FGM plates are developed in ANSYS easily in APDL rather the GUI and if you are working in Workbench then it may not be possible to do so. Yes the next major point as the FGM is the continually varying in thickness or in an other direction then you need to write a function related to the properties and in ANSYS it will be developed as layerwise process and the functions will be given based the discrete with the function and as the number of layer will increase it will be more precise.
Dr. Subrata's suggestion is good one, but also you can make the plate made of solid elements and across the thickness assign to each row or elements material with variable properties in the fashion you wish. This option allows for better capturing the thru-depth strain variation and shear, rather than use of layered plate elements that is limited to low order shear theories.
We can use layered elements with properties assigned to each layer or use a solid Shell element in ANSYS and assign element properties to each layer of elements. This process can be cumbursome , so one can write an APDL script to select all the elements on a given plane and assign one property to all of them and further on for each plane. By plane i mean all the elements at an equi-distance from a reference plane(origin). This works well for geometries with some kind of symmetry.. but for arbitrary geometries we have to think for something else.
The 'do' program is good for when the material variation in one direction but if the material is required to vary in Two directions then how to modify 'do' codes