In view of the problem of the redundancy of indicators and the poor accuracy of the evaluation results in the traditional transmission efficiency evaluation system of wind farms, the correlation analysis method is first cited to remove the more relevant indicators, and then the combined weight method is applied. Reasonable empowerment highlights the dominant factors, so as to screen and streamline a reasonable evaluation index system with 10 representative indicators, and construct a fuzzy neural network comprehensive evaluation system. The fuzzy neural network uses the GD method to train parameters with slow convergence rate and easy to fall into the local optimal solution. Therefore, the convergence speed and accuracy of GD, PSO and NPSO algorithms are compared and analyzed, so that the NPSO algorithm with optimal convergence performance is selected to solve the most. The parameters are updated and updated into the neural network. The test data shows that the actual output and the predicted output have a high degree of fit. Finally, the efficiency of transmission grid operation in city L, city T and city B of Gansu Province from 2011 to 2016 is evaluated and analyzed.