An active defense against an incoming missile requires information of it,
including a guidance law parameter and a first-order lateral time constant. To
this end, assuming that a missile with a proportional navigation (PN) guidance
law attempts to attack an aerial target with bang-bang evasive maneuvers, a
parameter identification model based on the gated recurrent unit (GRU) neural
network is built in this paper. The analytic identification solutions for the
guidance law parameter and the first-order lateral time constant are derived.
The inputs of the identification model are available kinematic information
between the aircraft and the missile, while the outputs contain the regression
results of missile parameters. To increase the training speed and the
identification accuracy of the Model, an output processing method called
improved multiplemodel mechanism (IMMM) is proposed in this paper. The
effectiveness of IMMM and the performance of the established model are
demonstrated through numerical simulations under various engagement scenarios.