Combined DWT Dynamic Data Reconciliation Research and Application
|School||East China University of Science and Technology|
|Course||Control Science and Engineering|
|Keywords||Discrete Wavelet Transform Data correction simpson integration method Mathematical programming method Kalman filtering|
This paper studies discrete wavelet transform (DWT) and traditional data correction theory combined to handle dynamic system data correction. Two ways to consider dynamic data correction processing. First, the first approach is the first dynamic system model equations into algebraic model equations, combined with a mathematical programming method for the correction value; The second method is the synchronization with the Kalman filter, and it is estimated the system state of the dynamic system estimates. In different ways in the course of the study, discrete wavelet transform analysis of dynamic data correction method to achieve the data corrected purpose. The main purpose of the use of discrete wavelet transform theory method for signal analysis and filtering of measurement data, and then solve the handling dynamic data correction problem is encountered shortcomings and deficiencies, such as polynomial rule is not suitable for handling dynamic amplitude larger The class dynamic process. Dynamic data correction method combined with wavelet transform analysis, a total of three. The first one, in the process of considering linear dynamic system to Simpson integration method the system dynamic model equations into algebraic equations, in turn, can take advantage of the more mature steady state data correction theory and methods to solve this dynamic data correction. Applying the discrete wavelet transform can be reduced the error generated in the conversion process, and can be pre-filtering the noise in the measurement data signal. The second method is to use a Kalman filter estimation method has the advantage of time recursive calculus, suitable to be used in online real-time dynamic data correction problem. Kalman filter augmented state variables often contain relatively large deviation, so the proposed the online filtering method based on discrete wavelet transform filtering method combined with the Kalman filter estimates, and to get better The estimated effects. Finally, the results of the measurement data signal as well as the use of the the Discrete Wavelet theory of scaling function to approximate the signal differential, and thus the system dynamic equations converted into algebraic equations using mathematical programming method processing dynamic data correction problem. Process parameter optimization, search the range of values ??where the range of values ??for the coefficients of the scaling function, which can reduce the number of search variables in the optimization process, improve computational efficiency and effective correction of the dynamic data purposes. For gross error detection and elimination method contains a significant error process, a gross error in the effective identification system is proposed in this article. In this paper, I will, by example simulation results further demonstrate that the proposed method, and discuss and compare the advantages and disadvantages of several methods.