62 results about "Linear discriminant analysis" patented technology

The invention relates to the technical field of intracranial pressure monitoring, in particular to a craniocerebral injury patient secondary brain injury early warning method based on multi-modal fusion, which comprises the following steps: acquiring intracranial pressure, cerebral blood flow, electroencephalogram signals and brain tissue oxygen partial pressure signals, carrying out denoising and normalization processing, constructing a multi-modal normalized time sequence data set, and carrying out multi-modal fusion on the basis of the multi-modal normalized time sequence data set; and inputting a linear discriminant analysis model to generate a time sequence difference vector, constructing a difference driving map based on a Bayesian dynamic network model, identifying abrupt change nodes, outputting a risk early warning section set, and carrying out trend stability analysis and reconstruction optimization. According to the method, the abnormal state recognition sensitivity is improved through normalized processing of multi-modal signals, enhancement of signal fusion capability, quantitative capture of direction change, reconstruction of spatial relationship between signals, modeling of dynamic transition probability, recognition of abrupt change nodes, combination of time difference vector change, evaluation of signal linkage trend, positioning of risk sections and extraction of parameter features; and the early warning and intervention precision is improved.

The invention discloses a lightweight optical fiber vibration intrusion event identification method and system for perimeter security and protection, and belongs to the technical field of optical fiber sensing technology and mode identification, and the method comprises the steps: collecting an original vibration signal of a perimeter monitoring region through a distributed optical fiber vibration sensing system; performing wavelet threshold de-noising preprocessing on the original vibration signal to obtain a de-noised signal; performing feature extraction on the denoised signal, and constructing a high-dimensional feature vector; performing dimension reduction processing on the high-dimensional feature vector by using a linear discriminant analysis method to obtain a low-dimensional classification feature vector; and inputting the low-dimensional classification feature vector into a pre-trained lightweight convolutional neural network model for classification and identification, and outputting a corresponding intrusion event category. According to the method, through cooperation of front-end LDA dimension reduction and a rear-end lightweight network, the model parameter quantity and calculation overhead are greatly reduced while high recognition precision is guaranteed, efficient real-time deployment on edge equipment is achieved, and the method is suitable for intrusion detection in perimeter security and protection.

The present invention relates to a technology for rest state-based user authentication by using an electromyogram signal and an inertia measurement device signal and provides a rest state-based user authentication device using an electromyogram signal and an inertia measurement device signal, which extracts a user's unique electromyogram signal and inertia measurement device signal during a rest state in which a specific operation is not performed, and inputs the extracted signal into a linear discriminant analysis and ensemble classification device to identify and authenticate the user, and a method thereof.

The invention relates to the technical field of urban rainfall pattern recognition, and aims to solve the problems that an existing method depends on an empirical formula, rainfall classification is unstable, and interpretability is insufficient. The method comprises the following steps: acquiring multi-site long-time actually measured rainfall data, and identifying an independent rainfall event based on a rainfall interval threshold value; multi-dimensional features such as rainfall duration and peak rainfall intensity are extracted and standardized; performing hierarchical clustering by adopting a Ward minimum variance method, and determining an optimal clustering number by combining a Clinski-Harabasz index and the like to obtain a rainfall mode; training a linear discriminant analysis (LDA) model by taking a clustering result as a label, and verifying and quickly classifying a new rainfall event; the result is used for flood control scheduling, flood risk assessment and the like. According to the method, rationality is improved based on measured data, classification stability is improved by fusing clustering and discriminant analysis, expandability is high, and intelligent disaster prevention is supported.

The present disclosure provides an array measuring method and interpretation device for ultrasonic detection of middle ear effusion, including an ultrasonic probe, an ultrasonic receiver, an analog-to-digital converter, and an analysis unit. The surface of the mastoid is divided into a plurality of measurement areas, and when ultrasonic is used for non-invasive detection of middle ear effusion, linear discriminant analysis is used for pre-training to find the best detection position and weighting parameters thereof to obtain the accurate evaluation value.

The present application relates to the field of artificial intelligence, and relates to a model trust region-based copyright protection method for smart grid deep learning models. The method comprises: acquiring a smart grid deep model set; for each model in the smart grid deep model set, searching an exclusive dataset for feature data samples having a model prediction value approximate to a model discrimination boundary, so as to obtain a trust region feature point set; performing dimensionality reduction on gradient vectors of the trust region feature point set on the model discrimination boundary according to a linear discriminant analysis method, so as to obtain perturbation vectors of the trust region feature point set; on the basis of predicted label changes of each model before and after the trust region feature point set is combined with the perturbation vectors, generating a model feature identifier set corresponding to the smart grid deep model set; and, on the basis of the model feature identifier set, training a copyright detection model to be trained, so as to obtain a pre-trained copyright detection model.

The invention provides a flower quality grading quality inspection method and system based on AI and image processing, and belongs to the technical field of computer vision and pattern recognizing.The method comprises the steps that an original image of the surface of a flower is collected through high-resolution imaging equipment, and high-frequency sub-band data representing tiny physical characteristics are extracted through multistage discrete wavelet transform; meanwhile, a sliding window is adopted to traverse the image, the color information entropy of a local area is calculated, and a color entropy graph is constructed. And after vectoring and splicing the two types of features, inputting the two types of features into a deep belief network model based on a multilayer restricted Boltzmann machine, and extracting deep abnormal feature codes. And finally, performing linear discriminant analysis on the code by utilizing a classification projection vector based on inter-class and intra-class distance optimization to generate an insect attack infection index, and realizing automatic grading quality inspection of the flower quality according to the insect attack infection index. According to the method, precise recognition and quantitative grading of tiny insect pests and recessive lesions on the surfaces of the fresh flowers are realized.

A method for rapidly detecting and identifying food-borne pathogens by using a two-dimensional / zero-dimensional MXene / GQDs Schottky heterojunction fluorescence sensor array based on machine learning comprises the following steps: synthesizing three nitrogen-doped graphene quantum dots through a hydrothermal method, determining the optimal excitation wavelength, concentration, reaction temperature and pH value of the nitrogen-doped graphene quantum dots, and further determining the fluorescence intensity of the nitrogen-doped graphene quantum dots under the optimal condition; constructing a fluorescence sensor array, mixing the three graphene quantum dots with target heavy metal ions, incubating, then measuring fluorescence intensity to obtain fluorescence response data, and normalizing the data to obtain the fluorescence response data. And analyzing by using machine learning methods of a decision tree, a support vector machine, K nearest neighbor, linear discriminant analysis, naive Bayes and a neural network. The prepared fluorescence sensor array can rapidly identify heavy metal ions, is excellent in accuracy, sensitivity and stability, provides an economical and efficient solution for heavy metal ion detection, and has remarkable application potential in environmental monitoring.

The application provides a lightweight identity recognition method combining voiceprint and earprint features. The method comprises the following steps: obtaining voice and ear canal echo signals of a known registered person and a person to be verified; extracting and fusing 13-dimensional mel-frequency cepstral coefficients (MFCC) of the voice and ear canal echo signals, i.e., obtaining voiceprint and earprint fusion features; inputting the fusion features into a lightweight identity recognition model to extract 128-dimensional embedding features of the known registered person and the person to be verified; calculating the similarity of the embedding features of the two types of persons by using a probabilistic linear discriminant analysis (PLDA) method; and determining whether the person to be verified is the known registered person according to the similarity. In the application, the fusion of earprint and voiceprint features improves the recognition performance of the classification model. The lightweight identity recognition model is obtained through a pre-training process, which reduces the equal error rate (EER) and greatly reduces the parameter quantity of the identity recognition model.

The invention discloses a heat-resistant steel failure diagnosis method based on a sound energy correction laser-induced breakdown spectroscopy technology, and belongs to the technical field of heat-resistant steel failure detection. The method comprises the following steps: firstly, establishing an LIBS experimental platform to synchronously acquire a spectral signal and a sound wave signal of a heat-resistant steel sample, and correcting an original spectrum in real time by utilizing a sound energy value obtained by sound wave main wave band integration to weaken a matrix effect and plasma fluctuation interference; carrying out dimensionality reduction on the corrected high-dimensional spectral data by adopting multiple threshold segmentation in combination with an average spectrum comparison mechanism; and finally, constructing a recursive feature elimination-linear discriminant analysis-particle swarm optimization driven radial basis function kernel support vector machine collaborative model to realize high-precision judgment of the aging grade of the heat-resistant steel. According to the method, the complementary advantage of the opto-acoustic multi-mode signals is utilized, and the method has the characteristics of nondestructive detection, strong anti-interference capability and good model generalization performance, can adapt to complex industrial scenes, and provides reliable technical support for safe operation and maintenance of high-temperature key parts in the industries of electric power, machinery and the like.

The application discloses a kind of composite gel breaking-directional adsorption purification combined special medical thickening component nitrate rapid detection method, belong to food detection technical field.The application is aimed at special medical food thickening component nitrate detection, colloid structure causes target extraction efficiency to be low, gel breaking effect and heat-sensitive component protection are difficult to give consideration to, method adaptability is poor, matrix interference is serious technical problem, first by LDA linear discriminant analysis model realizes sample classification and targeted gel breaking scheme matching, then adopts differentiating composite gel breaking system to cooperate entropy weight method-TOPSIS dynamic end point determination, by colloid polysaccharide characteristic fragment surface imprinting mesoporous nanometer silicon dioxide directional adsorption purification, finally by 15 N mark temperature-sensitive type isotopic internal standard is combined with three-wavelength in-situ calibration to complete quantitative detection.The method has good detection accuracy and repeatability, is suitable for multi-scene detection requirements, and can meet the quality control requirements of trace nitrate detection of special medical food.

The application discloses a kind of overhead transmission line insulator fault detection methods based on deep migration learning, including steps as follows: insulator image sample is obtained by unmanned aerial vehicle, and image sample is preprocessed;According to the method of migration learning, insulator image sample is divided into source domain and target domain, source domain is labeled normal insulator image sample, and target domain is unmarked insulator image sample of fault;Swin-Transformer is selected as feature extractor to extract the insulator features of source domain and target domain, and the insulator features are mapped to the same subspace, then the insulator features of the subspace are learned by linear discriminant analysis method, and then the nearest class prototype and structured prediction method is used to pseudo-label the target domain insulator fault sample subset;Finally, all insulator fault samples in the pseudo-labeled target domain are learned by iteration.The application can accurately and quickly detect the insulator fault in overhead transmission line.

The invention discloses a colorimetric sensing array based on hierarchical pore nano-enzyme as well as a construction method and application of the colorimetric sensing array, and belongs to the technical field of analysis and detection. According to the array, two cerium-based MOF nano-enzymes, namely micropore UiO-66 and hierarchical pore HP-UiO-66, are respectively combined with TMB (tetramethylbenzidine) and ABTS (2, 2, 6, 6-tetramethylbenzidine) chromogenic substrates to form a four-channel sensing system. The method comprises the following steps: synthesizing hierarchical pore HP-UiO-66 by a template method; collecting the absorbance of each channel under the characteristic wavelength by using an array, and generating a specific fingerprint spectrum of the to-be-detected object; by combining mode recognition technologies such as linear discriminant analysis and the like, rapid and synchronous distinguishing and quantitative detection of at least eight antioxidants with similar structures such as glycyrrhizic acid and caffeic acid in the traditional Chinese medicine decoction are realized. According to the method, the catalysis and mass transfer efficiency is improved by regulating and controlling the pore structure of the nano-enzyme, the problem of multi-target synchronous recognition in a complex matrix is solved by utilizing array cross response, and the method has the advantages of simplicity and convenience in operation, high flux and good accuracy.

The invention discloses a bearing fault classification method based on VMD-Kalman filtering fusion denoising and CNN-Transformer. The method comprises the following steps: 1) collecting original vibration signals of a bearing in normal and fault states; 2) decomposing an original signal through variational mode decomposition (VMD), screening an effective intrinsic mode function (IMFs), and eliminating a noise dominant mode; 3) performing secondary denoising on the effective IMFs by adopting Kalman filtering to obtain a fused denoised signal; 4) fusing signals through multiple strategies such as principal component analysis (PCA) and linear discriminant analysis (LDA), and screening an optimal fusion signal in combination with a three-dimensional evaluation criterion; and 5) constructing a CNN-Transform fusion model, cooperatively extracting local high-frequency features and global time sequence dependence features of the signals, and completing bearing fault classification. According to the method, noise is deeply suppressed by fusing a denoising technology, fault features are reserved, comprehensive feature extraction is realized by combining advantages of double models, the accuracy and robustness of bearing fault diagnosis under complex working conditions are remarkably improved, and the method is suitable for accurate fault diagnosis of bearings in industrial scenes.

This invention relates to a method for radio signal identification based on multidimensional feature extraction and feature fusion, belonging to the fields of signal recognition and deep learning technology. The invention collects radio signal data, segments the collected radio signals, and converts them into two-dimensional data. Four instantaneous features are extracted from the radio signals. Time-frequency transformation is performed on the radio signal data, and Shannon entropy feature values ​​are extracted. The extracted instantaneous features and time-frequency features are fused, and a linear discriminant analysis (LDA) dimensionality reduction algorithm is used to reduce the dimensionality of the fused feature data. The processed signal data is then input into a deep learning classification network model for signal identification to obtain its corresponding signal category. This invention combines the ideas of multidimensional feature extraction and feature fusion to achieve high-accuracy radio signal identification. Furthermore, by leveraging the advantages of deep learning technology, it can improve the speed and accuracy of radio signal identification.

The application discloses a PtIrRuCuPd high-entropy alloy nanoscale enzyme and a preparation method and application thereof, and belongs to the field of catalytic materials and analytical chemistry. The high-entropy alloy nanoscale enzyme is a nanoparticle composed of five elements of Pt, Ir, Ru, Cu and Pd, and has a quasi-spherical structure. The noble metal elements Pt, Ir, Ru and Pd are combined with the transition group metal element Cu, a multi-element synergistic effect is caused, the peroxidase activity, the oxidase activity and the laccase activity are improved, and the high-entropy alloy nanoscale enzyme has optimal comprehensive enzyme activity. The high-entropy alloy nanoscale enzyme prepared by the application realizes the synchronous enhancement of POD-like, OXD-like and LAC-like activities, a multi-channel sensing array is constructed, different responses of bisphenol substances are generated through different catalytic paths, and high-selectivity recognition and trace detection of bisphenol A in a complex sample are realized by combining hierarchical clustering analysis and linear discriminant analysis algorithms. The application provides a reference for the design of a novel multi-activity nanoscale enzyme and the development of an advanced sensing array.

The application relates to a method and system for establishing a prediction model of chemical liver toxicity, which comprises the following steps: (1) differentiating human embryonic stem cells or human induced pluripotent stem cells into hepatocyte-like cells and constructing a three-dimensional hepatocyte model; (2) taking miR-122, LDH and Cyto C in the three-dimensional hepatocyte as combined test indexes to judge liver toxicity of a liver toxicity mode compound; (3) classifying the liver toxicity mode compound into three categories through two canonical discriminant functions by a linear discriminant analysis modeling method; and (4) converting the canonical discriminant function into a Fishers discriminant function to obtain a liver toxicity prediction model. The application has the general characteristics of a conventional cell model for evaluating compounds, and can also evaluate the toxicity effect and health risk of the compounds in a high-throughput manner with relatively less manpower.

The invention provides a self-adaptive adjustment control method, device and equipment for an intelligent valve and a medium, and relates to the field of automatic control, and the method comprises the steps that according to the valve type or fault monitoring requirement, a multi-channel synchronous collection mode is adopted to obtain valve operation parameters; constructing a time domain-frequency domain-time-frequency domain three-dimensional feature extraction system, and extracting feature data of the valve operation parameters; performing selection and dimension reduction processing on the feature data through correlation analysis, principal component analysis or linear discriminant analysis to obtain dimension reduction data; based on the dimension reduction data, adopting three-level identification logic of threshold judgment, rule matching and clustering analysis, and combining diagnosis confidence to determine a fault diagnosis result; and based on the fault diagnosis result, a control signal is optimized through self-adaptive filtering, and self-adaptive adjustment control of the valve is achieved. In this way, the intelligent level of valve control can be remarkably improved, manual intervention is reduced, and the stability and reliability of valve operation are improved.

There is a desire for dimension compression of training data by linear discriminant analysis to be achieved by secure computation. An AI analysis model generation system according to the disclosed technology comprises a secure computation device and a user terminal. The secure computation device acquires data with n rows and m columns encrypted so as to be able to be securely computed, the data comprising n m-dimensional vectors to which class classifications are assigned. For each class, encrypted vectors belonging to the class are securely computed to obtain an encrypted sum vector and an encrypted sum-of-products matrix. The user terminal decrypts the encrypted sum vector and the encrypted sum-of-products matrix to obtain a plaintext sum vector and a plaintext sum-of-products matrix, uses the plaintext sum vector and the plaintext sum-of-products matrix to obtain a within-class scatter matrix and a between-class scatter matrix, and uses the within-class scatter matrix and the between-class scatter matrix to obtain a transformation matrix with m rows and d columns.

The invention discloses a ratio-type fluorescent probe based on dual-emission europium-containing carbon dots, a preparation method and application of the ratio-type fluorescent probe in detection of tetracycline antibiotics, and belongs to the technical field of fluorescent probes. Histidine and europium nitrate are used as precursors, the fluorescent powder is obtained through one-step synthesis with an ethanol-water solvothermal method, and the fluorescent powder has blue fluorescence located at 440 nm and Eu < 3 + > characteristic red fluorescence located at 615 nm. When the probe acts with oxytetracycline, the oxytetracycline as an antenna ligand and Eu < 3 + > are subjected to efficient energy transfer, so that red fluorescence at 615 nm is remarkably enhanced, and blue fluorescence at 440 nm is kept stable, thereby realizing ratio-dependent fluorescence response along with visual luminescence color change from blue to red luminescence. According to the method, fluorescence of the probe is combined with ultraviolet-visible absorption spectrum data, a multi-signal sensing platform is constructed by utilizing linear discriminant analysis, and accurate identification and quantitative analysis of various tetracycline antibiotics are successfully realized.

The invention relates to the technical field of water environment science, in particular to a lake wetland extreme hydrological situation quick response method and system.The method comprises the following steps that the light absorption spectrum and the three-dimensional fluorescence spectrum of colored soluble organic matter in a lake wetland are obtained, and optical characteristic parameters are obtained; obtaining an optical characteristic standardized data set as an input variable according to the optical characteristic parameters, and converting an extreme hydrological drought index and a flood index into binary processing variables as output variables; establishing a random forest model based on the input variables, and screening the optical characteristic parameters to obtain key optical parameters; and constructing a linear discriminant analysis model, obtaining the output variables corresponding to the key optical parameters, and realizing quick response to the extreme hydrological situation of the lake wetland. According to the invention, efficient and automatic identification of the lake wetland extreme hydrological events is realized, and technical support is provided for lake wetland management and disaster early warning.

The present invention proposes an algorithm that combines K-nearest neighbors (KNN), linear discriminant analysis (LDA), and linear regression (LR) models to achieve high recall performance in credit card fraud detection. By applying additional conditional instructions based on the predicted values ​​of each model to derive the final prediction, the invention achieves higher recall performance compared to individual models. This invention can be applied to various datasets, thereby improving the efficiency of credit card fraud detection systems and reducing financial losses.

The invention provides a structure prediction model suitable for alloy steel and an establishment method and application thereof, and the establishment method comprises the following steps: S1, obtaining component and structure type data of a plurality of steel grades covered by a component range, and calculating the statistical characteristics of each element by taking the mass fraction of the alloy element as a characteristic variable; s2, data processing is conducted, a decision boundary is obtained, and structure types are divided into single-phase austenite, austenite and ferrite or martensite, ferrite and / or martensite; and S3, predicting the model: performing dimensionality reduction on the obtained data by adopting the linear discriminant analysis in the step S2, combining softmax regression, performing linear discriminant analysis dimensionality reduction to obtain two LDA expressions, and obtaining a shaffler-like tissue prediction map according to the decision boundary straight lines of the corresponding three tissue types. Most material organizations can be correctly classified through the method, the accuracy is high, and the method is convenient to popularize and use in new materials and design components.

The invention discloses a copper-based nano-enzyme, a colorimetric sensing array and application of the copper-based nano-enzyme and the colorimetric sensing array, and belongs to the technical field of nano-material preparation. The copper-based nano-enzyme is prepared through a simple coprecipitation method, and sheet-shaped nano-enzyme (the molar ratio is 1: 55. 3) or circular-ring-shaped nano-enzyme (the molar ratio is 1: 6.176) with regular morphology can be prepared respectively by accurately regulating and controlling the molar ratio of metal salt to 2-methylimidazole. A four-channel colorimetric sensing array is constructed on the basis of the two nano enzymes, and high-selectivity distinguishing and quantitative detection of seven phenolic acids such as ferulic acid and chlorogenic acid are realized by capturing differential absorbance signals generated by catalytic oxidation of TMB (tetramethylbenzidine) under the wavelengths of 370 nm and 652 nm and combining chemometrics methods such as linear discriminant analysis (LDA). The method is simple and convenient to operate, high in sensitivity and strong in anti-interference capability, and is successfully applied to detection and distinguishing of phenolic acids in actual traditional Chinese medicine samples.

The invention belongs to the technical field of agricultural product variety identification, and particularly relates to a Tieguanyin variety identification method by combining high-resolution mass spectrometry with linear discriminant analysis. According to the method, 65 organic components in a tea leaf extracting solution are analyzed by adopting high-resolution mass spectrometry, characteristic components of the Tieguanyin variety are screened out by utilizing single-factor variance analysis, and a Tieguanyin variety discrimination model is established by combining linear discriminant analysis; the initial discrimination accuracy of the model reaches 99.0%, the cross validation accuracy is 97.1%, and the Tieguanyin variety can be effectively distinguished. The method provides important technical support for protection of the geographical indication product Anxi Tieguanyin, and has good practical application value.

The application relates to the technical field of odor recognition, and discloses a gas recognition method, device, equipment and medium suitable for a home environment, the method comprising the following steps: acquiring an odor real-time measurement value collected by a multi-factor gas-sensitive sensor, and constructing a real-time measurement value curve based on the odor real-time measurement value; performing legality verification on the real-time measurement value curve, and generating a virtual measurement value sequence based on a legality verification result; configuring a kernel function on the virtual measurement value sequence, and generating a kernel space distribution; and generating a home environment gas recognition result by using a kernel linear discriminant analysis method based on the kernel space distribution. The application realizes accurate recognition of gases in a home environment.

The application discloses a light-weighted optical fiber vibration intrusion event identification method and system for perimeter security, belongs to the field of optical fiber sensing technology and pattern recognition technology, and comprises the following steps: collecting original vibration signals of a perimeter monitoring area through a distributed optical fiber vibration sensing system; performing wavelet threshold denoising pretreatment on the original vibration signals to obtain denoised signals; performing feature extraction on the denoised signals to construct a high-dimensional feature vector; performing dimension reduction processing on the high-dimensional feature vector by using a linear discriminant analysis method to obtain a low-dimensional classification feature vector; inputting the low-dimensional classification feature vector into a pre-trained light-weighted convolutional neural network model for classification and recognition, and outputting corresponding intrusion event categories. The application cooperates the front-end LDA dimension reduction with the rear-end light-weighted network, greatly reduces the model parameter quantity and the calculation cost while ensuring high recognition accuracy, realizes efficient and real-time deployment on an edge device, and is suitable for intrusion detection in perimeter security.

The application discloses a single tobacco producing area and part identification method, a substitution method and a system, wherein the producing area and part identification method comprises the following steps: dividing single tobacco into multiple categories and setting category labels; constructing a training set; constructing a near-infrared spectrum matrix based on the near-infrared spectrum in the training set and performing standardization processing, calculating the covariance matrix of the standardized near-infrared spectrum matrix X; calculating the eigenvalue and eigenvector of the covariance matrix and performing sorting, taking the eigenvectors corresponding to the first k eigenvalues to form a dimension reduction matrix W; taking X*W and the corresponding category labels as training data to perform linear discriminant analysis classification training, and obtaining a single tobacco producing area and part identification model; and predicting the producing area and tobacco leaf part grade of a single tobacco to be predicted based on the single tobacco producing area and part identification model. The category label containing the single tobacco producing area and tobacco leaf part grade information can be predicted, the single tobacco producing area and part are accurately identified, and thus the accurate and efficient substitution of single tobacco in a formula leaf group is achieved.

This invention discloses an electronic nose system for kidney cancer detection, belonging to the field of medical testing technology. This invention aims to address the problems of existing kidney cancer screening methods, such as invasiveness, insufficient sensitivity, and difficulty in accurately distinguishing between kidney cancer patients, healthy individuals, and post-operative patients. The electronic nose system of this invention includes: a gas acquisition unit, a sensor array unit composed of multiple gas sensors and temperature and humidity sensors, a detection circuit unit, a main control unit, an output unit, and a data processing and classification unit. During detection, exhaled air is passed into the sensor array, multi-dimensional response features are collected, and after dimensionality reduction through linear discriminant analysis, a machine learning classification model is used to output the discrimination result. This invention achieves non-invasive, rapid, and low-cost screening for kidney cancer, with a classification accuracy rate of over 95% in real exhaled breath sample tests. The system also exhibits strong robustness and has good clinical applicability and prospects for widespread application.

The invention discloses a traditional Chinese medicinal material adulteration identification method based on feature selection and linear analysis, and relates to the technical field of intelligent identification of traditional Chinese medicinal materials. According to the method, rapid nondestructive detection is realized, the characteristics of high penetrability and low photon energy of terahertz waves are utilized, complex sample pretreatment is not needed, sample integrity is kept, detection efficiency is remarkably improved, accurate quantitative identification is realized, gradient adulterated samples are designed through a system, and high-specificity terahertz fingerprint spectrums are combined, so that the detection accuracy is improved. Accurate identification and distinguishing of adulterant types and adulteration proportions are realized, model robustness and generalization ability are improved, a random forest (RF) feature dimension reduction method is introduced to eliminate spectrum data redundancy, machine learning models of different principles of linear discriminant analysis and neural network classification are fused, a robust classifier is constructed, the problem of small sample overfitting is solved, and the accuracy of the model is improved. And the reliability of the method on samples with different batches and different sources is ensured.