The Study Of Nonlinear Compensation For Oil Tank Use Temperature Sensor Based On Artificial Neural Network

Large oil storage tank is very important to the national strategic reserve. Study on distribution of oil tank’s temperature can provide basic data for establish a safe and reasonable storage scheme in production. At present, the oil tanks use the temperature sensor for temperature test. It has stable performance and high reproducibility, but there exist some nonlinear error. To improve the accuracy of measurement, it’s necessary to compensate the oil tank use temperature nonlinear error. This paper proposes a nonlinear compensation method: nonlinear compensation for oil tank use temperature sensor based on artificial neural network.This paper first gives a brief analysis on the oil tank temperature field monitoring system and its temperature data, gives the distribution of oil tank’s temperature, then establish the oil tank internal space partition model, calculated multiple groups of examples, and used the weighted average method for influence factors, get the thresholds of oil tank interal space boundary. Divided the tank interal space into top area, central area and bottom area. Then according to the BP neural network theory and the temperature data, design the neural network model’s structure and function, the hidden layer nodes uses experimental method by comparing the output’s mean square error(MSE), the training data is measured by constant temperature trough and reprocessed before training.This paper established three BP neural networks for oil tank use temperature sensor nonlinear compensation by using MATLAB, and show the results of the three networks after training respectively. Through comparison and analysis of the output results with the true value measured in the field. The results show that the output of the BP neural network has high precision, good performance, and can be applied in oil tank temperature test.