Considering the nonlinearity, uncertainty, and rigid/elastic coupling, a state/parameter joint estimation method is essential for control system design or fault diagnosis of flexible hypersonic vehicles. With the goal of improving state/parameter estimation accuracy, this paper proposes a sensor placement strategy and a moving horizon estimation algorithm with a QR-decomposition-based arrival cost update strategy (MHE-QR). To enhance observability, a novel double sensor placement strategy, in which the sensor positions are obtained via solving a constrained nonlinear optimization problem, is developed. The MHE-QR algorithm transforms the arrival cost update problem into a least square problem and solves it utilizing QR decomposition. With this QR-decomposition-based arrival update strategy, the state/parameter estimation problem is solved as a nonlinear programming problem in the framework of moving horizon estimation. Finally, the performance of sensor placement strategy and MHE-QR is evaluated by Monte Carlo simulations in 10 different scenarios. Simulation results demonstrate that the sensor placement strategy and MHE-QR algorithm can effectively improve the estimation accuracy, convergence speed and computation rate. Additionally, the CPU time of MHE-QR validates its real-time applicability.