Study On Approaches To Processing Massive Real-time Data In The Internet Of Manufacturing Elements

The mixture of manufacturing technology and the rapid development of newgeneration of information techniques, including internet, cloud computing, and internetof things, is continuously forming new core technologies of the manufacturinginformatization, promoting the progress and development of manufacturing, which alsooffers great opportunity to the great-leap-forward development of manufacturing in ourcountry. Using the internet of things to promote traditional manufacturing industry, andto form manufacturing content group technology which promotes the manufacturinginformatization, is the integration and deeper infusion of advanced manufacturingtechnology, information technology and intellectual technology, and it also reflects thetrend from mechanization, automation, digitization to intellectualization of themanufacturing industry. It is currently the key area of industrial research to carryforward the integration of technology in the internet of manufacturing elements andmanufacturing procedure, to promote and reform the automation during manufacturingprocedure, and to create high-level production mode, finally reaching the “IntelligentManufacturing” in China’s manufacturing industry. Among those key technologies inthe internet of manufacturing elements, real-time monitoring, transmission anddistribution, processing and mixing toward massive data produced during manufactur-ing procedure, and so on, are essential to the real-time decision and control, and ensuresall the manufacturing procedure be made orderly, immediately, efficiently, and withminimum losses.This paper analyses the related fields such as existing information technology,manufacturing informatization and internet of things and so on, studies real-timesensing data modeling and processing method in manufacturing elements system.Aiming at manufacturing system’s features, key technologies in the internet ofmanufacturing elements and real-time monitoring requirements during producingprocedure, this paper does a thorough research on some involved problem, proposesnew or optimized methods, and applies them to specific production environments. Hereare the main research work and achievements of this paper:①Proposes the definition of real-time data in the internet of manufacturingelements, builds basic theoretical model to real-time data. This paper studies thereal-time data collecting process, proposes the collecting model, and analyses the key technologies in real-time data acquisition and storage. This paper analyses thechallenges during real-time data transition and distribution, proposes continuousmassive real-time data searching module, and comes up with the access model throughoptimized data access method, in order to deal with burst data and non-uniformreal-time data’s transition access problem.③Proposes a new real-time database storage mechanism based on main memory.This paper analyses the development, feature, related concepts and technologies of themain memory database, studies data organization of the memory-based database, querytechniques and its optimization, concurrency control and recovery mechanism, putsforward the CSB+tree-based index method to quickly locate the indexed real-time data.This paper also raises a new virtual memory pool technology based on virtual unit withsmart growth to meet the requirements of space utilization and system robustness ofmain memory database system. Based on the intelligent algorithm, this paper proposedcontinuous massive real-time data query techniques and its optimized algorithm. Tosolve conflicts while real-time transactions is dealing with concurrencies, this paperdesigns a time sorting algorithm, optimizes sorting algorithm by defining priorityalgorithm, solving the problem of priority-reversion in time sorting algorithm. Thispaper puts forward a real-time system data model in order to access real-time data morequickly based on the metadata hierarchical structure. This model divides real-timesensing data under the internet of manufacturing elements into levels, realizing thedata’s efficient organization based on metadata mapping. Metadata hierarchicalstructure realizes real-time data searching, and improves efficiency. Based on thestructure, this paper develops transfer strategy between real-time data and historical data,and can also accesses strategy performance. Various simulation and experimental resultsshow that this method could organize data efficiently and meet the requirements ofmassive real-time data storage for manufacturing elements.③Designs and realizes a practical real-time data access protocol to improvereal-time data system’s performance in the internet of manufacturing elements. Thispaper designs a data storage and delivery model based on double-buffer and promotingdata sending. Double-buffer model is in charge of receiving data from systemalternatively, storing these data into real-time database, ensuring the data integrity underdifferent network environments. To validate model’s performance, this paper proposedthe performance model and the prototype system of real-time data monitoring system,conducts various experiments which proves that this model is very strong in protecting data integrity. The model based on double-buffer and promoting data sending tackles theproblem of data loss during data collecting and sending process, and ensures dataintegrity. In order to optimize massive real-time data distributing efficiency, this paperstudies and proposes a real-time data distribution strategy based on smart multi-agentmodel and priority sorting algorithm, the performance analyze proves that this methodadvances the efficiency of massive real-time data distribution.④Analyses the model and general processing structure of massive real-time datafusion, uses u test method under resource-constrained network condition, andoptimizes fusion model under distributed testing environment. An integrated filtermethod that combined morphological filter and wavelet threshold filtering is proposedin this paper. This method combined the advantages of the two filtering methods to filterimpulse and white noise effectively. The comparative experiments with waveletthreshold filter and morphological filter show that this method not only couldeffectively filter out impulse noise but also white noise.⑤Proposes massive real-time data processing method’s application in theoperation management and safety monitoring system of petrochemical industry. Thisapplication organizes and manages real-time sensing data and leveled data model,transmits and distributes real-time data using distributed propulsion model, conductsreal-time fusion to local perception data in the terminals, and finally develops into theoptimized production and real-time monitoring platform, realizing performanceoptimization and improvements of the monitoring and management platform.In conclusion, this paper analyses and studies the key problems of the existing dataprocessing methods for the massive real-time data in the internet of manufacturingelements, designs and optimizes a serious of model and algorithm. Theoretical analysisand experiment results have shown that these related processing methods are real-timeand correct, they could process massive real-time data efficiently, and providespowerful guarantee for the real-time producing monitoring and management inmanufacturing industry.