Smart sensor for operational load measurement

Abstract

Operation load measurement is an important technique enabling machine health monitoring and fault localization; however, direct measurement is usually difficult or impossible. The paper deals with the development of a smart sensor, which estimates the load based on structure response. In case studies, a hardware neural network was used to obtain the load course from structure vibrations - shaker-excited in a steel frame, and arising during landing in a landing gear of an airplane. The paper describes the prototype of a smart sensor, and design and implementation of load identification algorithm. For monitoring purposes, signal processing had to be done in real-time; additionally the highest possible integration level is desired, and thus a field programmable gate array (FPGA) chip was selected as the hardware platform for load estimation. The methodology of signal processing algorithms, implementation in application-specific integrated circuit (ASIC)/FPGA was developed, which allows for automation of the most time-consuming and error-prone tasks. Moreover, the same code, being the starting point for hardware synthesis, can be used in a Matlab/Simulink environment, enabling high-level system simulation and algorithm verification. Details of the procedure are presented, along with the tools used and results obtained during its realization. Performance of the smart sensor during experiment is analysed and, finally, conclusions are shown.