When constraining robust optimization objective function, conventional methods ignore the auxiliary constraint conditions of photovoltaic inverter reactive power, which leads to high cost of light abandonment and power purchase after optimization of distribution network. For this reason, a robust optimization scheduling method connected with photovoltaic inverters is proposed. To minimize the total expected cost, a two-stage robust optimization objective function was calculated. In the first stage, scheduling costs such as photovoltaic energy and fluctuating load were selected, and in the second stage, penalty costs of light abandoning and load loss were selected. The objective function in the first stage was constrained through the restrictions of branch flow and charging and discharging power of distribution network. According to the photovoltaic inverter reactive auxiliary service and inverter access node voltage, etc., the objective function of the second stage is constrained to build a robust optimization model, search the optimal solution of the model, and select the scheduling decision that minimizes the expected cost. The power-heat combined distribution network was selected, 6 photovoltaic inverters were connected, and the comparative experiment was set. The results showed that, for bad scenes with different fluctuation ranges, the design method reduced the cost of light abandonment and power purchase, as well as the amount of light abandonment and load loss compared with the conventional method.