| The field of industrial control such as thermal power generation has been devoted to studying the valuable control laws contained in historical data to guide the optimization of equipment operation and improve economic efficiency.However,due to the complex composition and strong parameter coupling of the industrial control system,the analysis of such large-scale,complex and multivariate time-series control data requires a large number of progressive exploration attempts combined with domain knowledge.Visual analysis technology can provide flexible view interaction to support complex data iterative analysis processes,and it has become one of the efficient exploration methods for such iterative analysis problems.With the in-depth research on the visual analysis of industrial control data,the control processes and parameter types that need to be analyzed are increasing,the calculation part required to support visual analysis has also become complex and diverse,and the total development cost has continued to increase.In order to solve the series of problems such as low reusability and poor scalability in the low-level calculation process of visual analysis of industrial control data,the design idea of componentized computing process is introduced to support the data-driven visual analysis process in this thesis by summarizing and abstracting the data analysis requirements and challenges in the control field.First,the overall computing architecture is designed,and the key components are divided into data quality analysis components,parameter law exploration components,model analysis and evaluation components.Then,the internal computing units of each key computing component are refined and divided,and encapsulated into one a series of subcomponents,the assembly relationship of these subcomponents is designed from the perspective of data flow specifications.Next,each subcomponent is designed in detail,and the algorithm flow and data interface are improved and optimized from the perspective of supporting algorithm scalability and calculation parameter scalability,etc.,and its reusability is significantly improved.Finally,based on the current general data analysis tools and software libraries,the development of computing components is carried out,and a hierarchical component library development architecture is realized.Through applying the above computing components,a more complete visual analysis case was developed by taking the typical analysis process of thermal power control data as an example,the feasibility and usability of the computing components in this thesis are verified.This thesis has made meaningful attempts in the visual analysis research on the reusability and scalability of industrial control data,which provides a reference for the computing research of visual analysis in similar industrial control data. |
