摘要
为了提高模具表面缺陷检测的质量,采用二进制粒子群算法。首先建立测量模型,包括光学景深模型、成像系统景深模型、系统成像几何模型标定。接着对粒子位置的每一维分量被限制为1或0,对应模具表面检测过程中的有缺陷或无缺陷判决;然后利用sigmoid函数将粒子的速度转换到区间[0,1]上,并且修正粒子的位置;最后在二维平面内建立有限元模型和判决准则。实验仿真给出了各个方向的数据优化效果和不同算法空间数据迭代效果,相比其它算法新的算法对模具表面缺陷检测的过检率、正检率、漏检率指标较好。
In order to improve the quality of the detection of mold surface defects, binary particle swarm optimization algorithm was used. First, a measurement model was established including the doptical depth of field model, the depth of field of imaging system model, the system calibration of imaging geometric model. Second, each component of particle position was limited to 1 or 0, and corresponded to bad or good mold surface in the process of detecting defects. Third, the speed of particle conversion was transformed [0,1] by sigmoid function, and the particle＇s position was modified. Finally,finite element model and decision criteria were gave in the two-dimensional. simulation experiments. The data optimization iteration effect in each direction and the different algorithms of spatial data were obtained by the simulation. Compared twith the other algorithm, the proposed algorithm of the detection of mold surface defects has better index in detected rate,positive detection rate and false negative rate.
出处
《铸造技术》
CAS
北大核心
2016年第3期610-615,共6页
Foundry Technology
基金
四川信息职业技术学院课题
2011C6
2010C01
关键词
二进制
粒子群
缺陷检测
编码
特征
binary
particle swarm
defect detection
code
feature
