Energy harvesting multisensor wildfire monitoring system

Over the past 20 years, extreme wildfire activity has more than doubled globally. As fire seasons start earlier and end later, ecological, social, and economic impacts continue to grow. These costs can be reduced through robust monitoring and early detection. Current wildfire monitoring methods include unmanned aerial vehicles (UAVs), satellites, and sensor networks. UAVs and satellites collect data quickly over large areas but suffer from low spatial and temporal resolution. Remote sensor networks help fill these gaps but must operate nearly continuously, transmitting frequent data updates. This high energy demand leads to frequent battery depletion, and maintenance or replacement requires significant manual labor and logistical effort due to their remote locations. There is a clear need for a prolonged, self-sustaining wildfire monitoring system. UW Madison researchers have developed a self-sustained cyber-physical system that dynamically schedules monitoring sensor acquisition times to optimize sensor accuracy and energy storage, simultaneously. The monitoring system is made up of a sensor suite, including diagnostics for humidity, temperature, particulates, etc., an energy harvester and energy store. The schedule that is created is based on the information from an AI model trained with environmental parameters and harvestable power accumulated over a predefined time span. The training data also considers the stochastic nature of potential wildfire occurrences within that defined period. A power management circuit is responsible for handling this information as well as the real time data being accumulated to further optimize the schedule for higher sensor accuracy and efficient power usage.