为提高柴油机燃油系统故障诊断准确率，解决故障数据量不足的问题，提出了一种结合关联规则和一维卷积神经网络（1DCNN）的故障诊断方法。通过建立SC7H型柴油机仿真模型，在6个工况下模拟燃油系统四类典型故障：喷油不足、喷油过量、喷油正时提前和喷油正时滞后，并生成相应故障数据。使用Apriori算法将故障数据转化为关联规则，实现1DCNN训练数据量扩增，利用粒子群优化算法优化最小支持度和最小置信度。结果表明，该方法的故障诊断准确率达91.67%，优于单一的关联规则分类模型和1DCNN模型。
To enhance the accuracy of fault diagnosis in diesel
engine fuel systems and address limited fault data availability, a diagnostic
method integrating association rules and one-dimensional convolutional
neural networks (1DCNN) was proposed. The simulation model of SC7H diesel
engine was established to simulate the typical four kinds of fuel system faults
under six operating conditions, including insufficient injection, excessive
injection, advanced injection timing, and delayed injection timing. The Apriori
algorithm was employed to transform fault data into association rules, enabling
effective data augmentation for 1DCNN training. Particle swarm optimization
algorithm was employed to optimize the minimum support and confidence
thresholds. The experimental results demonstrate that the proposed method
achieves a fault diagnosis accuracy of 91.67%, outperforming standalone
association rule classification models and 1DCNN models.
