Dynamic Time Warping Research Articles

Similarity analyses of causative viruses for chronic obstructive pulmonary disease and asthma exacerbations

BackgroundThe representativeness of cohort studies compared to nationwide data is a major concern. This study evaluated the similarity and seasonality of causative respiratory viruses for chronic obstructive pulmonary disease (COPD) and asthma exacerbations between retrospective multicenter cohort study and nationwide data.MethodsWe compared data from the retrospective multicenter cohort study with Korean Influenza and Respiratory Surveillance System data between 2015 and 2018. Correlation, dynamic time warping (DTW), and seasonal autoregressive integrated moving average (SARIMA) analyses were performed.ResultsSpearman correlation coefficients [ρ] indicated very strong (respiratory syncytial virus [RSV] [ρ = 0.8458] and influenza virus [IFV] [ρ = 0.8272]), strong (human metapneumovirus [HMPV] [ρ = 0.7177] and parainfluenza virus [PIV] [ρ = 0.6742]), and moderate (rhinovirus [RV] [ρ = 0.5850] and human coronavirus [HCoV] [ρ = 0.5158]) correlations. DTW analyses showed moderate (PIV) and high (IFV, RSV, and HMPV) synchronicity between the two datasets, while RV and HCoV showed low synchronicity. SARIMA analyses revealed 12-month seasonality for IFV, RSV, PIV, and HMPV. The peak season was winter for RSV and IFV, spring to summer for PIV, and spring for HMPV.ConclusionsThis was the first study to report the synchronicity between a retrospective multicenter cohort study of viruses that can cause COPD or asthma exacerbations and nationwide surveillance system data.

MAFNet: Multimodal Asymmetric Fusion Network for Radar Echo Extrapolation

Radar echo extrapolation (REE) is a crucial method for convective nowcasting, and current deep learning (DL)-based methods for REE have shown significant potential in severe weather forecasting tasks. Existing DL-based REE methods use extensive historical radar data to learn the evolution patterns of echoes, they tend to suffer from low accuracy. This is because data of radar modality face difficulty adequately representing the state of weather systems. Inspired by multimodal learning and traditional numerical weather prediction (NWP) methods, we propose a Multimodal Asymmetric Fusion Network (MAFNet) for REE, which uses data from radar modality to model echo evolution, and data from satellite and ground observation modalities to model the background field of weather systems, collectively guiding echo extrapolation. In the MAFNet, we first extract overall convective features through a global shared encoder (GSE), followed by two branches of local modality encoder (LME) and local correlation encoders (LCEs) that extract convective features from radar, satellite, and ground observation modalities. We employ an multimodal asymmetric fusion module (MAFM) to fuse multimodal features at different scales and feature levels, enhancing radar echo extrapolation performance. Additionally, to address the temporal resolution differences in multimodal data, we design a time alignment module based on dynamic time warping (DTW), which aligns multimodal feature sequences temporally. Experimental results demonstrate that compared to state-of-the-art (SOTA) models, the MAFNet achieves average improvements of 1.86% in CSI and 3.18% in HSS on the MeteoNet dataset, and average improvements of 4.84% in CSI and 2.38% in HSS on the RAIN-F dataset.

Analysis of congestion key parameters, dynamic discharge process, and capacity estimation at urban freeway bottlenecks: a case study in Beijing, China

ABSTRACT Recurring bottlenecks significantly contribute to urban freeway congestion, making their analysis essential. This study examines six bottlenecks on Beijing’s Ring Road using multi-day data, identifying them via Dynamic Time Warping and Fuzzy C-Means Clustering (DTW+FCM). Key parameters—free-flow speed, critical speed, critical density, and jam density—are calibrated using fundamental diagram models. The Weibull distribution analyzes flow and speed patterns during congestion phases. The DTW+FCM method effectively identified bottlenecks and congestion levels. Severe congestion lasting over 10 hours on the West Second and Third Ring Roads averaged speeds of 15 km/h. The S3 model best fits data for the West Ring Roads, while the Van Aerde model suits the North Ring Roads. Different methods yield varying traffic capacity estimates, highlighting the need for nuanced approaches in urban expressway planning to maintain traffic quality and comfort. These findings offer valuable guidance for research and practical traffic management solutions.

Building electricity load forecasting based on spatiotemporal correlation and electricity consumption behavior information

Accurate prediction of building electricity load is essential for grid management and building optimization operations. This paper proposes a novel approach based on spatiotemporal correlations and electricity consumption behavior information. The K-Medoids algorithm and the Derivative Dynamic Time Warping (DDTW) distance are employed to explore the correlation between electricity consumption behaviors among different partitions and floors. Different partitions and floors are clustered and grouped, followed by modifying the adjacency matrix with electricity consumption behaviors. The hybrid model and K-Medoids-LSTM model are proposed separately for clusterable nodes and non-clustered nodes. For clusterable nodes, spatial-temporal features are extracted, trained, and predicted with the hybrid model based on graph neural networks (GNNs) and LSTM models. A K-Medoids-LSTM model based on the K-Medoids algorithm is proposed to predict the electricity load of the non-clustered nodes. To explore the model's practicality, we predicted the building electrical load under different dataset sizes. The model achieves an R2 above 0.89, and the MAE, MSE, and RMSE of the GCN-LSTM and GAT-LSTM models all remain below 0.1, indicating strong predictive capabilities. The results demonstrate that, without relying on other external features, the proposed method can accurately predict the building electricity load for different partitions and floors simultaneously.

Development and Application of an Advanced Automatic Identification System (AIS)-Based Ship Trajectory Extraction Framework for Maritime Traffic Analysis

This study addresses the challenges of maritime traffic management in the western waters of Taiwan, a region characterized by substantial commercial shipping activity and ongoing environmental development. Using 2023 Automatic Identification System (AIS) data, this study develops a robust feature extraction framework involving data cleaning, anomaly trajectory point detection, trajectory compression, and advanced processing techniques. Dynamic Time Warping (DTW) and the Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) algorithms are applied to cluster the trajectory data, revealing 16 distinct maritime traffic patterns, key navigation routes, and intersections. The findings provide fresh perspectives on analyzing maritime traffic, identifying high-risk areas, and informing safety and spatial planning. In practical applications, the results help navigators optimize route planning, improve resource allocation for maritime authorities, and inform the development of infrastructure and navigational aids. Furthermore, these outcomes are essential for detecting abnormal ship behavior, and they highlight the potential of route extraction in maritime surveillance.

Research on the railway multi-source homonymous geographical entity matching algorithm based on dynamic time warping

This study aims to explore an efficient technique for matching multisource homonymous geographical entities in railways to address the identification issues of homonymous geographical entities. Focusing on railway line vector spatial data, this research investigates the matching problem of multisource homonymous geographical entities. Building on statistical feature matching of attribute data, a curve similarity calculation method based on the DTW algorithm is designed to achieve better local elastic matching, overcoming the limitations of the Fréchet algorithm. The empirical study utilizes railway line layer data from two data sources within Beijing’s jurisdiction, fusing 6237 segment lines from source 2 with 105 long lines from source 1. The structural comparison between the two data sources is conducted through statistical methods, applying cosine similarity and the maximum similarity value of TF-IDF for text similarity calculation. Finally, Python is used to implement the DTW algorithm for curve similarity. The experimental results show an average DTW distance of 3.92, a standard deviation of 4.63, and a mode of 0.005. Similarity measurement results indicate that 95.53% of records are within the predetermined threshold, demonstrating the effectiveness and applicability of the method. The findings significantly enhance the accuracy of railway data matching, promoting the informatization of the railway industry, and hold substantial significance for improving railway operational efficiency and system performance.

Identification of Similar Weather Scenes in Terminal Areas Based on Multiresolution Spatiotemporal Windows

To accurately analyze the influence of similar weather scenes in the terminal area, a framework is proposed for identifying such scenarios based on the Multiresolution Spatiotemporal Window (MRSTW). The goal is to analyze the impact of similar weather patterns. This paper introduces a simple and effective method called the rasterized weather severity index (WSI) to reduce the dimensionality of data used for extracting air transport weather features, which can cause the loss of spatial information in an image. Additionally, the paper uses Dynamic Time Warping (DTW) and the Fuzzy C-mean (FCM) clustering algorithm to cluster time-series scenes influenced by convective weather in an unsupervised manner on a daily basis. The most similar weather scenes are then identified by searching for the same cluster within a multiresolution spatiotemporal window, using 4 h weather scenes as typical examples. Finally, the framework analyzes the impact of weather scenes on the operation of terminal area approach traffic flow by combining trajectory data. The findings demonstrate that this framework can effectively identify similar weather scenes and provide a more accurate reflection of their impact on the operation of terminal area approach traffic flow.

Interpreting the spatial distribution of soil properties with a physically-based distributed hydrological model

Digital soil maps are commonly data-driven as the development of physically-based models for soil mapping is difficult due to the complexity of soils. However, physically-based hydrologic models have been successful in simulating water dynamics. Since water movement is a major driver of pedogenesis, the physical rules that govern water movement might help explain and predict the spatial variation of soil properties. Here, we demonstrate the novel use of a physically-based, distributed hydrologic model to inform the spatial distribution of soil properties. The Distributed Hydrology Soil Vegetation Model (DHSVM) was utilized to simulate soil moisture content (SM) and water table depth (WTD) in two hillslope catchments under pasture and forest management wherein hydrologic model outputs were then compared with soil properties measured in situ. SM sensors and wells were installed in both catchments to validate simulations of soil water movement via Nash-Sutcliffe Efficiency (E). In-situ observations were made at 87 sites within both catchments to study the connection between simulated water movement (SM and WTD) and observed soil properties, namely the depth and thickness of the argillic (Bt), fragic (Btx), and C horizons, and the depth of redoximorphic features. The simulated time series of SM and WTD were also clustered per season using Dynamic Time Warping (DTW), which identified similarity among time series at varying timescales. Model validation suggested that simulations of surficial SM (0–20 cm) were reasonable (E = 0.45), however, simulated subsurface SM (45–60 cm) and WTD were not sufficiently accurate. The thickness of Btx horizons were spatially grouped into different populations by SM clusters from every season except spring. For the other properties, only SM dynamics of specific seasons grouped into significantly different populations, suggesting that the explanatory power of simulated water movement varies seasonally and was greater during winter. Here, we show clusters of simulated SM separated soil properties into statistically different populations, showing that hydrologic models could inform areas that followed different water dynamics related to pedogenic trajectories and related biogeochemical processes not necessarily simulated by the model. As such, physically-based modeling of water dynamics can, therefore, inform and advance digital soil mapping by linking water movement patterns stemming from hydrologic model outputs to spatial patterns of soil properties and pedogenesis.

Capacity estimation of lithium-ion battery based on soft dynamic time warping, stratified random sampling and pruned residual neural networks

Accurate battery capacity estimation is crucial for ensuring battery management systems' safe and reliable operation. Although deep learning algorithms have been widely applied in the field of image recognition, their application in battery diagnosis is relatively limited. Besides, obtaining complete cycling data during charging/discharging processes can be challenging in practical applications. This paper proposes a battery capacity estimation method based on partial charge curves and pruned residual neural networks. To leverage the aging information inherent in partial charge curves, the partial segment battery data is transformed into images using wavelet denoising and soft dynamic time warping algorithms. Subsequently, we present a stratified random sampling method to address the issue of imbalanced sample distributions in battery dataset partitioning, and employ residual neural networks to mitigate performance degradation caused by stacking multiple network layers. To validate the effectiveness of the proposed method, experiments are conducted on three types of battery data. During the experimental phase, the root mean square error and mean absolute error of estimated battery capacity are both below 0.79% and 0.54%. Furthermore, a hybrid-pruning algorithm is proposed to enable the deployment of the model on the battery management system, which reduces the model size and computational complexity without compromising accuracy. Compared to other pruning algorithms, the proposed hybrid-pruning algorithm outperforms diagnostic accuracy and computational efficiency after fine-tuning, reducing the model size by 77.40% and creating a more compact structure.

Reservoir oriented migrated seismic image improvement and poststack seismic inversion using well-seismic mistie

In seismic exploration of subsurface hydrocarbon reservoirs, migrated seismic image is the foundation for structural interpretation and subsequent seismic inversion for reservoir properties. Seismic events in migrated images can be mispositioned due to several factors including seismic velocity inaccuracy. The mispositioned events lead to poor well-seismic tie and degrades the accuracy of subsequent inversion and reservoir prediction. Particularly, the mistie can be quite obvious for the mid- and mature- phases of the reservoir development after many wells drilled. We propose to utilize the information from mistie between seismic images and synthetic traces to improve the accuracy of migrated seismic images and thus inversion results. Our method includes three major steps: firstly, calculate time shifts between the synthetic traces and the co-located seismic trace using the path-limited dynamic time warping (PDTW) algorithm; then compute the time shift volume for the entire migrated image using the seismic-driven modeling approach based on the time shift traces at the well locations; apply the time shift volume to update the migrated image. The improved seismic volume can be used for further processing such as seismic inversion, which makes the inversion results more consistent with seismic and well data. We apply our method to synthetic and field datasets. The results demonstrate that the updated seismic image has an improved seismic well tie and an improved inversion result. Our method indicates a venue for a better honor of seismic and well data in reservoir characterization.

Toroidal modified Miller-Turner CME model in EUHFORIA: Validation and comparison with flux rope and spheromak

Rising concerns about the impact of space-weather-related disruptions demand modelling and reliable forecasting of coronal mass ejection (CME) impacts. In this study, we demonstrate the application of the modified Miller-Turner (mMT) model implemented within EUropean Heliospheric FORecasting Information Asset (EUHFORIA) in forecasting the geo-effectiveness of observed coronal mass ejection (CME) events in the heliosphere. Our goal is to develop a model that not only has a global geometry, in order to improve overall forecasting, but is also fast enough for operational space-weather forecasting. We test the original full torus implementation and introduce a new three-fourths Torus version called the Horseshoe CME model. This new model has a more realistic CME geometry, and overcomes the inaccuracies of the full torus geometry. We constrain the torus geometrical and magnetic field parameters using observed signatures of the CMEs before, during, and after the eruption. We perform EUHFORIA simulations for two validation cases -- the isolated CME event of 12 July 2012 and the CME--CME interaction event of 8-10 September 2014. We performed an assessment of the model's capability to predict the most important $B_z$ component using the advanced dynamic time-warping (DTW) technique. The Horseshoe model predictions of CME arrival time and geo-effectiveness for both validation events compare well with the observations and are weighed against the results obtained with the spheromak and FRi3D models, which were already available in EUHFORIA. The runtime of the Horseshoe model simulations is close to that of the spheromak model, which is suitable for operational space weather forecasting. However, the capability of the magnetic field prediction at 1 AU of the Horseshoe model is close to that of the FRi3D model. In addition, we demonstrate that the Horseshoe CME model can be used for simulating successive CMEs in EUHFORIA, overcoming a limitation of the FRi3D model.

Full Waveform Inversion Based on Dynamic Time Warping and Application to Reveal the Crustal Structure of Western Yunnan, Southwest China

AbstractWe develop a 3D full waveform inversion (FWI) method based on dynamic time warping (DTW) to address the issue of cycle‐skipping, which can prohibit the convergence of conventional FWI methods. DTW globally compares data samples at different time steps in 2D matrices against the time shifts of waveforms. We introduce the concept of shape descriptors into softDTW, creating a soft‐shapeDTW objective function within our waveform inversion process to improve alignment accuracy. Additionally, including constraints from Sakoe‐Chiba bands in the inversion further enhances efficiency and overall performance. A synthetic test has shown that the soft‐shapeDTW inversion outperforms conventional waveform inversions in overcoming the cycle‐skipping challenges that arise from poor initial models. This method was applied to fit observed seismograms to reveal western Yunnan's crustal structure. Seismic waveforms were recorded by 88 broadband stations from 10 local earthquakes, which were then denoised using a continuous wavelet transform method. Generalized cut and paste waveform inversions were used to determine the source parameters of these seismic events. Our inversion well‐aligned various seismic phases in the selected time windows of seismograms, and the resolved velocity models well associate with local geological structure. Results suggest that the soft‐shapeDTW inversion offers a robust alternative to FWI, reducing the reliance on accurate starting models.

Temporal Directionality Between Symptoms During Treatment of Depressed Inpatients: A Dynamic Time Warp Network Analysis.

Depression is generally perceived from the perspective of the common-cause disease model. However, the network perspective assumes mutual influence of individual symptoms and stresses the importance of investigating symptom dynamics. Gaining a better understanding of symptom dynamics within individuals might contribute to more effective treatments. Current exploratory longitudinal research studied the associations and directionality between 43 symptoms from the generic questionnaire Symptom Questionnaire-48 (SQ-48) using dynamic time warp (DTW) analyses, in which trajectories with similar time-dependent patterns can be identified. Data from individuals were analysed first, yielding distance matrices for all symptom trajectories, after which the data were aggregated. The 148 included patients were all admitted for the treatment of their clinical depression. Undirected DTW analyses of three patients with longer time series but otherwise randomly chosen showed large variability among individuals. Group-level undirected DTW analyses showed numerous significant edges between symptoms, largely clustering symptoms according to the eight pre-existing subscales of the SQ-48. Group-level directed DTW analyses showed five symptoms with significant outstrength: 'hopeless', 'restless', 'down/depressed', 'feeling tense' and 'no enjoyment', meaning that change in these key symptoms preceded change in other symptoms. The 43 included symptoms of the SQ-48 primarily focus on internalizing problems in severely depressed inpatients, potentially limiting generalizability. DTW networks provided us with five key symptoms based on the dynamics of symptom scores. Future studies could explore whether process-based therapy targeted at symptoms with high outstrength might result in more effectivity as part of personalized treatment.

Label adversarial domain adaptation network for predicting remaining useful life based on cross-domain condition

Most data-driven methods for predicting remaining useful life assume that the data under different operating conditions follow the same distribution. However, this assumption rarely holds in real-world situation. Additionally, traditional methods do not fully utilize the hidden label information from the target domain or account for the transfer quality of source domain data. To address these issues, Label Adversarial Domain Adaptation (LADA) network is introduced in this paper. Specifically, LADA aims to filter the source domain data and maximize the use of hidden label information from the target domain. Firstly, a similarity measurement indicator based on the pearson correlation coefficient (PCC) and dynamic time warping (DTW) is employed to filter source domain data similar to the target domain data distribution. Then, in order to fully utilize the hidden label information from the target domain, the cloud model and golden section are utilized to create pseudo class labels. Furthermore, a feature difference module is established that minimizes the disparity between domain features. This is realized by using the maximum mean difference (MMD) and Kolmogorov–Smirnov (K–S) statistical test. The experimental results indicate that LADA has advantages in cross-domain RUL prediction.

Effect of Gait Speed and Dynamic Time-Warping on the prediction of Lower-Limb Joint Angles

Effect of Gait Speed and Dynamic Time-Warping on the prediction of Lower-Limb Joint Angles

Spatial–Temporal Similarity Fusion Graph Adversarial Convolutional Networks for traffic flow forecasting

Traffic flow forecasting is integral to the advancement of intelligent transportation systems and the development of smart cities. This paper introduces a novel model, the Spatial–Temporal Similarity Fusion Graphs Adversarial Convolutional Networks (STSF-GACN), which leverages advanced data preprocessing techniques to enhance the predictive accuracy and efficiency of traffic flow forecasting.The innovation of our approach lies in the meticulous construction of the spatial–temporal similarity matrix through the precise calculation of temporal and spatial similarities. This matrix forms the backbone of our model, serving as the generator in the integrated Generative Adversarial Network (GAN) architecture. The Spatial–Temporal Similarity Fusion Adaptive Graph Convolutional Network, developed as part of our GAN’s generator, utilizes cutting-edge techniques such as the Wasserstein distance and Dynamic Time Warping to optimize the adaptive adjacency matrix, enabling the model to capture latent spatial–temporal correlations with unprecedented depth and precision.The discriminator of the GAN further refines the model by evaluating the accuracy of the traffic predictions, ensuring that the generative model produces results that are not only accurate but also robust against varying traffic conditions. This cohesive integration of GAN into the model architecture allows for a significant improvement in prediction accuracy and convergence speed, moving beyond traditional forecasting methods.

Physicochemical Evaluation of Remote Homology in the Twilight Zone.

A fundamental problem in the field of protein evolutionary biology is determining the degree and nature of evolutionary relatedness among homologous proteins that have diverged to a point where they share less than 30% amino acid identity yet retain similar structures and/or functions. Such proteins are said to lie within the "Twilight Zone" of amino acid identity. Many researchers have leveraged experimentally determined structures in the quest to classify proteins in the Twilight Zone. Such endeavors can be highly time consuming and prohibitively expensive for large-scale analyses. Motivated by this problem, here we use molecular weight-hydrophobicity physicochemical dynamic time warping (MWHP DTW) to quantify similarity of simulated and real-world homologous protein domains. MWHP DTW is a physicochemical method requiring only the amino acid sequence to quantify similarity of related proteins and is particularly useful in determining similarity within the Twilight Zone due to its resilience to primary sequence substitution saturation. This is a step forward in determination of the relatedness among Twilight Zone proteins and most notably allows for the discrimination of random similarity and true homology in the 0%-20% identity range. This method was previously presented expeditiously just after the outbreak of COVID-19 because it was able to functionally cluster ACE2-binding betacoronavirus receptor binding domains (RBDs), a task that has been elusive using standard techniques. Here we show that one reason that MWHP DTW is an effective technique for comparisons within the Twilight Zone is because it can uncover hidden homology by exploiting physicochemical conservation, a problem that protein sequence alignment algorithms are inherently incapable of addressing within the Twilight Zone. Further, we present an extended definition of the Twilight Zone that incorporates the dynamic relationship between structural, physicochemical, and sequence-based metrics.

Transient modeling of stratified thermal storage tanks: Comparison of 1D models and the Advanced Flowrate Distribution method

Thermal energy storage (TES) is one of the key technologies for enabling a higher deployment of renewable energy. In this context, the present study analyzes the modeling strategies of one of the most common TES systems: stratified thermal storage tanks. These systems are essential to many solar thermal installations and heat pumps, among other clean energy technologies. Three different one-dimensional tank models are compared by their computing speed and resilience to long time steps. Two of the models analyzed are numerical, one being explicit and the other one implicit, and the other is analytical. The models are validated against data from experiments carried out considering small-scale stratified tanks, showing that their performance can be improved by using the Advanced Flowrate Distribution (AFD) method. The results show that the analytical model maintains its accuracy with longer time steps and is robust against divergence. Conversely, the numerical models show equivalent performance for short time steps, while the computation time is reduced. Although the AFD method shows promising results by achieving an improvement of 43% in terms of Dynamic Time Warping, its parameter optimization must be generalized for different tank designs, flow rates, and temperatures.

HYDROSAFE: A Hybrid Deterministic-Probabilistic Model for Synthetic Appliance Profiles Generation.

Realistic appliance power consumption data are essential for developing smart home energy management systems and the foundational algorithms that analyze such data. However, publicly available datasets are scarce and time-consuming to collect. To address this, we propose HYDROSAFE, a hybrid deterministic-probabilistic model designed to generate synthetic appliance power consumption profiles. HYDROSAFE employs the Median Difference Test (MDT) for profile characterization and the Density and Dynamic Time Warping based Spatial Clustering for appliance operation modes (DDTWSC) algorithm to cluster appliance usage according to the corresponding Appliance Operation Modes (AOMs). By integrating stochastic methods, such as white noise, switch-on surge, ripples, and edge position components, the model adds variability and realism to the generated profiles. Evaluation using a normalized DTW-distance matrix shows that HYDROSAFE achieves high fidelity, with an average DTW distance of ten samples at a 1Hz sampling frequency, demonstrating its effectiveness in producing synthetic datasets that closely mimic real-world data.

ECG arrhythmia classification based on the fast ant colony clustering algorithm with improved spatiotemporal feature perception ability

Electrocardiograph (ECG) is one of the most critical physiological signals used for arrhythmia diagnosis. In recent years, ECG arrhythmia classification devices consisting of multi-module sensors, clustering algorithms and neural networks play an important role in monitoring and diagnosing cardiovascular diseases. However, the commonly used ECG arrhythmia classification methods are still facing some problems such as the complex model structure and long running time. To address the above problems, this paper proposes an ECG arrhythmia classification method based on the fast ant colony clustering algorithm with improved spatiotemporal feature perception ability (SFP-FACC), which uses LSTM to fit the cluster centers and avoids the time consumption of updating the cluster centers during the classification process. The spatiotemporal feature perception ability of this model with the dynamic time warping (DTW) algorithm is improved. The classification is achieved by applying the combination of Euclidean distance and DTW. The convergence speed of the model is improved by using dynamic pheromone volatility coefficient; and finally the optimal solution of the model is determined by using radix sort. Based on the MIT-BIH arrhythmia dataset, the overall accuracy of the proposed classification method in this paper achieves 99.04 %, and even the accuracy of certain types of classification achieves 100 %, and the running time is about 3.5 times faster than that of the basic models. The experiments show that the method proposed in this paper has certain advantages.