CN-122024900-A - Multi-dimensional adsorption carbon performance evaluation method and system based on knowledge graph

CN122024900ACN 122024900 ACN122024900 ACN 122024900ACN-122024900-A

Abstract

The invention discloses a multidimensional adsorption carbon performance evaluation method and system based on a knowledge graph, and relates to the technical field of adsorption material performance evaluation. The method has the technical key points that by means of the correlation analysis result of the knowledge graph, firstly, the application scene is classified and the core requirement is clarified, then, a multi-dimensional evaluation index system is constructed, the different scene adaptation standards are matched for each index, the evaluation index can accurately meet the core requirements of different scenes, such as the tolerance and the service life of the industrial treatment scene for the heavy pollutant, the indoor purification scene is subjected to the side adsorption selectivity, the misjudgment problem caused by a cut standard is avoided, the evaluation result is more suitable for the actual application requirement, in addition, a double-self-adaptive weight adjustment model is constructed, the real-time working condition parameter and the scene type are combined to dynamically optimize each index weight to replace the fixed weight distribution mode of the rigidness of the prior art, the intelligent adaptation of the weight under the working condition and the scene change is realized, and the deviation of the evaluation result and the field actual effect is reduced.

Inventors

MA SHENGHUA
CHEN CHAOLAN
HAO CHANG
Xiang Jifu
XIANG JIQUAN

Assignees

江西奥瑞森新能源科技股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. The multidimensional adsorption carbon performance evaluation method based on the knowledge graph is characterized by comprising the following steps of: constructing a multi-source collaborative data acquisition system, and acquiring total associated original data related to the adsorbed carbon, wherein the total associated original data covers basic attribute data, dynamic experimental data, application scene environment data and full life cycle performance data of the adsorbed carbon; Constructing a dynamic associated adsorption carbon performance evaluation knowledge graph, defining an adsorption carbon related core entity and a cross-dimension association relation by adopting a structured association structure, introducing an incremental updating and quality evaluation mechanism, and carrying out real-time association completion and accuracy verification after new data access; combining the application scene classification of the adsorption carbon, and constructing a multi-dimensional evaluation index system which covers basic performance, adsorption efficiency, long-term stability, environmental adaptability and working condition adaptation by relying on the correlation strength analysis result of the knowledge graph, wherein each dimension index is correlated with a scene adaptation standard; Based on the associated time sequence characteristics of the knowledge graph and scene requirements, a double-self-adaptive weight adjustment mechanism is established, and the weights of all evaluation indexes are dynamically adjusted; Based on the total-quantity-associated original data, obtaining each evaluation index value by adopting a mode of combining single index calculation and multi-result fusion calibration, combining double self-adaptive weights, calculating an adsorption carbon performance evaluation value through weighted fusion and deviation correction, and synchronously generating an evaluation process tracing log; And outputting a hierarchical evaluation result, a performance short-board tracing report, a quantized optimization suggestion and a scene adaptation scheme according to the evaluation value and the tracing log.
2. The method of claim 1, wherein the steps of constructing a multi-source collaborative data acquisition architecture and acquiring a full amount of associated raw data are as follows: Acquiring basic attribute data of the adsorbed carbon by using detection equipment, wherein the data comprise material components, pore size distribution, specific surface area, pore volume and surface functional group related information, and the data are related to the detection equipment information and calibration records; dynamic experimental data are collected by combining multiple working condition combinations in an environment parameter range of an adaptation evaluation requirement through a controllable experimental platform, wherein the data comprise adsorption capacity, adsorption rate, desorption rate, balance time and adsorption selectivity under different working conditions, and the experimental process synchronously records equipment operation parameters and environment interference information; acquiring application scene environment data including temperature, humidity, pollutant concentration, airflow speed, air pressure and particulate matter content through distributed monitoring equipment, and correlating the monitoring position and equipment state; Invoking adsorption carbon performance evolution data, wherein the data comprises production information, use time length, performance attenuation condition, maintenance record, failure standard and historical optimization information, and all the data are associated with unique product traceability identification; establishing a multiple data verification mechanism, including equipment validity verification, data logic verification and cross-equipment consistency verification, re-acquiring or compliance-supplementing unqualified data after marking, storing full-quantity associated original data by adopting a standardized format, and carrying out data classified storage and quick retrieval by adopting a distributed storage architecture to associate data time sequence and tracing information.
3. The method of claim 1, wherein the step of constructing a dynamically associated adsorption carbon performance evaluation knowledge graph is as follows: Defining a structured ontology structure of a knowledge graph, and setting a core entity type, an attribute constraint rule, a time sequence association logic and a traceability association standard, wherein the core entity type comprises an adsorption carbon material entity, an experimental working condition entity, an environment parameter entity, a performance result entity and an optimization scheme entity, and all entity attributes are associated with a quantization index, data source verification information, a time sequence and traceability identifier; carrying out standardized pretreatment on the total associated original data, including data cleaning, de-duplication, space-time alignment, dimension unification and compliance verification; Importing a structured association information set to construct an initial map, establishing a map quality evaluation mechanism, carrying out quantitative evaluation from association accuracy, entity coverage, time sequence consistency and traceability integrity, setting up standard standards and regulating and controlling a map which does not reach the standard; and (3) deploying an incremental updating mechanism, identifying newly added entities, association relation and time sequence characteristics when new data access or working conditions and scenes change, complementing the newly added entities, association relation and time sequence characteristics to the existing map, and re-developing quality assessment.
4. The method of claim 1, wherein the step of constructing a multi-dimensional assessment index system is as follows: Classifying application scenes of the adsorption carbon, and setting core use requirements of various scenes, wherein the application scenes comprise industrial treatment, indoor purification, water quality purification and gas separation; Extracting the correlation strength analysis result of the basic property and the basic property of the adsorption carbon from the knowledge graph, and determining basic property dimension evaluation indexes including the specific surface area standard condition, the pore size distribution rationality, the pore volume sufficiency, the surface functional group activity and the microstructure stability by combining the scene core requirement; Extracting a correlation strength analysis result of experimental working conditions and adsorption efficiency, and determining an adsorption efficiency dimension evaluation index by combining scene pollutant characteristics, wherein the adsorption efficiency dimension evaluation index comprises unit mass adsorption capacity, adsorption rate up to standard conditions, desorption thoroughness, balance time rationality and target pollutant adsorption selectivity; Extracting a correlation strength analysis result of performance evolution data and long-term stability, and determining a long-term stability dimension evaluation index comprising a performance attenuation rate, a service life, a cycle resistance number, an attenuation curve smoothness and a regeneration performance recovery rate by combining the long-term requirement of a scene; Extracting a correlation strength analysis result of environmental parameters and environmental adaptability, and determining environmental adaptability dimension evaluation indexes including wide temperature range performance retention rate, high humidity adsorption efficiency, pollutant tolerance, air pressure suitability and anti-interference capacity by combining scene environment fluctuation characteristics; And extracting the correlation strength analysis results of the working condition, the scene and the performance adaptation, and determining the dimension evaluation index of the working condition adaptation, wherein the dimension evaluation index comprises the working condition switching response speed, the multi-working-condition adaptation accuracy, the extreme working condition performance retention and the scene parameter adaptation degree.
5. The method of claim 1, wherein the step of establishing a dual adaptive weight adjustment mechanism is as follows: Acquiring carbon adsorption performance data under different working conditions and different scenes, covering various carbon adsorption materials, application scenes and working condition combinations, and taking the carbon adsorption materials, the application scenes and the working condition combinations as training data of a weight adjustment mechanism, wherein the data are associated with complete evaluation indexes, actual performance results and scene and working condition identifiers; constructing a double self-adaptive weight adjustment model, defining the model to be input into real-time working condition parameters, scene types and knowledge graph correlation strength characteristics, and outputting the model as weight values of all evaluation indexes so as to minimize deviation between an evaluation result and actual performance and maximize scene demand adaptation degree as an optimization target; Training and iterative optimization are carried out on the model, a training set and a testing set are divided for verification, prediction deviation is monitored in real time, and when the deviation exceeds a preset range, the training is carried out again by supplementing data until the model converges; Establishing a model iteration updating mechanism, developing incremental training according to newly-added data, scene and working condition changes, recording triggering conditions, calculation processes and results of each adjustment, and performing logic reverse tracing.
6. The method according to claim 1, wherein the step of calculating the adsorbed carbon property evaluation value is as follows: The method comprises the steps of converting all evaluation index values into uniform interval standard values, eliminating dimension difference influence, recording the mapping relation between original values and standard values, carrying out product operation on each index standard value and double self-adaptive weights to obtain weighted scores, carrying out deviation correction on the weighted scores through a calibration model, and eliminating data noise and associated deviation influence; Setting a confidence level standard of the evaluation value, re-checking the data quality and weight adjustment result and re-calculating when the confidence level is not up to standard, generating a tracing log of the evaluation process, and recording data sources, calculation modes, weight values, correction parameters and confidence level information.
7. The method of claim 1, wherein the step of outputting the evaluation result and the optimization suggestion is as follows: Presetting a multi-gear carbon adsorption performance grade division standard, wherein the evaluation value corresponds to different performance grades, and each grade is explicitly adapted to a scene range, a use limit and a maintenance requirement; when the performance grade of the adsorption carbon is to be optimized or unqualified, the tracing function and the evaluation tracing log are associated by means of the knowledge graph, and core influencing factors of the performance short plates are positioned; searching for a case which is successfully adapted through similarity matching based on a historical optimization case library in the knowledge graph, and generating a quantized optimization suggestion by combining core influence factors and scene requirements; outputting a scene adaptation scheme, defining optimal use parameters, installation modes and matching equipment requirements of the adsorbed carbon in a target scene, and synchronously providing optimization effect prediction.
8. A knowledge-graph-based multidimensional adsorption carbon performance evaluation system, comprising: the multi-source collaborative collection module is used for obtaining the full-quantity associated original data and performing multiple data verification and standardized storage; The dynamic spectrum construction module is used for constructing a structured association knowledge spectrum by adopting an association mining method and an increment updating mechanism, integrating a spectrum quality evaluation and optimization mechanism and carrying out spectrum dynamic evolution and association; The scene index module is used for constructing a multi-dimensional evaluation index system based on knowledge graph correlation strength analysis and scene classification and configuring scene adaptation standards; the double self-adaptive weight module is used for establishing a double self-adaptive weight adjustment mechanism, and can perform model training, iterative updating and weight tracing and dynamically regulate and control weights; The comprehensive evaluation module is used for calculating an adsorption carbon performance evaluation value by adopting a mode of combining exclusive calculation, fusion calibration and deviation correction to generate an evaluation process tracing log; The intelligent output module is used for outputting a grading evaluation result, a performance short-board traceability report, a quantitative optimization suggestion and a scene adaptation scheme and carrying out case matching and optimization effect prediction; The multidimensional compatible module is used for adapting to different types of adsorption carbon materials, various detection devices and different application scenes and can perform standardized butt joint of device interfaces and data format conversion.
9. The system of claim 8, wherein the dynamic map construction module comprises: The ontology definition unit is used for defining a structured ontology structure of the knowledge graph, and defining core entity types, attribute constraint rules, time sequence association logic and tracing association standards; The data preprocessing unit is used for carrying out standardized preprocessing on the total associated original data and complementing the missing data; The association extraction unit is used for extracting structural association information by adopting an intelligent association mining method and combining the professional knowledge of the field of the adsorption material; And the quality evaluation unit is used for quantitatively evaluating the atlas from multiple dimensions, outputting an evaluation report and performing targeted optimization on the atlas which does not reach the standard according to the evaluation report.
10. The system of claim 8, wherein the integrated assessment module comprises: The single index calculation unit is used for configuring a dedicated calculation mode according to the type of the evaluation index to finish the calculation of the numerical value of the single evaluation index; The numerical value standardization unit is used for converting all the evaluation index numerical values into unified interval standard values and recording mapping relations; The weight fusion unit is used for carrying out product operation on the standard value and the double self-adaptive weights to obtain weighted scores of all indexes; The deviation correction unit is used for performing deviation correction on the weighted scores by adopting a multi-model calibration method and eliminating the influence of data noise and associated deviation; The confidence coefficient checking unit is used for calculating the confidence coefficient of the evaluation value and triggering re-evaluation when the evaluation value does not reach the standard; and the trace source log unit is used for generating an evaluation process trace source log and recording key information.

Description

Multi-dimensional adsorption carbon performance evaluation method and system based on knowledge graph Technical Field The invention relates to the technical field of adsorption material performance evaluation, in particular to a multi-dimensional adsorption carbon performance evaluation method and system based on a knowledge graph. Background In a plurality of key fields of environmental protection treatment, industrial production, civil health and the like, the carbon adsorbing material becomes a core material for pollutant removal and substance separation and purification by virtue of high specific surface area, rich pore structure and excellent adsorption selectivity, is widely applied to scenes such as industrial waste gas purification, indoor air treatment, drinking water purification, special gas separation and the like, and is a key premise for promoting the large-scale application and technical upgrading of the carbon adsorbing material by directly determining the optimization direction of the preparation process, the adaptation efficiency of the application scene and the control precision of the use cost of the carbon adsorbing material by accurately and comprehensively evaluating the performance of the carbon adsorbing material. With the continuous expansion of the application scene of the adsorption carbon material, the existing adsorption carbon performance evaluation technology gradually forms a typical implementation scheme, but has a plurality of problems in the actual landing process, and severely restricts the reliability and application guidance value of the evaluation result: When the performance of the adsorption carbon is not up to standard, the core influencing factors of the performance quality cannot be mined in the evaluation process, and the defects of the material, the insufficient adaptation of the working condition or the environmental interference are difficult to position, the follow-up optimization lacks a definite direction, the evaluation result only reflects whether the performance reaches the standard or not, but cannot explain why the performance reaches the standard, and the guiding value is high. When the method is implemented, the fixed weight distribution proportion is adopted, the same weight is used for calculating the comprehensive evaluation value no matter whether the application scene is industrial high-concentration waste gas treatment or indoor low-concentration air purification, or the unified index standard is adopted, the same specific surface area, pollutant tolerance and other thresholds are set for the adsorption carbon of all scenes, the different core requirements of different working conditions and the application scene are completely ignored, the industrial scene is high in pollutant tolerance and long in service life, the indoor scene is high in adsorption selectivity and low in interference, the weight distribution, the index standard and the actual performance influence rule are disjointed, the deviation of the evaluation result and the on-site application effect is obvious, or the adsorption carbon meeting the specific scene requirement is misjudged to be unqualified, or the adsorption carbon which does not meet the scene core requirement is misused up to standard, and the accurate performance reference cannot be provided for different scenes. Disclosure of Invention (One) solving the technical problems Aiming at the defects of the prior art, the invention provides a multidimensional adsorption carbon performance evaluation method and system based on a knowledge graph, which solve the technical problems of isolated data, fixed weight, poor scene suitability, inaccurate evaluation and weak pertinence of optimization suggestions in the conventional adsorption carbon performance evaluation. (II) technical scheme In order to achieve the above purpose, the invention is realized by the following technical scheme: a multi-dimensional adsorption carbon performance evaluation method based on a knowledge graph comprises the following steps: constructing a multi-source collaborative data acquisition system, and acquiring total associated original data related to the adsorbed carbon, wherein the total associated original data covers basic attribute data, dynamic experimental data, application scene environment data and full life cycle performance data of the adsorbed carbon; Constructing a dynamic associated adsorption carbon performance evaluation knowledge graph, defining an adsorption carbon related core entity and a cross-dimension association relation by adopting a structured association structure, introducing an incremental updating and quality evaluation mechanism, and carrying out real-time association completion and accuracy verification after new data access; combining the application scene classification of the adsorption carbon, and constructing a multi-dimensional evaluation index system which covers basic performance, adsorption efficiency, lo