Accurate evaluation of distributed photovoltaic (PV) system operation states is the primary prerequisite to ensure safe and stable system operation and improve the quality of power supply. This paper proposes a distributed PV system operation state evaluation method based on the TCN-LSTM-SE neural network. Firstly, we establish an indicator system that accurately reflects the operation condition, then we use the analytic hierarchy process to obtain the weights and get the operation state, and finally, we input the data into the TCN-LSTM-SE network to train and test the effect of the state evaluation. The experimental results demonstrate that the proposed method can accurately analyze the system state according to the indicators and has superior state evaluation performance in comparison to mainstream neural networks.