The Application Of Fuzzy Neural Network In Electronic Weighing System

The working temperature of electronic weighing system is between-20℃-80℃, the sensor is impacted by the high and low temperature, and it is non-linear. That is the output of the weighing system is different for the same weight at different temperatures. Therefore, we must consider how to eliminate the influence of temperature on the weighing system when the change is large. Otherwise, it will affect the accuracy of the sensor data acquisition, especially in the long-term use of electronic scale.The temperature compensation measures used in electronic weighing system of aluminum electrolytic enterprise are hardware compensation and software compensation. The hardware compensation eliminates the impact of temperature using hardware circuits, but it will drift and cannot achieve full compensation. Least-squares curve fitting is based on numerical analysis and is the most commonly used temperature compensation, but this calculation is easy access to local optima and cannot obtain the global optimal solution.The recent progress of the weighing systems in the aluminum electrolytic enterprise was reviewed in the paper. Fuzzy neural network was introduced and discussed. Fuzzy neural network was the combination of artificial neural network and fuzzy theory. Artificial neural network can obtain knowledge through learning and training. Is has a good ability to learn and to accurate fitting arbitrary non-linear function, in addition to the memory of known information. And it also can improve the accuracy of the fitting by learning the target system response data. The junction of fuzzy systems and artificial neural networks can play to their strengths effectively to improve the expression ability and learning ability of the whole system. Fuzzy neural network was used to the weighing systems, based on the temperature and internal code, developing a new approach for the weighing systems,p represents the inner code sensor, t represents the temperature and m represents output of the weighing sensor, i.e. m=f(p,t).168sets of data were divided into training set and test set randomly.12points worked at-14.1℃,-12.8℃,17℃, 17.8℃,18.8℃,18.8℃,18.8℃,53.1℃,54.6℃,58.5℃and59.5℃132sets of data were training set.12points worked at-13.5℃,18.9℃and56.8℃as test set, test fuzzy neural network generalization ability.Conclusions:Less than0.205%error weight were got through matlab simulationing, the output of electronic weighing system in low temperature, room temperature and high temperature can be normal. Fuzzy neural network has a good ability to accurate fitting arbitrary non-linear function, and it can solve the similar problem.