论文部分内容阅读
为将由不同农作物酿造的葡萄酒进行快速而准确地分类以满足市场的需求,从训练时间和分类正确率两个方面研究并比较了针对此类问题的BP网络和RBF网络性能。根据对葡萄酒中的13种代表性成分进行化学定量分析,将这13维定量成分向量经线性归一化后作为神经网络的输入向量,所属类别作为输出向量,在MATLAB中分别训练了RBF神经网络和BP神经网络,然后利用训练好的网络再对不同种的葡萄酒进行测试,并对测试结果进行了对比分析。仿真结果表明,RBF网络和BP网络的平均收敛时间分别为25.43s和2.93s;平均测试误差分别为0.22105和0.1684。BP网络收敛时间少,且分类准确率上也优于RBF网络。
In order to quickly and accurately classify wines brewed from different crops to meet market demand, the performance of BP network and RBF network for such problems are studied and compared from training time and classification accuracy. Based on the chemical quantitative analysis of 13 representative components in wine, this 13-dimensional quantitative component vector is linearly normalized and used as the input vector of the neural network. The category belongs to the output vector, and the RBF neural network And BP neural network, and then use the trained network to test different types of wine, and the test results were compared. The simulation results show that the average convergence time of RBF network and BP network are 25.43s and 2.93s, respectively. The average test errors are 0.22105 and 0.1684 respectively. BP network convergence time is less, and the classification accuracy is better than the RBF network.