基于BP神经网络的智能化玉米专家系统的构建

Architecture of Intelligent Corn Expert System Based on Neural Networks

传统专家系统的知识库是静态的,每一次维护和更新都要由农业专家和知识工程师共同完成。农业知识具有地域性和时效性,且不同的农业专家的知识经验也不尽相同。笔者提出利用神经网络和模糊处理算法相结合,借助BP神经网络大规模的并行分布处理结构完成自适应学习过程。利用BP神经网络自适应地产生和精炼这些知识,根据本地不同农业专家的经验采用BP处理去优化这些知识,可以使专家系统的知识库不断更新,且更加适宜于本地经验的积累和针对该地区的特点进行决策。

Traditional knowledge base of Expert System (ES) is static and can't update itself. Agricultural knowledge is limited to certain area and certain period of time and different experts may have different judgment to the same question. In this paper, an algorithm based on neural networks (NN) and Fuzzy system is presented. Self-learning process based on parallel Neural Networks has been achieved. Decisions can be made by ES utilizing NN based on the experiences of local agricultural experts. With the continuous updating of the ES knowledge base, local area experiences can be accumulated and more accurate decisions can be made.