The invention belongs to the technical field of production scheduling of flexible manufacturing systems, and particularly relates to a deadlock-free scheduling method for a flexible manufacturing system based on an improved genetic algorithm. The method comprises the following specific steps of: establishing a Petri net model of the flexible manufacturing system, determining genetic parameters, encoding and decoding, generating an initialized population, performing detecting and repairing, calculating processing time and fitness, judging whether a termination rule and genetic operation are metor not, and outputting an optimal individual; adjusting all chromosomes into control feasible chromosomes through a two-step forward looking method, and decoding the control feasible chromosomes intoa deadlock-free scheduling sequence; optimizing and improviing the genetic algorithm in the design process of the scheduling strategy. Meanwhile, in the variation process, the chromosome gene is divided into a path gene and a process gene, variation operation is conducted on the path gene and the process gene at the same time, the variation rates of the path gene and the process gene are the same, and therefore the operation steps are simple, the production efficiency is greatly improved, and the application environment is friendly.