The robustness testing method of the traditional multi MIMO photovoltaic power generation system, when the robustness of the detection system interference, does not score the robust evaluation index of the photovoltaic power generation system, and the weight stratification of the measured sample weight is not clear, which leads to the large error of the test results, and proposes a new method of robustness testing under the interference of the photovoltaic system. According to the structural characteristics of photovoltaic power generation system, the fault state of the system is divided, and the fault state of the photovoltaic power generation system is classified as the index of evaluating the robustness of the photovoltaic power system, and the test samples are weighted to obtain the weight of the stratified weight based on the comprehensive score, which is based on the test sample combined with the photovoltaic power generation system. The weight vector of the adaptive weighted BP neural network is optimized by the information and information change, and the robustness test of the photovoltaic power generation system is realized through the robust test model of the optical fiber power system based on the BP neural network. The experimental results show that the proposed method can accurately detect the robustness of PV system under interference, and has the advantage of high efficiency.