制冷机组蒸发器系统的Elman神经网络预测模型

Elman neural network prediction model for evaporator system of refrigeration unit

为解决制冷机组蒸发器系统故障预测要求高精度和高灵敏的问题,提出一种基于一步预测模式设置Adam算法改进的Elman神经网络故障预测模型,采用Matlab对其进行仿真。结果表明,所建立的预测模型,预测正确率相比于传统BP神经网络预测模型提高了30.43%,预测误报率降低了30.43%,可信度提高了18.33%,预测能力提高了33.33%,具有一定实用性。

This paper proposes an improved Elman neural network fault prediction model based on one-step prediction mode and Adam algorithm—a model aimed at a higher accuracy and sensitivity required for the fault prediction of evaporator system of refrigeration unit.The Matlab simulation shows that the proposed model promises certain practicability,thanks to a 30.43%higher prediction accuracy,a 30.43%reduced prediction false alarm rate,a 18.33%increased reliability,and a 33.33%higher prediction ability,than the traditional BP neural network prediction model.