Faults Detection And Factors Analysis For Equipment Based On Time Series Data

In the industrial production process,equipment shutdown caused by failure will lead to a large amount of economic losses.With the in-depth advancement of Industry 4.0 and the continuous development of Internet of Things technology,the detection and maintenance data of important industrial equipments are continuously collected and stored.The data contains a large amount of fault information and provids data prepare for fault detection and factor analysis of faults.The using of data mining technology to detect equipment failures is conducive to early detection of faults and avoiding more serious losses caused by fault expansion.Based on the equipment fault detection,further analysis of the fault factors and finding the causes of the equipment fault can guide the using and maintenance of industrial equipment scientifically.Factory can adjust the maintenance plan according to the fault factors to reduce accidents.This paper provides data-driven fault detection and factor analysis models.The significance of this research is to provide effective scientific theoretical guarantee for equipment maintenance and accurate fault detection.Aiming at the application characteristics of industrial equipment time series data,this paper proposes a complete process of time series data feature engineering for the need of fault detection and factor analysis.The feature engineering includes solution strategies of time series data missing,time series data aggregation and feature extraction,time series data standardization and sample imbalance problems.At the same time,in order to meet the requirements of multi-classification in industrial fault detection,an integrated multi-classification model based on DAG is proposed.The model can improve the fault detection efficiency by reducing the number of basic classifiers.In this paper,the random forest model is used to detect the fault.The correlation analysis is used for single fault factor analysis.In order to analyze the relationship between fault factors,this paper proposes a weighted CBA(WCBA)method to analysis correlation between fault factors.The WCBA model is used to detect equipment faults,and the pre-terms of classification rules are used to analyze the fault causes and their connection relation.Finally,the above models are verified by experiments.Experiments show that the models can not only have better fault detection accuracy,but also can analyze single factor and factors correlation between faults.The analyze providing theoretical support for fault repair and daily fault protection.In this paper,the time series data mining technology is applied to the fault detection and factor analysis of industrial equipment.The fault detection model is constructed from the equipment monitoring time series data and the fault factor analysis is carried out.The purpose is to improve the fault detection accuracy.The purpose of writing is to improve the current state of failure factor analysis that relies heavily on expertise.The research in this paper still has some shortcomings.The model proposed in this paper is based on lightweight data,and does not consider the factors of high concurrent big data;and the physical meaning behind the fault factors is slightly less explanatory.Therefore,the future work is to study fault detection and factor analysis under the premise of high concurrent big data sets,and further improve the efficiency and interpretability of the model.