ENHANCED ARTIFICIAL NEURAL NETWORK ALGORITHMS FOR FAST RADAR THREATS IDENTIFICATION | ||||
| The International Conference on Electrical Engineering | ||||
| Article 54, Volume 2, 2nd International Conference on Electrical Engineering ICEENG 1999, November 1999, Page 539-550 PDF (1.97 MB) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/iceeng.1999.62555 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
| E. E. Azzouz; M. Shoker; El-barbary El-barbary | ||||
| Egyptian Armed Forces. | ||||
| Abstract | ||||
| In this paper, an enhanced algorithm for radar threat identification and based on the artificial neural network (ANN) is proposed. Four radar parameters are used as the inputs for the suggested ANNs. These parameters are: 1) the radio frequency, 2) the pulse repetition frequency, 3) the pulse width, and 4) the scan rate. A lot of work has been done to select the suitable structure of the ANNs. The chosen ANNs achieve minimum sum square errors and short time training. Also, they provide the highest success rate over all the examined networks. It is found that, one can choose a single hidden layer ANN structure with 12 nodes in the hidden layer or a double hidden layer with six nodes in each hidden layer. These ANN provide 100% success rate. Due to the simplicity of the ANNs structure, it can be used for on-line analysis. To use the developed ANN algorithms for radar threat identification in the on-line analysis, the main requirement is to finish the training phase beforehand. | ||||
| Keywords | ||||
| Radar Identifications; Neural Networks | ||||
|
Statistics Article View: 124 PDF Download: 195 |
||||
