Multi-technology fusion plastic additive rapid detection method

Abstract

This application relates to a rapid detection method for plastic additives that integrates multiple technologies. The method includes: acquiring the original signal features of multiple additives; inputting them into a preset matrix interference model to classify and construct a signal separation framework; inputting the detection signal into this framework to decompose it into preliminary independent signal features; correcting these features using an adaptive particle swarm optimization algorithm to obtain independent signal features for multiple additives; extracting interaction data between additives based on these features, identifying the main interference patterns, constructing an interference prediction model using a random forest algorithm, optimizing extraction parameters based on the prediction results, and generating an extraction environment configuration scheme; detecting waste plastic samples according to this scheme to obtain detection data, comparing it with the benchmark values ​​of standard content samples, calculating accuracy data such as content determination deviation, recovery rate, and repeatability, and integrating the data to obtain component analysis data. This method reduces matrix and multiple additive superposition interference, balancing detection speed and accuracy, and provides technical support for plastic additive detection and resource recycling.

Description

Technical Field

[0001] This invention belongs to the field of component detection technology, and in particular relates to a rapid detection method for plastic additives that integrates multiple technologies. Background Technology

[0002] In plastic production, processing, and waste plastic recycling, the type, content, and distribution characteristics of plastic additives directly affect the performance, safety, and recycling value of plastic products. Therefore, rapid and accurate detection of plastic additives is crucial for ensuring product quality, meeting environmental compliance requirements, and promoting resource recycling. Current technologies for detecting plastic additives largely rely on single detection techniques (such as infrared spectroscopy, gas chromatography-mass spectrometry, etc.) or simple signal processing methods, presenting significant technical bottlenecks. On the one hand, the complex composition of the plastic matrix, coupled with interference from multiple additive signals and residual matrix interference, makes it difficult for single techniques to effectively separate target signals, leading to problems such as misidentification of additives and excessive deviations in content determination. On the other hand, traditional detection methods require cumbersome sample pretreatment procedures, have long detection cycles, are complex to operate, and lack targeted interference suppression and intelligent optimization mechanisms, making it difficult to balance detection speed and accuracy, and failing to meet the practical needs of industrial-scale screening and rapid on-site detection of waste plastics. Summary of the Invention

[0003] Therefore, it is necessary to provide a rapid detection method for plastic additives that can effectively improve the separation efficiency of target signals and interference signals, reduce the impact of matrix interference and the superposition of multiple additive signals on the detection results, and achieve targeted adaptation of detection conditions, in order to address the above-mentioned technical problems.

[0004] Firstly, this application provides a rapid detection method for plastic additives that integrates multiple technologies, including:

[0005] The original signal features of various additives are obtained, and the original signal features are classified based on a preset matrix interference model to obtain a signal separation framework.

[0006] Based on the signal separation framework, the detected signal is decomposed to obtain preliminary independent signal features. The preliminary independent signal features are then corrected using an adaptive particle swarm optimization algorithm to obtain independent signal features for various adjuvants.

[0007] Based on the independent signal characteristics, the interaction characteristics data between adjuvants are obtained and the main interference modes between adjuvants are determined. Random forest is used to predict the interference distribution of the main interference modes, and the extraction environment configuration scheme is obtained.

[0008] The waste plastic samples were tested according to the extraction environment configuration plan to obtain sample test data. The sample test data were compared with the determination benchmark value of the standard content sample to calculate the content determination accuracy data, and the component analysis data were integrated to obtain the component analysis data.

[0009] In one embodiment, the original signal features are classified based on a preset matrix interference model to obtain a signal separation framework, including:

[0010] Based on the original signal characteristics, a preset matrix interference model is used for feature classification processing, and a preliminary classification label is output. The preliminary classification label is used to distinguish the target additive signal characteristics from the initial matrix interference signal characteristics.

[0011] Independent component analysis (ICA) was used to separate the target additive signal component and the matrix interference signal component from the mixed signal in the preliminary classification label to obtain the independent additive signal.

[0012] Set a residual interference threshold and determine whether the intensity of the residual matrix interference signal component in the independent additive signal exceeds the residual interference threshold. If it does, use a convolutional neural network to perform fine classification of the residual matrix interference component and output fine interference classification labels.

[0013] An enhanced matrix interference model is constructed based on a fine-grained interference classification label. The independent additive signals are then subjected to secondary signal classification processing based on the enhanced matrix interference model, and an optimized signal separation framework is output.

[0014] In one embodiment, the detection signal is decomposed based on a signal separation framework to obtain preliminary independent signal features. These preliminary independent signal features are then refined using an adaptive particle swarm optimization algorithm to obtain independent signal features for various adjuvants, including:

[0015] Based on the signal separation framework, signal decomposition technology is used to process the superimposed interference in the detection signal, and preliminary independent signal characteristics of various additives are obtained.

[0016] Based on the preliminary independent signal features, a density clustering algorithm is used to group the signal features, generating signal cluster groups and corresponding preliminary labels of residual interference.

[0017] The signal clusters are matched with entries in the preset signal feature association rule base. Combined with the signal waveform similarity measurement method, the additive attribution label and interference component type of each signal cluster are identified. The entries in the signal feature association rule base include additive type, signal feature parameter range, interference component identifier and matching judgment conditions.

[0018] A signal purity verification model is built based on the additive attribution label and the type of interfering components. The signal purity assessment value of the preliminary independent signal characteristics of each cluster group is calculated by three indicators: signal peak stability, characteristic parameter consistency, and proportion of interfering components, and the purity assessment result is generated.

[0019] Based on purity assessment results, adjuvant attribution labels, and interfering component types, an adaptive particle swarm optimization algorithm is used to dynamically adjust the core parameters of the signal decomposition technique, and to perform targeted optimization and correction on the preliminary independent signal features, thereby obtaining the independent signal features of various adjuvants.

[0020] In one embodiment, the signal purity assessment value of the preliminary independent signal features of each cluster group is calculated using the following formula:

[0021]

[0022] in, This represents the signal purity assessment value. This indicates the peak stability index of the signal. , The standard deviation of the signal peak sequence is represented by... This represents the mean of the signal peak sequence. This represents the consistency index of characteristic parameters. , Indicates the dimension of the feature parameters. This represents the angle between the corresponding characteristic parameters of the signal to be evaluated and the standard signal. Indicator of the proportion of interfering components , The signal energy representing the interfering component, This represents the total energy of the signal in the corresponding cluster group. , , This represents a dynamic weighting factor that adaptively adjusts based on the adjuvant attribution label and the type of interfering components. Indicates the calibration coefficient. It is used to correct the inherent biases of different additive signal types.

[0023] In one embodiment, interaction feature data between adjuvants are obtained based on independent signal characteristics, and the main interference modes between adjuvants are determined. Random forest is used to predict the interference distribution of the main interference modes to obtain an extraction environment configuration scheme, including:

[0024] The correlation data reflecting the interaction between adjuvants is extracted from the independent signal features. The correlation data is then subjected to signal offset correction based on a preset correction algorithm to obtain the corrected signal feature values.

[0025] Based on the corrected signal characteristic values, a preset interference judgment threshold is used to determine the main interference modes among the additives.

[0026] The extraction condition control parameters, including extraction temperature, extraction time, and solvent ratio, were dynamically adjusted for the main interference modes to obtain optimized environmental configuration data.

[0027] Set predetermined standards for environment configuration, determine whether the optimized environment configuration data meets the predetermined standards, and if it does, generate the corresponding extraction and optimization scheme.

[0028] Based on the extraction optimization scheme, the random forest algorithm is used to analyze the interference trend among adjuvants and output the predicted interference distribution data.

[0029] The fine-tuning parameters of the environment configuration are adjusted based on the predicted interference distribution data to generate the final extraction environment configuration scheme.

[0030] In one embodiment, waste plastic samples are tested according to an extraction environment configuration plan to obtain sample test data. The sample test data is compared with the determination benchmark value of a standard content sample to calculate the content determination accuracy data. This data is then integrated to obtain component analysis data, including:

[0031] Perform testing operations on waste plastic screening samples according to the extraction environment configuration plan, and obtain the corresponding sample test data.

[0032] The distribution characteristics of additives in the sample detection data were quantitatively analyzed to obtain the spatial distribution characteristics and content gradient information of additives in the sample.

[0033] A signal quality stabilization algorithm is used to suppress and calibrate the signal noise and fluctuation components in the spatial distribution characteristics and content gradient information to obtain stable signal output data.

[0034] Based on the signal output data, the confidence level of each auxiliary agent signal feature is calculated by comparing it with the preset standard auxiliary agent signal feature library, and the auxiliary agent identification verification result is obtained.

[0035] By comparing the determination benchmark values ​​of the standard content samples, the determination deviation, recovery rate and repeatability of each auxiliary agent content are calculated, and the corresponding content determination accuracy data are obtained.

[0036] By integrating the results of auxiliary agent identification and verification with the accuracy data of content determination, component analysis data including auxiliary agent type, distribution location and content value are generated.

[0037] In one embodiment, the identification confidence level is calculated using the following formula:

[0038]

[0039] in, Indicates the confidence level of identification, where, Determined as a valid identification, This represents a multi-dimensional similarity factor, including peak feature similarity. Frequency feature similarity Waveform morphology similarity , This indicates adaptive weighting, dynamically assigned based on the adjuvant attribution label. This represents the signal quality correction factor. , This indicates the peak stability index of the signal. Indicates the uniqueness factor of the feature. , This represents the weighted similarity between the standard library and the signal to be identified. The number of non-target adjuvants.

[0040] Secondly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0041] Thirdly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned method.

[0042] The aforementioned multi-technology integrated rapid detection method, computer equipment, and storage medium for plastic additives acquires the original signal features of various additives. Using these original signal features as input, a classification process is performed based on a preset matrix interference model to construct a signal separation framework. Subsequently, the signal to be detected is input into the signal separation framework for decomposition, obtaining preliminary independent signal features. For these preliminary independent signal features, an adaptive particle swarm optimization algorithm is used to correct parameters, eliminating signal superposition interference and residual matrix interference, ultimately obtaining the independent signal features of various additives. Based on these independent signal features, interaction feature data between additives is extracted. The main interference modes between additives are determined through analysis of this data. For the main interference modes, a random forest algorithm is used to construct an interference prediction model to predict the distribution of additive interference. Based on the prediction results, parameters such as extraction temperature, time, and solvent ratio are optimized to generate an extraction environment configuration scheme. The waste plastic samples were tested according to the extraction environment configuration scheme to obtain sample test data. The sample test data was compared with the determination benchmark value of the standard content sample to calculate the content determination accuracy data such as the content determination deviation, recovery rate and repeatability of each additive. The content determination accuracy data was integrated with the previous additive identification results to form component analysis data covering additive types, distribution locations and content values. On the one hand, a signal separation framework was constructed by pre-setting a matrix interference model, and the initial independent signal features were corrected by an adaptive particle swarm optimization algorithm, which effectively improved the separation efficiency of target signals and interference signals, reduced the impact of matrix interference and the superposition of multiple additive signals on the detection results, and ensured the accuracy of independent signal features. On the other hand, based on the analysis of the interference patterns between additives based on independent signal features, the random forest algorithm was used to predict the interference distribution and optimize the extraction environment configuration, realizing targeted adaptation of detection conditions and balancing detection speed and accuracy. Meanwhile, by comparing and calibrating the sample test data with the standard benchmark value, the reliability of the content determination accuracy data is ensured. The final integrated component analysis data can comprehensively reflect the core information of additives in waste plastics, solving the problems of weak anti-interference ability of single technology, poor adaptability of test conditions, and unsystematic data integration in traditional detection methods. It can meet the actual needs of industrial-scale screening and rapid on-site detection of waste plastics, and provide technical support for plastic additive detection and resource recycling. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1A flowchart illustrating a rapid detection method for plastic additives integrating multiple technologies, provided in an embodiment of the present invention;

[0045] Figure 2 The flowchart illustrates the signal separation framework obtained by classifying the original signal features based on a preset matrix interference model, as provided in this embodiment of the invention. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0047] In one embodiment, such as Figure 1 As shown, this application provides a rapid detection method for plastic additives that integrates multiple technologies, which may include the following steps:

[0048] Step S101: Obtain the original signal features of various additives, classify the original signal features based on the preset matrix interference model, and obtain the signal separation framework.

[0049] Specifically, firstly, various known types of plastic additives are individually tested using standard laboratory testing methods to obtain the original signal characteristics of each additive. These original signal characteristics include quantifiable signal parameters such as peak value, frequency, amplitude, and phase. Subsequently, a pre-set matrix interference model is introduced. This model pre-integrates the interference signal characteristics and classification rules generated by common plastic matrices (such as polyethylene and polypropylene) during the testing process. Using the original signal characteristics as input, the model is classified according to the correlation between signal characteristics and matrix interference and the uniqueness of additive signals. Feature dimensions and separation logic that can effectively distinguish additive signals from matrix interference signals are selected, and finally, a signal separation framework is constructed. This framework clarifies the separation criteria, feature selection parameters, and basic processing procedures for target additive signals and interference signals in subsequent detection signals.

[0050] Step S102: Based on the signal separation framework, the detection signal is decomposed to obtain preliminary independent signal features. The preliminary independent signal features are then corrected using an adaptive particle swarm optimization algorithm to obtain independent signal features for various adjuvants.

[0051] The mixed detection signal of the plastic sample to be tested is input into the signal separation framework constructed in step S101. The framework splits the mixed detection signal according to the preset separation criteria and parameters, eliminates some matrix interference signals, and obtains the preliminary independent signal features corresponding to each additive. Since the preliminary independent signal features may still have a small amount of interference from the superposition of multiple additive signals and signal drift, an adaptive particle swarm optimization algorithm is used to correct them. This algorithm uses signal peak stability and feature parameter consistency as optimization objectives, dynamically adjusts the signal feature parameters, eliminates residual superposition interference and drift effects, and finally outputs the corrected independent signal features of multiple additives that can accurately characterize the unique properties of each additive. These features provide reliable data support for subsequent interaction analysis between additives and interference mode judgment.

[0052] Step S103: Obtain interaction feature data between adjuvants based on independent signal features and determine the main interference modes between adjuvants. Use random forest to predict the interference distribution of the main interference modes to obtain the extraction environment configuration scheme.

[0053] Based on the obtained independent signal features, interaction feature data between excipients is extracted through signal feature correlation analysis. This data covers quantitative information such as the superposition area of ​​signals from different excipients, the amplitude of feature parameter changes, and signal response delay. Combining this interaction feature data, the main interference modes between excipients, such as signal superposition interference and feature masking interference, are summarized through threshold determination and pattern matching. For the identified main interference modes, a random forest prediction model is constructed. Using the interference modes, environmental parameters, and interference distribution data in historical detection data as training samples, and inputting relevant information of the current main interference modes, the model predicts the distribution of excipient interference under different extraction conditions. Based on the prediction results, key parameters such as extraction temperature, extraction time, and solvent ratio are adjusted to optimize the extraction environment configuration scheme that can effectively suppress interference between excipients.

[0054] Step S104: Detect waste plastic samples according to the extraction environment configuration plan to obtain sample detection data, compare the sample detection data with the determination benchmark value of standard content samples to calculate the content determination accuracy data, and integrate to obtain component analysis data.

[0055] According to the obtained extraction environment configuration plan, standardized testing operations were performed on waste plastic screening samples, including sample pretreatment, detection parameter setting, signal acquisition, etc., to obtain sample detection data containing additive signals and a small amount of residual interference. The sample detection data was compared one by one with the preset standard content sample determination benchmark value, and the content determination accuracy data such as the determination deviation, recovery rate and repeatability of each additive content were calculated to quantitatively evaluate the accuracy and reliability of the test results. Finally, the obtained additive identification information, interference analysis results and the content determination accuracy data calculated in this study were integrated to form comprehensive component analysis data covering additive types, distribution locations, content values ​​and accuracy assessments.

[0056] The aforementioned multi-technology integrated rapid detection method for plastic additives acquires the original signal features of various additives, inputs them into a preset matrix interference model to classify and construct a signal separation framework, decomposes the detected signals into this framework to obtain preliminary independent signal features, and then corrects them using an adaptive particle swarm optimization algorithm to obtain independent signal features for various additives, based on these features to extract interaction data between additives, identify the main interference modes, construct an interference prediction model using a random forest algorithm, and optimize parameters such as extraction temperature, time, and solvent ratio based on the prediction results to generate an extraction environment configuration scheme, and then tests waste plastic samples according to this scheme to obtain detection data, compares it with the benchmark values ​​of standard content samples, calculates the accuracy data such as content determination deviation, recovery rate, and repeatability, and integrates the previous additive identification results to form component analysis data covering the types, distribution, and content of additives. This method improves signal separation efficiency and reduces interference from matrix and multiple additives by constructing a matrix interference model framework and correcting it with particle swarm optimization algorithm, thus ensuring signal accuracy. It optimizes detection conditions by combining interference mode analysis and random forest prediction, balancing detection speed and accuracy. The method is calibrated by comparison with standard samples to ensure data reliability. It solves the problems of weak anti-interference, poor condition adaptability and disordered data integration of traditional methods, meets the needs of industrial-scale screening and rapid on-site detection, and provides technical support for the detection and recycling of plastic additives.

[0057] In one embodiment, such as Figure 2 As shown, classifying the original signal features based on a preset matrix interference model to obtain a signal separation framework can include the following steps:

[0058] Step S201: Based on the original signal features, a preset matrix interference model is used to perform feature classification processing and output preliminary classification labels; the preliminary classification labels are used to distinguish the target adjuvant signal features from the initial matrix interference signal features.

[0059] Step S202: Independent component analysis is used to separate the target additive signal component and the matrix interference signal component of the mixed signal in the preliminary classification label to obtain the independent additive signal.

[0060] Step S203: Set a residual interference threshold, determine whether the intensity of the residual matrix interference signal component in the independent additive signal exceeds the residual interference threshold. If it does, use a convolutional neural network to perform fine classification of the residual matrix interference component and output fine interference classification labels.

[0061] Step S204: Construct an enhanced matrix interference model based on the fine interference classification label, perform secondary signal classification processing on the independent additive signals based on the enhanced matrix interference model, and output the optimized signal separation framework.

[0062] Specifically, the process uses the original signal characteristics of various additives as input, encompassing quantifiable core signal parameters such as peak value, frequency, amplitude, and phase. A pre-defined matrix interference model is employed for feature classification. This model integrates a feature library, feature matching rules, and classification algorithms for common plastic matrices (such as polyethylene, polypropylene, and polyvinyl chloride). By comparing the original signal characteristics with the matrix interference signal characteristics built into the model, preliminary classification labels are output. These labels contain two types of core information, corresponding to the signal characteristics belonging to the target additive and the signal characteristics belonging to the initial matrix interference in the original signal, thus achieving a preliminary distinction between the target signal and the initial interference signal. Based on this, independent component analysis (ICA) is used to separate the mixed signal corresponding to the preliminary classification labels. This technique, based on the principle of statistical independence of signals, decomposes the covariance matrix of the mixed signal and solves the optimal separation matrix to accurately extract the independent target additive signal components and matrix interference signal components from the mixed signal, ultimately obtaining independent additive signals with the main matrix interference preliminarily removed. Subsequently, a residual interference threshold is set based on the accuracy requirements of the detection scenario. This threshold is a critical value of residual interference intensity obtained from historical detection data statistics that does not affect the accuracy of subsequent detection results. The amplitude or energy of the residual matrix interference signal component in the independent additive signal is quantitatively calculated to determine whether it exceeds the preset residual interference threshold. If the result is that it exceeds the threshold, it indicates that the residual matrix interference in the independent additive signal may still affect the accuracy of subsequent detection. In this case, a pre-trained convolutional neural network is used to perform fine classification of the residual matrix interference component. This convolutional neural network uses residual matrix interference signal features of different types and intensities as training samples. Through multi-layer convolution, pooling, and fully connected layers, it accurately identifies the specific type (such as signal superposition interference, feature masking interference, noise interference, etc.) and intensity level of residual interference, and outputs fine interference classification labels containing information such as interference type, intensity, and range of influence. Finally, based on the refined interference classification labels, the residual interference features and classification rules contained therein are added to the initially preset matrix interference model to improve the interference feature library and classification algorithm of the model, and an enhanced matrix interference model is constructed. The enhanced matrix interference model is used to perform secondary signal classification processing on the aforementioned independent additive signals to further separate the residual matrix interference signal components, optimize the feature selection parameters, classification threshold and separation logic of signal separation, and finally output an optimized signal separation framework with stronger adaptability and higher separation accuracy.

[0063] This embodiment significantly improves the separation accuracy of target additive signals and matrix interference signals. Specifically, the preliminary classification of the preset matrix interference model provides clear directional guidance for subsequent signal separation, effectively reducing the separation difficulty of independent component analysis (ICA) and ensuring the integrity of the core features of the independent additive signals after preliminary separation. The application of ICA enables efficient separation of target and interference components in the mixed signal, significantly eliminating initial matrix interference and laying the foundation for subsequent residual interference judgment. The setting of the residual interference threshold enables quantitative determination of the degree of interference residue, avoiding resource waste caused by blindly carrying out fine processing and improving processing efficiency. The fine classification of the convolutional neural network can accurately locate the type and intensity of residual interference, providing accurate data support for model optimization. The enhanced matrix interference model built based on the fine classification results compensates for the initial model's insufficient adaptation to complex residual interference, and further improves the reliability of signal separation through secondary classification processing. The entire technical process effectively solves the problems of poor adaptability to complex matrix interference, difficulty in completely eliminating residual interference, and insufficient separation accuracy in traditional signal separation methods through iterative optimization of the model and multi-stage signal processing. The optimized signal separation framework can adapt to interference scenarios of different types of plastic matrices, ensuring the accuracy and effectiveness of subsequent detection data and helping to improve the overall efficiency and accuracy of plastic additive detection.

[0064] In one embodiment, the detection signal is decomposed based on a signal separation framework to obtain preliminary independent signal features. The preliminary independent signal features are then refined using an adaptive particle swarm optimization algorithm to obtain independent signal features for various adjuvants. This process may include the following steps:

[0065] Step S301: Based on the signal separation framework, signal decomposition technology is used to process the superimposed interference in the detection signal and separate the preliminary independent signal features of various additives.

[0066] Step S302: Based on the preliminary independent signal features, the signal features are grouped using a density clustering algorithm to generate signal cluster groups and corresponding preliminary residual interference labels.

[0067] Step S303: Match the signal cluster groups with the entries in the preset signal feature association rule library, and combine the signal waveform similarity measurement method to label the auxiliary agent attribution label and interference component type of each signal cluster group; the entries in the signal feature association rule library include auxiliary agent type, signal feature parameter range, interference component identifier and matching judgment conditions.

[0068] Step S304: Based on the additive attribution label and the type of interfering component, a signal purity verification model is built. The signal purity assessment value of the preliminary independent signal characteristics of each cluster group is calculated through three indicators: signal peak stability, characteristic parameter consistency, and proportion of interfering components, and the purity assessment result is generated.

[0069] Step S305: Based on the purity assessment results, the excipient attribution labels, and the types of interfering components, the core parameters of the signal decomposition technology are dynamically adjusted using an adaptive particle swarm optimization algorithm to optimize and correct the preliminary independent signal features in a targeted manner, thereby obtaining the independent signal features of various excipients.

[0070] Specifically, based on a pre-constructed signal separation framework, signal decomposition technology is used to split the detection signal containing superimposed interference from multiple additives. By analyzing the characteristic differences in different frequencies and amplitude dimensions of the detection signal, some matrix interference and superimposed components of multiple additive signals are separated and removed, obtaining preliminary independent signal features corresponding to each additive. These features initially reflect the core signal attributes of each additive, but a small amount of residual interference may still remain. Based on these preliminary independent signal features, a density clustering algorithm is used to group the signal features. The similarity of signal feature parameters is used as the clustering basis, and signal data with high feature similarity are grouped into one category, generating multiple signal clusters. At the same time, combined with the dispersion of signals within the clusters and the degree of deviation from the standard additive signal, the possible residual interference types and ranges are marked for each cluster, forming corresponding preliminary residual interference labels. Subsequently, each signal cluster group is matched against a pre-defined signal feature association rule base. This rule base contains multiple entries, each including the type of additive, the range of signal feature parameters (such as peak interval, frequency threshold, amplitude fluctuation range), the identifier of interference components, and matching judgment conditions (such as feature parameter matching degree threshold, waveform similarity threshold). During the matching process, signal waveform similarity measurement methods (such as dynamic time warping, cosine similarity calculation) are used simultaneously to quantify the similarity between the cluster group signals and the standard additive signals in the rule base. By combining the matching results and similarity measurement values, the additive attribution label (clearly identifying the type of additive) and the type of interference components (such as signal superposition interference, noise interference, feature masking interference) corresponding to each signal cluster group are accurately identified. Based on the calibrated adjuvant attribution labels and interfering component types, a signal purity verification model was built. This model uses signal peak stability, characteristic parameter consistency, and interfering component proportion as core evaluation indicators. Signal peak stability is quantified by calculating the coefficient of variation of the peak sequence, characteristic parameter consistency is determined by comparing the mean deviation of characteristic parameters between the cluster group signal and the standard signal, and the interfering component proportion is calculated by the ratio of interfering signal energy to the total signal energy of the cluster group. The model is used to quantify and evaluate the preliminary independent signal characteristics of each cluster group, calculate the signal purity evaluation value of each cluster group, and generate purity evaluation results covering the purity level and residual interference intensity of each cluster group. Based on the purity assessment results, adjuvant attribution labels, and interfering component types, an adaptive particle swarm optimization algorithm was used to dynamically adjust the core parameters of the signal decomposition technique (such as the number of decomposition layers, threshold parameters, and feature extraction dimensions). Targeted optimizations were implemented for clusters with different purity levels and different interference types: for clusters with lower purity and stronger interference, the frequency resolution of the signal decomposition and the interference removal threshold were adjusted; for clusters with higher purity and weaker interference, the feature extraction parameters were fine-tuned to retain more effective signal details. Through this optimization process, residual interference in the initial independent signal features was further eliminated, the core signal features of the adjuvants were strengthened, and ultimately, accurate and stable independent signal features for various adjuvants were obtained.

[0071] This embodiment significantly improves the accuracy and reliability of multi-additive signal separation through multi-stage progressive signal processing and optimization. Specifically, signal decomposition technology combined with a signal separation framework achieves initial removal of superimposed interference, laying the foundation for subsequent processing; density clustering algorithm achieves scientific grouping of signal features, and with the initial labeling of residual interference, narrows the scope of subsequent interference analysis; the combination of signal feature association rule base and waveform similarity measurement method achieves accurate labeling of additive attribution and interference type, providing a basis for targeted optimization; signal purity verification model quantifies signal quality through multi-dimensional indicators, clarifying the optimization direction; adaptive particle swarm optimization algorithm dynamically adjusts parameters according to purity evaluation results, additives, and interference types, achieving differentiated optimization correction and effectively eliminating residual interference. The entire process can accurately solve the problems of difficult separation of superimposed interference from multiple additives, ambiguous additive identification, and incomplete removal of residual interference in traditional signal processing. The final independent signal features have high accuracy and strong stability, significantly improving the overall accuracy and efficiency of plastic additive detection, and adapting to the actual needs of rapid detection of multiple additives in complex matrices.

[0072] In one embodiment, the signal purity assessment value of the preliminary independent signal features of each cluster group can be calculated using the following formula:

[0073]

[0074] in, This represents the signal purity assessment value. This indicates the peak stability index of the signal. , The standard deviation of the signal peak sequence is represented by... This represents the mean of the signal peak sequence. This represents the consistency index of characteristic parameters. , Indicates the dimension of the feature parameters. This represents the angle between the corresponding characteristic parameters of the signal to be evaluated and the standard signal. Indicator of the proportion of interfering components , The signal energy representing the interfering component, This represents the total energy of the signal in the corresponding cluster group. , , This represents a dynamic weighting factor that adaptively adjusts based on the adjuvant attribution label and the type of interfering components. Indicates the calibration coefficient. It is used to correct the inherent biases of different additive signal types.

[0075] This embodiment achieves signal peak stability through fusion ( ), feature parameter consistency ( ), proportion of interfering components ( Three core quantitative indicators are used to comprehensively evaluate the purity of the preliminary independent signal features of each cluster group, among which... The degree of signal fluctuation is quantified by the ratio of the standard deviation of the peak sequence to the mean. The feature matching degree is characterized by the cosine mean of the angle between multi-dimensional feature parameters and the standard signal. The interference percentage is determined by the ratio of interference signal energy to total signal energy, and these three factors work together to cover the core dimensions of signal stability, effectiveness, and anti-interference. The formula introduces a dynamic weighting factor that adaptively adjusts based on the adjuvant attribution label and the type of interfering component. It can flexibly allocate index weights based on different additive signal characteristics (such as peak-sensitive and frequency-sensitive) and interference component types (such as superposition interference and noise interference), solving the problem of insufficient adaptability of traditional fixed-weight evaluation to complex scenarios; at the same time, a calibration coefficient is added. (Value range 0.95-1.05) This formula specifically corrects the inherent biases of different additive signal types (such as infrared and fluorescence signals), further improving the universality of the assessment. The purity assessment results generated by this formula can accurately reflect the signal quality and interference level of each cluster group, providing a quantitative basis for subsequent dynamic adjustment of the core parameters of signal decomposition based on the adaptive particle swarm optimization algorithm. It enables targeted optimization and correction of the preliminary independent signal characteristics of different purities and different interference types, effectively improving the accuracy and stability of the final independent signal characteristics. This provides high-quality data support for subsequent additive interaction analysis, content determination, and other processes, helping to solve problems such as one-sided signal purity assessment and lack of accurate basis for correction in complex interference scenarios with multiple additives, and ensuring the overall accuracy and reliability of plastic additive detection.

[0076] In one embodiment, the interaction feature data between adjuvants is obtained based on independent signal characteristics, and the main interference modes between adjuvants are determined. The interference distribution of the main interference modes is predicted using random forest to obtain the extraction environment configuration scheme, which may include the following steps:

[0077] Step S401: Extract the correlation data reflecting the interaction between adjuvants from the independent signal features, and perform signal offset correction on the correlation data based on the preset correction algorithm to obtain the corrected signal feature values.

[0078] Step S402: Based on the corrected signal feature values, a preset interference judgment threshold is used to determine the main interference mode among the additives.

[0079] Step S403: Dynamically adjust the extraction condition control parameters, including extraction temperature, extraction time, and solvent ratio, for the main interference modes to obtain optimized environmental configuration data.

[0080] Step S404: Set the predetermined standard for environment configuration, determine whether the optimized environment configuration data meets the predetermined standard, and if it does, generate the corresponding extraction optimization scheme.

[0081] Step S405: Based on the extraction optimization scheme, the random forest algorithm is used to analyze the interference trend among adjuvants and output the predicted interference distribution data.

[0082] Step S406: Adjust the fine-tuning parameters of the environment configuration based on the predicted interference distribution data to generate the final extraction environment configuration scheme.

[0083] From the independent signal features of various excipients obtained in the early stage, correlation data reflecting the interactions between excipients is extracted. This correlation data covers quantifiable information such as the superposition range of signals from different excipients, the amplitude of characteristic parameter changes, signal response delay, and energy coupling degree. Using this correlation data as the processing object, a preset correction algorithm (such as baseline correction and phase correction algorithm) is used to correct signal offsets caused by factors such as the accuracy of the detection equipment and environmental fluctuations, eliminating systematic errors and obtaining accurate corrected signal feature values. Based on the corrected signal feature values, a preset interference judgment threshold (which is determined based on historical detection data statistics and covers the critical range of feature values ​​corresponding to different interference modes) is called for quantitative judgment. By comparing the matching relationship between the corrected signal feature values ​​and the threshold, the main interference modes existing between excipients, such as signal superposition interference, feature masking interference, and energy competition interference, are summarized. For the identified main interference patterns, the core control parameters in the extraction conditions are dynamically adjusted based on the interference type, intensity, and scope of influence. These include extraction temperature (adjusted based on the thermal stability of the interfering components), extraction time (optimized according to the dissolution rate of the excipients and the interference removal efficiency), and solvent ratio (adjusted based on the differences in solubility characteristics between the excipients and the interfering components), resulting in optimized environmental configuration data. A pre-defined environmental configuration standard is established, including core indicators such as extraction efficiency threshold, interference suppression rate threshold, and operational feasibility parameter ranges. The optimized environmental configuration data is compared with this pre-defined standard to determine if it meets the requirements of each indicator. If the result is that it meets the pre-defined standard, a corresponding extraction optimization scheme is generated based on the optimized environmental configuration data. Based on the extraction optimization scheme, a random forest prediction model is constructed. The extraction parameters, excipient types, and main interference patterns in the scheme are used as input variables, and the interference distribution in historical detection data is used as training samples. The random forest algorithm is used to quantitatively analyze the interference trends between excipients, outputting predicted interference distribution data under different extraction stages and parameter combinations. Based on the predicted interference distribution data, the parameter ranges where residual interference may exist are located, and the fine-tuning parameters in the environmental configuration (such as temperature fluctuation range, time gradient interval, solvent ratio accuracy, etc.) are adjusted in a targeted manner to further optimize the interference suppression effect, and finally generate a final extraction environment configuration scheme with strong adaptability and excellent interference suppression effect.

[0084] This embodiment significantly improves the ability of the extraction environment to suppress interference between additives. Specifically, the accurate extraction and signal offset correction of correlated data provide an accurate data foundation for subsequent interference pattern identification, avoiding interference judgment bias caused by errors in the original data; the preset interference judgment threshold enables the quantitative identification of major interference patterns, providing a clear direction for parameter adjustment; dynamic adjustment of core extraction parameters combined with predetermined environmental configuration standards ensures the feasibility and effectiveness of the optimization scheme, avoiding resource waste caused by blind adjustments; the random forest algorithm predicts interference trends, enabling forward-looking optimization of interference prevention and control, and avoiding potential interference risks in advance; finally, the accuracy of the scheme is further improved by fine-tuning parameters, ensuring the stability and integrity of additive signals during the extraction process. This process effectively solves the problems of strong subjectivity in parameter setting, poor adaptability to interference between additives, and lack of data support in traditional extraction environment configuration. The final extraction environment configuration scheme can accurately suppress interference between additives, improve the extraction efficiency and purity of target additives, provide a high-quality sample foundation for subsequent sample testing, content determination, and other stages, help improve the overall accuracy and reliability of plastic additive testing, and adapt to the testing needs of complex multi-additive systems.

[0085] In one embodiment, the process of obtaining sample test data by testing waste plastic samples according to the extraction environment configuration scheme, calculating the content determination accuracy data by comparing the sample test data with the determination benchmark value of the standard content sample, and integrating the results to obtain component analysis data may include the following steps:

[0086] Step S501: Perform testing on the waste plastic screening samples according to the extraction environment configuration plan to obtain the corresponding sample testing data.

[0087] Step S502: Quantitatively analyze the distribution characteristics of additives in the sample detection data to obtain the spatial distribution characteristics and content gradient information of additives in the sample.

[0088] Step S503: The signal quality stabilization algorithm is used to suppress and calibrate the signal noise and fluctuation components in the spatial distribution characteristics and content gradient information to obtain stable signal output data.

[0089] Step S504: Based on the signal output data, calculate the recognition confidence of each auxiliary agent signal feature with the preset standard auxiliary agent signal feature library to obtain the auxiliary agent recognition verification result.

[0090] Step S505: Compare the determination benchmark value of the standard content sample with the determination benchmark value, calculate the determination deviation, recovery rate and repeatability index of each auxiliary agent content, and obtain the corresponding content determination accuracy data.

[0091] Step S506: Integrate the auxiliary agent identification verification results with the content determination accuracy data to generate component analysis data including auxiliary agent type, distribution location and content value.

[0092] Strictly following the pre-defined extraction environment configuration, standardized testing procedures were performed on waste plastic screening samples, covering the entire process from sample pretreatment and detection parameter setting to signal acquisition. This yielded sample detection data containing additive signals, residual matrix interference, and detection noise. Using this sample detection data as the analysis object, feature quantification analysis was employed to extract information such as the spatial distribution range and density distribution of each additive in the sample, as well as the concentration gradient and distribution uniformity in the content dimension. This yielded the spatial distribution characteristics and content gradient information of the additives in the sample. To address signal noise and detection fluctuations present in the spatial distribution characteristics and content gradient information, signal quality stabilization algorithms (such as moving average denoising, wavelet threshold denoising combined with trend correction algorithms) were used for suppression and calibration, eliminating the influence of random noise, equipment drift, and other factors, resulting in stable signal output data with clear characteristics and controllable fluctuations. Based on stable signal output data, the signal characteristics of each auxiliary agent are matched with entries in a pre-set standard auxiliary agent signal feature library. Through multi-dimensional confidence calculation (such as a comprehensive confidence quantification integrating peak similarity, frequency consistency, and waveform matching), auxiliary agent identification verification results are obtained, including the identification type, confidence value, and effective identification judgment results for each auxiliary agent. Simultaneously, the measurement benchmark value of a pre-set standard content sample is retrieved, and the measurement results of each auxiliary agent content in the sample detection data are compared one by one with this benchmark value. The measurement deviation (absolute deviation, relative deviation), recovery rate (ratio of actual measured amount to theoretical added amount), and repeatability index (relative standard deviation of multiple parallel measurement results) of each auxiliary agent content are calculated to obtain content measurement accuracy data that can quantify the accuracy and stability of the detection. Finally, the auxiliary agent identification verification results (clarifying the auxiliary agent type) and content measurement accuracy data (verifying content reliability) are integrated, combined with the previously obtained spatial distribution characteristics of the auxiliary agents, to generate component analysis data covering the auxiliary agent type, specific distribution location, and precise content values.

[0093] This embodiment ensures the consistency of sample detection data by using a preset extraction environment configuration scheme, avoiding deviations in the original data caused by differences in detection conditions. Quantitative analysis of the distribution characteristics of additives enables precise characterization of the spatial and content dimensions of additives, providing a clear direction for subsequent signal processing. The application of a signal quality stabilization algorithm effectively eliminates noise and fluctuations, ensuring the reliability of the signal output data and laying the foundation for identification and verification. Multi-dimensional identification confidence calculation improves the accuracy of additive identification, reducing false positives and false negatives. The content determination accuracy data calculated by comparing with standard benchmark values ​​quantitatively verifies the accuracy and stability of the detection results. Finally, data integration provides a comprehensive presentation of additive type, distribution, and content information, solving problems such as data fragmentation, low identification accuracy, and insufficient reliability of content determination in traditional detection methods. The entire process is suitable for detection scenarios involving complex matrices of waste plastics, providing high-quality component analysis data for waste plastic classification, recycling, quality assessment, and resource recycling, ensuring the practicality and authority of the detection results.

[0094] In one embodiment, the identification confidence level can be calculated using the following formula:

[0095]

[0096] in, Indicates the confidence level of identification, where, Determined as a valid identification, This represents a multi-dimensional similarity factor, including peak feature similarity. Frequency feature similarity Waveform morphology similarity , This indicates adaptive weighting, dynamically assigned based on the adjuvant attribution label. This represents the signal quality correction factor. , This indicates the peak stability index of the signal. Indicates the uniqueness factor of the feature. , This represents the weighted similarity between the standard library and the signal to be identified. The number of non-target adjuvants.

[0097] Preferably, ,in, This represents the peak sequence of the signal to be identified. This indicates the peak sequence of the corresponding adjuvant in the standard library. , , These represent the frequency sequences of the signal to be identified and the standard signal, respectively. This represents the maximum DTW value for all auxiliaries in the standard library. Indicates the minimum value. , , These represent the normalized waveform sequences of the signal to be identified and the standard signal, respectively.

[0098] This embodiment achieves targeted matching of signal characteristics for different types of adjuvants, avoiding the limitations of single-dimensional similarity calculations. A signal quality correction factor is introduced into the formula. Establish data flow linkage with the preceding signal quality stabilization processing stage to ensure that confidence calculation reflects the stability of the signal itself, reducing the impact of noise or fluctuations on the recognition results; at the same time, incorporate feature uniqueness factors. This effectively avoids the risk of misjudgment caused by the confusion of similar adjuvant signals in the standard library. The identification confidence level is output through multi-parameter collaborative calculation. and with A value of ≥0.8 serves as an effective identification criterion, which can accurately quantify the reliability of the auxiliary agent identification results. This provides a clear and effective identification basis for subsequent calculation of content determination accuracy data and integration of component analysis data. It solves the problems of fixed weights, neglect of signal quality and feature uniqueness, and poor adaptability in traditional identification confidence calculation. This significantly improves the accuracy and stability of auxiliary agent identification and ensures the reliability of the identification link in the overall testing process.

[0099] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0100] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the multi-technology integrated rapid detection method for plastic additives as described above.

[0101] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0102] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0103] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A rapid detection method for plastic additives integrating multiple technologies, characterized in that, The method includes: The original signal features of various additives are obtained, and the original signal features are classified based on a preset matrix interference model to obtain a signal separation framework; Based on the aforementioned signal separation framework, the detection signal is decomposed to obtain preliminary independent signal features. The preliminary independent signal features are then corrected using an adaptive particle swarm optimization algorithm to obtain independent signal features for various adjuvants. Based on the independent signal characteristics, the interaction characteristic data between adjuvants are obtained and the main interference modes between adjuvants are determined. Random forest is used to predict the interference distribution of the main interference modes to obtain the extraction environment configuration scheme. The waste plastic samples were tested according to the extraction environment configuration scheme to obtain sample test data. The sample test data were compared with the determination benchmark value of the standard content sample to calculate the content determination accuracy data, and the component analysis data were integrated to obtain the component analysis data.

2. The method according to claim 1, characterized in that, The classification of the original signal features based on a preset matrix interference model to obtain a signal separation framework includes: Based on the original signal characteristics, a preset matrix interference model is used for feature classification processing to output preliminary classification labels; the preliminary classification labels are used to distinguish between the target adjuvant signal characteristics and the initial matrix interference signal characteristics. Independent component analysis (ICA) was used to separate the target additive signal component and the matrix interference signal component from the mixed signal in the preliminary classification label to obtain the independent additive signal. Set a residual interference threshold, determine whether the intensity of the residual matrix interference signal component in the independent additive signal exceeds the residual interference threshold. If it does, use a convolutional neural network to perform fine classification of the residual matrix interference component and output fine interference classification labels. An enhanced matrix interference model is constructed based on the refined interference classification labels. The independent additive signals are then subjected to secondary signal classification processing based on the enhanced matrix interference model, and an optimized signal separation framework is output.

3. The method according to claim 1, characterized in that, The detection signal is decomposed based on the signal separation framework to obtain preliminary independent signal features. These preliminary independent signal features are then refined using an adaptive particle swarm optimization algorithm to obtain independent signal features for various adjuvants, including: Based on the aforementioned signal separation framework, signal decomposition technology is used to process the superimposed interference in the detection signal, and preliminary independent signal characteristics of various additives are obtained. Based on the preliminary independent signal features, a density clustering algorithm is used to group the signal features, generating signal cluster groups and corresponding preliminary residual interference labels; The signal clusters are matched with entries in a preset signal feature association rule base. Combined with a signal waveform similarity measurement method, the additive attribution label and interference component type of each signal cluster are identified. The entries in the signal feature association rule base include additive type, signal feature parameter range, interference component identifier, and matching judgment conditions. A signal purity verification model is built based on the adjuvant attribution label and the type of interfering components. The signal purity evaluation value of the preliminary independent signal features of each cluster group is calculated through three indicators: signal peak stability, feature parameter consistency, and proportion of interfering components, and the purity evaluation result is generated. Based on the purity assessment results, adjuvant attribution labels, and interfering component types, an adaptive particle swarm optimization algorithm is used to dynamically adjust the core parameters of the signal decomposition technology, and to specifically optimize and correct the preliminary independent signal features to obtain the independent signal features of various adjuvants.

4. The method according to claim 3, characterized in that, The signal purity evaluation value of the preliminary independent signal features of each cluster group is calculated using the following formula: in, This represents the signal purity assessment value. This indicates the peak stability index of the signal. , The standard deviation of the signal peak sequence is represented by... This represents the mean of the signal peak sequence. This represents the consistency index of characteristic parameters. , Indicates the dimension of the feature parameters. This represents the angle between the corresponding characteristic parameters of the signal to be evaluated and the standard signal. Indicator of the proportion of interfering components , The signal energy representing the interfering component, This represents the total energy of the signal in the corresponding cluster group. , , This represents a dynamic weighting factor that adaptively adjusts based on the adjuvant attribution label and the type of interfering components. Indicates the calibration coefficient. It is used to correct the inherent biases of different additive signal types.

5. The method according to claim 1, characterized in that, The step of obtaining interaction feature data between adjuvants based on the independent signal features and determining the main interference modes between adjuvants, using random forest to predict the interference distribution of the main interference modes, and obtaining an extraction environment configuration scheme includes: Extract the correlation data reflecting the interaction between adjuvants from the independent signal features, and perform signal offset correction on the correlation data based on a preset correction algorithm to obtain the corrected signal feature values; Based on the corrected signal feature values, the main interference modes between the additives are determined by judging through a preset interference judgment threshold. The extraction condition control parameters, including extraction temperature, extraction time, and solvent ratio, are dynamically adjusted for the main interference modes to obtain optimized environmental configuration data. Set a predetermined standard for environment configuration, determine whether the optimized environment configuration data meets the predetermined standard, and if it does, generate a corresponding extraction optimization scheme. Based on the extraction optimization scheme, the random forest algorithm is used to analyze the interference trend among adjuvants and output the predicted interference distribution data. Based on the predicted interference distribution data, the fine-tuning parameters of the environment configuration are adjusted to generate the final extraction environment configuration scheme.

6. The method according to claim 1, characterized in that, The process involves obtaining sample testing data from waste plastic samples according to the described extraction environment configuration scheme, comparing the sample testing data with the determination benchmark value of the standard content sample to calculate the content determination accuracy data, and integrating the data to obtain component analysis data, including: Perform testing on waste plastic screening samples according to the extraction environment configuration scheme, and obtain the corresponding sample testing data; The sample detection data are quantitatively analyzed to determine the distribution characteristics of the additives, thereby obtaining the spatial distribution characteristics and content gradient information of the additives in the sample. A signal quality stabilization algorithm is used to suppress and calibrate the signal noise and fluctuation components in the spatial distribution characteristics and content gradient information to obtain stable signal output data. Based on the signal output data, the confidence level of each auxiliary agent signal feature is calculated by comparing it with a preset standard auxiliary agent signal feature library to obtain the auxiliary agent identification verification result. By comparing the determination benchmark values ​​of the standard content samples, the determination deviation, recovery rate and repeatability index of each auxiliary agent content are calculated, and the corresponding content determination accuracy data are obtained. By integrating the results of the adjuvant identification and verification with the content determination accuracy data, component analysis data including adjuvant type, distribution location, and content value are generated.

7. The method according to claim 6, characterized in that, The identification confidence level is calculated using the following formula: in, Indicates the confidence level of identification, where, Determined as a valid identification, This represents a multi-dimensional similarity factor, including peak feature similarity. Frequency feature similarity Waveform morphology similarity , This indicates adaptive weighting, dynamically assigned based on the adjuvant attribution label. This represents the signal quality correction factor. , This indicates the peak stability index of the signal. Indicates the uniqueness factor of the feature. , This represents the weighted similarity between the standard library and the signal to be identified. The number of non-target adjuvants.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images