The invention provides a WLAV (Weighted Least Absolute Value) robust estimation method based on a multi-predication-correction interiorpoint method and relates to an electric system operation and control method. Weighted least square state estimation is adopted for state estimation, but the estimation precision is difficult to further improve. The method comprises the following steps: 1) acquiring electric system parameters; 2) acquiring detection data; 3) initializing; 4) applying each quantity measurement Hessian matrix memory space and solving; 5) calculating a duality gap CGAP which is equal to alphI+betaU, judging whether meeting conditions that CGap is less than epsilon or K is less than Kmax; 6) predicating a step, namely setting a perturbation factor mu as zero, predicating according to formulae and solving out an affine direction; 7) correcting the step; 8) judging that a correction frequency counter t is less than 4; 9) solving out an equation as shown in the specification so as to obtain delta lambda co, and correcting; 10) calculating a new iteration step which is as shown in the specification and is greater than an original step, updating according to a formula of delta lambda new=delta lambda af + delta lambda co and t=t+1, and executing the step 8, otherwise, iterating a counter K which is equal to K+1, and executing the step 5. According to the technical scheme, the number of iterations is reduced, the processing speed is improved, and the convergence property of the algorithm is further improved.