One fundamental work of grid-connected photovoltaic (PV) system is to estimate its power generation. This paper introduces weighted Markov Chain estimation theory to build a grey-weighted Markov Chain estimation model after forecasting the overall generation trend of PV by applying the grey model. This model not only takes the advantage of GM(1,1) model when dealing with exponential series into account, but also considers the feature of fluctuation in power generation, which is described by state transfer probability matrix of relative residuals. After applying the model to forecast the generation of one PV power station in Hefei, the results indicate that the combination of grey model and weighted Markov Chain improves the precision when coping with generation data with greater fluctuation, verifying the feasibility and effectiveness of this model. This work is supported by National Natural Science Foundation of China (No. 51567018).