Modeling And Realization Of Networked Intelligent Sensing System Under IEEE1451Mixed Access Mode

This paper starts with the functions of intelligent sensor structure, mainly researches onobject-oriented information flow hierarchical dynamic modeling of networked intelligentsensors, intelligent sensing system plug and play strategy and realization under mixed accessmode, intelligent sensing system cluster server load balancing evaluation and realization, andapplies these theory and methods to networked intelligent weighing sensors system. It hasimportant academic value and practical significance for development of networked intelligentsensing technology, promotion modern manufacturing informatization. The research work issupported by the New Century Excellent Researcher Award Program from Ministry ofEducation of China (NCET-08-0211), High-level Talents Project of Guangdong higher school,Industrial Research Project of Guangdong Science and Technology Department(2008B010400043), and Technology Support Project of Guangzhou (2009Z2-D531).The dissertation researches on modeling and realization of networked intelligent sensingsystem under IEEE1451mixed access mode, summarizes the domestic and foreign researchprogress from IEEE1451standard, intelligent sensing system modeling, intelligent sensingsystem plug and play, and intelligent sensing system load balancing, and determines researchcontents. Primary work includes:⑴Combining UML static description with Petri Net dynamic analysis, researches onobject-oriented information flow hierarchical dynamic (IFHD) modeling method andapplication. Integrating generalized stochastic Petri Net and continuous time markov chain(CTMC), simplifies models, and realizes IFHD model dynamic performance evaluation. Themethod uses UML graphs descript information flow static attributes, establishes conversionrules, construct mapping table, and realizes the conversion of UML graphs to Petri Netmodel. Utilizing model optimization and simplification, static structure and dynamicbehaviors of sensors are described iconically and intuitively, and intelligent sensor structuredeployment, information flow direction, and its interactive relation with users and peripheralequipment are represented exactly. With the aid of CPN Tools, the reachability, boundedness,liveness, and deadlock features of IEEE1451networked intelligent sensor IFHD model are analyzed. Based on IFHD model application, interface performance improvement strategiesare proposed, such as adopting controllable continuous sampling, increasing valid data length,and adjusting data buffer mode. The regular influences of sensing information arrive interval,channel busy probability, and WTIM retransmission number on WTIM networkperformance are discussed. Under multi sensing channels data sampling method, TIM shouldadopt minimum delay priority scheduling (MDPS).⑵Aiming at serial port communication minimum pulse width difficult to be measureddirectly, and several TIM communicating simultaneously cause to data collision, sortingpulse width difference (SPWD) baud rate self-adaptation method and multi levels dynamicbackoff (MLDB) algorithm are put forwarded to realize data collision reasonable withdrawn,and plug and play of baud rate self-adapting wired sensing interface. The SPWD baud rateself-adaptation method acquires a number of communication pulses of serial bus by TIM,uses sorting pulse width difference calculate minimum pulse width, and with the help ofcommunication confirm mechanism, implemented TIM baud rate self-adaptation with NCAP.Based on different NCAP and TIM operation priority partition, MLDB algorithm adoptsdifferent deal with different data collision, selects contention window combine with TEDSparameters information, capable of realize data collision reasonable withdrawn, and averagethroughput increases47.14%, average delay reduces69.18%, and effectively optimizes theplug and play performance of IEEE1451intelligent sensor wired sensing interface.⑶Wireless sensing interface plug and play based on periodic association matchingcommunication (PAMC) and network interface plug and play mechanism are studied. Thesystem structure and flow of IEEE1451intelligent sensing based on ZigBee wireless accessare proposed, and PAMC is implemented through studying three key technologies, includesassociation information frame format define method, association configuration tableparameters selection and update mechanism, ZigBee network parameters real time saving.TEDS data structure simplification and ZigBee route algorithm improvement measures arestudied. With the help of ZigBee node neighbor table and characteristic parameters, wirelesssensing interface plug and play data optimal path transmission is realized. The experimentalresult indicates, after algorithm improvement, average hop decreases42.9%, average delayreduces28.1%, and data transmission efficiency and network real time property are enhanced greatly. Using UPnP device optimum seeking method based on information axiom, amount ofparameters information are distributed to NCAPs which meeting services request. Deviceoptimum seeking can be implemented while several NCAPs meeting services request, neednot adopting weight and normalized dispose, which reduces network interface data trafficoverhead, and improves UPnP device service efficiency.⑷The load balancing implementation method of networked intelligent sensing systembased on probabilistic preferred grey Markov chain prediction (PP-GMCP) is proposed, andnetworked intelligent sensing system load balancing simulation platform is also studied, usedto test and verify load balancing algorithm. This method aims at different NCAP web services,establishes different service request access probability table, determines service queuespriority, and avoids access traffic retention resulting from transferring data depend on TCPport number. NCAP load status is real time monitored using network bandwidth occupationratio, CPU occupation ratio, memory utilization, IO utilization, and process queuesoccupation ratio. NCAP load capacity is effectively forecasted combining gray GM(1,1) withmarkov chain prediction. Probabilistic preferred highest priority (PPHP) strategy is proposedto execute services scheduling, and probabilistic preferred minimum load (PPML) strategy isused to carry out services assignment, which offer a finer load granularity balancing method.⑸Networked intelligent weighing sensing system entirety design, high precisionweighing sensor design and parameters optimization, and system software platformdevelopment are carried out to verify whole application effect. Σ-Δ ADC, ratio measure, chopinput, synchronous suppression, temperature compensation, and self-calibration technologiesare adopted to improve intelligent weighing sensor precision. Effective number of bits(ENOB) is proposed to use as optimization aggregative indicator, and orthogonalexperimental method is employed to implement weighing sensor parameters optimizationconfiguration. Networked intelligent weighing sensor TIM and NCAP concrete structures areproposed from the view of functions, and function modeling are carried out incorporated withIFHD modeling method. Intelligent weighing sensor wired, wireless interface, and networkinterface plug and play test system are set up.The experiments indicate that PP-GMCP algorithm comprehensive considers differentservice request's influence on NCAP resource, distributes service requests to different NCAP evenly, and compared to weighted round robin (WRR) and least connection scheduling(LCS), its average service response delay drops11.1%and25.1%separately, averageresponse speed of data measure service increases35.0%and11.1%respectively, and theNCAP load variation is815bit/s~1300bit/s. The fluctuation range is smallest, and loadbalancing effect is better. After adopting PAMC, ZigBee sensing interface average firstaccess network time Tfen, repeat access network time Tren, fault disconnect network time Tcutdecreases2.33%,77.02%,1.18%respectively, and after employing MLDB, RS485sensinginterface average recognition time Treglower15%, recognition rate Pregincreases0.48%, andthe acquired plug and play performance test results are ideal. Networked intelligent weighingsensor and software platform are detected by Guangdong province quality supervision&inspection offices of measuring instruments and China CEPREI Laboratory, each index hasattain or superior to project requirement, and to certain extent, adequately manifests thecorrectness and validity of networked intelligent sensing system basic theory that researchedby this paper.