Research On Key Techniques Of Reliability Design And Analysis Of Space Information Processing Systems

As main components,space information processing systems(SIPS) in spacecrafts are with responsible for guise controlling,task management,data processing and other space tasks.In a narrow sense,SIPS are the computer systems that layout flight missions and store data.In a broad sense,SIPS generally refer to the computer systems that are concerned with information processing in spacecrafts.Since SIPS work in atrocious space environment and lack maintenance,it is incontestable that SIPS should own exceedingly high reliability.Therefore,more are required of the techniques of reliability design and analysis.The reliability design of SIPS includes two aspects,namely hardware design and software design.With a view to software reliability design,besides the methods in application level,fault-tolerant scheduling is also required in operating system level so as to ensure that systems should go on working if instant failures occur in applications. The thesis will research the fault-tolerant real-time scheduling algorithms from the aspect of operating systems,the aim of which is to improve the software reliability of SIPS.Reliability analysis is the process that builds reliability models for the reliability and failure properties of systems.It includes qualitative analysis and quantitative analysis.Reliability analysis provides a theoretical way to certify whether scheduled tasks can be fulfilled,which aims to avoid the mission lost caused by earlier design mistakes.With verity of reliability designs adopted,the architecture of SIPS shows itself many new characteristics,such as complex structures,phased missions,diversified parameters and so on.The thesis will consider those characteristics and research the reliability analysis models,which are adapted to SIPS,from the aspects including the failure modes of dynamic systems,the reliability analysis of phased-mission systems (PMS) and high reliable non-Markovian systems.The main contents and conclusions of the thesis are outlined as follows:1.Research on the techniques of software fault-tolerance in SIPSBy improving the software fault-tolerant model based on real-time operating systems,software fault-tolerance techniques are researched so as to improve the ability of SIPS to tolerate software faults.The thesis mainly focuses on the software fault-tolerant model that provides dual-version programs in uniprocessor.The processes of backward and forward scheduling in software fault-tolerant model are investigated. And the influences of backward and forward scheduling algorithms on fault-tolerant real-time scheduling algorithms are also researched.As a result,the cooperation relationships between backward scheduling and forward scheduling are revealed.In order to predict more accurately whether the primary of the fault-tolerant scheduling algorithm in software fault-tolerant model is executable,a new algorithm named prediction-table based algorithm(PTBA) is put forward,in which the executing situation of tasks between the current time and the notification time of alternate is fully considered.The executing situation of primary is accurately predicted according to the sequence of notification time in that time interval.If primaries do not fail,the task scheduling will be carried out based on the prediction-table.Simulation result shows that PTBA can acquire more execution time of primaries and decrease CPU time cost.2.Research on the failure modes of dynamic systems based on cut sequence setsUsing dynamic fault trees(DFT) as the tool to model SIPS,the thesis presents the concepts of cut sequence and cut sequence sets(CSS) to describe the failure modes of dynamic systems.CSS is the aggregate of all cut sequences and can be expressed by sequence failure expressions(SFE).CSS of DFT can be generated by converting dynamic gates to corresponding cut sequences.The thesis also investigates the quantification of CSS,and computes the probability of each cut sequence using multi-integration.Then,the system failure probability is the probability summation of all cut sequences in CSS.Since the CSS model takes the failure sequences into account and owns high computational complexity,some simplification methods,such as modular approach and probability-summation simplification,are provided so as to decrease the consumptions of time and space.The generation and quantification of CSS are new reliability models that can be used to research the failure modes of dynamic systems;and they solve the problems about locating the failure modes and computing the unreliability of dynamic systems.3.Research on the reliability analysis of dynamic PMS based on Bayesian networksThe thesis presents a Bayesian networks(BN) based PMS reliability model.Firstly, each phase of PMS is represented by a BN framework,named phase-BN.Then,in order to figure the dependences across the phases,all the phase-BN are combined by(1) connecting the root nodes that represent the same component but belong to different phases,and(2) connecting the leaf nodes of phase-BN with a new node representing the whole PMS mission.The new constructed BN is called PMS-BN.The reliability analysis of PMS is performed by a discrete-time BN model acting on PMS-BN.The PMS-BN based method provides a new efficient way to analyze the reliability of PMS, especially for those with dynamic phases.Moreover,it is also applicable to system diagnosis and sensitivity analysis,and has less computational complexity.4.Research on the reliability simulation of high reliable non-Markovian systemsThe thesis presents two simulation strategies for the reliability analysis of high reliable non-Markovian systems.In order to sample precisely the state transition time when simulating non-Markovian systems,a new sampling method named approximate sampling for residual distribution(ASRD) is put forward,in which several uniform distributions are used to approach the residual life distributions with general form.The approach-coefficient d is used to adjust the sampling precision and the efficiency.Based on ASRD,a simulation model named forced transitions based ASRD(FT-ASRD) is brought forward in order to simulate the non-Markovian systems with high reliability. FT-ASRD uses forced transitions strategy to get the unreliability in limited simulation times.Several examples show that,when 5≤d≤10,FT-ASRD can simulate high-reliable non-Markovian systems with higher efficiency and lower relative error compared with crude simulation methods.In conclusion,the thesis mainly makes researches on the techniques of SIPS from two aspects,namely,the reliability design(fault-tolerant real-time scheduling algorithms in software fault-tolerant model) and the reliability analysis(reliability models corresponding to different characters of SIPS).The proposed fault-tolerant real-time scheduling algorithm provides an available design scheme to improve the ability of tolerating software faults.And the different reliability models can be used to assess whether the scheduled tasks of SIPS can be fulfilled.