【摘要】 为了减少大型车辆转弯碰撞事故对车辆和行人的伤害，提出一种基于模糊综合评价法与GA-LSTM模型的碰撞预警方法。先确定模糊综合评价模型中影响碰撞危险度的因素，然后结合车辆碰撞理论确定各个因素的隶属度函数及权重的分配，通过模糊算子计算碰撞危险度(CRI)的大小，最后根据遗传算法-长短期记忆网络(GA-LSTM)模型预测未来某段时间CRI的大小对车辆侧方障碍物进行预警。实验结果表明：该方法准确率高，能有效对危险状况进行预警。更多还原

【Abstract】 In order to reduce the damage of the vehicles and pedestrians in collision accidents because of large vehicles turning, an early warning method, based on Fuzzy Comprehensive Evaluation Method and GA-LSTM model, is proposed. First, we should find out the factors affecting collision risk in the fuzzy comprehensive evaluation model, and then it’s time to determine the distribution of the membership function and weight of each factor in combination with the vehicle collision theory, and calculate the collision risk(CRI) through the fuzzy operator. Finally, it is possible that we make early warning of CRI to the vehicle side obstacle according to the genetic algorithm-long short-term memory network(GA-LSTM) model. The experimental results show that this method is of high accuracy and can establish the early warning of the dangerous condition effectively.更多还原