基于BP + Elman神经网络的草原土壤湿度预测研究

Grassland Soil Moisture Prediction Based on BP + Elman Neural Network

草原是我国重要的国土资源和物质财富,是发展多种产业经济的重要原料基地,能为畜牧生产的重要饲料来源。国家通过“退牧还草”政策,有效改善草原生态系统和民生,但是仍需寻找一种合理的牧场管理政策,以最大限度的提高每个地区的生产力,同时保持社会生态的长期可持续性。本文建立模型对保持放牧策略不变情况下对2022年、2023年不同深度草原土壤湿度进行预测,通过数据挖掘技术建立土壤湿度预测模型,并进行模型验证。首先,建立土壤湿度与各因素之间的映射关系,选用Elman神经网络模型和回归方程对影响土壤湿度的物理因素进行预测,并对比两者之间的准确性,最后选用BP神经网络预测模型对土壤湿度进行求解,对于草原放牧策略的制定以及土壤湿度的监测具有重要的现实意义。

Grassland is an important land resources and material wealth in our country. It is an important raw material base for the development of a variety of industrial economy, and an important feed source for livestock production. The state has effectively improved the grassland ecosystem and people’s livelihood through the policy of “returning grazing land to grassland”, but it still needs to find a reasonable grazing land management policy to maximize the productivity of each region and main-tain the long-term sustainability of social ecology. In this paper, a model was established to predict the soil moisture of grassland at different depths in 2022 and 2023 under the condition that the grazing strategy remained unchanged. A soil moisture prediction model was established through data mining technology, and the model was verified. First, the mapping relationship between soil moisture and various factors was established. The Elman neural network model and regression equation were used to predict the physical factors affecting soil moisture, and the accuracy between the two was compared. Finally, the BP neural network prediction model was used to solve the soil moisture. It is of great practical significance for formulating grazing strategy and monitoring soil moisture in grassland.