论文部分内容阅读
针对传统的土地复垦适宜性评价方法受人为主观影响较大的问题,该文通过分析神经网络结构、计算流程及误差传递规律,建立了基于人工神经网络的输变电土地复垦适宜性评价模型;然后根据土壤等级7个评价参数的特点确定输入层神经元数为7,根据土地复垦成4类的特点确定输出层神经元数为4,根据经验公式确定隐含层神经元数为9;配置成7-9-4的网络结构,采用Levenberg-Marquardt快速学习算法对网络进行训练、测试及矫正;最后采用输变电线路具有代表性实测样本对该算法进行验证。实验表明经过神经网络识别的土壤等级与实际综合评估土壤等级相符,说明了BP神经网络在输变电土地复垦项目中的可行性和实用性。
In view of the problem that the traditional land reclamation suitability evaluation method is greatly affected by man-made subjectivity, this paper sets up a suitability evaluation of reclamation power transmission and transformation land based on artificial neural network by analyzing the structure of neural network, calculation flow and error propagation law Then, the number of neurons in input layer was determined according to the characteristics of 7 evaluation parameters of soil grading. According to the characteristics of land reclamation into 4 types, the number of neurons in output layer was 4, and the number of neurons in hidden layer was 9; configured as a network structure of 7-9-4, adopting Levenberg-Marquardt rapid learning algorithm to train, test and correct the network; finally, the proposed method is verified by a typical measured sample of the transmission line. Experiments show that the soil grade identified by neural network is consistent with the actual comprehensive assessment of soil grade, indicating the feasibility and practicability of BP neural network in the power transmission and land reclamation project.