Development Of Soft Sensor Instrument For NO_x Based On Embedded System

In recent years,there are a great deal of demands and challenges realizing soft measuring algorithms such as neural network and least square support vector machine in embedded platform with the development of soft measuring technology and embedded technology.To deal with the lag problem of NO_x content signal in flue gas of coal-fired units,an embedded soft sensor instrument for NO_x is designed whose core is ARM11 microprocessor S3C6410.The least square support vector machine(LSSVM)algorithm is optimized and the parameters of soft measuring model are modified according to the size of prediction error to achieve real-time and accurate measuring for NO_x content.And the LSSVM algorithm is successfully transplanted into the embedded system,which makes hardware instrument of soft measuring technology realized.In this paper,the research and development process about embedded soft sensor instrument of NO_x is described in detail.The main work and innovations are as follows.(1)Hardware platform based on ARM11 microprocessor S3C6410 and software platform based on Linux embedded operating system are builded according to modular design concept.The hardware platform is mainly composed of S3C6410 microprocessor,human-computer interaction interface circuit,ethernet interface circuit,A/D conversion circuit and data transmission circuit,which can realize data acquisition and transmission of auxiliary variables,soft measuring of parameters to be measured and LCD display of measuring results cooperating with the software platform.(2)The LSSVM algorithm is applied to the soft measuring of NO_x content in flue gas.The parameters of LSSVM model l and b are calculated by iteration method and modified according to the size of prediction error to get the real-time and accurate measuring of NO_x content in flue gas under the condition of unit load fluctuating.(3)The embedded soft sensor instrument of NO_x is realized.LSSVM soft measuring algorithm is transplanted into embedded system on the basis of software and hardware platforms and cross-compiling environment of the embedded system,developing a soft sensor instrument of NO_x based on embedded system.The instrument can acquire auxiliary variables data through data acquisition module or from DCS through network interface.(4)The soft sensor application software of NO_x based on PC and soft sensor instrument of NO_x based on embedded system is tested in Huarun Fuyang power plant #2 unit.The test results show that quoted errors are less than 1.5% and response time is less than 1s.And the NO_x content of flue gas can be measured quickly and accurately when the unit load changes.The soft sensor instrument of NO_x based on embedded system provides a low cost,high accurate,high real-time measuring methods for NO_x content in flue gas,and also provides a new measuring way and solution for other thermal parameters of thermal power units.