论文部分内容阅读
讨论了人工神经网络方法在含氮量预报上的应用策略 ,并建立了一个 6 - 7- 1结构的三层 BP网络模型 ,进而分析了 BP网络模型在实际应用中存在的问题 ,对 BP网络算法进行了改进 ,在基于改进的神经网络算法基础上 ,使用 C语言实现了程序设计 ,采用收集的 6 7组实验数据进行了离线学习 ,完成了对网络的训练 ,并用训练好的网络模型对 1 2组样本进行测试 ,预测值误差在± 1 0× 1 0 - 6范围内时命中率为 74% .
The application strategy of artificial neural network in prediction of nitrogen content was discussed and a three - layer BP network model with 6 - 7 - 1 structure was established. Then the problems existed in practical application of BP network model were analyzed. The BP network Algorithm is improved. Based on the improved neural network algorithm, the program design is realized by using C language. The collected data of 67 experimental groups are used for offline learning and the network training is completed. With the trained network model 1 2 groups of samples were tested, the prediction error of ± 1 0 × 10 6 range hit rate was 74%.