Cloud computing-based alcohol biomass raw material characteristic data storage operation and maintenance system and method

Abstract

The present application relates to the technical field of biomass raw material data management, and particularly relates to a system and method for storing and operating characteristic data of alcohol biomass raw materials based on cloud computing, comprising: generating raw material characteristic data in a unified format by collecting characteristic data of biomass raw materials according to a hierarchical system and performing standardization processing; mapping the raw material characteristic data into a multi-dimensional feature vector to form a raw material performance vector space, and generating a hierarchical index by using an incremental dynamic clustering, while updating the feature vector and the hierarchical information; correlating and mapping the feature vector with alcohol processing parameters to form a cause-and-effect mapping result; performing nonlinear pattern recognition based on the feature vector, the hierarchical information and the cause-and-effect mapping result to generate a combination mode library and record the feature vector and the process parameters of each combination; and receiving an external request through an intelligent query interface and returning a structured analysis result. The present application realizes dynamic management of cross-regional and cross-type raw material performance, combination mode analysis and efficient retrieval.

Description

A Cloud Computing-Based Data Storage and Maintenance System and Method for Alcohol Biomass Raw Material Characteristics Technical Field

[0001] This invention relates to the field of biomass raw material data management technology, and in particular to a cloud computing-based system and method for storing and maintaining characteristic data of alcohol biomass raw materials. Background Technology

[0002] Performance data of alcohol biomass feedstocks are typically stored and managed using traditional database methods. Industrial analysis indicators, elemental composition indicators, component indicators, and physicochemical property indicators for different regions and feedstock types are entered separately and form a static data table structure. When feedstock batches are added or updated, the database needs to manually adjust the classification and hierarchical relationships to maintain the data retrieval and analysis order. At the same time, the analysis of feedstock combinations and performance patterns usually relies on manual summarization or linear calculation methods, lacking automated vectorized representation and dynamic organization methods.

[0003] The existing database cannot automatically adjust the classification and hierarchical relationship when new raw materials or batches are entered, resulting in low efficiency of cross-regional and cross-type data retrieval. It also limits the dynamic management and combination analysis capabilities of raw material performance space and cannot meet the needs of large-scale raw material characteristic management and intelligent decision-making. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides a cloud computing-based system and method for storing and maintaining the characteristics of alcohol biomass raw materials. It aims to improve the existing technology, which cannot automatically adjust the classification and hierarchical relationship when new raw materials or batches are entered, resulting in low efficiency of cross-regional and cross-type data retrieval and the inability to dynamically manage and combine raw material performance space.

[0005] In a first aspect, the present invention provides the following technical solution: a cloud computing-based alcohol biomass raw material characteristic data storage and maintenance system, comprising the following modules:

[0006] The data acquisition and standardization module is used to collect characteristic data of biomass raw materials in a hierarchical system and to standardize the collected data to form raw material characteristic data.

[0007] The feature vector construction module is used to map raw material characteristic data into multi-dimensional feature vectors to form a raw material performance vector space.

[0008] The self-organizing clustering and hierarchical indexing module is used to perform incremental dynamic clustering of feature vectors, generate hierarchical indexes, and update feature vectors and hierarchical information at the same time.

[0009] The process association mapping module is used to associate and map feature vectors with alcohol processing parameters to form causal mapping results;

[0010] The combined pattern recognition module is used to perform nonlinear pattern recognition based on feature vectors, hierarchical information and causal mapping results, generate a combined pattern library, and record the feature vectors and process parameters of each combination.

[0011] The intelligent query interface module is used to receive external query requests and return structured analysis results through hierarchical indexes, causal mapping results, and combined pattern libraries.

[0012] By adopting the above technical solution, multidimensional mapping and incremental dynamic clustering of raw material performance data are realized, enabling newly entered raw materials or batches to automatically adjust their classification and hierarchical relationships, thus solving the problem of low data retrieval efficiency across regions and types.

[0013] Preferably, the characteristic data of biomass feedstock collected according to the stratified system includes:

[0014] Biomass raw materials are divided into pre-defined regional groups according to geographical regions;

[0015] Establish raw material type layers in each regional group according to raw material type;

[0016] Under the raw material type layer, specific raw materials are selected to form a specific raw material layer, thus creating a multi-level hierarchical collection system based on region, raw material type, and specific raw material.

[0017] Data on the characteristics of each specific raw material are collected, including industrial analysis indicators, elemental composition indicators, component indicators, and physicochemical property indicators.

[0018] Preferably, the standardization process includes:

[0019] Standardize the units for the collected raw material characteristic data;

[0020] Normalize the raw material characteristic data;

[0021] Mark or fill in missing or abnormal data;

[0022] The processed data is formatted into a unified format to generate raw material characteristic data.

[0023] Preferably, mapping the raw material characteristic data into a multidimensional feature vector includes:

[0024] Determine the corresponding vector dimensions for each indicator in the raw material characteristic data, and set vector encoding rules;

[0025] The numerical values ​​of each indicator are converted into numerical representations in a vector space through linear or nonlinear mapping functions, forming multidimensional feature vectors.

[0026] Preferably, the incremental dynamic clustering includes:

[0027] Initialize cluster centers and hierarchical structure, and assign existing feature vectors to the initial cluster centers;

[0028] For a new input feature vector, calculate its vector distance to the existing cluster centers;

[0029] Based on vector distance and local density, new feature vectors are assigned to existing cluster centers or new cluster centers are generated;

[0030] Dynamically adjust the location of cluster centers and hierarchical structure;

[0031] Update the clustering identifier and corresponding hierarchical index information of the feature vector.

[0032] Preferably, the step of associating the feature vector with alcohol processing parameters includes:

[0033] Obtain a preset set of alcohol processing parameters, and quantify each parameter to obtain a process parameter vector;

[0034] The process parameter vector and feature vector are input into an association mapping model;

[0035] Perform dimension calibration on the input feature vector and process parameter vector to unify vector length, index position and data format;

[0036] The calibrated feature vectors and process parameter vectors are mapped to the same associated space through embedding transformation;

[0037] Generate associated data representing the correspondence between the two in the associated space;

[0038] A causal mapping result is formed based on the aforementioned associated data.

[0039] Preferably, the nonlinear pattern recognition includes:

[0040] The input feature vectors are combined to generate a set of raw material combinations to be identified;

[0041] The raw material combination set is matched with the corresponding hierarchical information to determine the hierarchical relationship of each combination;

[0042] The combined feature vector and the causal mapping result are input into the nonlinear pattern recognition model;

[0043] By extracting features from the input data through nonlinear transformation, a high-dimensional representation of the combined properties is obtained.

[0044] The high-dimensional representation is partitioned into patterns to obtain pattern data representing different combinations of characteristics;

[0045] Generate combination pattern entries to describe combination features, and record the combination pattern entries as structured combination pattern data.

[0046] Preferably, the generated combinatorial pattern library includes:

[0047] The combined pattern data are grouped according to a pre-defined pattern classification system;

[0048] Create unique identifiers for each combination pattern and establish index relationships;

[0049] The data structure that writes the combined pattern data into the combined pattern library according to the index relationship;

[0050] Record the corresponding feature vectors and process parameters for each combination mode.

[0051] Preferably, the structured analysis results include:

[0052] Parse the fields of the pattern entries in the combined pattern library and extract the basic field units;

[0053] The basic field units are labeled according to preset semantic categories to form a set of fields with category labels;

[0054] Based on the field set, identify the reference relationships and hierarchical relationships between fields, and generate a field relationship structure;

[0055] The field set is grouped and divided according to the field relationship structure to form structured fragments with a fixed organizational form;

[0056] The structured fragments are sequenced and combined to generate structured analysis results.

[0057] Secondly, the present invention provides the following technical solution: a cloud computing-based method for storing and maintaining characteristic data of alcohol biomass raw materials, the method comprising:

[0058] The characteristic data of biomass raw materials were collected according to a stratified system, and the collected data were standardized to form raw material characteristic data.

[0059] The raw material characteristic data is mapped into a multi-dimensional feature vector to form a raw material performance vector space;

[0060] Incremental dynamic clustering is performed on the feature vectors to generate hierarchical indexes, while updating the feature vectors and hierarchical information.

[0061] The feature vectors are correlated and mapped with alcohol processing parameters to form a causal mapping result;

[0062] Nonlinear pattern recognition is performed based on feature vectors, hierarchical information, and causal mapping results to generate a combined pattern library, and the feature vectors and process parameters of each combination are recorded.

[0063] It receives external query requests and returns structured analysis results through hierarchical indexes, causal mapping results, and a combined pattern library.

[0064] The present invention has the following beneficial effects:

[0065] 1. In this invention, by establishing a high-dimensional raw material performance vector space in the database and adopting an incremental dynamic clustering algorithm, the dynamic self-organization of the raw material performance space is realized. This solves the problem that when the performance indicators of different regions and raw material types are high-dimensional and complex in distribution, the database cannot automatically adjust the classification and hierarchical relationship when new raw materials or batches are entered, resulting in low efficiency of data retrieval and combination analysis.

[0066] 2. In this invention, by constructing a causal mapping model that combines rules and data-driven approaches, and by using embedded vector representation to map raw material indicators and process parameters to a unified vector space, non-explicit causal coupling between raw material characteristics and processing technology is achieved. This solves the problem that traditional databases, which only store indicators, cannot support raw material selection and process condition adaptation reasoning or optimization when there are complex nonlinear relationships between raw material performance and alcohol yield and by-product generation.

[0067] 3. In this invention, by constructing a high-dimensional combined index matrix and introducing a nonlinear pattern recognition algorithm, a searchable combined pattern library is formed, which realizes the identification of cross-regional raw material synergistic characteristics. This solves the problem that when there are complex nonlinear synergistic effects in cross-regional raw material combinations, traditional linear weighting or empirical methods cannot quantify potential advantages or risks, nor can they provide searchable references for alcohol yield optimization. Attached Figure Description

[0068] Figure 1 is an architecture diagram of the cloud computing-based alcohol biomass raw material characteristic data storage and maintenance system proposed in this invention.

[0069] Figure 2 is a flowchart of the cloud computing-based method for storing and maintaining characteristic data of alcohol biomass raw materials proposed in this invention. Detailed Implementation

[0070] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0071] Example 1:

[0072] In the first embodiment of the present invention, the present invention provides a cloud computing-based alcohol biomass raw material characteristic data storage and maintenance system, as shown in Figure 1, including the following modules:

[0073] The data acquisition and standardization module is used to collect characteristic data of biomass raw materials in a hierarchical system and to standardize the collected data to form raw material characteristic data.

[0074] Furthermore, the characteristic data of biomass feedstock collected according to the stratified system include:

[0075] Biomass raw materials are divided into pre-defined regional groups according to geographical regions;

[0076] Establish raw material type layers in each regional group according to raw material type;

[0077] Under the raw material type layer, specific raw materials are selected to form a specific raw material layer, thus creating a multi-level hierarchical collection system based on region, raw material type, and specific raw material.

[0078] Data on the characteristics of each specific raw material are collected, including industrial analysis indicators, elemental composition indicators, component indicators, and physicochemical property indicators.

[0079] Furthermore, standardization processes include:

[0080] Standardize the units for the collected raw material characteristic data;

[0081] Normalize the raw material characteristic data;

[0082] Mark or fill in missing or abnormal data;

[0083] The processed data is formatted into a unified format to generate raw material characteristic data.

[0084] Specifically, the data acquisition and standardization module first collects the characteristic data of biomass raw materials according to a preset hierarchical system;

[0085] The data collection process begins with the geographical origin of the raw materials, dividing them into several regional groups. Each regional group covers a certain geographical distribution to ensure that the raw material characteristic data reflects regional differences. Within each regional group, a raw material type layer is further established. Raw material types can be classified according to biomass origin, structural characteristics, or classification standards, such as forest trees, energy grasses, or crop straw. However, the specific type can be flexibly adjusted according to actual applications, thus forming a two-layer mapping relationship between the regional layer and the type layer. Under the raw material type layer, corresponding specific raw materials are selected to form a specific raw material layer. Each specific raw material layer can contain a representative and collectable sample set to ensure data coverage and diversity, forming a multi-level hierarchical data collection system of region, raw material type, and specific raw materials, making the data collection process hierarchical and systematic.

[0086] For each specific raw material, raw material characteristic data are acquired through sensors, analytical instruments, or data interfaces. This data includes, but is not limited to, industrial analysis indicators, elemental composition indicators, component indicators, and physicochemical property indicators. It can be automatically collected or manually entered through multi-source data interfaces to obtain comprehensive raw material information.

[0087] In the specific implementation process, the collected raw material characteristic data undergoes standardization processing. First, the units of various indicators are unified, converting data from different sources or with different measurement units into consistent units to eliminate data bias caused by unit differences. Then, the raw material characteristic data is normalized, which can be achieved using linear normalization or nonlinear mapping functions, for example, for indicator values. Using normalization function ,in and The minimum and maximum values ​​of the indicator are represented respectively, generating a standardized numerical representation that falls within a unified range;

[0088] During the normalization process, missing values ​​or outlier data are handled through labeling or imputation strategies, such as imputation using the mean of neighboring samples, interpolation, or substitution with preset default values, to ensure data continuity and integrity. Finally, the processed data is organized according to a unified data format, such as unified field names, data types, encoding rules, and storage structures. The standardized data is stored in a database or cache system, generating raw material characteristic data that can be directly used for subsequent feature vector construction and analysis, thus realizing an integrated process of data collection, cleaning, standardization, and storage.

[0089] To ensure the traceability and manageability of the data processing process, each collected data record includes the collection time, collection location, collection equipment information, and raw material identification code. During the standardization process, the system can record the normalization coefficient, filling strategy, and abnormal data handling method. This information is stored together with the raw material characteristic data so that it can be called in subsequent feature vector construction, cluster analysis, and process association mapping, realizing end-to-end data management and control.

[0090] The feature vector construction module is used to map raw material characteristic data into multi-dimensional feature vectors to form a raw material performance vector space.

[0091] Furthermore, mapping raw material characteristic data into multidimensional feature vectors includes:

[0092] Determine the corresponding vector dimensions for each indicator in the raw material characteristic data, and set vector encoding rules;

[0093] The numerical values ​​of each indicator are converted into numerical representations in a vector space through linear or nonlinear mapping functions, forming multidimensional feature vectors.

[0094] Specifically, firstly, the corresponding vector dimensions are determined based on the types and characteristic ranges of various indicators in the raw material characteristic data. Each indicator is assigned one or more dimensions to accurately represent its physical meaning, chemical composition, or physicochemical properties. In order to achieve vectorized representation, coding rules are set for each dimension. The coding rules can include direct numerical mapping, segmented mapping, or independent coding, etc., to ensure that different types of indicators are comparable and computable in the vector space.

[0095] In the vector encoding process, each indicator value of each raw material data is generated into a preliminary numerical representation according to the encoding rules. In order to ensure the consistency of data in multidimensional space, the numerical values ​​can be normalized or standardized so that they fall within a uniform numerical range.

[0096] In the specific implementation process, the numerical values ​​of raw material characteristic data are converted into a numerical representation in vector space through linear or nonlinear mapping functions, forming the final multidimensional feature vector; linear mapping can use linear scaling functions, such as... ,in Indicates the first The value of each indicator, Represents the mapped vector components. and These are adjustable scaling parameters used to ensure a balanced distribution of indicators across the vector space; nonlinear mappings can employ exponential functions, logarithmic functions, or activation functions, such as... or This highlights the nonlinear relationships and characteristic differences between indicators; during the mapping process, weights can be assigned to different indicators based on their importance. This forms a weighted eigenvector. ,in The total number of indicators can be weighted to enhance the influence of key features in subsequent analysis.

[0097] The generated multidimensional feature vectors are organized into a raw material performance vector space, with each vector corresponding to a raw material record. The vector space can be stored in a database or high-performance computing cache. The vector space can be used for subsequent incremental dynamic clustering analysis, process association mapping, and nonlinear pattern recognition, ensuring that raw material characteristic data can participate in the entire data processing and analysis process in a structured and computable form. At the same time, the vector construction process supports dynamic updates, that is, when new raw material data is input, new feature vectors are generated according to the same vector dimension and encoding rules, realizing the scalability and continuity of the raw material performance vector space.

[0098] The self-organizing clustering and hierarchical indexing module is used to perform incremental dynamic clustering of feature vectors, generate hierarchical indexes, and update feature vectors and hierarchical information at the same time.

[0099] Furthermore, incremental dynamic clustering includes:

[0100] Initialize cluster centers and hierarchical structure, and assign existing feature vectors to the initial cluster centers;

[0101] For a new input feature vector, calculate its vector distance to the existing cluster centers;

[0102] Based on vector distance and local density, new feature vectors are assigned to existing cluster centers or new cluster centers are generated;

[0103] Dynamically adjust the location of cluster centers and hierarchical structure;

[0104] Update the clustering identifier and corresponding hierarchical index information of the feature vector.

[0105] Specifically, the self-organizing clustering and hierarchical indexing module performs incremental dynamic clustering on the input multidimensional feature vectors to form a structured hierarchical index and updates the clustering identifiers and hierarchical information of each vector in real time. This module first initializes cluster centers on the existing feature vectors, and each cluster center is determined by calculating its center vector. Confirmed, among which Indicates the first cluster in the cluster. 1 eigenvector The number of vectors within a cluster. The cluster center vector is used to ensure that the cluster center accurately reflects the characteristic mean of the vectors within the cluster. At the same time, the hierarchical structure is initialized according to the distribution of the cluster centers to realize a coarse-to-fine hierarchical indexing framework, which provides a foundation for subsequent incremental clustering.

[0106] In the specific implementation process, when a new feature vector is input, the module first calculates its vector distance with the existing cluster centers. The distance can be calculated using methods such as Euclidean distance, cosine similarity, or weighted Manhattan distance. For example, Euclidean distance... ,in The first element of the new input vector is... One dimension, Represents the cluster center vector of the th One dimension, This represents the total dimension of the vector; by calculating the distance, the similarity between the new vector and each cluster center can be quantified; subsequently, local density information is combined. ,in With an adjustable density threshold, it is possible to determine whether a new feature vector should be assigned to an existing cluster center or a new cluster center should be generated, thereby realizing the dynamic expansion and updating of the cluster structure.

[0107] In the actual implementation, after the new vector is assigned, the cluster center positions are adjusted using an incremental update formula:

[0108] ;

[0109] in, The original cluster center vectors are the original cluster center vectors. This represents the number of vectors within a cluster before the update. For the newly added vector, This provides the updated cluster center vectors; simultaneously, it updates the hierarchical index information, including parent-child relationships, hierarchical depth, and unique identifiers, enabling synchronous maintenance of clustering and hierarchical indexes; each vector obtains a unique cluster identifier, ensuring accurate retrieval in subsequent process association mapping and combination pattern recognition.

[0110] The module supports batch and real-time input modes. In batch mode, multiple new feature vectors can be processed at once, improving computational efficiency. In real-time mode, each new vector is immediately subjected to distance calculation, density determination, and cluster assignment, and the hierarchical index is dynamically updated to achieve continuous self-organizing management of raw material characteristic data. The clustering results and hierarchical index can be stored in a cache or database, providing a structured index foundation for subsequent combined pattern recognition and intelligent query.

[0111] The process association mapping module is used to associate and map feature vectors with alcohol processing parameters to form causal mapping results;

[0112] Furthermore, the association mapping between the feature vector and alcohol processing parameters includes:

[0113] Obtain a preset set of alcohol processing parameters, and quantify each parameter to obtain a process parameter vector;

[0114] Input the process parameter vector and feature vector into the association mapping model;

[0115] Perform dimension calibration on the input feature vector and process parameter vector to unify vector length, index position and data format;

[0116] The calibrated feature vectors and process parameter vectors are mapped to the same associated space through embedding transformation;

[0117] Generate associated data representing the correspondence between the two in the associated space;

[0118] Causal mapping results are formed based on the associated data.

[0119] Specifically, the process first obtains a pre-defined set of alcohol processing parameters. Each parameter may include information such as temperature, pressure, reaction time, catalyst type, and concentration. Then, each parameter is quantified, converting categorical parameters into numerical representations while preserving the numerical characteristics of continuous parameters, resulting in a process parameter vector. ,in Indicates the first One sample, Indicates the dimension of process parameters. Indicates the first The first sample Each process parameter.

[0120] In the specific implementation process, the raw material feature vector With the corresponding process parameter vector Input association mapping model, where raw material feature vectors It is constructed from standardized raw material characteristic data. Specifically, it involves mapping various indicators in the raw material characteristic data according to preset vector dimensions and encoding rules, and converting the indicator values ​​into numerical representations in vector space through linear or nonlinear mapping functions, thus forming corresponding multidimensional raw material feature vectors. , Represents the dimension of the feature vector. Indicates the first The first sample Each raw material characteristic index; before input, the vector is dimensionally calibrated, including unifying the vector length, index position and data format, to ensure that the feature vector and process parameter vector are comparable in the same computational space.

[0121] The feature vector and process parameter vector are mapped to a unified correlation space by an embedding transformation function. The mapping function can be expressed as:

[0122] ;

[0123] in, The mapped association vector, To associate spatial dimensions, The eigenvector weight matrix, The process parameter vector weight matrix, As a bias vector, the matrix and vector dimensions are defined to ensure the mapping operation is correct.

[0124] Association vector It can be used to generate correlation data between raw material characteristics and process parameters. The correlation data can be obtained through vector similarity calculation, such as using cosine similarity.

[0125] ;

[0126] in, Indicates the first The sample and the first The degree of correlation between the samples Representing vectors The Euclidean norm, dot product, and norm are all clearly defined and have unique signs; associated vectors Indicates the first The mapping result of each raw material sample in a unified association space, and the association vector. Indicates the first The mapping results of each raw material sample in a unified association space; by calculating the similarity between the association vectors of different raw material samples, the similarity of different raw material samples at the process parameter association level is measured, thereby constructing the association relationship between samples and providing input basis for subsequent raw material combination pattern recognition.

[0127] The generated correlation data is organized into a causal mapping result matrix. ,matrix ,in The number of samples is represented by the association vector of one sample in each row, which is used for subsequent combination pattern recognition and intelligent query. The module supports batch processing and real-time mapping and updating of new raw material feature vectors, enabling dynamic management of raw material features and process parameters.

[0128] The combined pattern recognition module is used to perform nonlinear pattern recognition based on feature vectors, hierarchical information and causal mapping results, generate a combined pattern library, and record the feature vectors and process parameters of each combination.

[0129] Furthermore, nonlinear pattern recognition includes:

[0130] The input feature vectors are combined to generate a set of raw material combinations to be identified;

[0131] Match the raw material combination set with the corresponding hierarchical information to determine the hierarchical relationship of each combination;

[0132] The combined feature vector and the causal mapping result are input into the nonlinear pattern recognition model;

[0133] By extracting features from the input data through nonlinear transformation, a high-dimensional representation of the combined properties is obtained.

[0134] The high-dimensional representation is partitioned into patterns to obtain pattern data representing different combinations of characteristics;

[0135] Generate combination pattern entries to describe combination features, and record the combination pattern entries as structured combination pattern data.

[0136] Furthermore, the generation of the combinatorial pattern library includes:

[0137] The combined pattern data are grouped according to a pre-defined pattern classification system;

[0138] Create unique identifiers for each combination pattern and establish index relationships;

[0139] The data structure that writes the combined pattern data into the combined pattern library according to the index relationship;

[0140] Record the corresponding feature vectors and process parameters for each combination mode.

[0141] Specifically, it receives a set of feature vectors generated by the self-organizing clustering and hierarchical indexing modules. and corresponding hierarchical information Simultaneously, obtain the causal mapping result vector generated by the process correlation mapping module. ,in Indicates the first One raw material sample, For the feature vector dimension, For the associated spatial dimension;

[0142] In the specific implementation process, the feature vectors are combined to generate a set of raw material combinations to be identified. The combination construction includes: selecting multiple raw material feature vectors based on hierarchical index information, and forming a combined feature matrix through vector concatenation, weighted fusion, or matrix stacking. Each combination... Includes several raw material samples The combination construction can adopt a selection strategy based on hierarchical indexing, so that the raw materials in the combination are related in terms of region group, raw material type, and specific raw material level; then the combination vector is matched with the corresponding hierarchical information to form a combination hierarchical association matrix. Each row represents the hierarchical information of a certain raw material sample in the combination, ensuring that the combination structure can be correctly parsed in nonlinear pattern recognition;

[0143] The combined feature vector and the causal mapping result are input into a nonlinear pattern recognition model, which can employ a multi-layer nonlinear mapping function. Feature extraction and high-dimensional representation generation can be expressed as follows:

[0144]

[0145] in For the first A high-dimensional representation vector of a combination For the combined feature matrix, To combine causal mapping matrices, For the mapping weight matrix, For bias vectors, It is a non-linear activation function. The combination feature matrix is ​​obtained by arranging multiple raw material feature vectors formed in the combination construction step in a preset order to represent the quantity of raw materials in the combination; the combination causal mapping matrix is ​​obtained by arranging the correlation vectors that correspond one-to-one with the feature vectors in the combination in the same order; the combination feature matrix and the combination causal mapping matrix are consistent in sample order and dimension, and are used together as input to the nonlinear pattern recognition model.

[0146] In the specific implementation process, high-dimensional representation Pattern segmentation is performed, and clustering or classification algorithms are used to generate pattern data representing different combinations of characteristics. Each pattern data includes a combined identifier, feature vector characteristics, and associated process parameter information; subsequently, the generated combined pattern entries are grouped according to a preset pattern classification system, and a unique identifier is created for each entry. The data structure for constructing the composite pattern library is related to indexes. ;

[0147] The combination pattern library can be directly accessed by subsequent intelligent query interfaces, enabling rapid retrieval and analysis of raw material combination characteristics and corresponding process parameters. The module supports batch updates and real-time entry of new combinations, ensuring that the combination pattern library maintains dynamic consistency during system operation. At the same time, it ensures that the feature vector, causal mapping results, and hierarchical information of each combination are completely recorded, supporting subsequent structured analysis and intelligent scheduling applications.

[0148] The intelligent query interface module is used to receive external query requests and return structured analysis results through hierarchical indexes, causal mapping results, and combined pattern libraries.

[0149] Furthermore, the results of the structured analysis include:

[0150] Parse the fields of the pattern entries in the combined pattern library and extract the basic field units;

[0151] The basic field units are labeled according to preset semantic categories to form a set of fields with category labels;

[0152] Based on the field set, identify the reference relationships and hierarchical relationships between fields, and generate a field relationship structure;

[0153] The field set is grouped and divided according to the field relationship structure to form structured fragments with a fixed organizational form;

[0154] The structured fragments are sequenced and combined to generate structured analysis results.

[0155] Specifically, it first receives the query conditions input by the user. ,in Indicates the number of query conditions, each It can represent raw material type, region, characteristic index or process parameter. All symbols have unique definitions and internally parse query conditions to determine the query range and matching strategy.

[0156] In the specific implementation process, the composite pattern library The pattern entries are parsed to extract basic field units. ,in Indicates a composite identifier. This indicates the field number, and each field unit includes a feature vector value, a process parameter value, and hierarchical information; subsequently, the field units are labeled according to preset semantic categories, generating a set of fields with category labels. ,in Provide semantic category labels for fields, such as raw material type, chemical composition, processing conditions, etc.

[0157] Based on field collection Identify the reference relationships and hierarchical relationships between fields, and construct a field relationship structure matrix. Matrix elements Representation field Reference or related fields ,otherwise At the same time, parent-child relationships are established through hierarchical information, so that the field relationship structure contains both composite hierarchy and reference dependency, ensuring the logical integrity of subsequent grouping and partitioning;

[0158] Based on the field relationship structure The field set is grouped and partitioned to form structured fragments with a fixed organizational form. Each structured fragment contains several field units, which can be organized using a tree structure or a table structure to ensure consistency in the hierarchy and referencing relationships between fields; subsequently, the structured fragments are... Perform serialization and combination to generate the final structured analysis results. , can be adopted Alternatively, data can be encoded and transmitted using a custom data format so that external systems can read and parse it directly.

[0159] The intelligent query interface module can perform queries based on search criteria. The system filters and matches the combinatorial pattern library, and uses a hierarchical index to quickly locate relevant combinatorial pattern entries. And extract the corresponding feature vectors. With process parameters Subsequently, structured analysis results are generated through field parsing, semantic annotation, relation identification, grouping, and serialization operations. The system returns the data to the external calling system, enabling complete information retrieval and analysis from raw material characteristics to process parameters. The module supports dynamic updates and real-time queries, ensuring that the structured analysis results are updated synchronously with changes in the combined pattern library, thus ensuring data consistency and integrity.

[0160] Example 2:

[0161] In certain alcohol biomass processing scenarios, raw materials from different regions and of different types possess complex and high-dimensional performance indicators. When new raw materials or batches are entered, the existing database cannot automatically adjust its classification and hierarchical relationships, resulting in low data retrieval efficiency across regions and types. This also limits the dynamic management and combinatorial analysis capabilities of raw material performance space, posing challenges to alcohol processing technology optimization and raw material selection. To address these issues, this invention employs a cloud computing-based method for storing and maintaining the characteristics of alcohol biomass raw materials, the structure of which is shown in Figure 2. The specific implementation process of this method is as follows:

[0162] The characteristic data of biomass raw materials were collected according to a stratified system, and the collected data were standardized to form raw material characteristic data.

[0163] The raw material characteristic data is mapped into a multi-dimensional feature vector to form a raw material performance vector space;

[0164] Incremental dynamic clustering is performed on the feature vectors to generate hierarchical indexes, while updating the feature vectors and hierarchical information.

[0165] The feature vectors are correlated and mapped with alcohol processing parameters to form a causal mapping result;

[0166] Nonlinear pattern recognition is performed based on feature vectors, hierarchical information, and causal mapping results to generate a combined pattern library, and the feature vectors and process parameters of each combination are recorded.

[0167] It receives external query requests and returns structured analysis results through hierarchical indexes, causal mapping results, and a combined pattern library.

[0168] Specifically, the system systematically collects data on biomass raw materials from different regions and types through a data acquisition and standardization module. During the acquisition process, a pre-set multi-level hierarchical system is used, including regional groups, raw material type layers, and specific raw material layers. This system acquires industrial analysis indicators, elemental composition indicators, component indicators, and physicochemical property indicators for each specific raw material. Through standardization processing, the collected data is unified in units and normalized in value. Missing or abnormal data is marked or filled in, thereby forming raw material characteristic data with a unified format that can be directly used for subsequent analysis. In this process, the system can dynamically adapt to newly entered raw materials or batches, achieving data integrity and consistency management.

[0169] The feature vector construction module maps standardized raw material characteristic data into multi-dimensional feature vectors, forming a raw material performance vector space. In the specific implementation process, a vector dimension is assigned to each raw material indicator and a vector encoding rule is set. The numerical indicators are mapped through linear or nonlinear mapping functions. Convert to vector representation, where This indicates the numerical value of the raw material characteristic index. It represents the numerical values ​​of the corresponding dimension in the vector space, ensuring that different types of indicators can be compared and calculated in high-dimensional space; the generated raw material performance vector space can support similarity calculation and combination analysis of raw materials across regions and types.

[0170] Incremental dynamic clustering of feature vectors is performed using self-organizing clustering and hierarchical indexing modules. In the specific implementation process, the system initializes cluster centers and establishes a hierarchical structure, assigns existing feature vectors to the initial cluster centers, and calculates the vector distance between a new input feature vector and the existing cluster centers. and combined with local density The system determines whether to add to an existing cluster or generate a new cluster center, while dynamically adjusting the location of the cluster centers and the hierarchical index to maintain the dynamic updating of the raw material performance space and ensure the ability to quickly retrieve and combine data across regions and types.

[0171] In the process association mapping module, feature vectors are associated with alcohol processing parameters. Specifically, the process first acquires the set of alcohol processing parameters and quantifies them to generate process parameter vectors. Then, the raw material feature vectors and process parameter vectors are input into the association mapping model. The input vectors undergo dimensionality calibration and embedding transformation to map the two types of vectors to a unified association space. Finally, association data representing the correspondence between the two is generated through similarity or association calculations. ,in Represents the feature vector of raw materials. Represents a vector of process parameters. This represents the correlation matrix between raw materials and process parameters, thus forming a causal mapping result that supports the matching of raw material selection and process conditions.

[0172] The system achieves nonlinear pattern recognition of cross-regional raw material combinations through a combined pattern recognition module. In the specific implementation process, the system combines input feature vectors to construct a set of raw material combinations to be identified, and matches these combinations with hierarchical information to determine hierarchical relationships. Subsequently, the combined feature vectors and causal mapping results are input into the nonlinear pattern recognition model, and nonlinear transformations are performed. Extracting high-dimensional representations of combined features The system performs pattern partitioning on the high-dimensional representation to generate pattern data representing different combination characteristics. Finally, it generates structured combination pattern entries and stores them in the combination pattern library. At the same time, it records the feature vector and process parameters corresponding to each combination, providing a queryable basis for combination analysis and optimization.

[0173] The intelligent query interface module receives external query requests and returns structured analysis results. In its implementation, the system parses the fields of the combined pattern library entries, extracts basic field units, and labels them according to preset semantic categories to form a set of labeled fields. Then, based on the field set, it identifies the references and hierarchical relationships between fields to generate a field relationship structure. According to the relationship structure, the field set is grouped to form structured fragments with a fixed organizational form. Subsequently, the structured fragments are serialized and combined to generate structured analysis results that can be directly accessed and analyzed by external systems, ensuring efficient querying and intelligent decision support for cross-regional and cross-type raw material combinations.

[0174] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A cloud computing-based data storage and maintenance system for alcohol-based biomass raw materials, characterized in that: The system includes the following modules: a data acquisition and standardization module, used to collect characteristic data of biomass raw materials according to a hierarchical system and standardize the collected data to form raw material characteristic data; the hierarchical acquisition of characteristic data of biomass raw materials includes: dividing biomass raw materials into preset regional groups according to geographical regions; establishing raw material type layers in each regional group according to raw material type; selecting corresponding specific raw materials under the raw material type layer to form a specific raw material layer, forming a multi-level hierarchical acquisition system of region, raw material type, and specific raw material; collecting raw material characteristic data including industrial analysis indicators, elemental composition indicators, component indicators, and physicochemical property indicators for each specific raw material; a feature vector construction module, used to map the raw material characteristic data into multi-dimensional feature vectors to form a raw material performance vector space; the mapping of raw material characteristic data into multi-dimensional feature vectors includes: determining the corresponding vector dimension for each indicator in the raw material characteristic data and setting vector encoding rules; converting the numerical values ​​of each indicator into numerical representations in the vector space through linear or nonlinear mapping functions to form multi-dimensional feature vectors; and a self-organizing clustering and hierarchical indexing module, used to perform incremental dynamic clustering of feature vectors. The system comprises the following modules: a class for generating a hierarchical index and updating feature vectors and hierarchical information; a process association mapping module for associating feature vectors with alcohol processing parameters to form a causal mapping result; the associating feature vectors with alcohol processing parameters includes: obtaining a preset set of alcohol processing parameters and quantizing each parameter to obtain a process parameter vector; inputting the process parameter vector and feature vector into the association mapping model; dimensional calibration of the input feature vector and process parameter vector to unify vector length, index position, and data format; mapping the calibrated feature vector and process parameter vector to the same association space through embedding transformation; generating association data representing the correspondence between the two in the association space; and forming a causal mapping result based on the association data; a combined pattern recognition module for performing nonlinear pattern recognition based on feature vectors, hierarchical information, and causal mapping results to generate a combined pattern library and record the feature vector and process parameters of each combination; and an intelligent query interface module for receiving external query requests and returning structured analysis results through the hierarchical index, causal mapping results, and combined pattern library.

2. The cloud computing-based alcohol biomass raw material characteristic data storage and maintenance system according to claim 1, characterized in that, The standardization process includes: unifying the units of the collected raw material characteristic data; normalizing the raw material characteristic data; marking or filling in missing or abnormal data; and forming the processed data into a unified format to generate raw material characteristic data.

3. The cloud computing-based alcohol biomass raw material characteristic data storage and maintenance system according to claim 1, characterized in that, The incremental dynamic clustering includes: initializing cluster centers and hierarchical structure, and assigning existing feature vectors to the initial cluster centers; calculating the vector distance between a new input feature vector and the existing cluster centers; assigning the new feature vector to the existing cluster centers or generating new cluster centers based on the vector distance and local density; dynamically adjusting the position of the cluster centers and the hierarchical structure; and updating the cluster identifier and corresponding hierarchical index information of the feature vectors.

4. The cloud computing-based alcohol biomass raw material characteristic data storage and maintenance system according to claim 1, characterized in that, The nonlinear pattern recognition includes: combining input feature vectors to generate a set of raw material combinations to be identified; matching the raw material combination set with corresponding hierarchical information to determine the hierarchical relationship of each combination; inputting the combined feature vectors and causal mapping results into a nonlinear pattern recognition model; extracting features from the input data through nonlinear transformation to obtain a high-dimensional representation characterizing the combination characteristics; performing pattern partitioning on the high-dimensional representation to obtain pattern data representing different combination characteristics; generating combination pattern entries to describe the combination characteristics, and recording the combination pattern entries as structured combination pattern data.

5. The cloud computing-based alcohol biomass raw material characteristic data storage and maintenance system according to claim 1, characterized in that, The process of generating the combined pattern library includes: grouping the combined pattern data according to a preset pattern classification system; creating a unique identifier for each combined pattern and establishing an index relationship; writing the combined pattern data into the data structure of the combined pattern library according to the index relationship; and recording the corresponding feature vector and process parameters for each combined pattern.

6. The cloud computing-based alcohol biomass raw material characteristic data storage and maintenance system according to claim 1, characterized in that, The structured analysis results include: parsing the pattern entries in the combined pattern library to extract basic field units; labeling the basic field units according to preset semantic categories to form a field set with category labels; identifying the reference relationships and hierarchical relationships between fields based on the field set to generate a field relationship structure; grouping and dividing the field set according to the field relationship structure to form structured fragments with a fixed organizational form; and serializing and combining the structured fragments to generate structured analysis results.

7. A cloud computing-based method for storing and maintaining characteristic data of alcohol-based biomass feedstocks, characterized in that: The method for the cloud-based alcohol biomass raw material characteristic data storage and maintenance system according to any one of claims 1-6 includes: collecting characteristic data of biomass raw materials according to a hierarchical system, and standardizing the collected data to form raw material characteristic data; mapping the raw material characteristic data into multi-dimensional feature vectors to form a raw material performance vector space; performing incremental dynamic clustering on the feature vectors to generate a hierarchical index, and updating the feature vectors and hierarchical information simultaneously; associating the feature vectors with alcohol processing parameters to form a causal mapping result; performing nonlinear pattern recognition based on the feature vectors, hierarchical information, and causal mapping result to generate a combined pattern library, and recording the feature vectors and process parameters of each combination; receiving external query requests and returning structured analysis results through the hierarchical index, causal mapping result, and combined pattern library.

Images