To meet the radiosonde requirement of high sensitivity and linearity, this study designs and implements a monolithically integrated array-type piezoresistive intelligent pressure sensor system which is made up of two groups of four pressure sensors with the pressure range of 0⁻50 kPa and 0⁻100 kPa respectively. First, theoretical models and ANSYS (version 14.5, Canonsburg, PA, USA) finite element method (FEM) are adopted to optimize the parameters of array sensor structure. Combing with FEM stress distribution results, the size and material characteristics of the array-type sensor are determined according to the analysis of the sensitivity and the ratio of signal to noise (SNR). Based on the optimized parameters, the manufacture and packaging of array-type sensor chips are then realized by using the standard complementary metal-oxide-semiconductor (CMOS) and microelectromechanical system (MEMS) process. Furthermore, an intelligent acquisition and processing system for pressure and temperature signals is achieved. The S3C2440A microprocessor (Samsung, Seoul, Korea) is regarded as the core part which can be applied to collect and process data. In particular, digital signal storage, display and transmission are realized by the application of a graphical user interface (GUI) written in QT/E. Besides, for the sake of compensating the temperature drift and nonlinear error, the data fusion technique is proposed based on a wavelet neural network improved by genetic algorithm (GA-WNN) for average measuring signal. The GA-WNN model is implemented in hardware by using a S3C2440A microprocessor. Finally, the results of calibration and test experiments achieved with the temperature ranges from -20 to 20 °C show that: (1) the nonlinear error and the sensitivity of the array-type pressure sensor are 8330 × 10-4 and 0.052 mV/V/kPa in the range of 0⁻50 kPa, respectively; (2) the nonlinear error and the sensitivity are 8129 × 10-4 and 0.020 mV/V/kPa in the range of 50⁻100 kPa, respectively; (3) the overall error of the intelligent pressure sensor system is maintained at ±0.252% within the hybrid composite range (0⁻100 kPa). The involved results indicate that the developed array-type composite pressure sensor has good performance, which can provide a useful reference for the development of multi-range MEMS piezoresistive pressure sensor.
Keywords: MEMS pressure sensor; array-type; genetic algorithm wavelet neural network (GA-WNN); hysteresis compensation; intelligent sensor system; nonlinear error; sensitivity; temperature drift.