The increasing complexity in industrial products and systems entail appropriate optimization methods and automated design techniques. Typical optimization problems in real-world are often computationally hard and subject to multiple, conflicting and non-commensurate objectives. Traditional single-objective optimization techniques only offer one single solution to problems by maximizing/minimizing an overall goal function. Evolutionary algorithms (EAs) have proved their superiority to traditional optimization methods by their stronger global search ability in complex spaces. EAs are particularly suitable to solve multi-objective problems since they simultaneously process a population of possible solutions, which enabling finding a set of non-dominated tradeoff solutions in a single run. The set of non-dominated solutions can be presented to a decision maker/designer for the final choice based on her/his preference. The result of the project will be tested and applied in the industrial domain of analog circuit design. Making a “good” circuit is by no means a trivial task. It entails determination of optimal circuit structure together with many numerical parameters for components such as sizing parameters for resistor, reactor, capacitor as well as specifications for the control and protection. Further there are many conflicting objectives and constraints underlying the design procedure, and improving one factor would force others to worsen. This project will contribute an efficient software solution for hardware design taking into account different objectives and requirements.
|Research project manager
Adaptive Differential Evolution Supports Automatic Model Calibration in Furnace Optimized Control System (Jan 2017) Miguel Leon Ortiz, Magnus Evestedt , Ning Xiong Computational Intelligence (CI)
Towards a framework for online modelling and optimisation of airflow and temperature distribution in server rooms of data centers (Oct 2016) Ning Xiong, Xiaojing Zhang 42nd Annual Conference of the IEEE Industrial Electronics Society (IECON 2016)
Designing Optimal Harmonic Filters in Power Systems Using Greedy Adaptive Differential Evolution (Sep 2016) Miguel Leon Ortiz, Yigen Zenlander , Ning Xiong, Francisco Herrera 21st IEEE Conference on Emerging Technologies and Factory Automation (ETFA'16)