Loading...

Table of Content

    25 January 2019, Volume 30 Issue 02
    Big Data Driven Intelligent Manufacturing
    ZHANG Jie, WANG Junliang, LYU Youlong, BAO Jinsong
    2019, 30(02):  127-133,158. 
    Asbtract ( )   PDF (764KB) ( )  
    References | Related Articles | Metrics
    The industrial big data analytics is one of the most critical issues to enable the intelligent manufacturing.According to “the fourth paradigm: data-intensive scientific discovery”, the “connection-prediction-regulation” scientific paradigm of big data-driven intelligent manufacturing was proposed.According to the data processing processes, the method system of fusion processing, correlation analysis, performance prediction and optimization decision was summarized.The big data platform was designed around the edge layer, platform layer and application layer, and the enabling technology of big data-driven intelligent manufacturing was introduced. Then four typical application scenarios were illustrated and reviewed: design, planning and scheduling, quality optimization, and machinery health management.
    On Big Data Driving Manufacturing from “Internet Plus” to “AI Plus”
    YAO Xifan, LEI Yi, GE Dongyuan, YE Jing
    2019, 30(02):  134-142. 
    Asbtract ( )   PDF (744KB) ( )  
    References | Related Articles | Metrics
    “Internet+manufacturing” and “artificial intelligence (AI)+manufacturing” had been a strategy of manufacturing upgrading in China. Their development history and representative manufacturing models was elaborated, their fusion evolution driven by big data from the perspective of their historical origins was reviewed, and the smart manufacturing resulted from such fusion in the form of “internet+AI+manufacturing” was addressed, which is enclosed under the umbrella of the social cyber physical production systems.
    Study on General Data Models for Intelligent Control of Knitting Production
    ZHOU Yaqin, WANG Junliang, BAO Jinsong, ZHANG Jie
    2019, 30(02):  143-148,219. 
    Asbtract ( )   PDF (851KB) ( )  
    References | Related Articles | Metrics
    Based on the intelligent control function analysis of knitted fabric production, from the perspectives of three typical management and control functions, such as workshop planning and scheduling, product production processes and equipment management, an intelligent data management model for knitted fabric production was analyzed and established. The “data-model-key process-control” method was adopted to extract and analyze the data of the production processes of knitted fabrics. Product informations, equipment informations and production process informations were analyzed to describe the multi-dimensional data of high-end knitted fabric production. General data models for key processes such as production planning and scheduling, production execution and equipment management were analyzed and established to provide basic guidance for further intelligent control of weaving and dyeing processes in the key processes of knitting productions.
    Useful RFID Tag Identification and Automatic Association of Manufacturing Informations in Smart Job Shops
    WANG Chuang, JIANG Pingyu, YANG Xiaobao
    2019, 30(02):  149-158. 
    Asbtract ( )   PDF (941KB) ( )  
    References | Related Articles | Metrics
    A useful tag identification approach was proposed based-on electronic product code (EPC) and Auto-ID. Firstly, the manufacturing resources were classified in detail, and based on EPC Global encoding specification, the coding structure of RFID tag data was obtained  using custom fields of types and serial numbers. Secondly, according to the design information and production data of work in process (WIP), the manufacturing logic of WIP was generated. Based on that, the informations of manufacturing resources were automatically associated in group, space and time using database technology. Finally, the RFID tags and manufacturing informations of WIP were discussed to verify the feasibility and universality of the proposed models and methods.
    Big Data Modeling Analysis Method for Intelligent Production Maintenance
    LIU Weijie, JI Weixi, ZHANG Chaoyang,
    2019, 30(02):  159-166. 
    Asbtract ( )   PDF (854KB) ( )  
    References | Related Articles | Metrics
    Starting from the big data application characteristics of current discrete manufacturing processes in workshops, big data collection and processing analyses of the manufacturing processes were realized based on intelligent Agent of the workshops. A big data acquisition model of the manufacturing processes was established to realize the encapsulation of process data set. A clustering-based method was used to perform the outlier detection mark on the process data set to realize the description and quantification of production states of the manufacturing resources. A process level data feature model was established, and an improved similar element analysis method was used to realize the similarity evaluation of multi-level packaging description and manufacturing process of the process data. A underlying intelligent management system of the workshop was developed and applied to the actual productions of the enterprises.
    Modular Design of Intelligent Service Based on Industrial Big Data
    ZHANG Wei, DING Jinfu, JI Yangjian, XIA Wenjun, LAN Hu, ZHANG Jianhui
    2019, 30(02):  167-173,182. 
    Asbtract ( )   PDF (848KB) ( )  
    References | Related Articles | Metrics
    Through the analysis of service operation environment and service body requirements in the integration of manufacturing and serving, a modular intelligent service design method was created. This method is based on industrial environment of big data, adopted modular technical construction, covering intelligent service big data support, intelligent service modular decomposition and intelligent modular optimization. The relationships amony intelligent service applications, techniques and management were analyzed with industrial big data, the modular design strategy of intelligent service were made, the initialized modular decomposition of intelligent service was determined, and then intelligent service modules were optimized based on the structure matrix model. Through the applications of specific instances, the feasibility and the superiority of the modular design strategies of intelligent service under the environment of industrial big data were displayed.
    Method of Transform Customer Demands to Engineering Characteristic Weights in Green Design Based on Data Mining
    ZHANG Lei, ZHONG Yanjiu, YUAN Yuan, LI Jing, QIN Xu, DONG Wanfu
    2019, 30(02):  174-182. 
    Asbtract ( )   PDF (776KB) ( )  
    References | Related Articles | Metrics
    Due to the subjectivity, ambiguity and complexity of the transformation customers demands into engineering characteristics, a method  was proposed based on data mining, which converted customers demand data into engineering characteristic weights.This paper used fuzzy analytic hierarchy process to analyze customers demands and obtained the importance of customers demands.Single customers quality house was established to obtain engineering characteristic weights corresponding with customers demands.The model of MLP was built to mine the relationships between customers demand data and engineering characteristic weights.Finally, the feasibility and effectiveness of the proposed method were verified by the example of conversion from customers demands to engineering characteristics.
    Applications of Big Data in Equipment Health Status Prediction and Spare Parts Replenishment
    ZHANG Chen , LI Jia , WANG Haining, LI Siyue
    2019, 30(02):  183-187. 
    Asbtract ( )   PDF (528KB) ( )  
    References | Related Articles | Metrics
    A equipment health prediction and inventory optimization method was proposed herein. Firstly, a self-encoder was used to extract the features of monitoring signals. Based on that, a deep neural network model was used to predict the time series outcomes, and a equipment health indicator was also constructed. Secondly, a statistical distribution and parameter fitting prediction methods was used to achieve inventory optimization. Finally, the system provided active warnings for productions based on the information about the device health status and the number of spare parts. Example results show that the prediction accuracy of this method is higher than that of LSTM algorithm, which may accurately predict equipment failure. Early warning, and the reliability of the spare parts inventory optimization model is of 90.4%, which may effectively reduce spare parts inventory.
    Prediction of Optimal Rescheduling Mode under Machine Failures within Job Shops
    TANG Qiuhua, CHEN Shijie, ZHAO Meng, ZHANG Liping,
    2019, 30(02):  188-195. 
    Asbtract ( )   PDF (716KB) ( )  
    References | Related Articles | Metrics
    To predict the optimal scheduling mode under different failure scenarios, and achieve intelligent and effective scheduling decisions, an rescheduling mode prediction method was proposed integrating scheduling simulation and improved PNN. Firstly, in face of difficulty in sample collection and incompleteness in failure samples, the large-scale samples with randomness were generated on the ground of simulation methods. Moreover, after collecting the data about the cumulative changes of processing time, the numbers of changed operations and the changes of makespan, the label of each sample were generated. Particularly, all the tagged data were put into the PNN model, and the optimal rescheduling mode was predicted. The experimental results demonstrate that the accuracy rate of the proposed method reaches over 99.54%. On the premise of the specified processing workshop and production tasks, the failure machine and repair duration play a decisive role in the optimal rescheduling mode.
    Planetary Gearbox Fault Diagnosis Based on Multiple Feature Extraction and Information Fusion Combined with Deep Learning
    JIN Qi, WANG Youren, WANG Jun
    2019, 30(02):  196-204. 
    Asbtract ( )   PDF (889KB) ( )  
    References | Related Articles | Metrics
    According to the heavy noises of vibration signals and the difficulty of incipient fault diagnosis for planetary gearboxes using single classifier,a method of planetary gearbox fault diagnosis was proposed based on multiple feature extraction and information fusion combined with deep learning.The multiple excellent stacked denoising autoencoders(SDAEs) were acquired based on multi-objective evolutionary algorithm.Then,multi-response linear regression model was employed to integrate multiple SDAEs for building multi-obiective ensemble stacked denoising autoencoders (MO-ESDAEs),which was used to diagnose faults of planetary gearboxes.The experimental results show that the proposed method may enhance the fault diagnosis accuracy and stability.
    Gear Fault Diagnosis Based on DBNS
    CHEN Baojia, LIU Haotao, XU Chao, CHEN Fafa, XIAO Wenrong, ZHAO Chunhua
    2019, 30(02):  205-211. 
    Asbtract ( )   PDF (844KB) ( )  
    References | Related Articles | Metrics
    Aiming at the problems of gears and other parts in a gear transmission system that were prone to ault or failure, this paper presented a fault diagnosis method for gear transmissions based on deep learning theory. Firstly, the powerful feature self-extraction ability of DBNs was used to extract the features of the vibration signals of the gear transmission systems. Then the fault signals were identified by the complex map representation capability of DBNs. The diagnosis examples show that if the original time-domain signals of gear vibration are not extracted, the correct recognition rate may only reach about 60% when directly using DBNs to diagnose. If a simple Fourier transform is applied to the time domain signals, then DBNs may be used to diagnose the frequency spectrum of the processed vibration signals. The accuracy rate may reach 99.7%, which confirms the simplicity and effectiveness of the fault diagnosis method described herein.
    Intelligent Diagnosis on Health Status of Manufacturing Systems Based on Embedded CPS Method and Vulnerability Assessment
    GAO Guibing, YUE Wenhui, WANG Feng
    2019, 30(02):  212-219. 
    Asbtract ( )   PDF (722KB) ( )  
    References | Related Articles | Metrics
    Recently, the current health management technology was seldom applied in CPS, and few scholars explored the causes of the system failure from the perspective of vulnerability. According to the characteristics of forecast on health status in manufacturing systems, a forecasting health model embed on the key equipment was raised, which was aiming to achieve on-line forecast health and identification of critical factors in the whole service life cycles of the equipment. The research concentrated on the system framework of equipment health status online judgement, the abnormal judgment of manufacturing system performance based on data driven and the identification of equipment abnormal factors in sub-health states. Based on the performance changes during the operations of the equipment, the vulnerability status of equipment might be determined by the proposed method, and the changes of performance parameters in working processes were monitored by the equipment abnormality determination method, which might be adopted for finding the key factors that caused equipment abnormality. The results in a flexible manufacturing system show that those methods are effective ways to improve the efficiency of judgement of health status and to realize intelligent identification of critical factors.
    On-line Monitoring of Tool Wear Conditions in Machining Processes Based on Machine Tool Data
    LU Zhiyuan, MA Pengfei, XIAO Jianglin, WANG Meiqing, TANG Xiaoqing
    2019, 30(02):  220-225. 
    Asbtract ( )   PDF (685KB) ( )  
    References | Related Articles | Metrics
    To realize the on-line monitoring of tool wear conditions, and improve the feasibility of monitoring system in machining processes, an on-lime cutting tool condition monitoring method was proposed based on machine tool data in machining processes. OPC UA was used for NC machine tool data acquisition and storing, and the internal machine process informations related to tool wear conditions were collected. Based on the process informations and corresponding wear informations, convolutional neural network was used to establish a recognition model of tool wear conditions. The performance of proposed method was verified in machining cases, and compared with other tool wear condition monitoring methods. This method is more suitable for tool wear condition monitoring in practical machining processes.
    Feature Extraction and Defect Identification of Eddy Current Testing Signals on Narrow Lap Weld
    GE Liang, MIAO Rui, GE Qiuyuan, WU Yizhou
    2019, 30(02):  225-229. 
    Asbtract ( )   PDF (532KB) ( )  
    References | Related Articles | Metrics
    An effective eddy current testing method was presented to identify the defects of narrow lap joints. Firstly, the characteristic parameters of eddy current signals were extracted by EMD. Then, based on PCA, the characteristic parameters were reduced, the redundant informations were removed, and the main element characteristics of the welds were obtained. In the end, the main element features were used as the inputs of SVM to construct the multi-classifier, and the eddy current measurement signals of the narrow lap were identified. The results show that this method has high accuracy and low complexity. It may effectively identify different defects of weld seams, and has good engineering application values.
    Data Driven Wafer Pattern Defect Pattern Recognition Method
    YANG Zhenliang, WANG Junliang, ZHANG Jie, JIANG Xiaokang
    2019, 30(02):  230-236. 
    Asbtract ( )   PDF (676KB) ( )  
    References | Related Articles | Metrics
    Aiming at the characteristics of wafer map data angle and dimension diversity and quantity imbalance during wafer production processes, a wafer pattern defect recognition method was proposed based on generative adversarial networks. The two-stage wafer defect data pre-processing method was proposed to obtain standard wafer defect data, where Radon transform was designed to solve the multi-angle characteristics of the wafer map, and a resampling mechanism was used to realize the scaling of various data dimensions. The proposed wafer defect classification method used a generation mechanism to balance the number of samples of each defect type based on a generative adversarial networks, which could improve the defect pattern recognition accuracy. The experimental results show that this method may greatly improve the accuracy of small class samples, and the overall recognition rate is much better than the support vector machine and Adaboost algorithm.
    Latent Structure Modeling and Predictive Quality Control Based on Multi-source Data Streams in the Auto Body Assembly Processes
    LIU Yinhua, SUN Rui, WU Huan
    2019, 30(02):  237-243. 
    Asbtract ( )   PDF (621KB) ( )  
    References | Related Articles | Metrics
    This paper present a systematic review for the studies on assembly  accuracy insurance. Then, the shortcomings of the concurrent data-driven dimension assembly quality control methods were analyzed, and the latent structure modeling and predictive quality control method was proposed as to the characteristics of the measurement data of the process and product data sets. By extracting the principle components from the data, the partial least squares regression model for assembly deviation propagation were constructed. Furthermore, the assembly quality prediction and control under the conditions of current processes and product inspection strategies might be realized. A side rail assembly case was used to illustrate the proposed procedures. The partial least squares modeling, the quality qualification rate prediction of key features and optimizations of variation sources' variance were used, and the 6?? values of assembly key product features are decreased by about 25%.
    Data Analysis of Tyre Quality Based on Improved FP-Growth Algorithm
    LI Minbo, DING Duo, YI Yong
    2019, 30(02):  244-251. 
    Asbtract ( )   PDF (675KB) ( )  
    References | Related Articles | Metrics
    According to the problem analyses of abnormal quality in tyre manufacturing processes, tyre quality data acquisition, effective integration and data analysis processes were discussed. The structured data sets associated with production data and product inspection data were constructed based on Hive data warehouse. For the existing frequent pattern-growth (FP-Growth) algorithm, the performance of FP-tree was low, an improved FP-growth algorithm was proposed. A new tail attribute was added to the existing header table of frequent item and accelerate the construction of FP-tree. The experiments show that the improved FP-growth algorithm may effectively improve the correlation analysis efficiency of tyre quality abnormal data. The improved FP-growth algorithm is able to identify the factors that affect the quality of tire productions, and it is also suitable for large data mining.