Method, system and equipment for recognizing similarity of perishable garbage carbon and medium

Abstract

The invention discloses a perishable garbage carbon similarity recognition method, system and device and a medium, and relates to the field of perishable garbage resource.The method comprises the steps that multi-modal data of perishable garbage carbon to be recognized is obtained; performing cleaning, correction and assimilation processing on the multi-modal data to obtain assimilated multi-modal data; performing feature fusion on the assimilation multi-modal data to obtain an assimilation fusion feature vector; calculating the distance between the assimilation fusion feature vector and the center point of each class cluster so as to determine the similarity between the perishable garbage carbon to be identified and the perishable garbage carbon sample; the class cluster center point is determined in advance based on multi-modal data of a plurality of perishable garbage carbon samples. According to the invention, the accuracy and reliability of similarity identification of perishable garbage carbon are improved.

Description

Technical Field

[0001] The present application relates to the field of perishable waste resource utilization, and in particular to a method, system, equipment and medium for identifying perishable waste charcoal similarity. Background Art

[0002] Perishable waste, including restaurant and kitchen waste, as well as vegetable and fruit waste, rotting meat, meat scraps, eggshells, and livestock and poultry offal generated at farmers' and wholesale markets, is large and concentrated. Currently, landfilling, incineration, and composting are the primary methods for disposing of perishable waste. However, these methods pose secondary pollution risks, such as waste gas emissions and landfill leachate, and are limited by the large land occupation and low resource utilization.

[0003] Converting perishable garbage into perishable garbage charcoal through thermal decomposition and carbonization can not only effectively reduce waste pollution and greenhouse gas emissions, but the resulting perishable garbage charcoal also has good antioxidant and adsorption capacity. It can be used as a soil conditioner to increase soil organic matter content, improve soil structure, and promote plant growth, thereby achieving green, low-carbon and sustainable development of environmental protection and resource recycling.

[0004] Perishable waste has a complex composition, characterized by high moisture, salt, organic matter, and oil content. Furthermore, the resource and environmental properties of perishable waste vary significantly across regions. Furthermore, due to factors such as the production and discharge characteristics, spatiotemporal distribution, carbonization process, and pyrolysis conditions of perishable waste, the prepared perishable waste charcoals exhibit diverse and significant differences in physical and chemical properties such as structure, volatile content, ash content, pore volume, and specific surface area, leading to varying environmental effects.

[0005] Similarity recognition or identification of perishable waste charcoal is crucial for its resource utilization. However, related technologies typically rely on single-point physical and chemical property data, which fails to fully reflect the characteristics of perishable waste charcoal. This results in poor reliability of similarity recognition, low accuracy of type identification, and insufficient robustness. Summary of the Invention

[0006] The purpose of this application is to provide a method, system, equipment and medium for similarity identification of perishable garbage charcoal, which can overcome the problems of poor reliability, low category identification accuracy and insufficient robustness in the existing technology in the classification and category identification of perishable garbage charcoal.

[0007] To achieve the above objectives, this application provides the following solutions:

[0008] In a first aspect, the present application provides a method for similarity identification of perishable garbage charcoal, comprising:

[0009] Obtain multimodal data of perishable garbage charcoal to be identified;

[0010] Cleaning, correcting, and assimilating the multimodal data to obtain assimilated multimodal data;

[0011] Performing feature fusion on the assimilated multimodal data to obtain an assimilated fusion feature vector;

[0012] The distance between the assimilated fusion feature vector and the center point of each cluster is calculated to determine the similarity between the perishable garbage charcoal to be identified and the perishable garbage charcoal sample; the cluster center point is pre-determined based on the multimodal data of multiple perishable garbage charcoal samples.

[0013] In a second aspect, the present application provides a system for identifying similarity of perishable garbage charcoal, comprising:

[0014] A multimodal data acquisition module, used to obtain multimodal data of the perishable garbage charcoal to be identified;

[0015] A multimodal data processing module, configured to clean, correct, and assimilate the multimodal data to obtain assimilated multimodal data;

[0016] A feature fusion module, configured to perform feature fusion on the assimilated multimodal data to obtain an assimilated fusion feature vector;

[0017] A similarity identification module is used to calculate the distance between the assimilated fusion feature vector and the center point of each cluster to determine the similarity between the perishable garbage charcoal to be identified and the perishable garbage charcoal sample; the cluster center point is pre-determined based on the multimodal data of multiple perishable garbage charcoal samples.

[0018] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned method for identifying similarity of perishable garbage charcoal.

[0019] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned method for identifying similarity of perishable garbage charcoal.

[0020] According to the specific embodiments provided in this application, this application has the following technical effects:

[0021] The present application provides a method, system, equipment and medium for similarity identification of perishable garbage charcoal. By acquiring multimodal data of the perishable garbage charcoal to be identified and cleaning, correcting, assimilating and feature-fusing the multimodal data, similarity identification and category identification of perishable garbage charcoal are achieved, thereby improving the accuracy and reliability of similarity identification of perishable garbage charcoal, and providing basic support for ecological safety and environmental risk assessment for the carbonization treatment of perishable garbage and its application in soil improvement and remediation. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 This is a diagram of an application environment of a method for similarity identification of perishable garbage charcoal in an embodiment of the present application;

[0024] Figure 2 A schematic diagram of the overall process of a method for similarity identification of perishable garbage charcoal provided in one embodiment of the present application;

[0025] Figure 3 A detailed flowchart of a method for similarity identification of perishable garbage charcoal provided in one embodiment of the present application;

[0026] Figure 4 Schematic diagram of baseline correction and characteristic peak positions of an infrared spectrum image in one embodiment of the present application;

[0027] Figure 5 A schematic diagram of vector normalization and characteristic peak selection of an infrared spectrum image in one embodiment of the present application;

[0028] Figure 6 This is a schematic diagram of the similarity recognition results of perishable garbage charcoal by multimodal data assimilation and fusion in one embodiment of the present application;

[0029] Figure 7 A schematic diagram of the functional modules of a perishable garbage charcoal similarity recognition system provided in one embodiment of the present application;

[0030] Figure 8 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0032] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0033] The method for identifying similarity of perishable garbage charcoal provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the multimodal data of the perishable garbage charcoal to be identified to the server 104. After receiving the multimodal data, the server 104 cleans, corrects and assimilates the multimodal data to obtain assimilated multimodal data; performs feature fusion on the assimilated multimodal data to obtain an assimilated fusion feature vector; calculates the distance between the assimilated fusion feature vector and the center point of each cluster to determine the similarity between the perishable garbage charcoal to be identified and the perishable garbage charcoal sample. The server 104 can feedback the similarity between the perishable garbage charcoal to be identified and the perishable garbage charcoal sample to the terminal 102. In addition, in some embodiments, the perishable garbage charcoal similarity identification method can also be implemented separately by the server 104 or the terminal 102.

[0034] Terminal 102 may include, but is not limited to, various desktop computers, laptops, smartphones, tablet computers, IoT devices, and portable wearable devices. IoT devices may include smart speakers, smart TVs, smart air conditioners, and smart car devices. Portable wearable devices may include smart watches, smart bracelets, and head-mounted devices. Server 104 may be implemented as a standalone server or a server cluster consisting of multiple servers, or may be a cloud server.

[0035] In an exemplary embodiment, Figure 2 and Figure 3 As shown, a method for identifying similarity of perishable garbage charcoal is provided. The method is executed by a computer device, specifically a computer device such as a terminal or a server, or a terminal and a server. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used for explanation, including the following steps 201 to 204.

[0036] Step 201: Acquire multimodal data of the easily degradable garbage charcoal to be identified. The multimodal data includes physical and chemical property parameters and infrared spectrum images.

[0037] In a specific application example, step 201 includes the following steps 11 to 13.

[0038] Step 11: obtain the perishable garbage raw material to be identified, and pyrolyze and carbonize the perishable garbage raw material to be identified to obtain perishable garbage charcoal to be identified.

[0039] Step 12: Analyze the physical and chemical parameters of the easily degradable garbage raw materials and the easily degradable garbage charcoal to obtain physical and chemical property parameters. Specifically, through experiments or instrumental analysis, 43 physical and chemical property parameters characterizing the easily degradable garbage raw materials and the easily degradable garbage charcoal to be identified were obtained, including specific surface area (m 2 / g), porosity, pH value, conductivity (us / cm), organic matter content, water-soluble salt (g / kg), ash content (%), element content (%), polycyclic aromatic hydrocarbon content (ng / g); yield of easily degradable garbage charcoal (%), average pore size (nm), etc., are marked in sequence to form the attribute vector x = [x1, x2,…, x 43 ].

[0040] Step 13: Use Fourier transform infrared spectroscopy (FTIR) to obtain an infrared spectrum image of the easily degradable garbage charcoal to be identified.

[0041] Specifically, in order to characterize the chemical composition and structural characteristics of the biodegradable garbage charcoal, the FTIR method was used to obtain the infrared spectrum image of the biodegradable garbage charcoal, and the structural related information was obtained by analyzing the characteristic absorption peaks of surface functional groups (such as alkyl groups, aromatic groups, etc.).

[0042] Step 202: Clean, calibrate, and assimilate the multimodal data to obtain assimilated multimodal data. The assimilated multimodal data includes assimilated physical and chemical property parameters and assimilated infrared spectral images.

[0043] In a specific application example, to ensure the quality and reliability of multimodal data, data cleaning methods such as verification, denoising, duplication checking, missing value repair, and abnormal data identification are implemented for numerical data to reduce or correct errors and biases that may be introduced during the data acquisition process, thereby improving the accuracy, completeness, consistency, and validity of the data. For infrared spectral images, baseline correction and characteristic peak alignment are implemented to eliminate the effects of background noise and spectral shifts, thereby improving the signal-to-noise ratio, resolution, and comparability of the spectral images. Step 202 includes the following steps 21 to 24.

[0044] Step 21, the physical and chemical property parameters are sequentially verified, deduplicated, missing value repaired, and abnormal data eliminated to obtain preprocessed physical and chemical property parameters.

[0045] The K-nearest neighbor interpolation method was used to repair missing values ​​in the verified physical and chemical property parameters. The density-based spatial clustering of applications with noise (DBSCAN) algorithm was used to identify outliers in the physical and chemical property parameters after missing value repair to eliminate abnormal data.

[0046] Specifically, data validation is first performed, including format, type, and range checks, to eliminate invalid data and ensure data consistency and validity. Secondly, data deduplication is performed to identify and remove duplicate sample data to ensure data uniqueness. Next, missing values ​​for physical and chemical property parameters are filled using the K-nearest neighbor interpolation method to maintain data integrity, continuity, and availability. Finally, the DBSCAN algorithm is used to detect and remove outliers to reduce bias that may be introduced during data collection, thereby ensuring data accuracy.

[0047] The process of using the K-nearest neighbor interpolation method to fill missing values ​​is as follows: first, determine an appropriate k value, then, for each missing value, calculate the Mahalanobis distance between it and all non-missing values, and find the k nearest sample points. Finally, fill the missing value with the average value of the target variable of the k nearest neighbor sample points.

[0048] To test for abnormal outliers in the physical and chemical property parameters, we selected an appropriate neighborhood radius ε and minimum number of neighborhood points l, then used the DBSCAN algorithm to cluster the sample data and identify outliers. Data marked as outliers were then deleted from the original data to reduce bias during data collection and ensure data accuracy.

[0049] Step 22: performing baseline correction and characteristic peak alignment processing on the infrared spectrum image to obtain a preprocessed infrared spectrum image.

[0050] The infrared spectral images were baseline corrected using an adaptive iterative reweighted penalized least squares method to effectively eliminate background noise introduced by device drift or changes in experimental conditions, thereby improving the signal-to-noise ratio of the infrared spectral images. A dynamic time warping (DTW) algorithm was used to align the characteristic peaks in the baseline-corrected infrared spectral images to correct for spectral shift effects and improve the resolution and comparability of the infrared spectral images.

[0051] Step 23: standardize or normalize the pre-processed physical and chemical property parameters to obtain assimilated physical and chemical property parameters.

[0052] Step 24 , identifying characteristic peaks in the pre-processed infrared spectrum image, constructing characteristic peak data vectors, and performing vector normalization processing on the characteristic peak data vectors to obtain an assimilated infrared spectrum image.

[0053] In order to unify the scales of various variables and eliminate the differences in structure and modality between numerical data and infrared spectral images, this application performs scaling and assimilation on pre-processed multimodal data. Among them, the numerical data adopts standardization or normalization methods to eliminate the influence of the dimension and order of magnitude differences of different physical and chemical property parameters. Infrared spectral images use derivative transformation, characteristic peak selection and normalization methods to realize the digital transformation and representation of image features, reduce noise, eliminate intensity differences caused by different samples or measurement conditions, and enhance spectral peak resolution, thereby improving the accuracy, reliability and sensitivity of spectral data.

[0054] Step 203: Perform feature fusion on the assimilated multimodal data to obtain an assimilated fusion feature vector.

[0055] In a specific application example, step 203 includes the following steps 31 to 33.

[0056] Step 31: Merge the assimilated physical and chemical property parameters with the assimilated infrared spectrum image to obtain a set variable.

[0057] Step 32: Use the Uniform Manifold Approximation and Projection (UMAP) method to perform nonlinear dimensionality reduction on the set variables and extract characteristic variables.

[0058] In step 33, a graph neural network model incorporating a dual attention mechanism is used to optimize and screen the feature variables to obtain an assimilated fusion feature vector.

[0059] Step 204 calculates the distance between the assimilated fusion feature vector and each cluster center to determine the similarity between the identified perishable waste charcoal and the perishable waste charcoal sample. The identified perishable waste charcoal is classified into the perishable waste charcoal category with the smallest distance (highest similarity). The cluster center is pre-determined based on multimodal data from multiple perishable waste charcoal samples.

[0060] This application first integrates data from multiple sensors, including spectral analysis, thermogravimetric analysis, and elemental analysis, to comprehensively capture the physical, chemical, and thermal properties of perishable waste charcoal. It then denoises, normalizes, and standardizes the multimodal data to eliminate errors and interference, ensuring data consistency and reliability. Advanced data assimilation algorithms are then used to fuse multimodal data to construct a unified feature representation. The Unified Mapping (UMAP) method is further utilized, incorporating a dual-attention mechanism for nonlinear dimensionality reduction and feature screening. Node-level attention focuses on local structures, while feature-level attention selects global discriminant feature variables to synergistically enhance the ability to identify similarity between feature variables. Finally, the generated highly discriminative feature vectors are used to calculate similarity and identify perishable waste charcoal. The entire process exhibits excellent dynamic adaptability and can be flexibly adjusted and expanded based on different raw material characteristics, operating conditions, and algorithmic environments. It is suitable for a variety of application scenarios and has broad application prospects in solid waste resource utilization, including waste sorting, carbonization process optimization, and perishable waste charcoal quality control, providing technical support for solid waste resource utilization.

[0061] In a specific application example, the process of determining the cluster center point is as follows:

[0062] (1) Obtain multimodal data of perishable garbage charcoal samples, including physical and chemical property data matrix, infrared spectrum image set and category label column vector.

[0063] Specifically, various types of perishable waste samples, including restaurant waste, kitchen waste, and fresh food waste, were first collected from different geographical regions. The perishable waste samples were then pyrolyzed and carbonized at 500°C to produce biochar, yielding perishable waste charcoal samples. To fully understand their characteristics, multimodal characterization analysis was performed on the perishable waste samples and the perishable waste charcoal samples prepared from them. This included measuring physicochemical properties such as specific surface area, porosity, pH value, conductivity, and organic matter content, and using Fourier transform infrared spectroscopy to obtain infrared spectral images reflecting surface functional group information. Finally, for subsequent similarity recognition studies, each perishable waste sample was labeled with a corresponding category label based on its source type.

[0064] The physical and chemical properties of the i-th easily degradable garbage charcoal sample are recorded as x i , and different perishable garbage charcoal samples are arranged in row order i = 1, 2, ..., n, to obtain a physical and chemical property data matrix X of the perishable garbage charcoal samples. In this embodiment, the number of perishable garbage charcoal samples n = 117.

[0065] The FTIR method was used to obtain the wavelength of each biodegradable garbage carbon sample at 4000-400 cm -1 Wave number range, resolution is 0.5cm -1The infrared spectrum of the surface of the material is analyzed, and its structural information is obtained by analyzing the characteristic absorption peaks of different functional groups (such as alkyl, aromatic, hydroxyl, carboxyl, carboxyl, ketone, aldehyde, aromatic compounds, aliphatic compounds, CC, CO, CH, etc.) on the surface.

[0066] Based on the region and the type of perishable waste, the perishable waste charcoal samples were combined and coded into 9 categories, that is, ω = 9. Category c of perishable waste charcoal samples ν They are marked with numerical serial numbers ν = 1, 2, ..., 9. Thus, the category labels of each perishable garbage charcoal sample are marked one by one according to the row number of the physical and chemical properties data matrix X of the perishable garbage charcoal sample, forming the category label column vector c of the perishable garbage charcoal sample.

[0067] (2) The multimodal data of the perishable garbage charcoal samples are cleaned, corrected and assimilated to obtain the assimilated multimodal data of each perishable garbage charcoal sample.

[0068] Specifically, the physical and chemical property parameters of the perishable garbage charcoal samples are sequentially verified, deduplicated, repaired for missing values, and eliminated for abnormal data to obtain the preprocessed physical and chemical property parameters of the perishable garbage charcoal samples.

[0069] The infrared spectrum image of the perishable garbage charcoal sample is subjected to baseline correction and characteristic peak alignment processing to obtain the preprocessed infrared spectrum image of the perishable garbage charcoal sample.

[0070] For the pretreatment physical and chemical property parameters, Z-score standardization is used, that is, the formula j=1,2,…,p converts each pretreatment physical and chemical property parameter into a distribution with a mean of 0 and a standard deviation of 1; is the jth pretreatment physical and chemical property parameter of the i-th perishable garbage charcoal sample after standardization, x ij is the jth pretreatment physicochemical property parameter of the i-th perishable garbage charcoal sample before standardization, is the mean value of the jth pretreatment physical and chemical property parameter, s j is the standard deviation of the jth pretreatment physicochemical property parameter, and p is the number of pretreatment physicochemical property parameters of each perishable garbage charcoal sample; or the Min-Max normalization method is used to scale each pretreatment physicochemical property parameter to a specific range (usually [0,1] or [-1,1]) to unify the scale of each pretreatment physicochemical property parameter and eliminate the influence of the dimension and order of magnitude differences of different pretreatment physicochemical property parameters. The standardized physicochemical property parameter matrix is ​​recorded as

[0071] For the pre-processed infrared spectrum image, in order to eliminate the intensity differences caused by different perishable garbage carbon samples and measurement conditions and improve the spectral peak resolution, the central difference method is used to perform the first-order derivative transformation. Then, based on the extreme points of the first-order derivative, the significant characteristic peaks in the pre-processed infrared spectrum image, namely the maximum value (absorption peak) or minimum value (transmission peak), are detected to realize the digital representation of the spectral image features, such as Figure 4 As shown. The characteristic peak data vector of the i-th perishable garbage charcoal sample is recorded as x i,IR =[x′ i1 ,x′ i2 ,x′ i3 ,…,x′ iq ], q is the number of characteristic peaks, in this example q = 8, the characteristic peak vector matrix of the infrared spectrum image is recorded as X IR .

[0072] like Figure 5 As shown in Figure 2, the characteristic peak data matrix is ​​further vector normalized or sum normalized to eliminate the difference in absorbance scale caused by the thickness or concentration of the perishable garbage carbon samples and ensure the comparability of the data. The specific formula is x i,IR-Norm =x i,IR / ||x i,IR ||2 The characteristic peak data vector x of the i-th perishable garbage carbon sample i,IR The vector is normalized and the normalized spectral data matrix is ​​recorded as X IR-Norm .

[0073] (3) A feature fusion encoder is used to perform feature fusion on the assimilated multimodal data of each perishable garbage charcoal sample to obtain the assimilated fusion feature vector of each perishable garbage charcoal sample.

[0074] Specifically, in order to enhance information and achieve data complementarity, the normalized physicochemical property parameter matrix and the normalized spectral data matrix X IR-Norm Merge to form a collective variable sample data matrix that characterizes the characteristics of perishable garbage charcoal In this embodiment, X N The dimensions of and c are 117×51 and 117×1 respectively.

[0075] In order to reduce noise and redundant information, as well as reduce the correlation between variables, the UMAP method is used to perform nonlinear dimensionality reduction on the sample data matrix of the set variables and extract characteristic variables to reduce noise and redundant information, reduce model complexity and improve computational efficiency, and enhance the quality and similarity recognition ability of multimodal sample data.

[0076] A graph neural network model incorporating an attention mechanism was designed for feature fusion. Node-level attention focuses on local structure, while feature-level attention selects global discriminant feature variables. This synergistically enhances the similarity recognition capabilities of feature variables. This model captures complex correlations between sample data while preserving the structural characteristics of samples from different categories, enhancing both intra-class similarity and inter-class differences. Finally, cross-validation and results analysis were used to evaluate the effectiveness of the assimilated fusion feature vector in similarity recognition. During feature fusion optimization and cross-validation, the contribution of each feature was evaluated to select the optimal feature variables. Physical and chemical parameter numerical data and infrared spectral image characteristic peak data were fused at the feature level to form a unified feature variable matrix, achieving data complementarity and information enhancement.

[0077] Specifically, the physical and chemical property parameters x of the i-th perishable garbage charcoal sample i , use the k-NN algorithm to find its nearest k neighbors and calculate the distance d between it and the nearest neighbor i , and use the following formula to calculate and assign it to the neighbor x j The edge weights of:

[0078]

[0079] Among them, w ij is the edge weight between the i-th perishable garbage charcoal sample and the j-th perishable garbage charcoal sample, σ i is x i The local normalization parameter is used to adjust the weight distribution of edges in its neighborhood to ensure the continuity of the topological structure. Then, the local domain graph of the sample data is drawn through the weight matrix to realize the local structure representation of the data manifold. Then, the low-dimensional points [y1, y2, ..., y h ], h is the feature dimension. The midpoint y in the low-dimensional space i and y j The similarity q ij Using the modified t distribution: q ij =(1+a·||y i -y j || 2b ) -1 , a and b are hyper parameters. Then, the high-dimensional space adjacency relationship w between the i-th perishable garbage charcoal sample and the j-th perishable garbage charcoal sample is ij and the low-dimensional space adjacency relationship q ij , the cross entropy loss with the following formula is used to measure the structural similarity:

[0080]

[0081] Among them, L UMAPis the structural similarity between the i-th easily degradable garbage charcoal sample and the j-th easily degradable garbage charcoal sample.

[0082] Optimize L using momentum gradient descent UMAP , until the cross entropy loss function converges, and finally a low-dimensional embedding sample is obtained Preserve global and local structures of high-dimensional data.

[0083] Based on the data after dimensionality reduction, the Euclidean distance between sample points is calculated, and k neighbors of each sample point are selected to construct a weighted adjacency matrix A with weight A. ij =exp(-γ·d(y i ,y j ), where γ is a scale parameter. Furthermore, a graph attention network is used to calculate the attention coefficient between graph node i and its neighbor j and aggregate neighbor information. This node-level attention is used to model sample relationships to enhance intra-class similarity. Feature-level attention is then used to learn feature importance, assigning learnable weights to each feature dimension. Weighted feature representation is then achieved through fully connected layers.

[0084] Under the joint optimization goal of minimizing classification loss and topology preservation loss, feature sorting and selection are performed through k-fold cross validation. Accuracy, precision, recall and F1-score are used to quantitatively evaluate the stability and generalization of feature subsets, and finally the assimilation fusion feature matrix of the most discriminative m feature subsets is obtained. In this example, m=2.

[0085] (4) Based on the category label column vector c of the perishable garbage charcoal sample and the corresponding assimilation fusion feature matrix Calculate the center point z of each cluster v (v=1,2,…,c), which represents the characteristics of various clusters of perishable garbage charcoal.

[0086] In this embodiment, the similarity test results of the perishable garbage charcoal are as follows: Figure 6 As shown, the accuracy of each category (v = 1, 2, ..., 9) The accuracy of the overall sample data

[0087] Based on the same inventive concept, embodiments of the present application also provide a perishable garbage charcoal similarity identification system for implementing the aforementioned perishable garbage charcoal similarity identification method. The solution provided by this system is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more of the following perishable garbage charcoal similarity identification system embodiments can be found in the limitations of the perishable garbage charcoal similarity identification method described above and will not be further elaborated here.

[0088] In an exemplary embodiment, Figure 7 As shown, a perishable garbage charcoal similarity recognition system is provided, which includes: a multimodal data acquisition module 701, a multimodal data processing module 702, a feature fusion module 703 and a similarity recognition module 704.

[0089] The multimodal data acquisition module 701 is used to obtain multimodal data of the perishable garbage charcoal to be identified.

[0090] The multimodal data processing module 702 is used to clean, correct and assimilate the multimodal data to obtain assimilated multimodal data.

[0091] The feature fusion module 703 is used to perform feature fusion on the assimilated multimodal data to obtain an assimilated fusion feature vector.

[0092] The similarity identification module 704 is used to calculate the distance between the assimilated fusion feature vector and each cluster center point to determine the similarity between the perishable garbage charcoal to be identified and the perishable garbage charcoal sample. The cluster center point is pre-determined based on the multimodal data of multiple perishable garbage charcoal samples.

[0093] The perishable garbage charcoal similarity identification system provided in this application adopts a modular design, integrating key modules such as multimodal data acquisition, cleaning and correction, assimilation processing, feature fusion, and similarity identification. Through the collaborative work between modules, it overcomes the shortcomings of relying on a single data source for identification. The system can perform a unified feature expression for the high-dimensional data of the structure and properties of perishable garbage charcoal, and on this basis, achieve high-precision category similarity identification. In order to adapt to complex and changeable actual application scenarios, the system also has good dynamic adaptability and expansion capabilities, and can flexibly configure different raw material parameters, working conditions and algorithm models to provide technical support for applications in the field of solid waste resource treatment.

[0094] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 8As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store multimodal data of perishable garbage charcoal. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for identifying the similarity of perishable garbage charcoal is implemented.

[0095] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0096] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0097] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0098] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0099] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0100] In this application, all actions to obtain signals, information or data are carried out in compliance with the relevant data protection laws and policies of the country where they are located and with the authorization given by the owner of the corresponding device.

[0101] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0102] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0103] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0104] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for similarity identification of perishable garbage charcoal, characterized in that: The method comprises: Obtain multimodal data of perishable garbage charcoal to be identified; Cleaning, correcting, and assimilating the multimodal data to obtain assimilated multimodal data; Performing feature fusion on the assimilated multimodal data to obtain an assimilated fusion feature vector; The distance between the assimilated fusion feature vector and the center point of each cluster is calculated to determine the similarity between the perishable garbage charcoal to be identified and the perishable garbage charcoal sample; the cluster center point is pre-determined based on the multimodal data of multiple perishable garbage charcoal samples.

2. The method for similarity identification of perishable garbage charcoal according to claim 1, characterized in that: The multimodal data includes physical and chemical property parameters and infrared spectrum images; the assimilated multimodal data includes assimilated physical and chemical property parameters and assimilated infrared spectrum images.

3. The method for similarity identification of perishable garbage charcoal according to claim 2, characterized in that: Obtain multimodal data of the perishable garbage char to be identified, including: Obtaining a perishable garbage raw material to be identified, and pyrolyzing and carbonizing the perishable garbage raw material to obtain perishable garbage charcoal to be identified; Performing physical and chemical parameter analysis on the easily perishable garbage raw material to be identified and the easily perishable garbage charcoal to be identified to obtain physical and chemical property parameters; Fourier transform infrared spectroscopy is used to obtain an infrared spectrum image of the perishable garbage charcoal to be identified.

4. The method for identifying similarity of perishable garbage charcoal according to claim 2, characterized in that: The multimodal data is cleaned, corrected, and assimilated to obtain assimilated multimodal data, specifically including: The physical and chemical property parameters are sequentially verified, deduplicated, repaired for missing values, and eliminated for abnormal data to obtain preprocessed physical and chemical property parameters; Performing baseline correction and characteristic peak alignment processing on the infrared spectrum image to obtain a preprocessed infrared spectrum image; Standardizing or normalizing the pretreated physical and chemical property parameters to obtain assimilated physical and chemical property parameters; Characteristic peaks in the preprocessed infrared spectrum image are identified, characteristic peak data vectors are constructed, and vector normalization processing is performed on the characteristic peak data vectors to obtain an assimilated infrared spectrum image.

5. The method for similarity identification of perishable garbage charcoal according to claim 4, characterized in that: The K-nearest neighbor interpolation method was used to repair the missing values ​​of the physical and chemical property parameters after verification; the density-based noise application spatial clustering algorithm was used to identify outliers in the physical and chemical property parameters after missing value repair in order to eliminate abnormal data.

6. The method for identifying similarity of perishable garbage charcoal according to claim 4, characterized in that: An adaptive iterative reweighted penalty least squares method is used to perform baseline correction on the infrared spectrum image; and a dynamic time warping algorithm is used to align characteristic peaks in the infrared spectrum image after baseline correction.

7. The method for similarity identification of perishable garbage charcoal according to claim 2, characterized in that: Performing feature fusion on the assimilated multimodal data to obtain an assimilated fusion feature vector specifically includes: Merging the assimilated physicochemical property parameters with the assimilated infrared spectrum image to obtain a set variable; Using a unified manifold approximation and projection method to perform nonlinear dimensionality reduction on the set variables and extract characteristic variables; A graph neural network model with a dual attention mechanism is used to optimize and screen the feature variables to obtain an assimilated fusion feature vector.

8. A system for identifying similarity of perishable garbage charcoal, applied to the method for identifying similarity of perishable garbage charcoal according to any one of claims 1 to 7, characterized in that: The system comprises: A multimodal data acquisition module, used to obtain multimodal data of the perishable garbage charcoal to be identified; A multimodal data processing module, configured to clean, correct, and assimilate the multimodal data to obtain assimilated multimodal data; A feature fusion module, configured to perform feature fusion on the assimilated multimodal data to obtain an assimilated fusion feature vector; A similarity identification module is used to calculate the distance between the assimilated fusion feature vector and the center point of each cluster to determine the similarity between the perishable garbage charcoal to be identified and the perishable garbage charcoal sample; the cluster center point is pre-determined based on the multimodal data of multiple perishable garbage charcoal samples.

9. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for similarity identification of perishable garbage charcoal according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for identifying similarity of perishable garbage charcoal according to any one of claims 1 to 7 is implemented.

Images