呼吸机管路积水故障自动监测算法研究

Research on Automatic Monitoring Algorithm for Water Accumulation in Ventilator Pipeline

摘要: 呼吸机管路中的积水会导致管路内径变窄，增加气道阻力和患者的吸气做功，且当呼吸机管路积水严重时，甚至会造成患者发生窒息现象，而监测呼吸机管路积水故障能够有效提高呼吸机使用的安全性。为此，提出了呼吸机管路积水故障自动监测算法研究。利用小波变换去噪方法去除呼吸机管路信号中的噪声，以VC维理论和SRM准则为基础，建立支持向量机回归算法，通过拉格朗日函数将其优化为多输出支持向量机回归算法，采用改进的人工蜂群算法，优化多输出支持向量机相关参数，生成最优多输出支持向量机并加以训练。将去噪后信号输入至训练后的多输出支持向量机中，输出故障监测结果，实现呼吸机管路积水故障自动监测。实验结果表明，所提方法的故障监测准确率始终高于90%，证明该方法的故障监测精度较高，具有良好的故障监测效果。

Abstract: The accumulated water in the ventilator pipeline will cause the inner diameter of the pipeline to narrow, increase the airway resistance and the patient’s work of inspiration, and when the ventilator pipeline has serious accumulated water, it may even cause the patient to suffocate. Monitoring the accumulated water fault in the ventilator pipeline can effectively improve the safety of the use of the ventilator. Therefore, the research on automatic monitoring algorithm of ventilator pipeline water logging fault is proposed. The wavelet transform denoising method is used to remove the noise in the ventilator pipeline signal. Based on VC dimension theory and SRM criterion, the support vector machine regression algorithm is established. It is optimized into a multi output support vector machine regression algorithm through Lagrange function. The improved artificial bee colony algorithm is used to optimize the related parameters of the multi output support vector machine, generate the optimal multi output support vector machine and train it. The de-noising signal is input to the trained multi output support vector machine, and the fault monitoring results are output to realize automatic monitoring of ventilator pipeline water logging fault. The experimental results show that the fault monitoring accuracy of the proposed method is always higher than 90%, which proves that the fault monitoring accuracy of the method is high and has a good fault monitoring effect.