| Mixing is absolutely necessary and important in the process of rubber processing. The dispersion of rubber product’s important reinforcing fillers, such as carton black and white carton black, has great influence on rubber’s physical and mechanical properties and processing performance, even rubber product’s quality and service life. It’s easy to form big tiles when carton black’s dispersion is uneven, which is not only bad to the subsequent process, such as rolling, extruder, but also causes unbalance problem of products with great speed. The dispersion of carton black(including other fillers) shows dispersing condition of carton black in rubber, which is related with many factors, such as convertible-top peg’s press, rotor’s speed, cooling water’s temperature, rotor’s structure and type. Changing these factors will have an effect on mixing. It’s great meaningful for improving mixing process and quality of rubber product to research carton black’s dispersion theory and evaluate carton black’s dispersion.At present, the devices that can evaluate carton black’s dispersion automatically have been great application in rubber industry, but their problems are also obvious. for example, there are some deviations in recognizing carton black of rubber image because there is not obvious boundary between carton black and rubber. The method of evaluating dispersion is not enough complete and scientific, the result of evaluation just represents local carton black’s fragmentation, not shows the degree of carton black’s whole fragmentation and diffusion. The stability of carton black’s dispersion process is just characterized by process curve which shows instantaneous power of mixer.In accordance with these problems, this paper researched identifying carton black, characterizing carton black’s morphology, building the dispersion evaluation system, judging the stability of carton black’s dispersion process deeply, special works included:1. By the way of comparing several typical image segmentation algorithm combined with analysis of rubber image’s feature, k-means method was selected to process image. Thinking about the problems of k-means, two methods were supplied: the method based on symmetric distribution and the method based on inflection point. By comparing outcomes of experiments, the method based on inflection point was more reasonable. Combining the method based on inflection point, k-means algorithm solved the problem of distinguish carton black and rubber.2. Given the influence of carton black’s shape on rubber’s property, A new method was supported that the parameters, such as length diameter ratio, morphology complexity, were used to characterize carton black’s shape. The method of ellipse fitting and the method of minimum bounding rectangle solved the problem of calculating length diameter ratio. Experiments proved that the method of ellipse fitting was more reasonable. The method based on convex hull and the method based on perimeter solved the problem of calculating morphology complexity. Experiments proved that the method of perimeter was more precise.3. According to the difference between impurity and carton black, filamentous impurities and scratches in rubber image was disposed. According to the characteristics of long perimeter and arc about filamentous impurities, the ratio of perimeter to solidity of region and hole’s area were calculated to distinguish this type of impurity. Experiments proved that if hole’s area was larger than3pixels or the value of the ratio larger than160, the region was regarded as filamentous impurity. For linear impurity, such as scratch, experiments proved that if a region’s length-diameter ratio is larger than6.5, the region was regarded as scratch.4. Though analyzing standard images, the rule was found that the whole area and count of carton black changed smaller with the improvement of dispersion level. Combined with divided feature regions, this paper built four evaluation models about carton black’s microcosmic dispersion on the basis of the whole area and count of carton black. Experiments proved that the model had eight feature regions based on area was more reasonable.5. According to the theory that product’s feature values satisfied normal distribution in the stable process, on the basis of detecting carton black many times, SPC was used to judge whether canton black’s density satisfied the requirement of data distribution in the stable process, which decided the stability of carton dispersing process. According to the detected data, the feature values (including mean, skewness, kurtosis) were calculated, which described macroscopic fragmentation and diffusion of carton black.On the basis of identifying carton black more precisely, this paper built an evaluation model that represented the degree of fragmentation and diffusion from local and global perspectives,supplied a new method of evaluating carton black’s dispersion stability directly and developed corresponding software system. The project researching was helpful for researching the influence of changing process parameters on carton black’s dispersion, building the relationship between dispersion and rubber’s physical properties, which was used to optimizing mixing process. |
