摘要
为科学合理地预测大气污染物PM2.5颗粒物浓度变化规律,分析PM2.5颗粒物浓度变化历史数据,综合判断外部条件(温度、风速、天气状况)和内部条件(其它污染物的浓度)对PM2.5颗粒物浓度变化的影响。采用一种改进型PSO优化的模糊神经网络,将粒子群算法与模糊神经网络进行融合,发挥PSO算法全局寻优的特点,预测PM2.5颗粒物浓度的变化规律。对某市2013年PM2.5颗粒物浓度进行预测和验证,验证结果表明,该算法具备良好的预测精度。
To predict the changing rules of the PM2. s concentration in the atmosphere, the history data of the PM2. 5 concentration was analyzed. Based on the changing rules of the PM2. s concentration, the external factors which consisted of the temperature, the wind power, the weather situation and the internal factors which were mainly the concentrations of other air pollution parti- cles were taken into account to calculate their influences on the concentration of PM2. 5. To make the prediction more efficient, an improved particle swarm optimization algorithm which combined advantages of both the PSO and the fuzzy neural network was selected as the optimization method. Based on the actual data of the PM2. s concentration, the model is proved to he accurate in predicting the PM2.5 concentration.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第9期3258-3262,共5页
Computer Engineering and Design
基金
宁夏自然科学基金项目(NZ1151)
关键词
PM2.5浓度预测
改进型PSO算法
模糊理论
神经网络
模型参数
PM2.5 concentration prediction
improved particle swarm optimization
fuzzy theory
neural network
model parameters
