Multi-objective optimisation (MOO) has been used with an equation solver data reconciliation software to develop a tool for sensor system design based on modifying the sensitivity matrix of a simulated process. MOO enables searching for the best trade-off between two conflicting objectives: the cost of the system and the precision of key performance indicators (KPI) (variables that have to be measured or calculated). This methodology has been applied to design the sensor system of a two stage experimental air-water heat pump. Proper knowledge of modelling equations and constants helps to improve the estimation of the precision of variables, and lowers the cost of the system. Compared to single objective optimisation, the MOO strategy increases the number of solutions, yet the precision function still relates to different objectives for each KPI, and its formulation is shown to have an impact on the trade-off obtained.
[1]
David E. Goldberg,et al.
Genetic Algorithms in Search Optimization and Machine Learning
,
1988
.
[2]
Boris Kalitventzeff,et al.
Computer-Aided Design of Redundant Sensor Networks
,
2002
.
[3]
Geoffrey Basil Leyland,et al.
Multi-objective optimisation applied to industrial energy problems
,
2002
.
[4]
Charles Gide,et al.
Cours d'économie politique
,
1911
.
[5]
Lorenz T. Biegler,et al.
New strategies for flexibility analysis and design under uncertainty
,
2000
.
[6]
Boris Kalitventzeff,et al.
Sensitivity calculations and variance analysis in plant measurement reconciliation
,
1996
.