基于概率神经网络的液氨汽车罐车复合故障诊断

故障诊断在保证危险化学品汽车罐车运输安全方面具有重要意义。从国内交通运输安全的实际要求出发,依据液氨汽车罐车的结构特点及国家法律法规的要求,比较全面、系统地分析了液氨汽车罐车故障特征的相关参数,并将其作为概率神经网络的输入结点。根据实际可能发生的故障分类模式,考虑到故障诊断的容错能力和自适应能力,提出了基于概率神经网络的复合故障诊断模型。利用指标参数作为网络训练样本,对未知故障模式进行诊断,并以广西地区压力容器检验所液氨检测数据为例进行说明。理论分析和实例计算表明,该模型物理概念清晰,计算结果合理,精度较高,在危险化学品汽车罐车故障诊断中有很好的适用性。该项工作可为我国危险化学品汽车罐车故障智能诊断的深入开展提供参考依据。

Compound fault diagnosis of liquid ammonia tank car based on probabilistic neural network

Fault diagnosis is important to ensure the safety of hazardous chemical tank car＇ transportation.From the requirements of transportation security,based on the structural features of liquid ammonia tank car and the requirements of laws in domestics,the fault signature parameters of liquid ammonia tank car were analyzed all sidedly.The parameters were established as input nodes of probabilistic neural network.According to the fault classification in view of the facts,the fault tolerant and adaptive abilities of fault diagnosis were considered to structure the compound fault diagnosis model based on probabilistic neural network.Using index parameters as the network training samples,the unknown failure modes were diagnosed.The proposed method was applied to prove the liquid ammonia sense data which provided by pressure vessel inspection of Guangxi area.The result indicated that the model has the advantages of the clear physical concepts and high precision,so the used method is feasible and rational.The work specified in this paper can be as reference to the compound fault diagnosis of hazardous chemical tank car.