OPTIMAL H-2 AND H-INFINITY LQ STOCHASTIC FEEDFORWARD CONTROL