Fundamental Studies on Grain Drier Control by Learning Method

为了控制谷物干燥机从不可靠的概率控制元素中生产出高质量的谷物，我们提出了一种“学习方法”。学习方法从通过学习找到最优控制序列开始，在最优控制方式上跟踪实际控制。为了提高控制精度，有必要改进干燥前和干燥过程中谷物质量测量的传感技术。提出了一种测定高水分水稻品质的方法；采用叶轮式砻谷机对高水分稻谷进行脱壳，脱壳后的稻谷（糙米）用外质量检测仪检测外质量，用近红外分析仪检测内质量。这种方法的可用性得到了保证。该方法已被某公司用于制造商用测试仪。对叶轮脱壳机的动态特性进行了研究，得出了脱壳风机转速与脱壳机的比例规律。作为一种低成本生产高品质谷物的干燥方法，Nishiyama提出了吸气式气流储存干燥系统。该系统采用了学习控制方法。为节省实验时间，还采用了计算机模拟。通过计算机仿真，得到了最佳控制顺序，并通过控制干燥空气的平衡含水率进行了实际控制，控制精度在0.5%,w.b.s。

In order to control grain dryer to produce high quality of grain from unreliable probability control element, we proposed a "learning Method". Learning method starts from finding the best control sequence by learning and the real control is traced on the best control way.To improve the control accuracy, it is necessary to improve the sensing techniques for measuring quality of grain before and during drying. The proposed method of measuring quality of high moisture paddy rice is such; Dehulling high moisture paddy by impeller type rice husker, and dehulled rice (brown rice) is inspected on the outer quality by outer quality tester and inner quality by NIR analyzer. Availability of this way was assured. This method is adapted by a company to manufacture a commercial tester. Dynamic characteristics of impeller husker was investigated and the law of proportion in terms of r.p.m. of dehulling fan was found.As a drying method for producing high quality grain at low cost, Nishiyama proposed a suction air flow storage drying system. Learning control method was applied to this system. To save the experimental time, computer simulation was also used. By computer simulation, the best control sequence was obtained and real control was carried out by controlling the equilibrium moisture content of drying air and accuracies of controlling moisture content was within 0.5%,w.b..