Application of improved neural adaptive PSD algorithm in temperature control of INS

A neural adaptive proportion sum differential (PSD) algorithm with errors prediction is researched. It is applied in inertial navigation system (INS) temperature control. The algorithm do not need the process's precise mathematical model and can adapt to the process parameters changing, and can deal with the process with nonlinearity. According to the Smith predictor, a method is developed that takes the predicted process error and error change as neural adaptive PSD algorithm's input. The method is effective to the system with long dead time. The results of compute simulation show that this system has characters of quickly reaction, low overshoot and good stability. It can meet the requirements of temperature control of INS.