Temperature drift compensation for pressure sensor based on RBF network

In practical application, temperature drift exists in pressure sensors, which decreases the measurement precision of the sensor. A proper compensation method must be adopted to adjust the additive error and improve the accuracy of measurement. Aiming at the problem that hardware compensation has the shortcomings of high cost and low precision, a software compensation model based-on RBF is suggested. RBF network has good nonlinear mapping and generalization abilities. A gradient descending algorithm with memory factor is applied to adjust the parameters of RBF network. Experimental results indicate that RBF algorithm has a fast learning rate and high precision. The performance and measurement accuracy of pressure sensors are improved greatly and satisfactory results are achieved.