一种基于支持向量机回归的制氧系统寿命预估方法

A Lifespan Pre-Estimated Method for Oxygen Production Systems Based on Support Vector Machine Regression

随着我国老龄化程度的加深及后疫情时代大背景下国内医疗系统市场规模的不断扩大,制氧系统行业的发展进入快速扩容阶段。同时,制氧系统突发故障对吸氧人群生命与心理造成不可逆损伤的问题接踵而至。制氧行业迫切需要一种针对制氧系统寿命的预估方法,以解决制氧系统突发故障影响吸氧人群生命与财产安全的问题。该研究提出一种基于支持向量机回归(SVR)的制氧系统寿命预估方法。首先,基于SVR原理建立一种寿命预估模型,通过对某制氧系统3500h的氧浓度监测数据进行SVR训练,得到制氧系统的寿命预估模型。然后,规律选取300组氧浓度数据分别进行训练集预测和预测集预测,结果显示,该预测模型的准确性较高,且模型对预测集样本的预测结果与氧浓度监测的真实值基本保持一致。最后,对该制氧系统的吸附塔进行拆机验证,发现分子筛确有失效现象,经实际测量失效程度为6%,表明该模型可应用于制氧系统的寿命预估,并取得了良好的结果。因此,基于SVR的制氧系统寿命预估方法可以准确、有效地预估制氧系统的使用寿命,避免其突发故障,同时也为后续制氧系统的寿命预估方法提供了思路。

With the deepening of China's aging population and the continuous expansion of the domestic medical system market in the post-epidemic era,the development of the oxygen production system industry has entered a stage of rapid expansion.At the same time,with the problem of irreversible damage to the life and psychology of oxygen users caused by sudden failures in the oxygen production system,the oxygen industry urgently needs a method for estimating the lifespan of oxygen production systems to solve the problem of sudden failures affecting the safety of life and property of oxygen users.In this study,a lifespan pre-estimated method for oxygen production systems based on support vector regression(SVR)was proposed.Firstly,an life prediction model was established based on the principle of SVR.By training the oxygen concentration monitoring data of a certain oxygen production system for 3500 hours using SVR,a lifespan pre-estimated model for the oxygen production system was obtained.Then,300 sets of oxygen concentration data were lawly selected for training set prediction and prediction set prediction,respectively.The results showed that the prediction model has high accuracy,and the prediction results of the model on the prediction set samples were basically consistent with the true values of oxygen concentration monitoring.Finally,the adsorption tower of the oxygen production system was disassembled and verified,and it was found that the molecular sieve did indeed exhibit lose efficacy.The actual measurement of lose efficacy degree was 6%,indicating that the model can be applied to the lifespan prediction of the oxygen production system and achieve good results.Therefore,the life prediction method for oxygen production systems based on SVR can accurately and effectively estimate the service life of oxygen production systems,avoid sudden failures,and provide ideas for the subsequent lifespan prediction methods of oxygen production systems.