Ventilator fault prediction method based on IoT data and neural network
10.19745/j.1003-8868.2023174
- VernacularTitle:基于物联网数据和神经网络的呼吸机故障预测方法研究
- Author:
Ming-Kang TANG
1
;
Ke-Sheng WANG
;
Shuang-Shuang LI
;
Pei LIU
;
Xu-Guang PENG
Author Information
1. 电子科技大学机械与电气工程学院,成都 611731
- Keywords:
ventilator;
IoT;
neural network;
fault prediction;
data preprocessing
- From:
Chinese Medical Equipment Journal
2023;44(9):8-13
- CountryChina
- Language:Chinese
-
Abstract:
Objective To propose a neural network-based ventilator fault prediction method with a self-developed underlying data preprocessing method for ventilator IoT data.Methods Firstly the ventilator IoT data were sorted,categorized and cleaned.Secondly nondimensionalization and encoding of all the data were implemented via feature engineering methods based on data distribution.Thirdly data dimensionality reduction was carried out with a self-encoder.Finally,an artificial neural network was used for training with the abnormal data as the training label and predicting the abnormal data faults as the training objective,and the prediction performance of the neural network model was evaluated by calculating the accuracy,precision,recall rate,negative prediction value and specificity.Results The neural network model behaved well in learning with the accuracy being 99.68％,the precision being 99.66％,the recall rate being 99.99％,the negative prediction value being 99.95％and the specificity being 96.52％.Conclusion The proposed ventilator fault prediction model based on underlying data preprocessing and neural network can be used for the prediction of specific faults,and references are provided for IoT data-based medical equipment operation and maintenance management.[Chinese Medical Equipment Journal,2023,44(9):8-13]
