UC Santa Cruz

We prototyped and then created a hardware device (SEADS) that is able to sample high frequency current and voltage data using up to eight channels. SEADS was installed in the generic household with a variety of electrical appliances with two sensors on both lines of a single phase 240V circuit. The current and voltage measurements were taken applying bandpass filters at different frequencies of interest to isolate purely resistive and inductive loads. We identified the features of devices which consume most of the energy on the electrical panel and came up with algorithms to automatically identify when these devices are on or off. This information presents a great value to the end user since it allows to identify one's energy usage patterns and make more educated decisions. This is especially relevant in the states with time of use pricing that encourage the consumers to use energy at certain times of the day to reduce strain on the grid. In this work we created a practical solution to appliance identification in a real household using frequency analysis on the aggregate electrical current waveform. We were able to identify the most important appliances to effectively manage household energy consumption.