The problem of obtaining the correct optimal steady-state operation of an industrial process, despite the deficiencies in its available mathematical model, is addressed in the paper. The control method considered is the integrated system optimization and parameter estimation (ISOPE) technique, and a new algorithm, of Newton-like type, is presented in the paper for this technique. The algorithm is derived for the augmented version of the ISOPE technique and hence it is also applicable for non-convex problems. The algorithm does not require more measurement information than the previous simple relaxation-type algorithms. It is convergent in one iteration for the case of a linear process and quadratic performance function. A local convergence analysis is provided for the general non-linear case. Simulation results are also given, and comparison with the relaxation type algorithm is presented, indicating the superiority of the new approach for the examples considered