Data Reconciliation Based On Wavelet Packet Denoising Method With Bayesian

As the measurement error will inevitably exist, the measured value does not correctly reflect the actual chemical process, this phenomenon is known as the imbalance of the measurement, the solution of this problem is to use the data reconciliation. Current study of data reconciliation is mainly based on steady-state linear system. This paper studies a denoising method which is a combination of Bayesian, discrete wavelet transform and wavelet packet noise reduction method for the unstable conditions of actual production data. Combining a variety of statistical methods for data noise reduction, the method is used to minimize errors in the measured data and obtain the steady-state reconciliation data. This paper focuses on the use of Bayesian ideas to improve the computing speed and efficiency, with the strong ability of Bayesian judge, make full use of the database to obtain a large number of historical data prior knowledge, and as a basis for judging. This paper study the application of wavelet packet denoising to the production data of coking plant from the practical engineering point of view.Thesis as follows:Review wavelet transform, discrete wavelet transform, reviewed the current development of wavelet noise reduction.Discuses disadvantages of the Fourier transform of the significance and advantages of wavelet transform, analyze the principles of discrete wavelet transform in details.Propose a denoising method of wavelet packet with Bayesian. The method using Bayesian thinking, takes advantage of the rich history of industrial process data to improve the method of wavelet packet denoising.Take a coking plant production process as the object to study a representative sample of the process of data reconciliation issues. Reconcile the data after denoising and obtain more satisfactory results.Finally, summary the full-text content and forward views about the further study of the wavelet packet denoising technology.