Swadeshi Power Meter

The power meters that we have at our homes and at an industrial scale are the devices used by electrical agencies to monitor our power usage. This monitoring system is most often independent and unknown to the user namely the household/ industrial managers. In one word, we can say that the power meter is merely a “black-box” to the end-user. The user is unaware of why exactly he’s being charged and how exactly are devices consuming power.

 In this challenge, we as a team will be looking forward to developing something known as a Smart Power Meter (SP) (fondly called Aviral SPM v1). The fundamental motto of this device is to give the end-user a track and estimate how exactly are their devices consuming the electrical power. This device will monitor the power usage and consumption levels of the unit and will have a unique capability to interface it with the upcoming fields like the Internet of Things (IoT) and Machine Learning(ML). With the help of ML algorithms and training patterns, we look to give our device an additional capacity to predict the unusual power consumption beforehand and notify it the user, so that the damages and after-effects can be minimized.

For the analysis, we would like to interface our system with the three-phase lines which will run through the unit and collect various electrical data parameters like voltage, current, peak, and nominal power consumption levels along with some error predictors like the total harmonic distortion (THD). We plan to have a power electronics block that will acquire the above-mentioned data. This data collected is fed as input to the MCU block, where the whole processing of the data is done. Parallelly, the above-collected data is also sent as input to the DSP block which will be implemented on the FPGA chosen, and then the data processed from here is sent to the LED display device for real time data display. We plan to use appropriate ADC’s in between when and wherever the interface conversions from analog to digital domain are needed.

The MCU block sits at the core of our system, performing the essential tasks of data handling for IoT application and implementation of TCP/IP stack for cloud communication. Additionally, it will have an interface to the Wi-Fi/Ethernet module which will help the dumping of data to the cloud. One more essential duty of the MCU is to run the monitoring software which will notify the end user in case of any emergency.

Once the data is dumped into the cloud, we plan to implement powerful Machine learning algorithms on the dataset received with some predefined models are used to generate a predictive model about the consumption levels of each appliance/device. The algorithm will tend to have more accuracy as we increase the input datasets provided to the cloud. With the help of the ML model generated, the system will now have a capacity to estimate how the power consumption levels would be in the near future. If the predicted model and the real-time data obtained have large deviations, we can infer that there’s something going wrong with our power consumption level. Then, the algorithm will try to search and point out the device whose power consumption levels are not as per prediction and try to notify the user that this device may have caused the blackout, or this device may fail in near future.