Design And Implementation Of Equipment Fault Detection Method Based On Random Forest And LightGBM

With the rapid development of science and technology in the medical field,the medical equipment used is becoming more and more sophisticated.In order to ensure the normal and stable operation of the equipment,it is particularly important to timely and quickly find out equipment faults and accurately locate them.The traditional manual fault detection method has many problems such as long maintenance cycle and high cost.With the development of Internet technology,artificial intelligence and big data,various data of equipment are continuously collected.Adopting a set of effective equipment fault detection methods to analyze and detect equipment is conducive to accelerating the discovery of equipment fault information,timely maintenance and repair of equipment,and improving the use efficiency of equipment.This paper aims to design and implement a fault detection method based on PSO_RF(particle swarm optimization random forest)bidirectional feature selection and LightGBM(Light Gradient Boosting Machine).The main work includes:(1)Designed based on PSO_RF two-way feature selection and LightGBM equipment fault detection method.In this paper,the main idea of random forest for feature selection and the principle of LightGBM classification of equipment fault information were studied.The characteristics of representative medical image equipment data and their characteristic attributes were analyzed in depth.A device fault detection model based on PSO_RF bidirectional feature selection and LightGBM was designed.(2)Implemented PSO_RF bidirectional feature selection and LightGBM device fault detection method.This paper studies the background and practical significance of the equipment fault detection algorithm,applies the designed fault detection algorithm to the construction of the instrument sharing platform project,improves the utilization rate of the equipment,and improves the instrument sharing platform.(3)This paper analyzes the practical application of the designed fault detection algorithm and instrument sharing platform.In terms of feature selection,through comparison with CFS,the results show that the feature selection method used in this paper is superior to CFS;In terms of fault detection,random forest,GBDT and LightGBM are used to establish the model.Through experimental comparison,LightGBM has obvious advantages in accuracy and performance.Through these two aspects,the effectiveness of the PSO_RF bidirectional feature selection and LightGBM fault detection method proposed in this paper is verified.Finally,it was successfully applied to the implementation of the instrument sharing platform project,and achieved good application results.In this paper,the fault detection method is combined with the construction of the instrument sharing platform to realize the monitoring of the equipment state and the rapid and accurate positioning of the fault points,and the problem of fault detection of the instrument and equipment is successfully solved to realize the monitoring of the equipment state and the rapid and accurate positioning of the fault points.