Data Reconciliation Methods For Production Optimization

Petrochemical industry occupied a very important position in our country’s national economy, and the optimization techniques played an irreplaceable role in the energy saving, improving economic efficiency and other fields. Measurement data was the basis of process optimization techniques. Inaccurate measurement data would cause the error optimization strategy, and then the quality of the production would be affected, if serious, that would lead to accidents. Data reconciliation was based on existing measurement techniques to denoise the noise in measurement data using mathematical calculation methods. At last, the correction values must satisfy the law of material balance and energy balance for the production system. So we used data reconciliation technology to ensure the quality of the measurement data, only in this way can we develop correct optimization strategies and control schemes to achieve the objectives of low emissions, low energy consumption and high efficiency in industrial processes. The technologies related to the data reconciliation were researched in this subject.Problem description section discussed all the knowledge of data reconciliation, detailed calculation method of data reconciliation, gross error detection and covariance matrix. The objective function of data reconciliation process are mainly studied, the results of the least squares method and the coordination of robust were also compared in this section.Two parts were included in steady state test section. Firstly, the steady state test questions were defined in mathematical ways, and the on-line steady state test method proposed by Cao was mainly studied. Secondly, the test method based on wavelet packet filtering ideas was proposed, which pointed that the steady state test could be done after error of measurement data was denoised. The specific method was that random errors in the measurement data could be denosied by the wavelet packet filtering means, in this way, the correction value could reflect the operation trends very well, and then the state results would be known by the trends of the system.In simulation section, a fluid catalytic cracking unit(FCCU) was used as the research object, methods about steady state test, data reconciliation, parameter correction and simulation of the model were implemented. The steady model of the FCCU was described in detailed, the kinetic parameters were estimated based on the process data produced by the dynamic model of FCCU to correct the steady model.