With the increasing costs of warehousing, manpower, and aircraft downtime, precise control of aviation material inventory is becoming increasingly important. Accurately predicting inventory can achieve high-quality on-time delivery and reduce the possibility of order defaults, can reduce storage expenses and effectively arrange maintenance tasks, can help airlines make scientific decisions based on sales and inventory, implement dynamic real-time pricing strategies, and reduce decision-making costs. This article first selects 1 080 inventory data of A aviation materials and normalizes them after desensitization, and provides the processing results. Subsequently, the LSTM time series prediction solution method was introduced, taking advantage of the strong learning ability, gradient explosion prevention, and controllable information memory function of LSTM, analyzing and designing parameters such as error function, activation function, optimizer, and batch processing. Subsequently, the optimized LSTM algorithm will be used in the TF framework for long-term inventory prediction, and the reasonable inventory quantity of A aviation materials for the next time node will be obtained. The predicted values will be used for aviation material procurement and maintenance planning.