Cel
This project concerns multivariate on-line process control for reduction of high dimensional data with special emphasis on two innovative features: the design of a well structured object oriented database of process and output parameters and the extension from traditional process performance parameters to include environmental emissions. The methodology is model based with high quality case-study data and extensive modelling capability leading to accurate multivariate calibration models and associated control bounds. The overall objective is to demonstrate how multivariate methods could be used in any kind of process to improve both product quality and reduce environmental impact. The expected results are useful both for end users, the European industry, and to consultants within Europe working on further development and commercialisation of multivariate concepts for statistical process control.
Objectives:
The overall objective of the proposed project is to reduce the environmental impact from industrial processing, with improved product quality and quantity. Designing and validating an innovative cocktail of multivariate SPC methodology and in-line systems implemented in real-time software that attain the innovative objective obtain this.
To implement the full multivariate methodology for process optimisation and cleaner production following specific objectives will be achieved
(i) To construct conceptual model of a process to systemise the measured parameters and the generated data;
(ii) To develop data mining methods and construct an object-oriented database to guarantee data quality and prepare data for multivariate modelling;
(iii) To demonstrate the effectiveness of dynamic multivariate modelling for evaluation of complex process data and for on-line process monitoring and control;
(iv) To disseminate widely to industry the new knowledge regarding multivariate methods in real time systems for statistical process control via validated modelling solutions and prototype software.
Work description:
The purpose of the work is to demonstrate the applicability of multivariate methods to high dimension data to extract process information for optimisation and cleaner production. The work will be carried out at 3 different sites representing a range of processes. Data generated from these sites will be utilised for multivariate evaluation. Starting from a conceptual model, describing each process, the identification of important process parameters will lead to the efficient installation of sensors and instruments. The conceptual model will also provide the structure of the object-oriented database. Each process factor will be associated to the production output, sampled at specific times. Thus an object-oriented database is adjusted to the production rate and will substantially reduce the data overload. The data mining methods will be developed and evaluated, to secure data quality and prepare data for multivariate modelling. The next key step is the use of multivariate dynamic modelling to reduce the dimensionality of data for extraction of important process information. These models will in turn be used for real-time systems for statistical process monitoring and control. Thus, targeted data acquisition and reduction combined with multivariate model-based process monitoring and control will improve and optimise the process and reduce the environmental impact. Environmental concerns will be strongly featured. The project will, from the start, monitor all the relevant environmental regulations and legislations. Parameters measuring the environmental impacts caused by the activity of the manufacturing process will be especially identified. The environmental impact will thus be evaluated before and after implementation of the multivariate SPC system.
Conceptual modelling, database management modelling and innovative data mining will produce high quality data for multivariate modelling. The object-oriented database minimises the data overload. The multivariate calibration models extract essential process information. Executing the models on-line improve the ability of the process operators to run and control the process in a proper way. The multivariate SPC prototype will be demonstrated at three different sites with various processes and implemented within a number of different companies for testing.
Dziedzina nauki
- engineering and technologymechanical engineeringmanufacturing engineering
- natural sciencescomputer and information sciencessoftware
- natural sciencescomputer and information sciencesdatabases
- natural sciencescomputer and information sciencesdata sciencedata mining
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsensors
Zaproszenie do składania wniosków
Data not availableSystem finansowania
CSC - Cost-sharing contractsKoordynator
100 31 STOCKHOLM
Szwecja