-
Notifications
You must be signed in to change notification settings - Fork 5
Home
The SEAD project aims to create a power system monitoring device which measures real/apparent power, power factor, and non-intrusively detects when devices are operational. While other non-intrusive load disaggregation techniques currently exist, our project will leverage cutting edge research and new embedded processing solutions to create a system capable interfacing with a cloud service without relying on said service for processing and analysis of data.
The objective of this project is to create a system which measures power usage in a system, and reports to an online database. This system will sample and buffer current and voltage data at a frequency at least 20kHz. Analysis of the spectral energy distribution as a stochastic source will be completed in the embedded system and fits of known device power signatures will be matched and reported on. These events are sent to an online database for aggregation and presentation to an end user. A database and web-app will be developed alongside the hardware to demonstrate/test the scalability of the system and API specifications.
There is no doubt that fields relating to the internet of things (IoT), big data, and smart home devices are growing. While these fields are rapidly being researched and developed by large companies, it is my opinion that no matter how saturated the consumer base may be, that many of the solutions which exist today are built upon a paradigm which either relies upon a quality of internet connectivity which is unattainable to many, or is built upon a technology which offers low-fidelity data with tractable internet requirements. In addition to power, this project can be viewed through the lens of internet connectivity for the IoT, with a focus on developing a framework for creating an efficient internet connected device which can perform high-speed digital data acquisition without saturating one’s connection with traffic and without an always-on connection.
In short, to create a system which is accessible to a larger population (access to a new market base), and cloud based energy application developers, there needs to be a new approach. This project aims to explore that approach through the lens of power and will combine the reliability and ease of integration of low-frequency event based power monitoring, and high bandwidth frequency distribution based monitoring. Combining these two approaches can be achieved through leveraging the capabilities of modern multi-core digital signal processors and proper integration of high-quality analog front-ends.
- Continuous sampling, buffering, and FFT analysis of current and voltage data.
- Intelligent DSP windowing and filtering of input data.
- Adaptive analog filtering circuits based off of automatic gain control circuits and non-volatile digital potentiometer in filter cascades.
- Multi-core digital signal processing.
- Web-based data visualization and power signature database.
Sampling and FFT analysis of current and voltage data over a set bandwidth with a given sample period. The ‘set bandwidth’ and ‘sample period’ are dependent on hardware buffering and processing capabilities and will change as hardware and software for the project becomes more refined. Minimum specs are for the current single core microcontroller implementation and will be a 20kHz sample rate with a sample window of 5mS every one second. Early spring multi-core DSP integration should yield continuous sampling and analysis with sample rates above 20kHz.
The analog filtering section should remove the large fundamental power harmonics while amplifying interference produced by devices on the network, AGC’s and intelligent filtering design (noise propagation considerations) will prevent clipping of signals and ensure that the signal to noise ratio of the system is as close to the theoretical maximum which the analog front end should supply given its quantization noise floor.
A flexible internet communications stack will be created which will allow for multiple physical layers to be used with the system. PHY’s like WiFi, ZigBee, Ethernet, and BLE are all possibilities.
The dimensions of this system should be no larger than 5cm x 10cm x 5 cm.