Study On Detection Of Gross Errors For Process Data

In process industry, the laws of materials or energy conservation can't always be satisfied for instrument readings. So it is important to reconciliate measurements by data reconciliation technique. Instrument biases and malfunctioning instruments are two main reasons for biased measurements. Leaks and biases are called gross errors. If measurements are adjusted to satisfy the laws of conservation in the presence of gross error, all the measurements are greatly affected by such leaks and biases and would not generally be reliable indicators of the state of the process. So gross errors must be detected and either rectified or discarded before data reconciliation.This dissertation concerns on gross error detection and estimation in steady state linear process. The main contents of this dissertation are outlined as follows,(1) A novel data classification algorithm based on matrix projection is presented. When matrix projection is not unique, data classification can not be classified perfectly by the existing matrix projection algorithm. To solve the problem, a pre-classification algorithm of measured data introduced and a new algorithm is developed to divide unmeasured data into determinable and indeterminable data, using matrix projection algorithm. For the new algorithm, only two projections of two incidence matrices are required to solve and all data can be classified perfectly.(2) An NT-MT combined method based on nodal test (NT) and measurement test (MT) is developed for gross error detection and data reconciliation for industrial application. The NT-MT combined method makes use of both NT and MT tests and this combination helps to overcome the defects in the respective methods. It also avoids any artificial manipulation and eliminates the huge combinatorial problem that is created in the combined method based on the nodal test in the case of more than one gross error for a large process system. Serial compensation strategy is also used to avoid the decrease of thecoefficient matrix rank during the computation of the proposed method.(3) A novel mixed integer linear programming model for detection of gross errors is presented in this paper. Yamarura designed a model for detection of gross errors and data reconciliation based on Alaike information criterion. But much computational cost is needed due to its combinational nature. A mixed integer linear programming (MILP) approach was performed to reduce the computational cost and enhance the robustness. But it loses the super performance of maximum likelihood estimation. To reduce the computational cost and hold up the merit of maximum likelihood estimation, the simultaneous data reconciliation method in a frame work of MILP is decomposed and replaced by a novel mixed integer linear programming (NMILP) subproblem and a quadratic programming (QP) or a least squares estimation (LSE) sub-problem.(4) For gross error detection, there are no effective approaches. Many methods are presented, but few considers the identification of gross errors. Based on the previous results, series of theory of identification of gross error presented.(5) The measurement test(MT) method performs poorly in practical industry. Several modified methods are presented to improve the performance of MT. But they all reduce the redundancy of measurements. Two new modified MT are presented. In the new methods, system structure are not changed during computation.The conclusion and perspective are given at the end of the dissertation.