It's critical to select weighing matrixes of Q and R when designing Linear Quadratic Regulator(LQR). This paper put forward a method of LQR control whose weighing matrixes are optimized by elitist preserving genetic algorithm, the best individual which enter directly into the next generation replaces the worst one during the copy process of genetic algorithm, the direct entering guarantees it won't be destroyed by the following processes, including crossover and mutation, so the fitness of the best individual of each generation increase or at least keeps the same as the step of evolution goes by, the weighing matrixes are optimized each generation. The last part of the thesis gives an example for simulation which testifies the validity of the algorithm.