The machining of wood products has acquired great importance in recent years, identified as being different from that for homogeneous metal removal in the need to avoid creating splintering, delaminating or burning. The various properties of the wood fiber and their orientations have a significant effect on the machining process. The study of the wood machining process is around the chip formation, tool wear, work piece surface quality, crack initiation and propagation into the wood, with taking into account several factors such as humidity, temperature, and vibrations that can affect the wood during the machining process. Rotary peeling veneer is a very specific machining process, where the chip is the final product. The fact that works related to this manufacturing process are rare, our objective is to investigate on the optimal cutting parameters, tool edge geometry, through the use of Teaching-Learning based optimization (TLBO) algorithm  in order to obtain the best quality with the desired thickness of the veneer product. A study is carried out to identify the objective function that best characterize the machining parameters to be optimized. The challenge is to maintain the best possible quality of peeled veneer with the control of the pre-splitting condition and the veneer thickness variation. The developed algorithm, implemented in Matlab, used in this study is described through two flowcharts: main algorithm and the TLBO algorithm. In the main algorithm, the whole resolution procedure is prescribed. The second algorithm is dedicated to the description of all steps of the TLBO technique. Preliminary numerical results obtained from TLBO algorithm are consistent with the experimental ones. The proposed numerical model allows us to predict the characteristic tool angles for different chip thicknesses and friction coefficient. The need to use a pressure bar to produce a quality veneer is numerically proved.