Further improvement of optimisation method for mass transit signalling block-layout design using differential evolution

Abstract

The paper describes the ongoing development of optimisation methods for the layout design of equi-block n-aspect mass transit signalling systems. The authors previously applied genetic algorithms (GAs) in place of conventional gradient search methods for solving the problem. Being theoretically and empirically sound for providing multiple-point search, the GA-based approach simplifies the gradient search approach, broadens the scope for dealing with changes of either the objective function or signalling scheme, and provides robust and global convergence in complex search spaces. The GA-based formulation divides an inter-station run into three sections: a constraint section, a stretchable section and a critical section. Since each of these sections was prescribed with a different design criterion, the GA optimises separately the layout of block joints in each section and their positions within each section. The solution time for optimal signalling design is further reduced with the use of differential evolution (DE) algorithms. The DE-based method combines the three subordinate objective functions in the original GA algorithm into one composite objective function for one single search. The performance improvements of the DE algorithms over the GA-based method are shown. The choice of variants for solving the DE-based signalling design problem is also discussed.