Online Detecting Of Wood Dry Kilns’Parameters Based On Embedded System

Lumber is widely used in construction, decoration furniture, manufacturing. One of the keysteps of wood processing is drying, the complete wood drying equipment include wood dryingkiln’s detection and control system. This paper choose the wood drying kiln detection as theresearch background, which involve sensor technology, embedded technology, statistical learningtheory and multi-sensor data fusion technology, develope an intelligent on-line wood moisturecontent real-time detection system.Assimilating the achievement of aboard and domestic, combining with the support of thetimber company, this paper develop an on-line wood kiln’s detection base embedded system, whichmonitor temperature and relative humidity of kiln, resistivity of the samples real-time. Processed bythe signal conditioning circuit, the multi-channel signals transmit into embedded system, which hasadvanced on high speed processor, good compatibility and convenient transplant. The digitalfiltering part use Grubbs to remove abnormal data and smooth the results with weighted averagesalgorithm on embedded system that transmit the results to PC by CAN bus. There obtain the bestestimates of wood moisture content with online modeling, which can improve detection accuracy.Considering of the special ambience of wood kiln, this paper analysis the interference source ofsystem and plan the signal transmission lines and design PCB layout rationally, to improve theability of the EMC immunity to electromagnetic interference.When wood becoming dry completely, its resistance will up to100. The module of timberresistivity measurement can switch the range automatically in order to meet the accuracyrequirement. For multi-channel temperature detection, which be affected by the resistance oflong-distance signal transmission line, paper propose an automatic compensation circuit to leadresistance by digital potentiometer in three wire PT100balancing bridge. According to the moisturecontent curve of wood is continuous with no mutations in drying process, Least squares supportvector regression (LS-SVR) based on “Ping-Pang” model with Parameter optimization by randomweight Particle swarm optimization (PSO) algorithm not only reflect the dynamic characteristics ofwood moisture content, but also it can ensure the prediction accuracy and shorten the processingtime effectively. Random weight PSO algorithm has the capabilities of balancing global and localsearching optimal. Collecting the data of oak in the experiment, its results show that both theaccuracy and stability of the system are able to meet the requirements.