CN-121996971-A - Chemical process simulation data management system and method based on mathematical model

CN121996971ACN 121996971 ACN121996971 ACN 121996971ACN-121996971-A

Abstract

The invention discloses a chemical process simulation data management system and method based on a mathematical model, and relates to the technical field of data management. The system comprises a data characteristic system construction module, a dynamic mapping rule generation module, a cross-scale cooperative data processing module and a matching relationship self-adaptive optimization module. The method comprises the steps of constructing a multi-scale data quantization characteristic system and generating characteristic vectors, establishing a dynamic mapping mechanism and an association matching rule set, wherein the dynamic mapping mechanism and the association matching rule set comprise cross-scale cooperative weights and working condition dynamic threshold factors, checking data in multiple dimensions, complementing missing attributes and pushing the multi-dimensional check data according to requirements, and iteratively optimizing the characteristic vectors and the mapping rules through deviation analysis. The method solves the problem of hidden mismatch between the data and the model, adapts to the fluctuation of the working condition, improves the accuracy and the robustness of the cross-scale simulation, and realizes the safe and efficient data transfer.

Inventors

ZHU FENG
YANG RONGYE
WANG MEIMEI

Assignees

青岛三瑞节能环保技术有限公司

Dates

Publication Date: 20260508
Application Date: 20260129

Claims (10)

1. A chemical process simulation data management method based on a mathematical model is characterized by comprising the following steps: S1, respectively extracting granularity level, precision threshold and space-time dimension characteristics of corresponding data according to chemical process mathematical models of different scales, and generating data attribute feature vectors of the models of all scales after carrying out quantization definition on the characteristics; S2, constructing a mapping relation table based on the feature vector generated in the step S1, defining the matching priority and the screening threshold of each scale model and the data attribute, and forming a callable association matching rule set; S3, identifying the currently-started target model scale when the trans-scale model is in collaborative simulation, calling the association matching rule set in the step S2, checking the multi-dimensional attribute of granularity, precision and space-time dimension of the input chemical simulation original data, filtering unqualified data, complementing missing attribute, and pushing the checked data to a simulation operation module in a directional manner.
2. The method for managing chemical process simulation data based on the mathematical model of claim 1, wherein the step S1 of defining the quantization of the features and generating the feature vectors of the data attributes specifically comprises the following steps: Dividing scale levels of the chemical multi-scale mathematical model, distinguishing, and respectively quantizing three types of characteristics of granularity levels, precision thresholds and space-time dimensions of data corresponding to each scale model, wherein the space-time dimension quantization is obtained by combining time sampling intervals, data coverage actual space ranges and corresponding model total space ranges through weight distribution calculation; And respectively processing the quantized results of the three types of features by adopting a maximum and minimum normalization method, normalizing to obtain normalized results of the features, and combining to generate data attribute feature vectors of the scale models.
3. The method for managing chemical process simulation data based on the mathematical model of claim 1, wherein the establishing of the dynamic mapping mechanism and the forming of the association matching rule set in S2 specifically comprise: Constructing a mapping relation table based on the feature vector generated in the step S1, wherein elements in the table are basic adaptation degrees of each scale model and data attributes, and the basic adaptation degrees directly represent matching priorities; Quantizing the data association strength between scales by the ratio of covariance of the attribute feature vector of the adjacent scale model data to the product of the corresponding standard deviation to obtain cross-scale cooperative weight, and setting a differential data sharing or isolation strategy according to the weight value; combining the deviation degree of the actual load relative to the rated load of the process and the deviation of the components of the raw materials, introducing a weight coefficient to construct a working condition dynamic threshold factor, determining a screening threshold based on the basic adaptation degree and the working condition dynamic threshold factor, and finally forming an association matching rule set containing a mapping relation table, cooperative weights and the screening threshold.
4. The chemical process simulation data management method based on the mathematical model of claim 1, wherein the multi-dimensional attribute verification, missing attribute completion and directional pushing processes in S3 specifically comprise the following steps: identifying the scale type of the target model, calling an association matching rule set, and associating adjacent scale collaborative models based on the application rule of the cross-scale collaborative weights; Extracting feature vectors of original data, respectively checking granularity, precision and space-time dimension features by adopting differential checking conditions, judging the data which does not meet any condition as unqualified data, filtering, and marking missing attribute items; The method comprises the steps of carrying out collaborative complementation on missing attribute items through adjacent scale qualified data, setting boundary conditions, namely adopting another adjacent scale data to carry out weighted complementation if adjacent single scale data are missing, adopting the same scale historical data to carry out mean complementation if adjacent double scale data are missing, and carrying out normalization processing on the complemented data; The load rate of the simulation operation module is collected, the load rate is the ratio of the current operation task number to the maximum load carrying task number, and different pushing strategies are adopted according to the load rate, including direct pushing, pushing to a standby module and scheduling load balancing pushing after buffering.
5. The chemical process simulation data management method based on the mathematical model of claim 1, further comprising a self-adaptive iterative optimization step, specifically: S4, acquiring a deviation value of a simulation operation output result and actual chemical process operation data, and reversely correcting a matching priority and a screening threshold in a mapping relation table based on a minimum deviation principle when the deviation value exceeds a preset threshold, synchronously updating data attribute feature vectors, and completing self-adaptive iterative optimization of a model and data matching relation.
6. A mathematical model-based chemical process simulation data management system, which is applied to the mathematical model-based chemical process simulation data management method according to any one of claims 1 to 5, and is characterized in that the system comprises a data characteristic system construction module, a dynamic mapping rule generation module and a trans-scale collaborative data processing module; The data characteristic system construction module extracts granularity levels, precision thresholds and space-time dimension characteristics of corresponding data and quantifies the granularity levels, the precision thresholds and the space-time dimension characteristics of the corresponding data aiming at chemical process mathematical models with different scales to generate data attribute characteristic vectors of the models with different scales; The dynamic mapping rule generation module builds a mapping relation table based on the data attribute feature vector, defines a matching priority and a screening threshold, calculates a cross-scale cooperative weight, sets a working condition dynamic threshold factor and forms an association matching rule set; and the trans-scale collaborative data processing module identifies the scale of the target model and invokes the association matching rule set, performs multi-dimensional attribute verification and missing attribute completion on the chemical engineering simulation original data, and performs data directional pushing by combining the load rate of the simulation operation module.
7. The chemical process simulation data management system based on the mathematical model of claim 6, wherein the data characteristic system construction module comprises a scale level dividing unit and a characteristic quantization and vector generation unit; The scale level dividing unit divides the scale levels of the chemical multi-scale mathematical model and distinguishes the scale levels, and the simulation targets of the scale models and the characteristic boundaries of corresponding data are defined; The feature quantization and vector generation unit quantizes three types of features of granularity level, precision threshold and space-time dimension of data corresponding to each scale model respectively, and performs maximum and minimum normalization processing on the quantized results of the three types of features to generate data attribute feature vectors of each scale model in a combined mode.
8. The chemical process simulation data management system based on the mathematical model as set forth in claim 6, wherein said dynamic mapping rule generating module comprises a basic fitness calculating unit and a cooperative weight and threshold setting unit; the basic adaptation degree calculation unit constructs a mapping relation table based on the characteristic vector of the data attribute, calculates the basic adaptation degree of each scale model and the data attribute in the table, sets constraint conditions, and if the summation result of each component of the characteristic vector is zero, each characteristic dimension is processed according to the equal priority; the collaborative weight and threshold setting unit calculates the cross-scale collaborative weight and defines an application rule thereof for controlling a data sharing strategy between adjacent scale models, and meanwhile, the process load ratio, the raw material component deviation and the weight coefficient are combined to calculate a working condition dynamic threshold factor, and a screening threshold is determined based on the basic adaptation degree and the threshold factor to form a complete association matching rule set.
9. The chemical process simulation data management system based on the mathematical model of claim 6, wherein the trans-scale collaborative data processing module comprises a data verification and missing complement unit and a load perception data pushing unit; The data verification and deletion complement unit extracts feature vectors of original data, respectively verifies granularity, precision and space-time dimension characteristics, filters unqualified data and marks deletion attribute items, and sets boundary conditions by weighting and complementing deletion attribute of adjacent scale qualified data, namely adopting another adjacent scale data to complement when adjacent single scale data is deleted and adopting the same scale historical data to complement when adjacent double scale data is deleted, and re-executing normalization processing and verifying threshold value on the complemented data; the load sensing data pushing unit collects the load rate of the analog operation module, the load rate is the ratio of the current operation task number to the maximum bearing task number, and different pushing strategies are adopted according to the load rate, including direct pushing, pushing to the standby module and scheduling load balancing pushing after buffering.
10. The chemical process simulation data management system based on the mathematical model of claim 6, further comprising a matching relation self-adaptive optimization module, wherein the matching relation self-adaptive optimization module is used for collecting deviation values of simulation output results and actual process data, judging deviation sources through attribution analysis, and realizing self-adaptive iteration of the matching relation based on minimum deviation principle layering optimization feature vectors and mapping rules.

Description

Chemical process simulation data management system and method based on mathematical model Technical Field The invention relates to the technical field of data management, in particular to a chemical process simulation data management system and method based on a mathematical model. Background Design optimization, working condition prediction and production regulation of chemical processes are highly dependent on cooperative application of multi-scale mathematical models such as a molecular physical model, a unit equipment model, a full-flow coupling model and the like. In the prior art, management of chemical analog related data focuses on basic layers such as unified format, storage capacity expansion and query efficiency, and deep design is not developed aiming at the differentiated data requirements of a multi-scale model. The mathematical models with different scales have obvious differences on the granularity, precision and space-time dimension requirements of the data, for example, a molecular physical model needs microscopic and high-precision molecular parameter data, a full-flow coupling model needs macroscopic and wide-time domain process operation data, and the existing data management mode does not establish a mapping association mechanism of model scales and data attributes. In the process of collaborative simulation of a trans-scale model, the defects are easy to cause hidden mismatching of data call, so that the input data of the model is not matched with the actual demand, the reliability of simulation operation is reduced, the simulation result is obviously deviated from the actual chemical process running state, and the accurate optimization and scientific decision of chemical production are difficult to support. Therefore, it is needed to propose a chemical process simulation data management scheme adapting to a multi-scale mathematical model. Disclosure of Invention The invention aims to provide a chemical process simulation data management system and method based on a mathematical model, so as to solve the problems in the background technology. In order to solve the technical problems, the invention provides the following technical scheme: a chemical process simulation data management method based on a mathematical model comprises the following steps: S1, respectively extracting granularity level, precision threshold and space-time dimension characteristics of corresponding data according to chemical process mathematical models of different scales, and generating data attribute feature vectors of the models of all scales after carrying out quantization definition on the characteristics; S2, constructing a mapping relation table based on the feature vector generated in the step S1, defining the matching priority and the screening threshold of each scale model and the data attribute, and forming a callable association matching rule set; S3, identifying the currently-started target model scale when the trans-scale model is in collaborative simulation, calling the association matching rule set in the step S2, checking the multi-dimensional attribute of granularity, precision and space-time dimension of the input chemical simulation original data, filtering unqualified data, complementing missing attribute, and pushing the checked data to a simulation operation module in a directional manner. Further, the process of defining the quantization of the feature and generating the feature vector of the data attribute in S1 specifically includes: Dividing scale levels of a chemical multi-scale mathematical model, distinguishing, and respectively quantifying three types of characteristics of granularity levels, precision thresholds and space-time dimensions of data corresponding to each scale model, wherein the smaller the precision threshold quantization result value is, the higher the data precision is represented; And respectively processing the quantized results of the three types of features by adopting a maximum and minimum normalization method, normalizing to obtain normalized results of the features, and combining to generate data attribute feature vectors of the scale models. Further, the process of establishing the dynamic mapping mechanism and forming the association matching rule set in S2 specifically includes: constructing a mapping relation table based on the feature vector generated in the step S1, wherein elements in the table are basic adaptation degrees of each scale model and data attributes, and the basic adaptation degrees directly represent matching priorities; Quantizing the data association strength between scales by the ratio of covariance of the attribute feature vector of the adjacent scale model data to the product of the corresponding standard deviation to obtain cross-scale cooperative weight, and setting a differential data sharing or isolation strategy according to the weight value; combining the deviation degree of the actual load relative to the rated load of the p