Description
This module will provide a wide set of data driven approaches that are relevant for solving process engineering problems. While large amount of data is routinely collected in chemical plants, it is mainly used for control and automation purposes and then it is archived. This module will enable analysis of such data and how that analysis can be used for design and operation of complex process systems to improve process performance. For novel process systems with limited or no data, generation of new data can be expensive and hence experiments have to be designed carefully to maximise the information obtained by carrying out the experiments; this module will impart skills for generation of data using design of experiments. The data can also be generated from first-principles based models to obtain low-complexity, approximate models for use in optimisation and control, especially where fast evaluation of the model is required. In this module, surrogate modelling techniques for obtaining approximate process system models and their use in model-based optimisation will be presented. Another very useful tool for analysing the data is based upon identifying the patterns in the data using data classification techniques. In this module, the formulation of this problem will be presented as a mathematical programming problem and then solved using rigorous optimisation techniques. For highly nonlinear chemical process systems, artificial neural networks provide a powerful machine learning technique for analysing the data and developing models suitable for steady-state as well as dynamic optimisation. In this module, an introduction to artificial neural networks will be provided and their applications in process engineering will be presented.
Aims:
Aims of the module are:
- To provide students with an understanding of surrogate modelling for complexity reduction of process systems.
- To enable students to formulate and solve data analysis problems using mathematical programming approaches.
- To teach students concepts of design of experiments for the generation of data for chemical engineering problems.
To provide students with skills for solving process engineering problems using artificial neural networks.
Learning outcomes:
After successfully completing this module the students will be able to:
- Obtain surrogate or approximate models from complex first-principles process engineering models such that the surrogate models are amenable for optimisation and/or control.
- Understand and implement mathematical programming approaches for process data analysis.
- Iteratively generate data for chemical engineering problems using design of experiments techniques such that the information obtained from the process data is maximised.
Solve process engineering problems using artificial neural network techniques.
Module deliveries for 2024/25 academic year
Last updated
This module description was last updated on 19th August 2024.
Ìý