Abstract

Due to the nonlinearity, strong coupling and hysteresis of the parameters used in measuring the grain drying process, it is always a challenge to perform accurate control of the drying system. The purpose of this paper is to describe the design of a suitable neural network model to control the grain dryer more effectively. First, the operating mechanism of a continuous grain dryer and the principle of the heat and mass transfer that can be used in the process of grain drying was analyzed before an intelligent control system was designed accordingly. Second, an intelligent control system based on the Back Propagation Neural Network (BPNN) was developed. The BPNN was the optimal model selected based on a series of comparative test results. According to the BPNN prediction of the moisture content of dried rice, the system could adjust the rate of grain discharge of the dryer, and then control the drying process accurately. Finally, the neural network control model was simulated using computer simulation technology, and was optimized by comparing analysis results with experimental results. The results showed that the optimized intelligent control system using BPNN has the advantage of strong stability and good noise handling, and could have great potential for future implementation studies.

摘要

由于用于测量谷物干燥过程的参数的非线性、强耦合和滞后性，对干燥系统进行精确控制一直是一个挑战。本文的目的是描述一种合适的神经网络模型的设计，以更有效地控制谷物干燥机。首先，分析了连续式粮食烘干机的工作原理和粮食烘干过程中可利用的传热传质原理，并据此设计了智能控制系统。其次，开发了基于反向传播神经网络（BPNN）的智能控制系统。BPNN是基于一系列对比测试结果选择的最优模型。根据 BPNN 对干米水分含量的预测，该系统可以调节烘干机的出粮速度，从而准确控制烘干过程。最后，利用计算机仿真技术对神经网络控制模型进行了仿真，并通过分析结果与实验结果的对比对神经网络控制模型进行了优化。结果表明，采用BPNN优化的智能控制系统具有稳定性强、噪声处理好等优点，在未来的实施研究中具有很大的潜力。