Research On Multi-energy Pattern Recognition Method For Wireless Sensor Node

With the advent of the information age,wireless sensor networks are widely used in military,transportation,medical and many other fields,and have a profound impact on the human society.With the increasingly powerful functions of wireless sensor nodes,how to maintain the nodes to work stably and reliably for a long time has become one of the bottlenecks in the Internet of Things technology.For this reason,with the continuous development of various low-power technologies,micro-energy technologies suitable for wireless sensor nodes are booming,such as micro-solar cells,micro-vibration energy harvesters,and radio-frequency energy harvesters.At present,common wireless sensor nodes with energy harvesting function usually only use a single energy harvesting technology(usually solar cells),while using lithium batteries as energy storage devices.In some specific vibration environments,there are also nodes that use various vibration energy harvesters to collect energy.The output of micro-energy devices is greatly affected by environmental factors.Using a single energy harvesting scheme,the lifespan of wireless sensor nodes is greatly restricted.To this end,this paper designs a composite micro-energy management platform and its adaptive algorithm to cope with the simultaneous input of multiple different micro-energy sources,and uses this platform to manage a micro weather station node.First of all,this project samples the open circuit voltage of micro-energy under different working conditions,and establishes a micro-energy feature vector data set;builds a micro-energy recognition algorithm based on BP neural network,and the recognition accuracy of the algorithm can reach more than 94%;and A power management algorithm with supercapacitor priority is designed.Secondly,this subject carries out the hardware design of the composite micro-energy management platform,transplants the micro-energy recognition algorithm based on BP neural network and the power management algorithm with super capacitor priority to FPGA,and conducts FPAG prototype verification.The composite micro-energy management platform can flexibly adapt to two micro-energy inputs and has a flexible energy flow path.While providing a stable power output for the load,it also provides the load with micro-energy input information and system energy status.Finally,this subject conducted a detailed test on the output performance of the composite micro-energy management platform to verify the accuracy of the identification algorithm and the reliability of the power management algorithm.The micro-weather station node based on STM32 L and BME680 is used to verify the composite micro-energy system.The verification result shows that the node can dynamically adjust the working mode of the node according to the energy input information provided by the composite micro-energy management platform and the system energy state to maximize the node life.This subject realizes a composite micro-energy management system based on BP neural network.Its solar energy management efficiency can reach 85%,and the maximum output power of vibration energy can reach 7.4mw.Compared with the traditional power management system,the system can be adaptive External input broadens the range of energy harvesting,enriches the energy flow path,and provides abundant energy information for the load,forming a new composite micro-energy management scheme.