Import and export coal quality intelligent evaluation and classification system and method

Abstract

The present application relates to the technical field of coal quality evaluation, and discloses an import and export coal quality intelligent evaluation and classification system and method, which comprises a data regularization module, a feature sorting module, a weight fusion module, a compliance evaluation module, a trade classification module and a report generation module. The system performs data regularization on the multi-source data of import and export coal to obtain a regularized feature sequence. Based on historical evaluation results, the regularized feature sequence is sorted according to feature priority to obtain a feature priority sequence. Based on the feature priority sequence, the regularized feature sequence is adaptively weighted and fused to obtain a weighted feature sequence. The weighted feature sequence is evaluated for compliance to obtain a coal quality evaluation grade and a preliminary classification result. According to the preliminary classification result of the coal quality, a coal trade classification conclusion is generated. According to the coal quality evaluation grade and the trade classification conclusion, a coal quality evaluation and classification report is generated. The present application can improve the efficiency of import and export coal quality evaluation and classification.

Description

Technical Field

[0001] This invention relates to the field of coal quality assessment technology, and in particular to an intelligent assessment and classification system and method for the quality of imported and exported coal. Background Technology

[0002] In the field of import and export coal quality assessment and classification, existing technologies have significant shortcomings when processing multi-source data. Due to the complexity and heterogeneity of coal quality-related data sources, existing technologies lack efficient standardization mechanisms, making it difficult to achieve deep integration and standardized processing of different types of data. This can easily lead to data bias or redundancy, resulting in insufficient accuracy of characteristic information used for subsequent assessments. Consequently, these technologies cannot provide reliable data support for quality assessments, thus affecting the credibility of the assessment results.

[0003] Existing technologies lack dynamic adaptability in feature prioritization and weight allocation. They often employ fixed sorting rules and weight settings, failing to adjust feature importance in real time based on historical evaluation results. This leads to insufficient attention being paid to key quality features, while secondary features interfere with the evaluation process. This not only results in an inaccurate preliminary coal quality classification but also significantly increases the time required for the evaluation process, making it difficult to meet the actual needs of rapid coal quality assessment and classification in import and export trade scenarios. Therefore, improving the efficiency of data report generation has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides an intelligent assessment and classification system and method for the quality of imported and exported coal, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, this invention provides an intelligent assessment and classification system for the quality of imported and exported coal. The system comprises a data normalization module, a feature sorting module, a weight fusion module, a compliance assessment module, a trade classification module, and a report generation module, wherein:

[0006] The data normalization module is used to normalize the multi-source data of imported and exported coal to obtain the normalized feature sequence of the multi-source data.

[0007] The feature sorting module is used to sort the regular feature sequence by feature priority based on the historical assessment results of the imported and exported coal, so as to obtain the feature priority sequence of the regular feature sequence.

[0008] The weight fusion module is used to perform adaptive weight fusion on the regular feature sequence based on the feature priority sequence to obtain a weighted feature sequence of the regular feature sequence.

[0009] The compliance assessment module is used to perform compliance assessment on the weighted feature sequence to obtain the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence.

[0010] The trade classification module is used to generate a coal trade classification conclusion based on the preliminary coal quality classification results.

[0011] The report generation module is used to generate a coal quality assessment and classification report for the imported and exported coal based on the coal quality assessment grade and the coal trade classification conclusion.

[0012] In a preferred embodiment, when the data normalization module performs data normalization on multi-source data of imported and exported coal to obtain a normalized feature sequence of the multi-source data, it is specifically used for:

[0013] Heterogeneous data fusion is performed on multi-source data of imported and exported coal to obtain an intermediate dataset of the multi-source data.

[0014] The intermediate dataset is standardized by dimension to obtain the standardized dataset of the intermediate dataset;

[0015] Feature extraction is performed on the standardized dataset to obtain a regularized feature sequence of the standardized dataset.

[0016] In a preferred embodiment, when the feature sorting module performs feature priority sorting on the regularized feature sequence based on the historical assessment results of the imported and exported coal to obtain the feature priority sequence of the regularized feature sequence, it is specifically used for:

[0017] Based on the historical assessment results of imported and exported coal, data association analysis is performed on the regularized feature sequence to obtain the association feature set of the regularized feature sequence;

[0018] Priority scores are calculated on the associated feature set to obtain the priority score distribution of the associated feature set, wherein the calculation formula for the priority score distribution is as follows:

[0019] ;

[0020] In the formula, For the first Priority scores for each feature. For the first The correlation strength coefficient between each feature and the historical evaluation results, For the first The frequency of occurrence of each feature in historical assessments For the first The coefficient of variation of each feature in the current batch of data. The maximum value of the coefficient of variation of all features. It is the natural logarithm function. The preset association strength weighting coefficients, The preset frequency weighting coefficients, These are preset stability weighting coefficients;

[0021] Based on the priority score distribution, the regularized feature sequence is sorted to obtain the feature priority sequence of the regularized feature sequence.

[0022] In a preferred embodiment, when the feature sorting module performs data association analysis on the regularized feature sequence based on the historical assessment results of the imported and exported coal to obtain the associated feature sequence of the regularized feature sequence, it is specifically used for:

[0023] The historical assessment results of the imported and exported coal are filtered for timeliness to obtain a valid historical dataset of the historical assessment results;

[0024] Dynamic correlation analysis is performed between the effective historical dataset and the regularized feature sequence to obtain the correlation strength distribution between the effective historical dataset and the regularized feature sequence;

[0025] Based on the correlation strength distribution, the regular feature sequence is optimized and recombined to obtain the correlation feature sequence of the regular feature sequence.

[0026] In a preferred embodiment, when the weight fusion module performs adaptive weight fusion on the regularized feature sequence based on the feature priority sequence to obtain a weighted feature sequence of the regularized feature sequence, it is specifically used for:

[0027] The feature priority sequence is subjected to priority distribution analysis to obtain the feature distribution structure of the feature priority sequence;

[0028] Based on the aforementioned feature distribution structure, adaptive weight calculation is performed on the feature priority sequence to obtain a weight value sequence of the feature priority sequence, wherein the calculation formula for the weight value sequence is as follows:

[0029] ;

[0030] In the formula, For the first The weight values ​​of each feature, For the first Priority scores for each feature. For the first Priority scores for each feature. The minimum score in the feature priority sequence, The maximum score in the feature priority sequence. The preset weight distribution adjustment factor, These are the preset high-priority feature enhancement coefficients. For indicator functions, This refers to the set of high-priority features in the priority distribution analysis. The total number of features, It is an exponential function;

[0031] The regular feature sequence and the weight value sequence are weighted and fused to obtain the weighted feature sequence of the regular feature sequence.

[0032] In a preferred embodiment, when the weight fusion module performs a weighted fusion of the regularized feature sequence and the weight value sequence to obtain a weighted feature sequence of the regularized feature sequence, it is specifically used for:

[0033] Based on the weight value sequence, the regularized feature sequence is subjected to sequence alignment verification to obtain the weight alignment feature mapping of the regularized feature sequence;

[0034] The weighted aligned feature map is fused in layers to obtain a weighted intermediate representation of the weighted aligned feature map;

[0035] The weighted feature intermediate representation is reconstructed to obtain the weighted feature sequence of the regular feature sequence.

[0036] In a preferred embodiment, when the compliance assessment module performs a compliance assessment on the weighted feature sequence to obtain the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence, it is specifically used for:

[0037] The weighted feature sequence is subjected to multi-level threshold determination to obtain a multi-dimensional conformity status identifier of the weighted feature sequence.

[0038] Confidence prediction is performed on the multi-dimensional compliance status identifiers to obtain the comprehensive confidence rating of the multi-dimensional compliance status identifiers;

[0039] Based on the comprehensive confidence rating and the multi-dimensional compliance status identifier, the coal quality assessment level and preliminary coal quality classification results of the weighted feature sequence are generated.

[0040] In a preferred embodiment, when the trade classification module generates a coal trade classification conclusion based on the preliminary coal quality classification result, it is specifically used for:

[0041] The preliminary coal quality classification results are subjected to multi-standard compliance verification to obtain a summary of the compliance judgment of the preliminary coal quality classification results.

[0042] The compliance determination summary is integrated with decision factors to obtain a trade adaptability plan for the compliance determination summary;

[0043] A risk situation analysis was conducted on the aforementioned trade adaptability plan to obtain risk management recommendations for the plan.

[0044] Based on the aforementioned risk management recommendations, the trade adaptability scheme is optimized to obtain the coal trade classification conclusion based on the preliminary coal quality classification results.

[0045] In a preferred embodiment, when the report generation module generates a coal quality assessment and classification report for the imported and exported coal based on the coal quality assessment grade and the coal trade classification conclusion, it is specifically used for:

[0046] By mapping the coal quality assessment grades to the coal trade classification conclusions, the core topology of the import and export coal report is obtained.

[0047] Based on the core topology of the report, the preset report template is dynamically and structurally filled to obtain the initial draft report of the imported and exported coal.

[0048] Enhanced indexing was performed on the initial draft report to obtain the coal quality assessment and classification report for the imported and exported coal.

[0049] To address the aforementioned problems, this invention also provides a method for intelligent assessment and classification of imported and exported coal quality, the method comprising:

[0050] S1. Perform data normalization on the multi-source data of imported and exported coal to obtain the normalized feature sequence of the multi-source data;

[0051] S2. Based on the historical assessment results of the imported and exported coal, the regularized feature sequence is sorted by feature priority to obtain the feature priority sequence of the regularized feature sequence;

[0052] S3. Based on the feature priority sequence, perform adaptive weight fusion on the regular feature sequence to obtain the weighted feature sequence of the regular feature sequence;

[0053] S4. Perform a conformity assessment on the weighted feature sequence to obtain the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence;

[0054] S5. Based on the preliminary coal quality classification results, generate a coal trade classification conclusion based on the preliminary coal quality classification results.

[0055] S6. Based on the coal quality assessment grade and the coal trade classification conclusion, generate a coal quality assessment and classification report for the imported and exported coal.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] 1. The intelligent assessment and classification system and method for import and export coal quality of this invention achieves significant results in data processing and feature optimization. The data normalization module can perform heterogeneous fusion, dimensional standardization, and feature extraction on multi-source coal data to generate a normalized feature sequence, ensuring data accuracy and consistency and providing high-quality data support for subsequent assessments. The feature sorting module combines historical assessment results to perform data correlation analysis and determines the feature priority sequence through priority score calculation, which can accurately lock in key quality features, making the assessment focus more in line with actual needs and improving the accuracy of the assessment direction.

[0058] 2. This invention offers significant advantages in evaluation efficiency and result reliability. The weight fusion module adaptively calculates weights based on feature priority and performs weighted fusion, making the weighted feature sequence more aligned with the core of quality assessment. The compliance assessment module outputs accurate quality assessment levels and preliminary classification results through multi-level threshold determination and confidence prediction. The trade classification module optimizes trade classification conclusions through compliance verification and risk analysis. The report generation module dynamically generates assessment reports. The entire process is highly efficient and coherent, significantly improving the efficiency and reliability of import and export coal quality assessment and classification, and meeting the practical needs of trade scenarios. Attached Figure Description

[0059] Figure 1 A system architecture diagram of an intelligent assessment and classification system for the quality of imported and exported coal provided in an embodiment of the present invention;

[0060] Figure 2 This is a flowchart illustrating an intelligent assessment and classification method for the quality of imported and exported coal, provided as an embodiment of the present invention.

[0061] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments belong to some, but not all, embodiments of the present invention. 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.

[0063] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “said” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0064] Depending on the context, the word "if" or "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0065] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0066] In practice, the server-side equipment deployed in an intelligent assessment and classification system for the quality of imported and exported coal may consist of one or more devices. This intelligent assessment and classification system for the quality of imported and exported coal can be implemented as: a business instance, a virtual machine, or hardware devices. For example, this intelligent assessment and classification system for the quality of imported and exported coal can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, this intelligent assessment and classification system for the quality of imported and exported coal can be understood as software deployed on a cloud node, used to provide an intelligent assessment and classification system for the quality of imported and exported coal to various user terminals. Alternatively, this intelligent assessment and classification system for the quality of imported and exported coal can also be implemented as a virtual machine deployed on one or more devices in a cloud node. This virtual machine contains application software for managing various user terminals. Alternatively, this intelligent assessment and classification system for the quality of imported and exported coal can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more hardware devices configured to provide an intelligent assessment and classification system for the quality of imported and exported coal to various user terminals.

[0067] In terms of implementation, the intelligent assessment and classification system for import and export coal quality and the user terminal are mutually compatible. That is, if the intelligent assessment and classification system for import and export coal quality is implemented as an application installed on a cloud service platform, then the user terminal is a client that establishes a communication connection with the application; or if the intelligent assessment and classification system for import and export coal quality is implemented as a website, then the user terminal is implemented as a webpage; or if the intelligent assessment and classification system for import and export coal quality is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0068] like Figure 1The diagram shown is a system architecture diagram of an intelligent assessment and classification system for the quality of imported and exported coal provided in an embodiment of the present invention.

[0069] The intelligent assessment and classification system 100 for import and export coal quality described in this invention can be set up in a cloud server. In terms of implementation, it can be used as one or more service devices, or as an application installed in the cloud (e.g., a mobile service operator's server, server cluster, etc.), or it can be developed as a website. Depending on the functions implemented, the intelligent assessment and classification system 100 for import and export coal quality may include a data normalization module 101, a feature sorting module 102, a weight fusion module 103, a compliance assessment module 104, a trade classification module 105, and a report generation module 106. The modules described in this invention can also be called units, referring to a series of computer program segments that can be executed by an electronic device's processor and perform a fixed function, stored in the electronic device's memory.

[0070] In this embodiment of the invention, in an intelligent assessment and classification system for the quality of imported and exported coal, each of the above-mentioned modules can be implemented independently and can call other modules. Here, "calling" can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. In the intelligent assessment and classification system for the quality of imported and exported coal provided by this embodiment of the invention, the applicable scope of the system architecture can be adjusted by adding modules and directly calling them without modifying the program code, achieving cluster-based horizontal expansion to quickly and flexibly expand the system. In practical applications, the above-mentioned modules can be set in the same device or different devices, or they can be set in a virtual device, such as a service instance in a cloud server.

[0071] The following describes, with reference to specific embodiments, each component and specific workflow of an intelligent assessment and classification system for import and export coal quality:

[0072] The data normalization module 101 is used to normalize the multi-source data of imported and exported coal to obtain the normalized feature sequence of the multi-source data.

[0073] In this embodiment of the invention, when the data normalization module performs data normalization on multi-source data of imported and exported coal to obtain the normalized feature sequence of the multi-source data, it is specifically used for:

[0074] Heterogeneous data fusion is performed on multi-source data of imported and exported coal to obtain an intermediate dataset of the multi-source data.

[0075] The intermediate dataset is standardized by dimension to obtain the standardized dataset of the intermediate dataset;

[0076] Feature extraction is performed on the standardized dataset to obtain a regularized feature sequence of the standardized dataset.

[0077] Specifically, we first collected multi-source data on imported and exported coal. This data included physical property data recorded during the mining process, component analysis data obtained from laboratory tests, and category and origin data submitted during trade declarations. The data formats varied, with some being structured data in tabular form, some unstructured data in text form, and some single-item test data in numerical form. We then processed the data according to different formats. The unstructured text data was broken down into identifiable key information and converted into a structured format. The field names and order of the structured tabular data were standardized. The single-item test data in numerical form was supplemented with corresponding feature identifiers. Then, we checked for duplicates in all the processed data, deleted completely identical data entries, and removed abnormal data that clearly exceeded the reasonable range. Finally, we integrated these standardized, deduplicated, and anomaly-removed data into a single data framework to form an intermediate dataset of multi-source data.

[0078] Furthermore, each data item in the intermediate dataset is checked one by one, and the specific physical or chemical meaning of each data item is analyzed in depth. For example, some data items reflect the moisture content of coal, some reflect the calorific value of coal, and some show the content of harmful components in coal. Through such analysis, the unit of measurement used for each data item is clarified. Then, the name, corresponding physical or chemical meaning, and unit of measurement of all data items are recorded one by one and compiled into a dimensional list of the intermediate dataset to ensure that the dimensional information of each data item can be clearly grasped in subsequent processing.

[0079] Furthermore, referencing industry-standard practices in the field of imported and exported coal quality assessment, and considering the data format requirements of subsequent feature sorting and weight fusion processes, a unified unit of measurement is pre-defined for each type of data item with the same physical or chemical meaning. For example, for data items reflecting the calorific value of coal, a unit of measurement widely recognized in the industry and convenient for subsequent calculations is pre-defined. Then, based on the conversion relationship between the current unit of measurement of each data item and the pre-defined unified unit of measurement, the corresponding conversion coefficient is determined to ensure that the value under the current unit of measurement can be accurately converted to the value under the pre-defined unified unit of measurement through this coefficient.

[0080] Furthermore, according to the determined conversion coefficients, a numerical conversion operation is performed on each data item in the intermediate dataset. During the conversion process, for each data item, its current value is first extracted, and then calculated with the corresponding conversion coefficient to obtain the new value of the data item under the preset unified dimension. At the same time, the source information of each data item is checked in real time during the conversion process to ensure that the conversion operation corresponds to the correct data item, avoid conversion errors caused by data item confusion, and ensure that the numerical conversion of each data item is accurate.

[0081] Furthermore, after all data items have completed numerical transformation, the transformed dataset is subjected to secondary verification. First, it is checked whether the units of each data item have been unified to the preset units to confirm that there are no inconsistencies in units. Then, a portion of data items are randomly selected, and their transformation process is recalculated to verify whether the transformed values ​​are accurate. At the same time, it is checked whether there are any data items that have been omitted from the transformation in the dataset. After such comprehensive verification, it is confirmed that all data items have completed the correct unit transformation and the values ​​are accurate, and the standardized dataset of the intermediate dataset is obtained.

[0082] In summary, the process begins with a comprehensive review of the standardized dataset to identify data items directly relevant to the quality assessment of imported and exported coal. These items include, but are not limited to, key indicators reflecting coal quality, such as ash content, volatile matter content, sulfur content, calorific value, fixed carbon content, and caking index. For each selected key data item, its data source, testing methods, and the quality significance of its value are meticulously reviewed. For example, the ash content data is clarified to originate from laboratory ignition tests, with higher values ​​indicating more non-combustible impurities in the coal. The numerical range and normal fluctuation interval for each key data item are also recorded to ensure a clear and comprehensive understanding of its attributes. Finally, following the logical flow of imported and exported coal quality assessment, these reviewed key data items are arranged in an orderly manner, from basic physical properties to core chemical components and then to key quality indicators. This forms a clearly structured and logically coherent feature set, which constitutes the regularized feature sequence of the standardized dataset.

[0083] The feature sorting module 102 is used to sort the regular feature sequence by feature priority based on the historical assessment results of the imported and exported coal, so as to obtain the feature priority sequence of the regular feature sequence.

[0084] In this embodiment of the invention, when the feature sorting module performs feature priority sorting on the regular feature sequence based on the historical assessment results of the imported and exported coal to obtain the feature priority sequence of the regular feature sequence, it is specifically used for:

[0085] Based on the historical assessment results of imported and exported coal, data association analysis is performed on the regularized feature sequence to obtain the association feature set of the regularized feature sequence;

[0086] Priority scores are calculated on the associated feature set to obtain the priority score distribution of the associated feature set, wherein the calculation formula for the priority score distribution is as follows:

[0087] ;

[0088] In the formula, For the first Priority scores for each feature. For the first The correlation strength coefficient between each feature and the historical evaluation results, For the first The frequency of occurrence of each feature in historical assessments For the first The coefficient of variation of each feature in the current batch of data. The maximum value of the coefficient of variation of all features. It is the natural logarithm function. The preset association strength weighting coefficients, The preset frequency weighting coefficients, These are preset stability weighting coefficients;

[0089] Based on the priority score distribution, the regularized feature sequence is sorted to obtain the feature priority sequence of the regularized feature sequence.

[0090] When the feature sorting module performs data association analysis on the regularized feature sequence based on the historical assessment results of the imported and exported coal to obtain the associated feature sequence of the regularized feature sequence, it is specifically used for:

[0091] The historical assessment results of the imported and exported coal are filtered for timeliness to obtain a valid historical dataset of the historical assessment results;

[0092] Dynamic correlation analysis is performed between the effective historical dataset and the regularized feature sequence to obtain the correlation strength distribution between the effective historical dataset and the regularized feature sequence;

[0093] Based on the correlation strength distribution, the regular feature sequence is optimized and recombined to obtain the correlation feature sequence of the regular feature sequence.

[0094] Specifically, the historical assessment results of imported and exported coal are first compiled, and valid historical data that is complete and consistent with the current assessment scenario is selected. This includes the quality assessment grade, trade classification conclusion, and characteristic data used in the assessment for each batch of coal in the past. Then, each feature in the regularized feature sequence is matched with the valid historical data one by one, and the correspondence between the changes in feature values ​​and the assessment results is observed. If the change in a certain feature value can stably affect the assessment results, it is determined that its correlation is strong. The qualified features are extracted and classified according to the coal quality dimension to form a set of associated features of the regularized feature sequence.

[0095] Furthermore, for each feature in the associated feature set, the correlation strength coefficient is obtained by collecting data on its correlation with historical evaluation results. Valid historical evaluation records are reviewed one by one to establish a correlation statistical ledger for each feature, recording the specific numerical value and corresponding evaluation level for each occurrence of the feature. The frequency of changes in evaluation level when the feature value changes significantly is then calculated, and the proportion of this frequency to the total frequency of occurrences is calculated. If the proportion exceeds a set percentage, the correlation is considered strong, and this proportion is converted into a quantified correlation strength coefficient, i.e., the coefficient of correlation. The correlation strength coefficient of each feature.

[0096] Furthermore, the frequency of occurrence of features in historical assessments is collected to determine the first The frequency of occurrence of each feature is determined by: first, counting the total number of valid historical assessments; then, verifying each assessment record and counting the total number of times the feature was explicitly included in the assessment analysis; dividing the total number of included features by the frequency of occurrence of the total number of assessments; and simultaneously recording the corresponding assessment scenarios. The occurrence count, frequency, and scenario information are then compiled and archived.

[0097] Furthermore, the stability of features in the current batch of data is collected to calculate the coefficient of variation and the maximum value of the coefficient of variation for all features. The total number of samples in the current batch covering different sampling points and transportation periods is determined. The values ​​of features in each sample are extracted and listed. The list is sorted, and the difference between the maximum and minimum values ​​is calculated. Then, the mean, the deviation of each value from the mean, and the squared deviation are calculated. These are summed and divided by the total sample size to obtain the mean of the squared deviations. The square root is taken to obtain the standard deviation. The standard deviation is divided by the mean to obtain the nth... The coefficient of variation of each feature is calculated, and the coefficients of variation of all features involved in the evaluation are calculated using the same steps, and the maximum value is selected.

[0098] Furthermore, pre-defined weighting coefficients for correlation strength, frequency, and stability are determined. Referring to industry standards for assessing the quality of imported and exported coal, and considering the actual impact of these three dimensions on the conclusions of previous assessment projects, the importance of each dimension is converted into fixed coefficients according to the degree of impact, which are then used as weighting coefficients for correlation strength, frequency, and stability, respectively.

[0099] Further, calculate the priority score for each feature. First, multiply the association strength coefficient by the association strength weight coefficient to obtain the association strength dimension score; the larger the coefficient, the higher the score, and the higher the priority score. Next, add 1 to the occurrence frequency, perform a natural logarithm operation, and multiply by the frequency weight coefficient to obtain the occurrence frequency dimension score; the higher the frequency, the higher the score, and the higher the priority score. Finally, divide the feature's coefficient of variation by the maximum coefficient of variation, subtract the result from 1, and multiply by the stability weight coefficient to obtain the stability dimension score; the larger the coefficient of variation, the lower the score, and the lower the priority score. Summing the scores of the three dimensions yields the feature priority score. During calculation, each score item should be checked to avoid errors.

[0100] Furthermore, the impact of changes in the maximum value of the feature's coefficient of variation and its weighting coefficients is analyzed. If the coefficient of variation of a certain feature remains constant, an increase in the maximum coefficient of variation may increase its stability dimension score, and consequently, its priority score. If the coefficient of variation of a certain feature increases proportionally to the maximum coefficient of variation, the stability dimension score remains unchanged, and the contribution of that dimension to the priority score remains constant. When the weighting coefficients of association strength, frequency, and stability increase, the corresponding dimension scores have a more significant impact on the priority score, highlighting the importance of closely associated, frequently used, and stable features, respectively.

[0101] Furthermore, construct the priority score distribution of the associated feature set. Create a score distribution table, with the first column listing the complete names of all features and the second column containing the corresponding priority scores. After filling in the table, check whether the feature names cover the associated feature set. Randomly select features to review the score calculation process. Once the values ​​are confirmed to be correct, the priority score distribution of the associated feature set is formed.

[0102] Specifically, first, the timeliness standard for historical assessment results of imported and exported coal is determined. Referring to the general industry standards for the timeliness of historical data in the field of imported and exported coal quality assessment, a reasonable time range is determined in combination with the type of coal to be assessed. Historical assessment results outside this time range are directly marked as invalid because they cannot reflect the current changes in coal quality characteristics. Next, the data integrity of historical assessment results within the time range is checked. Each result is confirmed to contain a complete quality assessment grade, the corresponding trade classification conclusion, and all characteristic data supporting the assessment conclusion. If there is any missing or ambiguous information, such results are removed. Finally, all historical assessment results that simultaneously meet the time range and data integrity requirements are collected. The collection of these results is the valid historical dataset of historical assessment results.

[0103] Furthermore, the valid historical dataset is first categorized and split according to the type of evaluation conclusion, clarifying the range of coal characteristic data corresponding to each type of evaluation conclusion. For example, under the "high-quality" evaluation conclusion, the specific ranges of the minimum value of coal calorific value and the maximum value of sulfur content are defined, while under the "qualified" evaluation conclusion, the corresponding characteristic value ranges are defined. Then, each feature in the regularized feature sequence is matched one by one with the split evaluation conclusions, recording the total number of times each feature appears under a certain type of evaluation conclusion. At the same time, it is continuously observed whether the evaluation conclusion changes when the value of the feature changes. For example, when the value of a feature increases from the upper limit of the qualified range to the high-quality range, the frequency of the evaluation conclusion changing from qualified to high-quality is recorded. The correlation between features and evaluation conclusions is comprehensively recorded.

[0104] Furthermore, based on the total number of occurrences of each feature under different evaluation conclusions and the frequency of changes in evaluation conclusions, the degree of association between the feature and the valid historical dataset is comprehensively determined: if a feature occurs frequently under a certain type of evaluation conclusion, and its numerical changes frequently cause changes in the evaluation conclusion, it indicates that the feature has a significant impact on the evaluation conclusion, and the degree of association is determined to be high; if a feature occurs infrequently, and its numerical changes have almost no impact on the evaluation conclusion, the degree of association is determined to be low. The degree of association of all features is quantified and labeled according to the unified standard of "high", "medium" and "low", and then the feature names and their corresponding degrees of association are matched one by one to form a structured list. This list is the distribution of the association strength between the valid historical dataset and the regular feature sequence.

[0105] Furthermore, the association strength label of each feature is extracted from the association strength distribution. Using the association strength of "high", "medium" and "low" as the primary sorting criteria, all features in the regular feature sequence are initially sorted to ensure that features with "high" association strength are all at the beginning of the sequence, features with "medium" association strength follow closely behind, and features with "low" association strength are at the end of the sequence. Then, combined with the conventional logical process of coal quality assessment, the sequence after the initial sorting is checked. If there are any unreasonable logical orders, the sequence is locally adjusted to ensure that the feature arrangement conforms to the assessment logic.

[0106] Furthermore, the integrity of the adjusted sequence is verified by checking whether each feature in the regularized feature sequence has been included in the current sequence to ensure that no features are omitted or duplicated. Then, it is verified whether the arrangement order of features in the sequence can provide effective support for subsequent priority ranking and weight fusion. For example, whether features with high correlation strength can be prioritized in subsequent steps, and whether the position of each feature in the sequence can accurately reflect its correlation value with historical evaluation results. The ordered set of features after integrity verification and rationality verification is the correlation feature sequence of the regularized feature sequence.

[0107] In summary, the process begins by identifying and verifying the timeliness and integrity of historical datasets. Then, the distribution of correlation strength is obtained through classification, feature matching, and correlation quantification. Finally, the sequence is optimized and recombined based on correlation strength ranking, logical adjustment, and integrity confirmation to obtain the correlation feature sequence. The entire process is supported by clear operational standards and evaluation logic, ensuring that the products of each step accurately meet the requirements and laying a reliable foundation for the feature processing stage of subsequent import and export coal quality assessment.

[0108] In summary, each feature score is extracted from the priority score distribution, and the associated features are sorted from high to low scores. The remaining features that are not included in the associated feature set are found from the regular feature sequence, sorted from most frequent to least frequent in historical evaluations, and appended to the associated features. The overall order of the regular feature sequence is then adjusted, and the resulting ordered feature set is the feature priority sequence of the regular feature sequence.

[0109] The weight fusion module 103 is used to perform adaptive weight fusion on the regular feature sequence based on the feature priority sequence to obtain a weighted feature sequence of the regular feature sequence.

[0110] In this embodiment of the invention, when the weight fusion module performs adaptive weight fusion on the regularized feature sequence based on the feature priority sequence to obtain a weighted feature sequence of the regularized feature sequence, it is specifically used for:

[0111] The feature priority sequence is subjected to priority distribution analysis to obtain the feature distribution structure of the feature priority sequence;

[0112] Based on the aforementioned feature distribution structure, adaptive weight calculation is performed on the feature priority sequence to obtain a weight value sequence of the feature priority sequence, wherein the calculation formula for the weight value sequence is as follows:

[0113] ;

[0114] In the formula, For the first The weight values ​​of each feature, For the first Priority scores for each feature. For the first Priority scores for each feature. The minimum score in the feature priority sequence, The maximum score in the feature priority sequence. The preset weight distribution adjustment factor, These are the preset high-priority feature enhancement coefficients. For indicator functions, This refers to the set of high-priority features in the priority distribution analysis. The total number of features, It is an exponential function;

[0115] The regular feature sequence and the weight value sequence are weighted and fused to obtain the weighted feature sequence of the regular feature sequence.

[0116] When the weight fusion module performs a weighted fusion of the regularized feature sequence and the weight value sequence to obtain a weighted feature sequence of the regularized feature sequence, it is specifically used for:

[0117] Based on the weight value sequence, the regularized feature sequence is subjected to sequence alignment verification to obtain the weight alignment feature mapping of the regularized feature sequence;

[0118] The weighted aligned feature map is fused in layers to obtain a weighted intermediate representation of the weighted aligned feature map;

[0119] The weighted feature intermediate representation is reconstructed to obtain the weighted feature sequence of the regular feature sequence.

[0120] Specifically, all features in the feature priority sequence are first split into different groups according to their priority, strictly following the original sorting order in the sequence. The feature that appears first in the sequence is assigned to the highest priority group, which becomes the set of high-priority features involved later. Moving sequentially, features in the middle of the sequence are assigned to the medium-priority group, and features at the very end of the sequence are assigned to the low-priority group. The number of features within each group is then counted, and the feature names for each group are recorded. Simultaneously, the coal quality assessment dimensions pointed to by the features within each group are analyzed. For example, it is determined whether the features in the highest-priority group reflect core coal quality indicators such as calorific value and sulfur content, and whether the features in the lower-priority group are auxiliary assessment indicators such as appearance and color uniformity. Finally, the grouping results, the number of features in each group, the feature names in each group, and the corresponding assessment dimensions are compiled into a well-organized document. This document is the feature distribution structure of the feature priority sequence.

[0121] Furthermore, the first Priority scores of each feature With the Priority scores of each feature These values ​​are specific values ​​obtained through comprehensive calculation based on the correlation strength between the corresponding features and historical assessment results, the frequency of occurrence in historical assessments, and the stability in the current batch of data. These values ​​directly reflect the priority of the features in coal quality assessment. Covering the feature priority sequence except for the first The priority scores of all features other than the first feature provide data support for the summation of the denominator in subsequent weight calculations; the minimum score in the feature priority sequence. It is a priority score for all features in the sequence, including , All priority scores, including the highest score, are compared one by one, and the score with the smallest value is selected from all scores; the highest score in the feature priority sequence is also considered. This means selecting the score with the largest value from all priority scores. and Together, they are used to eliminate the magnitude difference between scores of different priorities and ensure that the calculation results are within a reasonable range.

[0122] Furthermore, the preset weight distribution adjustment factor This is a fixed value determined with reference to the conventional requirements for feature weight differences in the field of imported and exported coal quality assessment. If past assessment projects required a more significant difference in weights between different priority features to highlight the influence of core features, a larger value was set. If a smoother weight distribution and a reduction in weight differences between features were desired, a smaller value was set. The setting process requires verification of the adjustment effect based on multiple batches of assessment results; the preset high-priority feature enhancement coefficient... The value is a fixed value determined based on the importance of high-priority features in the evaluation. If the impact of high-priority features on the evaluation conclusion needs to be significantly higher than that of other features, a larger value should be set. If only a slight enhancement of their importance is needed, a smaller value should be set. When setting the value, it is necessary to review the actual role of high-priority features in the final conclusion in historical evaluations.

[0123] Furthermore, indicator functions The value must first be determined by identifying the set of high-priority features. This set is precisely the highest priority group formed by previously splitting the feature priority sequence, used to determine the first... Does this feature belong to...? At that time, it is necessary to check the first Does the name of the feature exist? In the feature list, if a feature exists, the indicator function takes a value of 1; otherwise, it takes a value of 0. The total number of features... This value is the count of all features in a regularized feature sequence. During the count, each feature in the sequence must be checked individually to ensure no double counting and no feature is omitted. This value is used to define the range of the subsequent summation operation in the denominator, i.e., from the first feature to the... One characteristic.

[0124] Further, calculate the first The weight values ​​of each feature First, construct the molecular part. The molecule consists of two parts, and the first part needs to be calculated first. minus The result is then divided by... minus The result is a normalized value, which is then multiplied by a preset weight distribution adjustment factor. Finally, the results are subjected to exponential calculation. This exponential calculation amplifies the differences in scores after normalization between different features, making the advantages of high-priority scoring features more obvious. The second part is the preset high-priority feature enhancement coefficient. With indicator functions The product of , if the first Each feature belongs to When the indicator function takes a value of 1, the numerator increases. The value is incremented by 0 if it does not belong to a certain category, thus emphasizing the importance of high-priority features; then the denominator is constructed, which is a subset of all... Features, from the first to the second The sum of the numerators of each feature is obtained by adding them one by one, ensuring that the sum of the weights of all features equals 1; finally, the numerator is divided by the denominator to obtain the result of the first feature. The weight values ​​of each feature Weights of all features Arranging the features one by one in their original order, the resulting ordered set of weights is the weight value sequence of the feature priority sequence. The core significance of this process is to accurately reflect the importance of each feature relative to other features in coal quality assessment through quantitative calculation.

[0125] Furthermore, when the first Priority scores of each feature When it increases, minus The value will increase accordingly, and the normalized value will also increase accordingly. After adjustment and exponential operations, the value of the first part of the numerator will increase significantly, while the value of the second part will increase significantly while the denominator remains unchanged. The weight values ​​of each feature It will increase accordingly; when the preset weight distribution adjustment factor When it increases, Multiplying by the normalized value will increase the result; the exponential operation has a more significant amplifying effect on score differences, leading to a decrease in the weight of high-priority score features. Further increase the weight values ​​of low-priority scoring features. Further reduction makes the weight differences between features more prominent; when the preset high-priority feature enhancement coefficient is reduced... When it increases, if the first Each feature belongs to The value of the second part in the molecule will increase, and the weight value will increase. It will increase accordingly, if it does not belong to Weight value Then it will not be affected, ultimately making The weight differences between the features in the data and those in other features are more significant.

[0126] Furthermore, when the minimum score in the feature priority sequence When decreasing, minus As the value increases, the normalized value also increases. After adjustment and exponential operation, the value of the first part of the numerator increases, and the weight value... It will increase accordingly; when the maximum score When it increases, minus Divide the result by minus The normalized value obtained will decrease after... After adjustment and exponential operation, the value of the first part of the numerator decreases, and the weight value... It will decrease accordingly; when the first This feature never belongs to Become to belong to At that time, the molecular portion of this feature will increase. The value of the fraction increases as the denominator increases, if the first... Each feature belongs to Its weight value It will usually remain at a high level; if it does not belong to the category, the weight value will be... It will decrease as the denominator increases.

[0127] Further, after generating the weight value sequence, the regularized feature sequence and the weight value sequence are weighted and fused. First, the two sequences are placed side-by-side, and the positions of each element in the sequence are checked one by one to ensure that the first feature of the regularized feature sequence corresponds to the first weight value of the weight value sequence, and the second feature corresponds to the second weight value, until all features and weight values ​​are perfectly aligned. If a mismatch is found between a feature and a weight value, the sequence sorting process is immediately checked, corrected, and realigned. Then, for each aligned feature and weight value, the specific numerical value of the feature is extracted, and the value is multiplied by the corresponding weight value. During the calculation, the matching relationship between the numerical value and the weight value is carefully checked to avoid incorrect or missed multiplications, and the result of each feature multiplication is recorded. Finally, the results of multiplying all features are arranged sequentially according to the original order of the regularized feature sequence, forming a new ordered feature set. This new ordered feature set is the weighted feature sequence of the regularized feature sequence.

[0128] Specifically, first, all feature names in the regularized feature sequence are listed one by one, and simultaneously, all weight values ​​in the weight value sequence are listed in their original order, forming two parallel lists. Then, starting from the first feature name and the first weight value, the positions of elements in the two lists are checked one by one to confirm whether a feature in the regularized feature sequence matches a weight value in the weight value sequence. If a mismatch is found, such as a feature name of sulfur content but a weight value corresponding to moisture content, the feature priority sorting and weight value generation process is immediately traced back to find and correct the deviation, and then the check is repeated. Once all features and weight values ​​are accurately matched, each feature name and its weight value are recorded as key-value pairs and organized into a structured document. This document is the weight-aligned feature mapping of the regularized feature sequence.

[0129] Furthermore, the weight value of each feature is extracted from the weight-aligned feature map. By comparing the magnitude of all weight values, the features are divided into three levels: the one with the largest weight value is assigned to the high-weight level, the one with the middle weight value is assigned to the medium-weight level, and the one with the smallest weight value is assigned to the low-weight level. During the division, it is ensured that each feature belongs to only one level, with no duplication or omission.

[0130] Furthermore, integration rules are set according to the differences in the impact of each level on coal quality assessment: high-weight level features have the greatest impact on the assessment conclusion, and the integration method of "characteristic value × weight value and then summing" is adopted to fully amplify its value; medium-weight level features have the next greatest impact, and the integration method of "characteristic value × weight value × fixed proportion and then summing" is adopted to reflect the value without exceeding the high-weight level; low-weight level features have the least impact, and the integration method of "characteristic value × weight value × lower fixed proportion and then summing" is adopted to ensure reasonable integration without interfering with the core features.

[0131] Furthermore, the results of each level are calculated according to the fusion rules: the specific value of each feature of the high-weight level is extracted, multiplied by the corresponding weight value to obtain the weighted value, and the sum of all weighted values ​​is the fusion result of the high-weight level; similarly, the fusion result of the medium-weight level is calculated according to the medium-weight rule, and the fusion result of the low-weight level is calculated according to the low-weight rule.

[0132] Furthermore, the fusion results of high-weight, medium-weight, and low-weight levels are arranged in the order of "high-medium-low", and the corresponding level name is marked next to each fusion result to facilitate subsequent tracing of the source. Such an ordered set of fusion results is the weighted feature intermediate representation of the weighted alignment feature mapping.

[0133] Furthermore, the document recording the original arrangement order of the regularized feature sequences is retrieved, and the position of each feature in the original sequence is identified, forming an original sequence position lookup table. Then, the fusion results of each level are extracted from the intermediate representation of the weighted features, and precisely mapped to the corresponding position in the original sequence using the lookup table. For example, the fusion result of the heat output of the high-weight level corresponds to the first position in the original sequence.

[0134] Furthermore, supplementary explanations are added after each fusion result, including the weight level and fusion rules, to ensure that subsequent steps clearly understand the calculation logic. After supplementation, it is checked whether each feature of the regularized feature sequence has a corresponding fusion result to confirm that there are no omissions or mismatches. Then, all fusion results are arranged into a new ordered feature set according to the original sequence order. This set is the weighted feature sequence of the regularized feature sequence.

[0135] In summary, by verifying, correcting, and organizing key-value pairs between the regularized feature sequence and the weight value sequence, a weight-aligned feature mapping is obtained; by setting weight value hierarchies, hierarchical fusion rules, and sorting the results, a weighted feature intermediate representation is obtained; and by comparing the original sequence, corresponding the fusion results, supplementing the explanation, and verifying, a weighted feature sequence is obtained. The entire process focuses on the accurate matching and value quantification of features and weights, providing structured data for subsequent coal quality assessment.

[0136] In summary, by splitting the feature priority sequence, statistically analyzing group information, and sorting out evaluation dimensions, a feature distribution structure is obtained through priority distribution analysis. This structure contains a set of high-priority features. Determined based on feature priority scores , , Set up in combination with industry needs , Through numerator and denominator calculations and exponential operations, adaptive weight calculations are performed to obtain a weight value sequence, reflecting the influence trend of different parameter changes on the weight values. By aligning the sequence, multiplying the feature values ​​with the weight values, and integrating the results, a weighted feature sequence is obtained through weighted fusion. The entire process is based on feature priority and correlation value, with clear standards for each step to ensure that the product, including the feature distribution structure, weight value sequence, and weighted feature sequence, accurately corresponds to the requirements, providing reliable weighted feature data for subsequent coal quality compliance assessments.

[0137] The compliance assessment module 104 is used to perform compliance assessment on the weighted feature sequence to obtain the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence.

[0138] In this embodiment of the invention, when the compliance assessment module performs a compliance assessment on the weighted feature sequence to obtain the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence, it is specifically used for:

[0139] The weighted feature sequence is subjected to multi-level threshold determination to obtain a multi-dimensional conformity status identifier of the weighted feature sequence.

[0140] Confidence prediction is performed on the multi-dimensional compliance status identifiers to obtain the comprehensive confidence rating of the multi-dimensional compliance status identifiers;

[0141] Based on the comprehensive confidence rating and the multi-dimensional compliance status identifier, the coal quality assessment level and preliminary coal quality classification results of the weighted feature sequence are generated.

[0142] Specifically, the process begins by referencing industry standards for assessing the quality of imported and exported coal, and combining this with past qualified assessment cases corresponding to multiple batches of weighted feature sequences. Multi-level threshold standards are then determined for each feature dimension. For example, the calorific value dimension is set with excellent, qualified, and unqualified thresholds, and the sulfur content dimension is similarly set with corresponding multi-level thresholds. Next, the specific values ​​for each feature dimension are extracted from the weighted feature sequence, and each value is compared with the corresponding multi-level threshold. If a feature dimension value is higher than the excellent threshold, the dimension is marked as excellent; if the value is between the qualified and excellent thresholds, it is marked as qualified; and if the value is lower than the qualified threshold, it is marked as unqualified. Finally, the qualification statuses of all feature dimensions are organized in dimensional order, and the resulting ordered set of statuses is the multi-dimensional qualification status identifier for the weighted feature sequence.

[0143] Furthermore, collect matching data between past batches of multi-dimensional compliance status labels and the final actual coal quality assessment results. This data must include the specific content of the multi-dimensional compliance status labels in each assessment, as well as the corresponding actual coal quality assessment conclusions, to ensure that the data is complete and can reflect the correlation between the labels and the actual results.

[0144] Furthermore, the collected historical matching data is classified according to the type of multi-dimensional compliance status identifier. The total number of times each type of identifier appears in the historical data is counted. Then, the number of times the actual evaluation result corresponding to each type of identifier is consistent with the identifier status is counted. The number of consistent results is divided by the total number of occurrences of that type of identifier to obtain the single-dimensional confidence score of each type of identifier. Each single-dimensional confidence score can reflect the reliability of the matching between the corresponding identifier and the actual result.

[0145] Furthermore, all the single-dimensional confidence scores corresponding to all feature dimensions in the current multi-dimensional compliance status identifier are extracted, summarized, and the average level of these single-dimensional confidence scores is calculated. In the calculation, all single-dimensional confidence scores are added together and then divided by the total number of single-dimensional confidence scores to obtain the average confidence score value.

[0146] Furthermore, referring to the conventional standards for confidence rating in the field of import and export coal quality assessment, a high confidence standard and a medium confidence standard are set. If the calculated average confidence value is higher than the high confidence standard, the overall confidence rating is determined to be high confidence; if the average confidence value is between the medium and high confidence standards, it is determined to be medium confidence; if the average confidence value is lower than the medium confidence standard, it is determined to be low confidence. The result is the overall confidence rating of the multi-dimensional compliance status indicator.

[0147] Furthermore, establish corresponding rules for comprehensive confidence rating, multi-dimensional compliance status indicators, and coal quality assessment grades. The rules should cover combinations of different confidence levels and different status indicators. For example, when the confidence level is high and all characteristic dimensions are excellent, the corresponding coal quality assessment grade is super grade; when the confidence level is high and most dimensions are excellent and a few dimensions are qualified, the corresponding assessment grade is level one; when the confidence level is medium and all dimensions are qualified, the corresponding assessment grade is level two; when the confidence level is low or any dimension is unqualified, the corresponding assessment grade is unqualified.

[0148] Furthermore, the current comprehensive confidence rating and multi-dimensional compliance status indicators are compared one by one with the established corresponding rules to find the rule entries that match perfectly. Based on these entries, the corresponding coal quality assessment level is determined to ensure that the assessment level and the combination of confidence and status indicators are completely consistent and there is no mismatch.

[0149] Furthermore, referring to the classification standards in the field of import and export coal trade, and combining the impact of coal quality assessment grades on trade uses, a correspondence rule between assessment grades and preliminary coal quality classifications is established. For example, the preliminary coal quality classification corresponding to the special grade and first grade assessment grades is the high-quality trade classification, which is applicable to high-end industrial uses; the classification corresponding to the second grade assessment grade is the regular trade classification, which is applicable to ordinary industrial uses; and the classification corresponding to the unqualified assessment grade is the restricted trade classification, which requires further processing before it can be used.

[0150] Furthermore, based on the determined coal quality assessment grade, the corresponding classification rule entries are found to determine the preliminary coal quality classification result corresponding to the assessment grade. Then, the coal quality assessment grade and the preliminary coal quality classification result are organized into a document according to the correspondence between "assessment grade - classification result". The document should clearly list the status indicator, comprehensive confidence rating, final assessment grade and classification result of each feature dimension. The resulting document is the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence.

[0151] In summary, by referencing industry standards to determine thresholds and comparing numerical values ​​to mark status, multi-level threshold judgments are completed to obtain multi-dimensional compliance status indicators; by collecting historical data, calculating single-dimensional and average confidence levels, and matching rating standards, confidence level predictions are completed to obtain a comprehensive confidence rating; by establishing corresponding rules, matching evaluation levels, determining classification results, and organizing documents, coal quality evaluation levels and preliminary classification results are generated. The entire process is supported by industry standards and historical data to ensure that the products at each stage accurately correspond to the requirements, providing a reliable quality evaluation basis for import and export coal trade decisions.

[0152] The trade classification module 105 is used to generate a coal trade classification conclusion based on the preliminary coal quality classification results.

[0153] In this embodiment of the invention, when the trade classification module generates a coal trade classification conclusion based on the preliminary coal quality classification result, it is specifically used for:

[0154] The preliminary coal quality classification results are subjected to multi-standard compliance verification to obtain a summary of the compliance judgment of the preliminary coal quality classification results.

[0155] The compliance determination summary is integrated with decision factors to obtain a trade adaptability plan for the compliance determination summary;

[0156] A risk situation analysis was conducted on the aforementioned trade adaptability plan to obtain risk management recommendations for the plan.

[0157] Based on the aforementioned risk management recommendations, the trade adaptability scheme is optimized to obtain the coal trade classification conclusion based on the preliminary coal quality classification results.

[0158] Specifically, the first step is to determine the basis for multi-standard compliance verification, including industry technical standards for imported and exported coal quality, entry inspection standards of the target trading country, and environmental emission-related standards. These standards are then compiled into a clear verification checklist to ensure coverage of core dimensions such as quality, safety, and environmental protection. Next, key information is extracted from the preliminary coal quality classification results, including quality indicators such as calorific value, sulfur content, and ash content, as well as the trade uses corresponding to the classification results. Then, each extracted quality indicator is compared with each standard in the verification checklist. If an indicator meets all standard requirements, it is marked as compliant; if it does not meet any standard requirements, it is marked as non-compliant, and the specific non-compliant standard clause is recorded. Finally, the compliance markings, non-compliant clauses, and corresponding classification results of all indicators are compiled into a structured document, which is the summary of the compliance determination of the preliminary coal quality classification results.

[0159] Furthermore, the specific types of standards required for multi-standard compliance verification should be clearly defined. Industry technical standards should refer to the coal quality grading documents issued by the state or industry. Entry inspection standards should refer to the latest inspection requirements published by the official agencies of the target trading country. Environmental emission standards should be based on the local environmental protection department's restrictions on coal combustion emissions, ensuring that all standards are current and valid versions.

[0160] Furthermore, when extracting information from the preliminary coal quality classification results, it is necessary to check each quality indicator recorded in the classification results one by one to confirm the accuracy of the indicator values. For example, whether the calorific value is consistent with the test report, whether the sulfur content is accurate to the specified decimal places, and whether the trade purpose is clearly marked as industrial use, power generation, or other specific categories, so as to avoid affecting subsequent comparisons due to incomplete or incorrect information.

[0161] Furthermore, when comparing indicators with standards, the process is carried out one by one in the order of the verification checklist. First, the industry technical standards are compared, then the entry inspection standards are compared, and finally the environmental emission standards are compared. The result is recorded after each standard comparison is completed. If an indicator does not meet the requirements of any standard, it is immediately marked as non-compliant, and the clause number and specific requirements of the standard are accurately recorded to ensure that the reasons for non-compliance are traceable.

[0162] Furthermore, when summarizing the compliance assessment, a structured format is adopted. First, the core information of the preliminary coal quality classification results is listed. Then, the compliance status of each indicator is presented in chapters. Each indicator corresponds to one line of record, including the indicator name, standard requirements, actual value, compliance status, and non-compliant clauses. Finally, the overall compliance conclusion is summarized to ensure that the document is clear and easy to use later.

[0163] Furthermore, the specific types of decision factors are clarified, including coal quality factors, compliance factors, and market adaptability factors. Quality factors focus on the compliance status of core indicators such as calorific value and sulfur content. Compliance factors directly adopt the compliance status in the compliance judgment summary. Market adaptability factors focus on the degree of matching between the classification results and the target market demand. The three together constitute the basis for judging trade adaptability.

[0164] Furthermore, when extracting information from each decision factor, the specific records of the compliance factors are copied from the compliance judgment summary, including the overall compliance conclusion and the compliance status of individual indicators; the specific values ​​and compliance levels of the quality factors are extracted from the preliminary coal quality classification results, as well as the demand description of the target market in the market adaptation factors, to ensure that each factor has corresponding original data support and no additional assumptions are added.

[0165] Furthermore, when determining the importance of decision factors, the conventional priorities of import and export coal trade are referenced. Quality and compliance factors directly determine whether coal can enter the target market and are listed as core factors. Market suitability factors affect the efficiency of trade transactions and are listed as auxiliary factors. The weight of core factors in the formulation of the plan is higher than that of auxiliary factors to ensure that core needs are met first.

[0166] Furthermore, when integrating information from various factors to form a trade adaptation solution, the information is arranged in the order of core factors first and auxiliary factors last. First, the basic adaptation range is divided based on quality and compliance factors. For example, compliant and high-quality coal corresponds to high-end demand, while compliant but of medium quality corresponds to mid-range demand. Then, the specific customer types are refined in combination with market adaptation factors to form adaptation suggestions for different scenarios. The collection of these suggestions is the trade adaptation solution.

[0167] Furthermore, we need to analyze the types of risks in the trade adaptation plan. Quality risks correspond to the stability of quality indicators in the plan, trade compliance risks correspond to the timeliness of standards in the target country, and market risks correspond to adapting to changes in customer needs. Each type of risk needs to be precisely linked to specific aspects of the plan. For example, quality risks are linked to the re-inspection requirements for high-quality coal, and compliance risks are linked to the applicability of entry inspection standards.

[0168] Furthermore, when assessing the probability of a risk occurring, historical data from similar trade projects over the past three years are retrieved to calculate the frequency of each risk in similar scenarios. This data is then combined with the fluctuation records of indicators in the current scenario. If the historical frequency of occurrence is high and the current indicator fluctuates significantly, the risk is classified as high-probability; otherwise, it is classified as low-probability. This ensures that the assessment is based on actual data.

[0169] Furthermore, when analyzing the degree of risk impact, it is assessed from two aspects: economic loss and time cost. Economic loss includes possible return shipping costs and fines, while time cost includes the time spent on re-inspection and reclassification. Based on the amount of loss and the length of time, it is divided into three levels: high, medium, and low. For example, high economic loss and long time are considered high impact risk, while low loss and short time are considered low impact risk.

[0170] Furthermore, when formulating risk management recommendations, priority should be given to designing control measures for high-probability, high-impact risks, such as adding pre-shipment re-inspection for quality risks and setting up a standard update early warning mechanism for compliance risks; for medium- and low-probability risks, routine measures should be formulated, such as regularly tracking changes in market demand, and all measures should be classified and organized according to risk type to form a risk management recommendation list.

[0171] Furthermore, adjust the trade adaptability plan based on risk management recommendations. If it is recommended to add re-inspection, specify the implementing agency, testing items, and qualification standards for re-inspection in the plan. If it is recommended to track standard updates, supplement the tracking process with designated personnel and the adjustment plan after the update to ensure that each control measure corresponds to the specific modification content of the plan.

[0172] Furthermore, the revised plan is compared with the preliminary coal quality classification results. The focus is on checking whether the modified parts deviate from the core quality attributes of the original classification. For example, if the original classification was "high-quality coal", the revised plan still maintains the requirements for high calorific value and low sulfur content, ensuring that the adjustment of the plan does not change the essential characteristics of the preliminary classification.

[0173] Furthermore, when determining the final trade classification, specific details are clarified in conjunction with the adjusted plan, including target customer types, quality protection measures during transportation, inspection procedures after entry, and basic pricing range. These details are then integrated with the original preliminary classification results to form a complete classification description.

[0174] Furthermore, when compiling the coal trade classification conclusions, a standardized document format is adopted. First, the original preliminary classification results and compliance basis are summarized. Then, the key points of the adjusted trade adaptability plan are presented. Finally, the final trade classification, applicable scenarios, and control measures are clarified to ensure that the document contains all key information and meets the needs of trade decision-making.

[0175] In summary, by refining and defining standards step by step, extracting information, and comparing and organizing data, a compliance verification summary is obtained; by identifying factors, extracting information, determining weights, and forming recommendations, decision-making factors are integrated to obtain a trade adaptability plan; by identifying risks, analyzing probabilistic impacts, and formulating measures, a risk situation analysis is conducted to obtain risk management recommendations; and by adjusting the plan, verifying and integrating it, strategy optimization is completed to obtain a coal trade classification conclusion. The entire process is based on original data and standards without adding any extra content, ensuring that the product accurately supports trade decisions.

[0176] The report generation module 106 is used to generate a coal quality assessment and classification report for the imported and exported coal based on the coal quality assessment grade and the coal trade classification conclusion.

[0177] In this embodiment of the invention, when the report generation module generates a coal quality assessment and classification report for imported and exported coal based on the coal quality assessment grade and the coal trade classification conclusion, it is specifically used for:

[0178] By mapping the coal quality assessment grades to the coal trade classification conclusions, the core topology of the import and export coal report is obtained.

[0179] Based on the core topology of the report, the preset report template is dynamically and structurally filled to obtain the initial draft report of the imported and exported coal.

[0180] Enhanced indexing was performed on the initial draft report to obtain the coal quality assessment and classification report for the imported and exported coal.

[0181] Specifically, the process begins by determining the mapping dimensions between coal quality assessment grades and coal trade classification conclusions. These dimensions include the assessment grade name, the corresponding quality indicator range, the trade classification type, and the suitable customer groups. These dimensions cover the core points of correlation between the two. Next, the assessment grade name and the compliance range of core quality indicators are extracted from the coal quality assessment grades, and the trade classification type and suitable customer group information are extracted from the coal trade classification conclusions, ensuring that the extracted information is complete and consistent with the original results. Then, a mapping relationship is established, directly corresponding the assessment grade name and trade classification type for the same coal batch, the quality indicator range and the quality requirements in the classification conclusions, and the suitable customer groups and the quality positioning of the assessment grade. Finally, these mapping relationships are organized into a hierarchical structure according to the dimensions. The upper layer represents the direct correspondence between the assessment grade and the classification conclusions, while the lower layer contains the detailed correlation content of each dimension. This structure forms the core topology of the import and export coal reporting system.

[0182] Furthermore, when determining the mapping dimensions, refer to the conventional framework of import and export coal reports. The assessment grade name must be completely consistent with the previously determined names such as "Special Grade" and "Grade 1". The trade classification type must correspond to the classification results such as "High Quality" and "Regular" to ensure that the dimension settings meet the needs of subsequent reports.

[0183] Furthermore, when extracting information, the quality indicator ranges in the assessment level are checked one by one to confirm whether they include the specific ranges of key indicators such as calorific value and sulfur content. The applicable customer groups in the trade classification conclusion are clearly marked as industrial customers, power generation customers, etc., to avoid missing information affecting the accuracy of mapping.

[0184] Furthermore, after establishing the mapping relationship, check whether there are any corresponding deviations, such as whether the trade classification is a high-quality classification when the evaluation level is top grade, and whether the quality index range matches the index requirements of the high-quality classification, to ensure that there are no logical contradictions at each related point.

[0185] Furthermore, when organizing the topology, a hierarchical list format is used. First, the overall correspondence between the evaluation level and the classification conclusion is listed, and then the detailed relationship of each dimension is explained point by point, making the topology clear and intuitive, which is convenient for filling in the template later.

[0186] Furthermore, the structure of the pre-set report template is analyzed to identify the modules it contains, such as the cover, core conclusion page, quality indicator details page, and trade recommendation page. Each module has a fixed information filling position. Then, the information in the core report topology is broken down by module. The core conclusion page is filled with the correspondence between the assessment level and the classification conclusion. The quality indicator details page is filled with the association between the indicator range and the classification requirements. The trade recommendation page is filled with information suitable for the target customer group. Next, the information is filled in according to the template format requirements to ensure that the text description is consistent with the original template format, such as the title font and paragraph spacing conforming to the specifications. Finally, it is checked whether each module is completely filled in and no information is missing. The resulting report is the initial draft report for import and export coal.

[0187] Furthermore, when analyzing the template structure, check the annotations of each module in the template, confirm the title position of the core conclusion page and the table format of the quality indicator details page, and ensure that subsequent filling does not change the original layout of the template.

[0188] Furthermore, when splitting the topology information, information is filtered according to module requirements to avoid filling in information that is not needed by a certain module. For example, the cover only needs the coal batch number and does not need quality indicator details, ensuring that the content is accurate.

[0189] Furthermore, after filling in the information, the text descriptions are checked to ensure that the assessment level names and classification types are completely consistent with the original results, with no typos or expression deviations, thus ensuring the accuracy of the initial draft report.

[0190] Furthermore, the initial draft report is enhanced with indexing, and the indexing types are determined, including keyword indexing and citation annotation. Keyword indexing selects the assessment grade name, trade classification type, and core quality indicators as keywords. Citation annotation is used to indicate the original test report number of the quality indicator data. Then, indexing is added to the corresponding positions in the report. Keywords are bolded and marked when they first appear, and citation annotations are marked with annotation symbols after the indicator data and the report number is noted in the footer. Finally, the accuracy of the indexing is checked, whether the keywords cover the core content, and whether the citation annotation is consistent with the original report number. The report after indexing is the coal quality assessment and classification report for imported and exported coal.

[0191] In summary, the core topology of the report is obtained by determining the mapping dimensions, extracting information, and establishing relationships; the initial draft report is obtained by analyzing the template, splitting information, and filling in and verifying it; and the final evaluation and classification report is obtained by determining the indexing type, adding indexes, and verifying accuracy. The entire process is based on the original results, with clear steps that meet the report generation requirements, ensuring that the final report is complete and standardized.

[0192] Reference Figure 2 The diagram shown is a flowchart illustrating an intelligent assessment and classification method for the quality of imported and exported coal according to an embodiment of the present invention. In this embodiment, the intelligent assessment and classification method for the quality of imported and exported coal includes:

[0193] S1. Perform data normalization on the multi-source data of imported and exported coal to obtain the normalized feature sequence of the multi-source data;

[0194] S2. Based on the historical assessment results of the imported and exported coal, sort the regular feature sequence by feature priority to obtain the feature priority sequence of the regular feature sequence;

[0195] S3. Based on the feature priority sequence, perform adaptive weight fusion on the regular feature sequence to obtain the weighted feature sequence of the regular feature sequence;

[0196] S4. Perform a conformity assessment on the weighted feature sequence to obtain the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence;

[0197] S5. Based on the preliminary coal quality classification results, generate a coal trade classification conclusion based on the preliminary coal quality classification results.

[0198] S6. Based on the coal quality assessment grade and the coal trade classification conclusion, generate a coal quality assessment and classification report for the imported and exported coal.

[0199] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0200] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0201] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent assessment and classification system for the quality of imported and exported coal, characterized in that, The system includes a data normalization module, a feature sorting module, a weight fusion module, a compliance assessment module, a trade classification module, and a report generation module, wherein: The data normalization module is used to normalize the multi-source data of imported and exported coal to obtain the normalized feature sequence of the multi-source data. The feature sorting module is used to sort the regular feature sequence by feature priority based on the historical assessment results of the imported and exported coal, to obtain a feature priority sequence of the regular feature sequence, including: Based on the historical assessment results of imported and exported coal, data association analysis is performed on the regularized feature sequence to obtain the association feature set of the regularized feature sequence; Priority scores are calculated on the associated feature set to obtain the priority score distribution of the associated feature set, wherein the calculation formula for the priority score distribution is as follows: ； In the formula, For the first Priority scores for each feature. For the first The correlation strength coefficient between each feature and the historical evaluation results, For the first The frequency of occurrence of each feature in historical assessments For the first The coefficient of variation of each feature in the current batch of data. The maximum value of the coefficient of variation of all features. It is the natural logarithm function. The preset association strength weighting coefficients, The preset frequency weighting coefficients, These are preset stability weighting coefficients; Based on the priority score distribution, the regularized feature sequence is sorted to obtain the feature priority sequence of the regularized feature sequence; The weight fusion module is used to perform adaptive weight fusion on the regularized feature sequence based on the feature priority sequence to obtain a weighted feature sequence of the regularized feature sequence, including: The feature priority sequence is subjected to priority distribution analysis to obtain the feature distribution structure of the feature priority sequence; Based on the aforementioned feature distribution structure, adaptive weight calculation is performed on the feature priority sequence to obtain a weight value sequence of the feature priority sequence, wherein the calculation formula for the weight value sequence is as follows: ； In the formula, For the first The weight values ​​of each feature, For the first Priority scores for each feature. For the first Priority scores for each feature. The minimum score in the feature priority sequence, The maximum score in the feature priority sequence. The preset weight distribution adjustment factor, These are the preset high-priority feature enhancement coefficients. For indicator functions, This refers to the set of high-priority features in the priority distribution analysis. The total number of features, It is an exponential function; The regular feature sequence and the weight value sequence are weighted and fused to obtain the weighted feature sequence of the regular feature sequence; The compliance assessment module is used to perform compliance assessment on the weighted feature sequence to obtain the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence. The trade classification module is used to generate a coal trade classification conclusion based on the preliminary coal quality classification results. The report generation module is used to generate a coal quality assessment and classification report for the imported and exported coal based on the coal quality assessment grade and the coal trade classification conclusion.

2. The intelligent assessment and classification system for the quality of imported and exported coal as described in claim 1, characterized in that, When the data normalization module performs data normalization on multi-source data of imported and exported coal to obtain the normalized feature sequence of the multi-source data, it is specifically used for: Heterogeneous data fusion is performed on multi-source data of imported and exported coal to obtain an intermediate dataset of the multi-source data. The intermediate dataset is standardized by dimension to obtain the standardized dataset of the intermediate dataset; Feature extraction is performed on the standardized dataset to obtain a regularized feature sequence of the standardized dataset.

3. The intelligent assessment and classification system for the quality of imported and exported coal as described in claim 1, characterized in that, When the feature sorting module performs data association analysis on the regularized feature sequence based on the historical assessment results of the imported and exported coal to obtain the associated feature set of the regularized feature sequence, it is specifically used for: The historical assessment results of the imported and exported coal are filtered for timeliness to obtain a valid historical dataset of the historical assessment results; Dynamic correlation analysis is performed between the effective historical dataset and the regularized feature sequence to obtain the correlation strength distribution between the effective historical dataset and the regularized feature sequence; Based on the correlation strength distribution, the regular feature sequence is optimized and recombined to obtain the correlation feature set of the regular feature sequence.

4. The intelligent assessment and classification system for the quality of imported and exported coal as described in claim 1, characterized in that, When the weight fusion module performs a weighted fusion of the regularized feature sequence and the weight value sequence to obtain a weighted feature sequence of the regularized feature sequence, it is specifically used for: Based on the weight value sequence, the regularized feature sequence is subjected to sequence alignment verification to obtain the weight alignment feature mapping of the regularized feature sequence; The weighted aligned feature map is fused in layers to obtain a weighted intermediate representation of the weighted aligned feature map; The weighted feature intermediate representation is reconstructed to obtain the weighted feature sequence of the regular feature sequence.

5. The intelligent assessment and classification system for the quality of imported and exported coal as described in claim 1, characterized in that, When performing a compliance assessment on the weighted feature sequence to obtain the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence, the compliance assessment module is specifically used for: The weighted feature sequence is subjected to multi-level threshold determination to obtain a multi-dimensional conformity status identifier of the weighted feature sequence. Confidence prediction is performed on the multi-dimensional compliance status identifiers to obtain the comprehensive confidence rating of the multi-dimensional compliance status identifiers; Based on the comprehensive confidence rating and the multi-dimensional compliance status identifier, the coal quality assessment level and preliminary coal quality classification results of the weighted feature sequence are generated.

6. The intelligent assessment and classification system for the quality of imported and exported coal as described in claim 1, characterized in that, When the trade classification module generates a coal trade classification conclusion based on the preliminary coal quality classification results, it is specifically used for: The preliminary coal quality classification results are subjected to multi-standard compliance verification to obtain a summary of the compliance judgment of the preliminary coal quality classification results. The compliance determination summary is integrated with decision factors to obtain a trade adaptability plan for the compliance determination summary; A risk situation analysis was conducted on the aforementioned trade adaptability plan to obtain risk management recommendations for the plan. Based on the aforementioned risk management recommendations, the trade adaptability scheme is optimized to obtain the coal trade classification conclusion based on the preliminary coal quality classification results.

7. The intelligent assessment and classification system for the quality of imported and exported coal as described in claim 1, characterized in that, When the report generation module generates a coal quality assessment and classification report for imported and exported coal based on the coal quality assessment grade and the coal trade classification conclusion, it is specifically used for: By mapping the coal quality assessment grades to the coal trade classification conclusions, the core topology of the import and export coal report is obtained. Based on the core topology of the report, the preset report template is dynamically and structurally filled to obtain the initial draft report of the imported and exported coal. Enhanced indexing was performed on the initial draft report to obtain the coal quality assessment and classification report for the imported and exported coal.

8. A method for intelligent assessment and classification of imported and exported coal quality, characterized in that, The method includes: S1. Perform data normalization on the multi-source data of imported and exported coal to obtain the normalized feature sequence of the multi-source data; S2. Based on the historical assessment results of the imported and exported coal, the regularized feature sequence is sorted by feature priority to obtain the feature priority sequence of the regularized feature sequence, including: Based on the historical assessment results of imported and exported coal, data association analysis is performed on the regularized feature sequence to obtain the association feature set of the regularized feature sequence; Priority scores are calculated on the associated feature set to obtain the priority score distribution of the associated feature set, wherein the calculation formula for the priority score distribution is as follows: ； In the formula, For the first Priority scores for each feature. For the first The correlation strength coefficient between each feature and the historical evaluation results, For the first The frequency of occurrence of each feature in historical assessments For the first The coefficient of variation of each feature in the current batch of data. The maximum value of the coefficient of variation of all features. It is the natural logarithm function. The preset association strength weighting coefficients, The preset frequency weighting coefficients, These are preset stability weighting coefficients; Based on the priority score distribution, the regularized feature sequence is sorted to obtain the feature priority sequence of the regularized feature sequence; S3. Based on the feature priority sequence, perform adaptive weight fusion on the regularized feature sequence to obtain a weighted feature sequence of the regularized feature sequence, including: The feature priority sequence is subjected to priority distribution analysis to obtain the feature distribution structure of the feature priority sequence; Based on the aforementioned feature distribution structure, adaptive weight calculation is performed on the feature priority sequence to obtain a weight value sequence of the feature priority sequence, wherein the calculation formula for the weight value sequence is as follows: ； In the formula, For the first The weight values ​​of each feature, For the first Priority scores for each feature. For the first Priority scores for each feature. The minimum score in the feature priority sequence, The maximum score in the feature priority sequence. The preset weight distribution adjustment factor, These are the preset high-priority feature enhancement coefficients. For indicator functions, This refers to the set of high-priority features in the priority distribution analysis. The total number of features, It is an exponential function; The regular feature sequence and the weight value sequence are weighted and fused to obtain the weighted feature sequence of the regular feature sequence; S4. Perform a conformity assessment on the weighted feature sequence to obtain the coal quality assessment grade and preliminary coal quality classification result of the weighted feature sequence; S5. Based on the preliminary coal quality classification results, generate a coal trade classification conclusion based on the preliminary coal quality classification results. S6. Based on the coal quality assessment grade and the coal trade classification conclusion, generate a coal quality assessment and classification report for the imported and exported coal.

Images