Study On Load Measurement Of Shovel Loader Based On Neural Networks

A shovel loader is one kind of wide-used mechanical equipment for construction to load or unload stuff. It’s necessary to know the mass of stuff carried by the loader in order to improve work efficiency and to reduce transportation costs. But unfortunately, all accessible products suffer from the wrong assumption that the shovel moves at uniform velocity. A dynamical measurement system to measure the mass of stuff carried at non-uniform velocity by shovel loader was designed in this thesis, which satisfies the requirement of accuracy without additional sensor, can resist the harsh environment to some extent and will not change the working procedure of the shovel loader.By analyzing the working procedure of the shovel loader firstly, the finding that pressure generated by hydraulic cylinder determines the movement of shovel was proposed in this thesis. Then the non-linear, uncertainty-involving relation between the mass of load and the pressure was obtained by analyzing the mechanical property of the loader. This thesis proposes a load measuring algorithm based on learning artificial neural networks, to approximate the relation between the mass of load and the pressure, so as to estimate the mass of the load. By numerical experiments, the approximated relation between the pressure and the mass was authenticatedAn embedded load measurement system was designed and developed in this thesis, which implements load-measuring algorithm based on learning neural networks. The system consists of the signal process module, the control and compute module, the human-machine interface (HCl) module and the power modules. The signal process module is based on integrated operational amplifier LM324 and high speed analog-digital converter AD7888, aiming to regulate the input signal. The control and compute module is based on STM32F103, controls the other modules and realizes the load-measuring algorithm. The HCI module includes input module based on the matrix keyboard and output module based on LCD. The HCI module displays the results of measuring and accept the order from users. The power module consists of AMS1117-3.3 and X7805, supplying power to the other modules.At last, a test experiment verifies that the measurement system in this thesis can work at non-uniform velocity and can satisfy the requirement of accuracy.