The Research Of Iron Core Surface Defect Detection System Based On Machine Vision

Iron cores are one of the most important parts in transformer,which usually form an electromagnetic induction system with coils wound on it to provide magnetic circuit.The type is various,in the process of production,due to the influence of equipment conditions,production technics and material characteristics,the iron core surface will appear cracks,pits and cutting unevenness.If the core with surface detect is used in actual power transformer,it will lead to the change of core compaction,which may result in poor grounding of iron core or insufficient earthing capacity,thus causing discharge failure and transformer damage.Even lead to the collapse of the power system,causing huge economic losses.Therefore,we should efficiently remove the surface detect core,to minimize the possibility of a security failure.The traditional manual detection method is slow in speed and low in efficiency.In addition,it is easy to produce visual fatigue during long working hours,which results in missing detection of core surface defects,false detection.Greatly reduces the product quality,but also can’t guarantee the inspection efficiency and the inspection precision.In this paper,we take E core in automatic production line of Guangdong Zhaoqing microelectronics Co.,Ltd as the research objects,design a set of core surface detect detection system based on machine vision.The system realizes the online intelligent detection of the products,improves the detection efficiency and production detection rate and reduce the production cost.The main research contents and results are as follows:(1)The machine vision detection system is built.Through the evaluation and precision requirement of the detection object,,choose the CMOS sensor camera and high resolution camera lens,using LCD ring light source illuminate directly,to obtain a clear iron core image.(2)An image processing based detection algorithm for iron core surface detects is developed by using HALCON toolbox.Gaussian filter is used to preprocess the image;The image is segmented based on Sobel operator,and the ROI region extracted by morphological processing and region filling technique;The defect region is preliminarily located by gray scale analysis,and pseudo-defect is eliminated by means of area analysis of connected region.Experimental results show that the algorithm can accurately detect the defects of vacancy blocks and pits..(3)In order to implement the classification of iron core surface defects,the classifiers based on multilayer perceptron(MLP)and support vector machine(SVM)are designed respectively.The SVM classifier and the MLP classifier are compared in recognition accuracy and detection time.The accuracy and average detection time of MLP classifier are 95%and 0.64s respectively,which is superior to SVM classifier.Finally,MLP classifier is selected to identify the surface detect of iron core.(4)A software system for the detection of iron core surface defects is developed.Based on the investigation of the user’s demand,the whole scheme of software system is designed.Using the visual development tool Visual Studio 2010,the framework of dialog box based on MFC framework,combining with HALCON algorithm development kit,adopting C++ programming language,synthesizing multi-thread thought,dynamic link library technology to realize database transfer and image collection,processing,display,storage and so on four main tasks.The experiment results show that the designed core surface defect detection system can realize the real time and non-contact detection classification of E iron stably and effectively.It has great practical value for enterprise to realize the full automatic inspection production line and provides reference for other kinds of iron core detection research.