摘要
【目的】针对现有拖拉机耕作能量利用率低、不能根据实际情况合理分配发动机功率的问题.通过研究拖拉机耕作的耕深、速度与牵引阻力之间的复杂关系,并最终对该过程构建牵引阻力预测模型.【方法】基于此提出1种改进模糊神经网络(FNN)的牵引阻力预测方法,以耕作时的耕深、速度和牵引阻力为研究对象,通过基于密度的噪声应用空间聚类(DBSCAN)聚类算法确定FNN的初始结构和模糊规则数,并设计使用最小二乘法与反向传播算法组成的混合学习算法来实现模型的训练.【结果】将优化的模糊神经网络模型与支持向量机(SVM)、随机森林(RandomForest)对比后发现,DBSCAN优化后的FNN平均相对误差为4.36%,比其他2种模型分别低1.68%、2.40%.【结论】利用改进的FNN能够精确的预测出耕作时的牵引阻力大小,为后续的拖拉机能量管理和功率分配的研究提供了基础.
【Objective】In view of the fact that the power consumption of existing tractor farming has low energy efficiency and the engine power cannot be properly distributed according to actual conditions.Through the study of the complex relationship among the depth,speed and traction resistance of tractor tillage,the traction resistance prediction model is constructed for this process.【Method】An improved prediction method of traction resistance based on fuzzy neural network(FNN)is proposed.The research on the depth,speed and traction resistance of tillage is taken.The density-based noise application spatial clustering(DBSCAN)clustering algorithm is used to determine the initial FNN.Structure and fuzzy rule numbers,and design a hybrid learning algorithm consisting of least squares and backpropagation algorithms to achieve model training.【Result】The optimized fuzzy neural network model is compared with support vector machine(SVM)and random forest.The experimental results show that the average relative error of FNS after DBSCAN optimization is 4.36%,which is 1.68%and 2.40%lower than the other two models.【Conclusion】The improved FNN can accurately predict the power consumption during tillage,which provides a basis for the subsequent research on tractor energy management and power distribution.
作者
张波
周俊
ZHANG Bo;ZHOU Jun(College of Engineering,Nanjing Agricultural University,Key Laboratory of Intelligent Agricultural Equipment of Jiangsu Province,Nanjing 210031,China)
出处
《甘肃农业大学学报》
CAS
CSCD
北大核心
2020年第1期213-220,228,共9页
Journal of Gansu Agricultural University
基金
国家重点研发计划项目“电动拖拉机智能化操控与作业关键技术研究及核心零部件研制”(2016YFD0701003).
关键词
牵引阻力预测
模糊神经网络
自适应神经模糊推理系统
支持向量机
随机森林
traction resistance prediction
fuzzy neural network
adaptive network-based fuzzy inference system
support vector machine
random forest