摘要
针对数值信息与定性领域知识相互融合的推理问题,提出了一种基于云变换的混合推理神经网络。利用逆向正态云发生器可实现论域中定量数值到定性概念描述间不确定关系的转换,建立基于云变换的混合信息推理逻辑和神经网络模型。将数值信息通过云变换在概率意义下转化为定性概念谓词,把推理规则表示为神经元,利用神经网络的学习性质来实现定量与定性混合信息的自适应计算推理。以油田开发水淹层判别为例,实验结果验证了模型和算法的有效性。
Aiming at the reasoning problems of the mutual integration betw een the numerical information and qualitative domain knowledge,a Reasoning Neural Network（ RNN） based on cloud transformation was proposed. Using the reverse normal cloud generator can achieve the conversion of the uncertain relationship betw een the quantitative values and qualitative concept description,and build the mixed information reasoning logic and RNN model based on cloud transformation. Then transform the numerical information into qualitative concept in the sense of probability through the cloud transformation,and express the inference rules as neurons,and use the learning nature of the neural networks to achieve adaptive processing of mixed quantitative and qualitative information. Taking pumping unit balance diagnostic for example,the experimental results verify the validity of the model and algorithm.
出处
《计算机技术与发展》
2015年第3期118-121,共4页
Computer Technology and Development
基金
国家自然科学基金资助项目(61170132)
中国石油科技创新基金项目(2010D-5006-0302)
关键词
信息融合
云模型
人工神经网络
推理
学习算法
information fusion
cloud model
artificial neural network
reasoning
learning algorithm
