The invention relates to a circular form deviation evaluation method based on a grouping teaching and learning algorithm. The method includes the steps that firstly, a solving mathematical model for circular form deviation evaluation is established based on the minimum area principle; then a target function of a question is established by acquiring measured data of measured circular form elements and combining with the established mathematical model; finally the target function is solved through the grouping teaching and learning algorithm, the main steps include the parameter initialization process of the algorithm, the teaching stage of student marks, mutual learning between students and the like, aiming at the problems that the standard teaching and learning algorithm process is not high in solving precision, low in convergence speed and the like, a population grouping strategy is adopted, the global optimization capacity of the algorithm is enhanced, the calculation precision is fully improved accordingly, and finally solving is carried out. Fewer parameters are needed by the method, the algorithm stability is good, the measured data can be fully applied, the algorithm application level is novel, a test function is more accurate compared with other common algorithms, the iteration speed is higher, the solving process completely conforms to the minimum area principle in the international standard, and therefore the precision of the calculation result is higher.