RESEARCHES ON FUNCTION-LINK ARTIFICIAL NEURAL NETWORK BASED LOAD CELL COMPENSATION
Zhu Zijian
Abstract:
A new approach to load cell compensation modeling based on a function link neural network is discussed in this paper. It firstly introduces the function-link neural network to compensate both linearity and temperature effect of a load cell. An example is given to illustrate the proposed method. Various of coefficients of this network is discussed including the different compensation results on three functional expansion, the relationship between initial learning step and compensation accuracy and etc. A proper network is worked out to compensate load cell up to OIML C10 degree in this paper. This neural network compensation of the above errors was achieved via micro controller. Results in this paper indicate that with above compensation the accuracy of a transducer could be improved greatly. This approach for sensor modeling is superior to the existing techniques. It has a potential future in the field of measurement.
