The Development Of High-precision Intelligent Signal Conditioning System

Piezoresistive sensors have many features, such as high sensitivity, large scale of measurement and frequency, little volume etc. They also can easy to be miniaturized and integrated, so they can be batch produced. But the accuracy that the piezoresistive sensors' measured is influenced by temperature. As the development of field-bus and auto-control technology, the demands of the pressure sensors' accuracy, nonlinearity, temperature drift and time drift are further increased. But the demand of out signal conditioning circuit is higher. Therefore, to develop and explosive pressure of signal conditioner with high precision is very important to the high accuracy and intelligence of sensors.To measure the pressure and condition the sensors, we use the computer to communicate with the signal conditioning circuit . And it realized the digital signal conditioning, the integration of signal conditioning and data acquisition. Then the signal conditioner's performance is increased. There are many compensation methods at present. Two-dimensional regression analysis, binary interpolation and neural network are studied in this paper. The main contents to be studied are shown as following:First, the two-dimensional regression analysis, binary interpolation and neural network that based on two sensors' data confusion are studied. The neural network is selected as the compensation method by comare the methods. The BP neural network method includes an input level, an implication level and an output level. The computation rules and the error back propagation transition are studied in this paper. Then the detailed procedure is generalized. And the program flowchart is draw, the program is written by the C language.Second, the functional diagram, the comunacation protocol and the orperation principle of MAX1452 are studied in this paper. And based on the studies, the hardware operation system is created. Then the compensation test is carried out. Last, the advantage of neural network and the practicability of the selected scheme are verification.