Development Of A Set Of On-line Detection System For Steel Materials

Angle steel is the main component of the power tower.A mixture of different material steel will produce a great impact on the performance to the tower.In machining process of iron tower,it needs to operate to the angle steel including clamping,punching and shearing.Different material of steel have different punching sound and shear stress.Therefore,this paper developed a material on-line inspection system for the angle steel mixing problem in the process of iron tower,which using the punching sound signal and shear pressure signal in the machining process,to achieve real-time online detection of the angle steel material.The main work accomplished in this paper are as follows: first of all,the research status and methods of sound signal and pressure signal are analyzed,then the reasonable technical route and the detection scheme are selected to complete the overall design of the system.The hardware components are selected,including sound and pressure sensors,electromagnetic switches,signal isolators,data acquisition card,relays,etc.The interface connection between the hardwares has been complicated,and the packaged the hardwares reasonably.Secondly,After compared a variety of feature extraction methods and pattern recognition methods,the paper decide to follow the energy ratio method.And at the same time,the Euclidean distance identification method and a neural network method based on Mel frequency spectrum parameters(MFCC)have been used.The method of setting the mean of shear stress threshold is used to identify the material of angle.Then,use the LabVIEW programming language to complete the preparation of the software part.The software part mainly includes three main interface: real time acquisition and detection interface,offline browsing analysis interface,template threshold setting interface.The off-line neural network is programmed by MATLAB.The installer and application of the test software are generated.And the integrated software and hardware are installed to a reasonable location.To collect a certain amount of sample signal in the processing field,and through offline processing to save the signal as the original template and use to angle detection.With the increasing of the sample,the template library is gradually improved.The test results of the system at the process scene show that it is Simply,fully functional,and provides a fast and accurate method for the detection of steel material.Through on-line detection of each angle without increasing the workload,this system solves the traditional detection methods which are time-consuming and difficult to adaptto the needs of the production and reduce the occurrence of mixing phenomenon.