Design of output data fusion soft sensor based on adaptive extended Kalman filtering algorithm

Soft sensor technology has attracted wide attention as an important method for the acquisition of critical quality variables in chemical processes.The researches on soft sensor nowadays mainly focus on the modeling technique.However,due to the complexity and diversity of chemical processes,there are always unsatisfactory results,such as unstable estimations,large random mistakes and so on,when using soft sensor models to directly estimate the critical quality variables.Aimed at this problem,several ameliorative algorithms have been reported,but they still have drawbacks of heavy calculation burden and poor applicability.Thus the authors propose a new soft sensor method,the output data fusion soft sensor design method based on adaptive extended Kalman filter (EKF)algorithm,which fuses the model estimations and field measurements to calibrate the deviations in modeling results by Kalman filtering.And a noise statistics estimator with attenuation factors is also developed under the condition of data fusion soft sensor.By integrating the noise estimator with EKF algorithm,an adaptive extended Kalman filter is constructed,which can effectively improve the accuracy and anti-interference capability of the EKF-based data fusion soft sensor.The effectiveness of the proposed algorithm is analyzed in depth through simulations.The algorithm is also used in a lab experiment to validate its practicability and applicability.