection dataset is output. S43: Compare the gradient concentration standard samples with the corrected detection dataset to establish a precise mapping relationship between the signal and the true concentration, realize the upgrade from semi-quantitative data to precise quantification, and output the calibrated quantitative data; S44: Optimize the component-concentration-toxicity correspondence graph based on the calibration and quantification data, label the influence weight of operating conditions, and output the quantitative detection results.

8. The quality testing method for a low-odor wire harness material according to claim 1, characterized in that: In step S5, the specific steps for obtaining the early warning model signal are as follows: S51: Based on the quantitative detection results, extract the concentration of VOCs components, toxicity values ​​and corresponding time series data, perform data cleaning and normalization, and construct a multi-dimensional time series dataset containing operating condition parameters; S52: Using the multi-dimensional time series dataset as training samples, build an N-BEATS model architecture, set the number of iterations and learning rate parameters, train the model to discover the critical threshold and mutation law of VOCs toxicity exceeding the standard, and output a converged basic model. S53: Develop an embedded adapter module for the IoT real-time monitoring interface based on the convergence basic model, complete the protocol docking between the model and sensors and data transmission devices, and output the embedded early warning model; S54: Based on the embedded early warning model, set the VOCs toxicity risk level judgment rules, simulate real-time monitoring scenarios to conduct stress tests, optimize the early warning response delay, and output the early warning model signal.

9. The quality inspection method for a low-odor wire harness material according to claim 8, characterized in that: In step S52, the specific steps for outputting the convergent basic model are as follows: The core architecture of the N-BEATS model, consisting of stacked fully connected layer modules, is constructed. Trend and fluctuation decomposition branches are divided to adapt to the characteristics of VOCs toxicity data and output the initial N-BEATS model. Based on the initial N-BEATS model, the number of iterations and learning rate are set, the input and output dimensions of the multi-dimensional time series dataset are bound, and the model to be trained is output. The multi-dimensional time-series dataset is input into the model to be trained for iterative training. The weights are optimized through backpropagation to explore the critical threshold and mutation law of VOCs toxicity exceeding the standard and output a set of model performance indicators. Verify whether the loss value of the model performance index set reaches the convergence threshold and whether the pattern recognition accuracy meets the standard. If yes, output the converged basic model.

10. The quality testing method for a low-odor wire harness material according to claim 1, characterized in that: In step S6, the specific steps for outputting the treatment acceptance report are as follows: S61: Extract VOCs toxicity level, exceeding components and concentration information from the early warning model signal, match the corresponding control scheme according to risk level-component type, and output the VOCs treatment implementation scheme; S62: According to the VOCs treatment implementation plan, targeted adsorption is initiated for highly toxic trace components, the temperature and humidity of the sealed chamber and the ventilation rate are adjusted to inhibit the continuous slow release of VOCs, and real-time control condition data is output. S63: Restart the VOCs quantitative detection based on the real-time control condition data linkage detection system, verify the removal effect of the control measures on toxic components, and output the VOCs quantitative detection results; S64: Compare the quantitative test results before and after regulation to evaluate the effectiveness of the governance plan. If the standard is not met, feedback is sent to the strategy library to optimize the plan and repeat the regulation. If the standard is met, a complete closed loop is formed and a governance acceptance report is output.

Images