Way to compensate temperature for high-precision electronic balance

The circuit structure of a bidirectionally forced electronic balance is unsuitable for temperature compensation if using hardwares. An attempt is therefore made to apply the neural network to the balance for temperature compensation. It is found that the drift of mass indicating values changes nonlinearly if placing the balance in a full temperature field where the mass remains unchanged. On the other hand, the drift changes linearly if placing the balance in a full mass field where the temperature remains unchanged. Based on the two important characteristics, the neural network is applied to the balance for temperature compensation. The result indicates that the new way to compensate temperature is highly practical and precise to improve effectively the metering accuracy of electronic balance.