针对船舶汽轮机故障诊断问题,提出一种基于沙猫群算法(SCSO)优化概率神经网络(PNN)的船舶汽轮机故障诊断方法。首先,采用最小-最大(Min-Max)标准化方法对采集的汽轮机运行数据进行标准化处理,得到汽轮机故障特征向量；其次,将标准化的汽轮机故障特征向量分为PNN的训练集和测试集；进而,构建PNN故障诊断模型,并通过沙猫群算法对PNN故障诊断模型实施优化获取最优平滑因子；最后,输入测试集到优化后的PNN故障诊断模型,输出故障诊断结果。仿真实验结果表明,经过沙猫群算法优化的PNN故障诊断模型,训练正确率由92%提升到96%,测试正确率由80%提升到93.3%,说明所提方法能够有效提升故障诊断准确度,而且该方法结构简单、迭代次数较少、运行时间短,具有较高的运行性能和执行效率,能够较好提升船舶汽轮机故障诊断水平。

Aiming at the fault diagnosis problem of marine steam turbine, a fault diagnosis method of marine steam turbine based on sand cat swarm optimization (SCSO) optimization probabilistic neural network (PNN) was proposed. Firstly, the Min-Max method was used to standardize the collected steam turbine operation data, and the steam turbine fault feature vector was obtained. Secondly, the standardized steam turbine fault feature vector was divided into training set and test set for PNN. Then, the fault diagnosis model of probabilistic neural network was constructed, and the optimal smooth factor was obtained by using sand cat swarm optimization. Finally, the test set was input into the optimized probabilistic neural network fault diagnosis model, and the fault diagnosis result was output. The simulation experiment results show that the training accuracy of the probabilistic neural network fault diagnosis model optimized by the sand cat swarm optimization is increased from 92% to 96%, and the test accuracy is increased from 80% to 93%, indicating that the proposed method can effectively improve the fault diagnosis accuracy, and the method has a simple structure, few iterations, short running time and high operating performance and execution efficiency. It can improve the fault diagnosis level of marine steam turbine.