mavalderrama / ai-flood-prediction-system

This is an IoT Edge computing system powered by an ANN to predict floods on cities near Mississippi river

AI Powered Flood Prediction System

This implementation of an IoT Edge Computing System can predict a flood and dinamically reconfigure its hardware based on the data load

VIVADO HLS Hardware ANN IP Core

Hardware implementation of a fully conected Artificial Neural Network using Xilinx Vivado HLS Available at Vivado/HLS

Features

• Input Layers:
  • Reconfigurable N number of inputs
• Hidden Neurons & Layers:
  • Reconfigurable N number of Layers
  • Reconfigurable N number of Neurons
• Ouput Layers:
  • Reconfigurable N number of outputs
• Training:
  • Load and download weights
  • Hardware feed forward back progragation method
• Activation Function:
  • Sigmoid
• Buses
  • AXI Full for data I/O
  • AXI Lite for management and configuration

Artificial Neural Network (Software implementation)

Features

Same features as HW counterpart but slower Available at Zynq7K/Software

Petalinux on ZYNQ7000

Fork of ARM Linux for ZYNQ7000 devices This version was compiled to work with an Artificial Neural Network Accelerator implemented in the ZYNQ7000 FPGA hw portion

Features

• Compiled with an Ubuntu 16 LTS File System
• The Device Tree Source includes UIO (User Inputs Outputs) and a gigabit Ethernet drivers Zynq7K/Device Tree Source
• Boot files available on Zynq7K/Petalinux Boot Files