Abstract: In order to quickly predict the risk of head injury of occupants in road traffic accidents and solve the relationship between occupant injury and influencing factors, construct a driver-side injury prediction model based on modified whale optimization algorithm-back propagation (MWOA-BP) neural network. Select the minivan as the research model, the study condition is in frontal collision, and construct the injury prediction model which selects the initial velocity of the vehicle collision, seat belt use, and airbag deployment as inputs, and the driver's head abbreviated injury scale (AIS) as the prediction target. The MWOA-BP prediction model is trained, and the damage prediction effect is compared with the traditional back propagation (BP) neural network model. The results show that the MWOA-BP prediction model has good test effect, and the test accuracy is up to 90%. Combined with real accidents, the damage prediction model is applied to prove that the damage prediction model can be applied to actual accidents.

Key words: Minivan; Frontal collision; BP neural network; MWOA; Injury prediction

摘要： 为了快速预测道路交通事故中乘员的头部损伤风险，求解乘员损伤与影响因素之间的 关系，构建一种基于改进鲸鱼算法优化反向传播（MWOA-BP）神经网络的驾驶员侧损伤预 测模型。选用微型面包车为研究车型，研究工况为正面碰撞，构建以车辆碰撞初速度、安全 带使用情况、安全气囊展开情况为输入，以驾驶员的头部简明伤害等级（AIS）为预测目标的 损伤预测模型。训练 MWOA-BP 预测模型，并与传统的反向传播（BP）神经网络模型进行对 比损伤预测效果。结果表明，MWOA-BP 预测模型有良好的测试效果，其测试准确率达到 90%。 结合真实事故，将损伤预测模型进行应用，证明该损伤预测模型可以应用到实际事故中。

关键词: 微型面包车；正面碰撞；BP 神经网络；MWOA；损伤预测