福建技术师范学院学报

智能交通系统考虑人—车—路—环境协同关系，在复杂路段考虑人流、车流、路况、环境等因素影响，具有随机性、不确定性和模糊性等特点，难以建立精确的系统模型。BP神经网络模型预测具有处理多变量数据、车辆动态性复杂结构和参数映射函数条件高等优点，特别适合在盲区路段调整网络权重比例和阈值大小来满足约束条件，从而提高控制精度和车辆稳定性。将具有模糊理论的预测方法与BP神经网络引入智能交通系统动态建模。模拟盲区路况条件下，车辆在实验环境变化时，建立模糊数据预测与决策方法，提升行车自主适配能力和数据预测能力。Matlab-Simulink仿真表明，该决策方案具有较高的模型连续性，自学习数据推理能力，大幅提升了智能交通的安全性和驾驶越过盲区精准度。

Intelligent transportation systems(ITS) consider the coordinated relationship among people, vehicles, roads, and environment. However, in complex road sections, factors such as pedestrian and vehicle flow, road conditions, and the environment make it difficult to establish an accurate system model, exhibiting characteristics of randomness, uncertainty, and fuzziness. On the one hand, the BP neural network prediction model offers advantages in handling multivariate data, capturing the dynamic complexity of vehicle structures, and meeting the high demands of parameter mapping functions. On the other hand, especially for blind spot sections, adjusting the input-output weight ratios and thresholds to satisfy constraint conditions can effectively improve control accuracy and vehicle stability. Therefore, this study introduced a prediction method based on fuzzy theory along with a BP neural network into the dynamic modeling of intelligent transportation systems. Under simulated blind spot conditions, a fuzzy data prediction and decision-making method was established. This method leveraged human control experience and reasoning to adapt to experimental environments and data variations, thereby improving vehicular autonomous adaptability and data prediction capability. Matalab-simulink simulation experiments verified that the proposed decision scheme maintains high model continuity and strong self-learning reasoning ability, which significantly improves the safety of intelligent transportation and the accuracy of vehicle passing through blind spots.