Multi-level and multi-objective decision-making problems are prone to information defects and other factors affecting decision-making in the process of decision - making, and the subjective experience and judgment of decision-makers can provide important information for them. In order to evaluate such problems, a fuzzy analytic hierarchy process based on triangular fuzzy number and analytic hierarchy process is proposed. This method changes the common evaluation matrix into a fuzzy evaluation matrix, changes the common number into triangular fuzzy number without checking the consistency of the matrix, and can directly calculate the model of each index weight and the final evaluation result. In order to ensure the effectivity of the result, the evaluation result of the model is used as the input of BP neural network to check whether there is any error in the application of the model constructed by fuzzy analytic hierarchy process, and if the error is not within the specified range, the fuzzy evaluation matrix is reconstructed. In addition, the model is analyzed from both subjective and objective aspects, ensuring the scientificity and rationality of the method and providing a useful model for multi-level and multi-objective decision-making problems.