Active vibration isolation controller using extended minimal resource allocation neural network (EMRAN)

In this paper a previously developed extended minimal resource allocation network (EMRAN) is applied as a controller to actively isolate the vibration from a source to the foundation. The source is mounted by using two passive elastomeric vibration isolators located at the front and rear of the source and an active force actuator acted as counter force to cancel the vibration is placed in parallel with the passive isolators. The source of vibration is assumed to be sinusoidal and vibration isolation is achieved when the total transmitted force to the foundation is minimized. By training the controller online EMRAN has the ability to add and prune the number of hidden layer neurons of its network based on a certain criteria that needs to be satisfied. This vibration model can be a representation of an engine mounted to its chassis and vibration isolation is needed to improve the passengerspsila ride comfort or the mounting of machinery to a rigid floor in a production plant where vibration isolation is essential not to influence other machineries.