中文摘要：

文中提出一种新的齿轮箱噪声预测方法，它与传统的仅用声压计测量方法不同：该方法采用多个加速度和噪声传感器对齿轮箱进行监测并获取振动和噪声信号；先通过成熟的“以振代噪”技术，计算振动信号的噪声评估向量；然后根据组合预测方法，采用应用线性神经网络算法预测出齿轮箱的声压级，并且通过实验验证其准确性有效性。

英文摘要：

A new method for noise prediction of gearbox is put forward. Different from the traditional measurement method by using sound level meter, several vibration and acoustics pressure sensors are used together during the monitoring process. The vibration and noise signals in different monitoring positions can be acquired simultaneously. Using the developed technology ＂replacing noise by vibration＂, a noise estimation vector for vibration signal can be computed. Then, using the hybrid predication method and linear neural network algorithm, the noise level of the gearbox is .predicted. Through the experiment, the accuracy and the effectiveness of this method are testified.