基于多源信号融合的往复式压缩机气阀健康评估

压缩机气阀测点处信号是多个激励源的综合响应,因此,仅依靠单一信号难以准确评估压缩机的健康状态。针对这一问题,提出了一种基于多源信号融合的往复式压缩机气阀健康状态评估方法。首先,以某型四级高压往复式压缩机为研究对象,通过故障模拟实验,获取了其进排气阀不同健康状态的热工参数和声发射信号;然后,提取了不同信号源的时域特征和频域特征及热工参数,以气阀健康状态下各特征参数的均值作为健康基准,计算了气阀不同健康状态样本与健康基准的马氏距离(MD),基于多源信号融合理论,将不同信号计算所得的马氏距离相融合,进行了样本重构;最后,基于决策树构建了压缩机气阀健康状态评估模型,评估了气阀的健康状态。研究结果表明:各单一信号源的评估准确率分别为70.6%、87.2%和85.2%,而基于多源信号融合重构后,样本不同健康状态的区分度显著提高。基于决策树构建的气阀健康状态评估模型可以有效识别气阀的健康状态,识别的准确率可达100%,具有良好的健康状态评估效果。

Data mining-based study on sub-mentally healthy state among residents in eight provinces and cities in China

OBJECTIVE: To apply data mining methods to research on the state of sub-mental health among residents in eight provinces and cities in China and to mine latent knowledge about many conditions through data mining and analysis of data on 3970 sub-mentally healthy individuals selected from 13385 relevant question naires.METHODS: The strategic tree algorithm was used to identify the main mani festations of the state of sub-mental health. The backpropogation artificial neural network was used to analyze the main mani festations of sub-healthy mental states of three different degrees. A sub-mental health evaluation model was then established to achieve predictive evaluationresults.RESULTS: Using classifications from the Scale of Chinese Sub-healthy State, the main manifestations of sub-mental health selected using the strate gictree were F1101(Do you lack peace of mind?),F1102(Are you easily nervous when something comes up?), and F1002(Do you often sigh?). The relative intensity of manifestations of sub-mental health was highest for F1101, followed by F1102,and then F1002. Through study of the neural network, better differentiation could be made between moderate and severe and between mild and severe states of sub-mental health. The differentiation between mild and moderate sub-mental health states was less apparent. Additionally, the sub-mental health state evaluation model, which could be used to predict states of sub-mental health of different individuals, was established using F1101, F1102, F1002, and the mental self-assessment totals core.CONCLUSION: The main manifestations of the state of sub-mental health can be discovered using data mining methods to research and analyze the latent laws and knowledge hidden in research evidence on the state of sub-mental health. The state of sub-mental health of different individuals can be rapidly predicted using the model established here.This can provide a basis for assessment and intervention for sub-mental health. It can also replace the relatively outdated approaches to research on sub-health in the technical era of information and digitization by combining the study of states of sub-mental health with information techniques and by further quantifying the relevant information.