Abstract: Airport noise prediction plays an important role in airport noise controlling, flight planning and airport designing. The airport noise prediction models?are usually built based on aircraft noise distance curve（NPD）, and the NPD curves are little by little revised to the noise propagation model under the specific airport environmental conditions by using a variety of mathematical models. In this way, there are shortcomings of the high cost and great prediction error. This paper presents an airport noise prediction model for particular airport environmental conditions. The proposed model applies BP neural network and history data of the airport noise monitoring to modifying the NPD curves. Experiment results show that in particular specific airport environmental conditions, the accuracy rate of noise prediction is more than 91.5% in the case of ±0.5 dB error. The proposed model has the features of lower cost and high accuracy.

Key words: Back Propagation（BP） neural network, airport noise, prediction model, Noise-Power-Distance（NPD）

摘要： 机场噪声预测对机场噪声控制、航班计划制定和机场规划设计具有十分重要的作用。现有的机场噪声预测模型都是以飞机的噪声距离曲线（NPD曲线）为核心，用相应的数学模型将其修正至与具体机场的特定环境条件相关的噪声传播模型，存在预测成本高和误差大的缺点。针对这种情况，提出一种使用BP神经网络利用机场噪声历史监测数据进行NPD曲线修正计算方法，从而建立适用于特定机场环境条件的机场噪声预测模型。实验表明，在特定机场的特定环境条件下，允许误差为0.5 dB时，该模型预测准确率高达91.5%以上，具有预测成本小、准确度高的特点。

关键词: 反向传播（BP）神经网络, 机场噪声, 预测模型, 噪声距离曲线（NPD）