家电中混合塑料的模糊聚类智能红外分选及其循环再利用

Plastics consumed by home appliances is about 10--20％ of its total volume. The recycling of waste home appliances and waste plastics is a green and efficient solution to the saving raw material and reducing energy consumption. It is an old problem of seperating different kind’s plastics after manual dismantling and mechanic shredding. This project tries to develop the improved version of patterns recognization based on the fuzzy classification and intelligence, and guide lines for these typical plastics by the infrared technology. ..We will try to separate the typical plastics such as PP, PS，ABS and etc. There are several steps of improving fuzzy classification and intelligent pattern recognation, testing and verifying substantial samples from the recycling plant. Therefore, we draw a reasonable pattern recognation based on fuzzy classification and intelligence, and guide lines for these typical plastics. It will help enrich the thoery of fuzzy classification and increase the accurate rate of samples. Some plastics, which contain hazardous material restricted in RoHS directive and bio-degradable PLA, will be explored as well...The new patterns recognization based on the fuzzy classification and intelligence will distinguish 98% from the mixed plastics particles accurately. It will meet the requirement of the existed recycling plants. We have planed to experiment it in a plant to confirm the function and realiability. Research method, content and the application of this project are consistent and innovative. The cooperation among enterprise, university and institute is encouraged by the government and this model will be put into effact in the project. The close relation has provided foundation for the future popularization. Some postgraduates and engineers will benefit from the research process.

（1）家电用塑料约占其年消耗量的10-20%，对其循环再利用是一项节材、减耗的绿色经济发展之路。针对废旧家电塑料在回收拆解之后面临的识别分选难题，揭示其中的模糊分类智能识别模型及红外分选特征判据等相关机理。（2）针对PP、PS、ABS等典型塑料，改进模糊聚类分选过程中的智能识别模型，通过大量实测样本进行新模型的检测、验证和完善，得出基于改进的模糊聚类智能识别模型，及不同种类塑料的分选特征判据，从而达到丰富模糊聚类通用理论的目的，并提高红外分选的准确度。对含有RoHS指令限制使用有害物质的塑料、PLA等做有益的分选探索。（3）模糊聚类智能识别模型对典型塑料识别准确率将达98%以上，达到混合塑料分选的要求，并在1家处理工厂测试。项目在研究方法、内容及研究结果的应用方面均体现出创新性，一致贯彻基础性理论及模型的改进、分析优化、验证与试制推广的产学研联合开发之路，并培养该领域后备创新人才。

随着化工技术的发展，塑料产量与消费量也逐年增长，如果按体积计算已经接近或超过世界钢铁的产量；另一方面，全球资源能源相对短缺，废弃后的各种塑料带来了严重的环境压力。以红外分选技术为基础的物理分选技术是实现各类废塑料分类循环再利用的有效途径之一，可望实现节约原材料、减少能耗、发展循环经济良好的社会和经济效益。项目试制开发了一套家电废弃塑料分选硬件设备，及对应的智能分类识别模型，对PP、PE、ABS、PS典型塑料的识别准确率达98%以上。.（1）项目开发了一套家电废弃塑料分选硬件设备，主要包括上料、物料传送、识别、喷吹分离、料仓以及相关的控制软硬件。通过对喷嘴进行多因素的正交实验分析，得出了一组优化的结构参数：收缩比1.5，收缩段长度4mm，喷嘴直径5mm，喷嘴总长度选择尽量短，内壁光滑；流体参数：出入口压力比2；还对各因素之间的交互作用进行了优化计算，结果表明喷嘴出口半径、收缩比和收缩段长度这3个因素对喷嘴射流核心段长度、噪声均有显著性影响。.（2）分类识别智能算法方面，对生产一线过程中的实验样本PP、PC、ABS、PS、PVC的红外光谱并进行分析，开发了实验光谱数据的预处理、PCA算法程序、综合遗传算法和BP神经网络的优化模式识别程序等，总结出分选适应性规律，实现优化的分类识别智能算法及分选特征判据，识别准确率达98%以上。. 分析对比了国内外两个不同的红外光谱传感器，国外高价的第一个主成分累计贡献率就已经达到了95%以上，国内低价的前三个主成分才能够达到95%的累计贡献率，加阶后都可以满足分选的精度要求。.（3）废弃塑料分选硬件设备和分类识别智能算法这两项核心技术，申请的发明专利都获得了授权；发表科研论文8篇，培养研究生6名，获得了包括山东省技术发明二等奖在内的科技奖励3项。项目定量指标都超过了申请书中的预期成果。

内容获取失败，请点击重试