Deep-reinforcement-learning-based controller design for pantograph and catenary system
摘要
The articulated suspension system of a pantograph and catenary system has a power collection junction, i.e., a contact network, between the pantograph and the catenary. For stable current collection and maintaining control contact force, the junction should be reliable and safe with respect to external disturbances and irregularity at the contact point, which may vary due to the speed of the train. This study proposes a novel fuzzy-based controller to control the contact force. The proposed controller, fuzzy-based fractional order proportional integral derivative (F2OPID), uses fuzziness to evaluate control contact force. The error functions of the proposed controller are monitored by a bilateral long short-term memory network. The gains of the F2OPID controller are evaluated using an Aquila optimizer. The system is analyzed on Matlab platform based on variations in train speed, stiffness coefficient, and length of the span. From the analysis, it has been observed that oscillation in contact force has been reduced using the proposed controller compared with other methods.