Development Of Soft Sensor Platform Based On ARM

With the continuous development of science and technology and the continuous improvement of the manufacturing craft, the Complexity is increasing for process of production field, whether industrial production factors can be accurately measured or not directly affect the production efficiency and operating stability. But some production conditions are not allowed to participate in the external factors, so some variables are difficult to messure. In order to meet market demand, a wide range of measuring instrument came into being. But they are mostly the traditional measuring instrument, they have many shortcomings not to mesure exactly, which hamper efficient production.In view of this situation, a soft sensor is developed and proposed based on embedded system in this paper, The system uses the S3c2440processor, linux2.6.32core, Qtopia desktop. The key algorithm is Back-propagation neural network(BP) or Support Vector Machine(SVM), which can predict the of data production field in the Linux environment. Through Applications Qt library collection to display human-computer interface, Nandflash, Norflash, Sdram memory and other logic controller were extended, using Nec3.5-inch touch screen and other external devices to expand its detection. The establishment of the soft measurement platform include hardware design, transplantationd of linux environment, and the most important part is the development of software, which includes interface design, data pretreatment, modeling and so on. Compile this program into the Flash memory cure. Through the collection of data related to primary air in power plant, the model of primary air is constructed and tested The experiments show that the error between predicted results and actually result is only three percent, which meet the demands of this paper.