卡尔曼滤波在预测管廊甲烷体积分数中的应用

Application of Kalman filter in predicting methane volume fraction in utility tunnel

为提高综合管廊环境监控系统的预警性,保障管廊安全运行,必须解决燃气舱甲烷体积分数在预测过程中存在的偏差问题。通过加入温度参数对预测数据进行补偿,建立卡尔曼滤波甲烷体积分数预测模型,并对比神经网络与卡尔曼滤波相结合的经典模型。仿真结果表明:加入温度补偿后的卡尔曼滤波方法得到的平均误差为6.18%,神经网络结合卡尔曼滤波方法得到的平均误差为8.79%。对比分析表明:温度补偿后的卡尔曼滤波方法预测效果更优,具有较好的跟踪能力和反应速度,预测值更接近测量值。

In order to improve the early warning performance of the environmental monitoring system in utility tunnel and ensure the safe operation of the utility tunnel,it is necessary to solve the problem of deviation of methane volume fraction in the gas cabin in the prediction process.In this paper,the prediction data is compensated by adding temperature parameters,and the prediction model of methane volume fraction by Kalman filter is established,and compare the classical model combining neural network and Kalman filter.The simulation results show that the average error of Kalman filter method with temperature compensation is 6.18%,and that of neural network combined with Kalman filter method is 8.79%.Comparative analysis shows that Kalman filter method with temperature compensation has better prediction effect,better tracking ability and response speed,and higher prediction value.