Genetic algorithms (GAs) are applied to determine the worst-case overvoltage caused by nonsimultaneous energisation of mass rapid transit (MRT) power distribution systems. Two GA-based optimisation methods are compared, using case studies performed on a typical MRT system. Simulation results show that the objective function of the GA problem is highly multimodal, discontinuous and noisy, which makes it difficult for the traditional sequential search method to obtain global optimisation. Although the effectiveness of the GA approach is verified, one drawback of the approach is that it can be CPU time-intensive. The GA approach performs many executions of the Electromagnetic Transient Program for function evaluations. The micro-GA (/spl mu/GA) is proposed as an alternative to the simple GA. Results show that the /spl mu/GA performs fewer function evaluations as it searches over the response surface more efficiently than the simple GA.
[1]
N. Mullineux,et al.
Computation of power system transients
,
1976
.
[2]
David E. Goldberg,et al.
Genetic Algorithms in Search Optimization and Machine Learning
,
1988
.
[3]
Kalmanje Krishnakumar,et al.
Micro-Genetic Algorithms For Stationary And Non-Stationary Function Optimization
,
1990,
Other Conferences.
[4]
J. W. Feltes,et al.
The effect of switching surges on 34.5 kV system design and equipment
,
1990
.
[5]
G. D. Allen,et al.
Transient overvoltages and overcurrents on 12.47 kV distribution lines: computer modeling results
,
1993
.
[6]
G. Sheblé,et al.
Refined genetic algorithm-economic dispatch example
,
1995
.