改进灰狼算法在土壤墒情监测预测系统中的应用

发布时间：2018-05-25 19:39

本文选题：土壤墒情预测系统 + 灰狼优化算法　；参考：《计算机应用》2017年04期

【摘要】：针对现有的固定端传感器土壤墒情监测预测系统架设成本高、传感器易损坏、预测精度较低等问题,设计并实现了基于非固定无线传感器组网与改进灰狼算法优化神经网络的土壤墒情监测预测系统。系统使用非固定即插即用式传感器蓝牙组网收集墒情数据,使用高精度多源定位接入融合方法进行广域室外高精度定位。在算法方面,针对灰狼算法在迭代中后期易陷入局部最优等问题,提出一种基于末尾探索者策略的改进灰狼算法。首先,根据种群个体适应度值排名,在原有算法个体类型中增加探索者类型。然后,将种群搜索分为三个时期:活跃探索期、周期探索期和种群回归期。最后,在每个时期使用特有的位置更新策略进行探索者位置调整,使得算法在探索初期更具随机性,在探索中后期依然保持一定的解空间搜索能力,从而增强算法的局部最优回避能力。使用标准函数进行算法性能测试,并将该算法应用于优化土壤墒情神经网络预测模型问题,使用某市2号试验田的数据进行实验。实验结果表明,所提算法与直接神经网络预测模型相比,相对误差下降约4个百分点;与传统灰狼算法、粒子群优化(PSO)算法优化模型比较,相对误差下降约1至2个百分点。所提算法拥有更小的误差,更好的局部最优回避能力,能有效提高墒情的预测质量。
[Abstract]:Aiming at the problems of high installation cost, easy damage of sensors and low precision of prediction, the existing fixed end sensor system for monitoring and forecasting soil moisture content is high. A soil moisture monitoring and forecasting system based on non-fixed wireless sensor networking and improved gray wolf algorithm optimization neural network is designed and implemented. The system uses non-fixed plug and play sensor Bluetooth network to collect moisture data, and uses high-precision multi-source positioning access fusion method to carry out wide-area outdoor high-precision positioning. In the aspect of algorithm, an improved gray wolf algorithm based on the end seeker strategy is proposed to solve the problem that the gray wolf algorithm is prone to fall into local optimum in the middle and late stage of iteration. Firstly, according to the rank of population individual fitness, the seeker type is added to the individual type of the original algorithm. Then, the population search is divided into three periods: active exploration period, periodic exploration period and population regression period. Finally, the special location updating strategy is used to adjust the location of the explorers in each period, which makes the algorithm more random in the initial stage of exploration, and still maintains a certain ability of solving space search in the middle and late stages of exploration. Thus the local optimal avoidance ability of the algorithm is enhanced. The standard function is used to test the performance of the algorithm, and the algorithm is applied to optimize the prediction model of soil moisture. The experimental results show that the relative error of the proposed algorithm is about 4 percentage points lower than that of the direct neural network prediction model, and that of the particle swarm optimization (PSO) algorithm is about 1 to 2 percentage points lower than that of the traditional grey wolf algorithm. The proposed algorithm has smaller error and better local optimal avoidance ability, which can effectively improve the prediction quality of soil moisture.
【作者单位】：北京邮电大学电子工程学院;
【基金】：国家科技支撑计划项目(2014BAD10B06)~~
【分类号】：TP18;TP212.9

【相似文献】