Weak signal detection based on strongly coupled duffing-van der pol oscillator and long short-term memory

This study presents a weak-signal detection system based on a strongly coupled Duffing-Van der Pol (DVP) oscillator. The simulation shows that under all noise conditions with different noise variances, this system achieves better performance compared with a single DVP oscillator and weakly coupled DVP oscillator. Furthermore, a weak signal is detected through the transition from a chaotic to a periodic or intermittent chaotic state. Previous studies could generally only identify chaotic and periodic states, whereas in actual weak-signal detection, most of the cases need to be distinguished between intermittent chaotic states and chaotic states. To enhance the discrimination accuracy between these three states as well as address low signal-to-noise ratios (SNRs), this detection system uses a long short-term memory (LSTM) neural network for recognition and classification. The results indicate that the application of the strongly coupled DVP oscillator and LSTM method can be used to detect weak signals at very low SNRs with high accuracy.