UC Santa Cruz

Extensive deployment of power electronic loads in naval ship power systems indicate full ship electrification is inevitable. Next generation warships require high power density weapons drawing pulse power from a dc power grid. A particularly concerning issue is that these pulse loads draw large currents in short periods of time and are similar in behavior to a fault; and therefore may be indiscernible from a fault. This dissertation introduces novel machine/deep learning based algorithms, including long short-term memory recurrent neural network based autoencoders and data-driven clustering based machine learning approaches to detect dc faults and monitor load conditions applied to naval pulse loads. Two feature extraction methods are also implemented including the short-time Fourier transform and stationary wavelet transform. The novel load monitoring solution presented herein can be applied to any load profile that exhibits repetitive transients during normal operation. The frequency-domain features of the load current are extracted for the network training to set the network weights and biases. Once the network training is completed, the machine/deep learning approach will predict both signal classification and fault identification. Finally, the method is demonstrated in a low power laboratory system meant to mimic naval shipboard power systems.