基于PSO优化RBF神经网络的溶解氧预测算法研究

摘要

The concentration of dissolved oxygen is an important index for the safety of fishery water quality,and it is also a key factor affecting the health of aquaculture products,and it is of great significance to monitor and predict the water quality in real time.Affected by the environment parameters such as pH value,dissolved oxygen is adopted in this paper on the changes of dissolved oxygen MI_PSO_RBF(mutual information_particle swarm_RBF neural network) algorithm to predict the dissolved oxygen content in the fishery breeding environment,first using MI reduce the statistical correlation between two random variables,and then using RBF neural network algorithm for fishery breeding dissolved oxygen in water environment to predict the change trend.Finally,PSO was used to optimize the model parameters of RBF neural network,and the model was used to predict the change trend of the dissolved oxygen in fishery.The experimental results show that the stability of multi-parameter remote monitoring system is good,and the prediction algorithm based on MI_PSO_RBF is good,which provides a good reference value for fishery cultivation.%溶解氧浓度是渔业养殖水质安全的重要指标,也是影响养殖水产品健康的关键因素,对其进行实时监测和预测具有重要意义.溶解氧受环境中pH值等参数影响,针对溶解氧的变化情况该文采用MI_PSO_RBF(互信息_粒子群_RBF神经网络)算法对渔业养殖环境溶解氧含量进行预测,首先采用互信息理论MI降低两个随机变量统计的相关性;然后采用径向基函数RBF神经网络算法对渔业养殖水环境中溶解氧变化趋势进行预测;最后采用粒子群算法PSO对RBF神经网络的模型参数进行优化,并利用该模型对渔业养殖溶解氧变化趋势进行预测.经实验验证表明,多参数远程监测系统稳定性好,基于MI_PSO_RBF的溶解氧预测算法预测效果良好,为渔业养殖提供了良好的参考价值.