Based On Neural Network Sensor Nonlinear Compensation Technology And Its Applications

The accuracy of the sensor plays a decisive role to the performance of the detection system. The research on the linearization and non-linear compensate to the sensor's output-input characteristic already become an important subject studied both at home and abroad.This paper uses the neural network to carry on non-linear compensation to the temperature sensor, and have improved the accuracy of the sensor greatly. In theory, the neural network model set up in this paper, through adjusting the network parameter, and imply layer include enough neuron, can approach wanton curve by the wanton precision. The algorithm of studying bases on the foundation of the descent method most rapidly, and was improved. The article analyzes any kinds of factor that influence the confirming of network structure, the popularization of performance and learning process, too. And has carried on emulation on the MATLAB.This paper applies the neural network structure model set up to the system of measure and controlling of the Platen Vulcanizing Presses further. The article uses one-chip computer to replace PLC (programmable controller) to design the system of measure and controlling of the Platen Vulcanizing Presses. The system is designed carefully, practically, and is intelligent. It works steadily. Adopting neural network to carry on non-linear compensation to temperature sensor in the system is innovation, and the whole performance of the system was improved greatly. Have also explained in detail in the article how one-chip computer realizes the neural network.The result of practice indicate, on the non-linear compensation of the sensor, the neural network has advantages even more than compared with least square method.