One of the most crucial emerging challenges in lithography is achieving rapid and accurate alignment under a wide variety of conditions brought about by the different overlying films occluding the marks. The problem is exacerbated by planarizing processes such as chemical mechanical polishing that reduce the topographical contrast used to view the marks and by distortion of the wafer and of the stage. Thus, an effective learning process is needed to rapidly acquire the best possible positional information from an array of the marks across the wafer. In this article, a neural network model for global alignment is described. The wafer and stage distortions can be incorporated into the model. The algorithm finds the best fit for the wafer distortion and at the same time compensates for both stage and wafer distortion. The algorithm can also learn to identify and ignore any alignment marks that yield significantly erratic signals. A few common distortion functions will also be used to test the model. Preliminary simulation results show alignment errors <5 nm in the presence of Gaussian noise with sigma=30 and 500 nm sinusoidal stage distortion. Also the simulations show that increasing the magnitude of stage distortion has no impact on the results. It will be shown that considering wafer distortion can reduce the alignment error more than 10 times in the case of bowed and distorted wafers.