基于分布式集成方法的区间二型模糊模型设计

Distributed integration method for the design of interval type-2 fuzzy model

基于分布式集成方法的区间二型模糊模型设计可以降低模型的计算复杂度,提升模型的学习速度。文章采用该方法将数据集划分为多个子数据集,通过并行处理策略得到每个子数据集对应的一型模糊模型并进行集成,构建初始区间二型模糊模型;利用最小二乘法对二型模糊模型进行参数优化以确保区间二型模糊模型性能;对比分析了自适应模糊模型、反向传播神经网络以及基于差分进化的区间二型模糊模型等经典方法。结果表明:所提模型预测精度与其他3种方法的相近;分布式集成方法在保证模型性能基础上加快了学习速度,其在风力发电和地铁人流量预测中的训练时间分别为0.6708和0.9204 s,较最优势的模型速度分别快了17和13倍。

The design of the interval type-2 fuzzy model based on distributed integration can reduce the learning complexity of the model and improve the training speed.With this method,the dataset is divided into multiple sub-datasets,and the corresponding type-1 fuzzy models are obtained by parallel processing strategy and integrated to construct the initial interval type-2 fuzzy model.The least square method is used to optimize the parameters of the interval type-2 fuzzy model to ensure its performance.Classical methods such as the popular adaptive fuzzy model,the back propagation neural network,and the interval type-2 fuzzy model based on differential evolution are compared and analyzed.Experimental results demonstrate that the distributed integration method proposed in this study accelerates the learning speed on the basis of ensuring the performance of the interval type-2 fuzzy model.Its training time in wind power forecasting and subway traffic estimation problems is 0.6708 s,0.9204 s,respectively,17 and 13 times faster than the most advantageous model.