We consider an uplink time division duplex cell-free massive multiple input multiple output (MIMO) system in which many user equipments (UEs) are simultaneously served by many access points (APs) via simple matched filtering processing. The propagation channel is modeled via the Ricean distribution, which includes a dominant line of sight component on top of diffuse scattering. The Ricean K factor of each link varies with the UE location (relative to the locations of the AIs). The system performance in terms of the spectral efficiency is investigated taking into account imperfect channel knowledge. Power and AP weighting control is exploited to maximize the lowest spectral efficiency across all UEs. This optimization problem can be efficiently solved via the bisection method by solving a sequence of linear feasibility problems together with the generalized eigenvalue problem. We show that by optimally selecting the power control and AP weighting coefficients, the per UE throughput increases significantly. Furthermore, we propose an AP selection scheme to reduce the backhaul requirements in a cell free massive MIMO system, with slight compromise in performance.