Investigation On The Topography Characterization Of The Cross-Hatched Liner Surface In The Diesel Engines

Internal combustion engine is a main equipment in modern advanced warship and transportation engines,and is used to convert energy and transfer power.Internal combustion engine has received continuous innovation support from the relevant sectors of each country.In the internal combustion engine,the liner topography usually affect greatly the tribological and service reliable performance of the engine.In order to meet the demands that the internal combustion engine requests the measurement technology of the cross-hatched liner topography when the engine is developing toward high efficiency and high service reliable performance,this dissertation aims to investigate the machine vision perception on the roughn ess of the cross-hatched liner surface,the physical features characterization of the cross-hatched line on the liner surface,and the image recognition on the angle of the cross-hatched line.The present research can provide valuable reference and technology support on the tribological design and reliable design of the internal combustion engine.The main contents are listed as follows:(1)Based on the gray image of the cross-hatched liner surface,to avoid the local distortion problem of the reference surface for roughness evaluation caused by the cross-hatched li ne on the liner surface,a curvelet transform method is used to separate the image information in multi-direction and multi-scale,and to extract the reference plane(i.e.,the image information related to the processing error and waviness)for roughness evaluation.On this basis,the effectiveness and feasibility of the curvelet transform method to reconstruct the roughness reference plane is analyzed.(2)On the basis of the roughness reference plane,an Abbott-Firestone curve is used to characterize the high-frequency information(it is related to the roughness)of the gray image.Then,by using the mind evolutionary algorithm(i.e.,MEA)and back propagation neural network(i.e.,BPNN)method,a mapping relationship model between the Abbott-Firestone curve parameters and the measured roughness is established to evaluate the roughness on the cross-hatched liner surface effectively.(3)For the cross-hatched line with non-uniform size and distribution form,the gray image information related to its physical features(i.e.,geometry and distribution forms)is obtained using the curvelet transform.On this basis,the gray level co-occurrence matrix(i.e.,GLCM)is adopted to characterize the physical features of the cross-hatched line.The effectiveness of curvelet transform to extract the physical features of the cross-hatched line is investigated,and the relationship between the parameters of GLCM and the physical features of the cross-hatched line is also discussed.(4)By conducting frequency domain transformation and binarization for the gray image of the cross-hatched liner surface,the non-contact recognition on the angle of the cross-hatched line is implemented using the Canny algorithm and the modified Hough transform algorithm.On this basis,the effectiveness and feasibility of the recognition method for the angle of the cross-hatched line is analyzed.