基于随机权重粒子群优化极限学习机的土壤湿度预测

Prediction of Soil Humidity Based on Random Weight Particle Swarm Optimized Extreme Learning Machine

土壤湿度的预测对农业生产和科学研究都有着重要的意义.针对极限学习机(ELM)回归模型预测结果受输入参数影响的问题,本文将随机权重的粒子群优化算法(RandWPSO)应用于ELM回归模型中,提出了一种基于随机惯性权重的粒子群优化极限学习机的土壤湿度预测方法.该方法是利用传感器测出的土壤温度和光照强度数据进行数据预处理,构建出训练样本集,并且建立ELM回归模型,利用随机权重的粒子群算法优化ELM中的输入权值和阈值,避免陷入局部最优,从而建立起基于RandWPSO-ELM土壤湿度预测模型.利用MATLAB仿真软件,构建随机权重的粒子群优化ELM的预测模型,并对呼兰地区大棚甜菜的土壤湿度进行实验.结果表明:该方法的精度高且稳定性好,能够为大棚甜菜的生长提供有效的参考依据.

The high quality prediction of soil moisture is of great significance to agricultural production and scientific research.To solve the problem that the prediction result of the limit learning machine(ELM) regression model is affected by the input parameters,the random weight particle swarm optimization algorithm(RandWPSO)is applied to the ELM regression model.This method is to use the sensor to measure soil temperature and light intensity data for data preprocessing,build a training sample set,and establish the ELM regression model,the random weighting of the particle swarm algorithm to optimize input weights and threshold of ELM,avoid falling into local optimum,thus built up based on RandWPSO-ELM soil moisture forecast model.MATLAB simulation software was used to construct the prediction model of ELM with random weight,and the experiment was conducted on the soil humidity of greenhouse beet in Hulan area.The experimental results show that the method has high precision and good stability,and can provide effective reference for the growth of greenhouse beet.