Predicting wildfires by artificial intelligence to mitigate the threat
As climate change intensifies and the weather becomes hotter and drier, wildfires are expected to become more frequent and intense, with a global increase of extreme fires of up to 14 percent by 2030 according to a report by UNEP and GRID-Arendal. To sooner prevent wildfires from happening and loss of life and economic damages, today artificial intelligence can help prevent and detect wildfires. Axiomtek provides lightweight and rugged edge systems compatible with kinds of sensors and can run AI inference that can be deployed in harsh environments for wildfire prediction. Seeing wildfires as early as possible and rapidly responding is essential to preventing a potentially deadly disaster. This is where artificial intelligence, machine learning, and big data come into play.
The wildfire prediction and detection solutions usually consist of an edge system that integrates the data from multiple sensors. While the processing power is physically closer to the sensors collecting data, it gives real-time insights. There are sensors monitoring smoke and temperature respectively to analyze if it is the possible timing for wildfire to flare up; sometimes the user can even add a small weather station to the edge to obtain a more complete picture of the weather, such as wind speed and direction, to assist firefighters with putting off the fire quickly. Moreover, there are cameras connected to the systems that constantly monitor the presence of flames particularly in fire-prone areas. The continuously captured images can be compared with the model of pictures taken before the wildfires by AI algorithm. In addition to a typical camera, adding an infrared camera that can detect heat and fire makes the solution more accurate. Once the forecast AI model processes all the data from cameras and sensors in real-time and finds the possibility or presence of a wildfire, it can warn the local fire department in advance to get rid of the “fuel” of the wildfire.
AI-driven visual detection can protect important facilities and venues like electricity wires as well. In wildfire season, the predictive mechanism can locate the spots accumulating too much “fuel” like dry leaves. Once the fuel is cleaned beforehand, the chance of wildfire is reduced.
For this solution, in terms of the system requirement, the system must be based on a capable processor to run the AI inference to identify and compare the landscape images. To be connected to different sensors, the system must provide multiple COM, LAN and USB ports for data transmission from sensors and cameras. Lastly, the system must be able to function under harsh conditions in extreme weather. Moreover, the system also must support different connections like 4G LTE or Bluetooth not only to receive data from wireless sensors, most important of all but also to send data to the central server for comprehensive and additional analysis and AI model training.
DIN-rail Fanless Embedded System with Intel® Celeron® Processor N3350 or Pentium® Processor N4200, Isolated COM, LAN, and DIO
DIN-rail Fanless Embedded System with Intel® Celeron® Processor N3350, COM, CAN, PoE LAN, and DIO
DIN-rail Fanless Embedded System with Intel® Celeron® Processor N3350, COM/CAN/DIO, 2 LAN, and 2 USB
To build wildfire prediction and detection solutions, Axiomtek recommends its ICO300-83B, a rugged DIN-rail embedded system with low power and fanless design. Its operating temperature ranges from -40°C to +70°C, which makes it a perfect fit as a gateway outdoor. It features Intel® Pentium® processor N4200 and one DDR3L-1866 SO-DIMM with up to 8GB of memory. Being compatible with multiple sensors and peripherals it provides four isolated COM ports (RS-232/422/485), two LAN ports and four USB 3.0 ports.
For wired sensors, they can connect to COM ports via RS-485 - Modbus RTU for long-distance and fast transmission. Users can furtherly take the advantage of RS-485, with signal repeaters, multiple devices can be connected to maximize the number of sensors. Moreover, sensors can also connect to the system via LAN with Modbus TCP. For wireless sensors, an additional Bluetooth module enables the system to read the real-time status. Two full-size PCI Express Mini Card slots accompanied by an internal SIM slot enables the system to communicate with the central server or control center without latency with a 4G LTE module.
Considering the remote deployment location of the system, in the future, Axiomtek will release the Out-of-band (OOB) module to allow operators to monitor the status of the system to save labor force and time for on-site maintenance; if the system freezes, it can be turned on remotely through OOB. Since the system deployed in the wild relies on solar power, power saving is another crucial issue. The system can turn to sleep mode via OOB to minimize power consumption and turns on again when it is necessary. In terms of software, with the services from our Cloud Service Partner, Allxon, the software of the system can be updated remotely, accomplishing predictive maintenance and remote management.
If the user wants to integrate and deploy low power-consuming peripherals in the solution, Axiomtek has the ICO320-83C serving with four RJ-45 Gigabit PoE ports that meets the demands. It can operate under wide temperatures ranging from -40°C to 70°C and withstand vibration up to 2G. Additionally, the system also provides one RS-232/422/485 port, two USB 3.0 ports, one VGA port, one CANbus 2.0 A/B port, and one 8-bit programmable DIO.
More than that, Axiomtek’s super lightweight (0.45 kg) version of the gateway, the ICO120-E3350, features the Intel® Celeron® processor N3350, two COM ports and two LAN ports. With the same extended operating temperature of -40°C to +70°C, the biggest feature is its flexible I/O to support different peripherals. Axiomtek can customize the two COM ports to CAN and DIO ports according to the demands; what’s more, an extension module that provides COM, CAN, LAN and DIO can be added to the ICO120 to maximize its workloads and connectivity. The ICO120 is certified by Microsoft Azure, which ensures its cloud compatibility and is open to more analytical services.
In the era of extreme weather, to fight against frequent wildfires, we need more efficient and smarter ways. The systems from Axiomtek can integrate AI-driven functions and multiple sensors and provide real-time insights, making the most use of data to retrain the AI model to help civilians detect anomalies in the environment and prevent wildfires.