The invention relates to an intelligent reflector-assisted multi-cell robust joint transmission beam optimization method, which comprises the following steps of: firstly, considering that an error exists in intelligent reflector-user link channel estimation, constructing an optimization model of joint multi-base station transmission beam forming and intelligent reflector shift, and considering the fairness of cell users, ensuring the minimum achievable rate of cell edge users according to the invention; secondly, due to variable height coupling, decomposing the problem into two problems of base station beam forming and intelligent reflecting surface shifting, converting the base station beam forming problem into a convex problem by adopting an S-Procedure theorem and convex semi-definite relaxation, and providing an iterative algorithm based on dichotomy to obtain an optimal solution of the intelligent reflecting surface shifting; finally, alternately solving the two problems until convergence. According to the method provided by the invention, the total transmitting power of the base station can be minimized under the condition that all possible channel errors are realized, and meanwhile, the minimum achievable rate requirement of the worst user is met.