Optimization of actions in activation timed influence nets