On the innovated indirect method to determine the weighted coefficient matrix of data reconciliation

For the problem of worse performance induced by improper weighted coefficient matrix in data reconciliation algorithm, an innovated indirect method to determine the weighted matrix for data reconciliation is proposed. The method deduces the measurement variances with the estimated constraint residual errors based on the redundancy character. When singular matrix appears in the calculation process of matrix inversion, block matrix together with the known variances from empirical or direct method are applied to compensate the vacant elements of the weighted matrix. Thus the known information is utilized as more as possible to deduce the unknown measurement variances. Compared with the empirical method or direct method, the innovated indirect method shows obvious advantages, and it solves the problem of singular matrix inversion in indirect method. The result of data reconciliation for the flow rate measurement of steam network in an iron & steel corporation shows the method can largely improve the effect of data reconciliation.