基于D-S证据理论与Markov模型的安全仪表系统可靠性研究

发布时间：2018-11-24 14:38

【摘要】：现代工业生产的大规模发展,为人类提供更丰富的物质生活产品的同时,也带来了灾难性事故危害的潜在风险。为了提高工业生产的安全性,安全仪表系统(Safety Instrumented System, SIS)被越来越广泛地应用于现代流程工业,以便在事故发生前能执行其安全功能,避免事故的发生或减少事故发生后造成的危害。安全仪表系统作为流程工业安全生产的关键设施,对其自身的安全可靠性要求很高,但由于自身结构、软件、硬件、周围环境及维护等原因,安全仪表系统本身也存在着安全性问题。为了确保安全生产、避免事故的发生,需要对安全仪表系统的可靠性进行准确的评估,因此,其可靠性评估方法的研究成为功能安全领域研究的热点问题。目前,对安全仪表系统可靠性评估的主要方法有五种,即：故障树、可靠性框图、简化公式法、PDS及Markov模型。前四种模型计算简单,但一次建模所考虑的性能指标有限,存在过多的条件假设,无法动态的反应安全仪表系统各状态之间的动态变化。而Markov模型克服了前四种模型的不足,但也存在未考虑安全仪表系统的参数不确定性、模型的不完整性及人为因素等不确定性因素,使得评估结果仍有不足。由于人为因素较难量化,本文暂不考虑,而参数不确定是指Markov模型中假设各个状态参数值为点值,但在实际应用中各个状态参数会随着使用时间的推移而变化,是一个区间值。其次,为了提高安全仪表系统的可用性和可靠性系统通常采用冗余结构,而在多重冗余结构中对安全可靠性影响最大的是共因失效因子,但Markov模型中采用的β因子模型,不同冗余结构使用相同的共因失效因子β,模型存在不完整性。为了解决Markov可靠性评估模型存在的不确定性问题,本文对现有的可靠性评估模型进行了分析研究。针对Markov模型应用到SIS系统可靠性评估中存在的参数不确定性问题,本文将D-S证据理论的方法引入Markov模型,提出了DS-Markov模型,该模型通过引入D-S证据理论的信度函数和似真度函数来计算由各个状态组成的辨识框架中失效率的区间值,得到两个不同的Markov模型状态转移矩阵,根据不同的状态转移矩阵计算单元或系统的平均要求时失效概率,最终得到一个平均要求时失效概率的区间值。其次,针对SIS可靠性评估中Markov模型的模型不完整性问题,本文结合β因子模型和多重β因子模型提出了一个新的共因失效因子模型—β*模型,β*模型不仅采用共因失效修正因子CMOON来区分不同的冗余结构的共因失效因子,还考虑了安全仪表系统的自诊断性。本文通过DS-Markov模型和Markov模型对常见的冗余结构(1oo1、1oo2、2oo2、2oo3、1oo2D)单元进行可靠性评估,并对结果进行了分析：通过β*模型与β模型对2003冗余结构单元评估并分析了评估结果；再把改进了的模型应用到一个完整的安全仪表系统中验证对其可靠性评估的准确度。实验验证表明本文提出的DS-Markov模型评估均值高于Markov模型评估得到的点值,DS-Markov模型评估准确度更高；本文提出β*模型评估结果均高于β模型,对安全仪表系统的可靠性评估更准确。
[Abstract]:The large-scale development of the modern industrial production also brings the potential risk of the disastrous accident hazard to the human being supplied with a richer material life product. In order to improve the safety of the industrial production, the safety instrument system (SIS) is widely used in the modern process industry so as to be able to carry out its safety function before the accident happens, to avoid the occurrence of the accident or to reduce the harm caused by the accident. As the key facility for safe production of the process industry, the safety instrument system is very high in its own safety and reliability, but the safety instrument system itself has security problems due to its own structure, software, hardware, surrounding environment and maintenance. In order to ensure the safety production and avoid the occurrence of accidents, the reliability of the safety instrument system needs to be accurately evaluated. Therefore, the reliability evaluation method is a hot issue in the field of functional safety. At present, there are five main methods to evaluate the system reliability of the safety instrument, namely, the fault tree, the reliability block diagram, the simplified formula method, the PDS and the Markov. The first four models are simple to calculate, but the performance index considered by one-time modeling is limited, there are too many condition hypotheses, and the dynamic reaction safety instrument system can not be used dynamically The Markov model overcomes the shortcomings of the first four models, but there are uncertainties such as the uncertainty of the parameters of the safety instrument system, the incompleteness of the model and the human factors. Not enough. This paper is not considered in this paper because the human factors are hard to quantify. The parameter uncertainty is the assumption that each state parameter value is a point value in the Markov model, but the state parameter in the actual application will change with the time of the use. It is a region. Secondly, in order to improve the availability and reliability system of the safety instrument system, the redundant structure is usually adopted, and the most important to the safety reliability in the multiple redundant structure is the failure factor, but the factor of the redundancy used in the Markov model The model and the different redundant structures use the same common cause failure factor, and the model is not finished. In order to solve the uncertainty of the Markov reliability evaluation model, this paper makes an analysis of the existing reliability evaluation model. In this paper, the method of D-S evidence theory is introduced into the Markov model and the DS-Mar is put forward. The model of the kov model, which is based on the reliability function and the degree of similarity function of the D-S evidence theory, is used to calculate the interval value of the failure efficiency in the identification frame, which is composed of various states, and two different Markov models are obtained. The state transition matrix is used to calculate the failure probability when the average requirement of the unit or system is calculated according to the different state transition matrix, and an average requirement is finally obtained. Secondly, based on the model of the Markov model in the reliability evaluation of the SIS, this paper presents a new common cause failure for the model of the Markov model in the reliability evaluation of the SIS. In this paper, the model of the sub-model is not only used for distinguishing the common failure factors of different redundant structures due to the failure correction factor CMOON, but also the safety instrument The self-diagnosis of the system is presented in this paper. In this paper, the reliability evaluation is carried out on the common redundant structures (1o1, 1o2, 2oo2, 2o3, and 1o2D) through the DS-Markov model and the Markov model, and the results are analyzed: the 2003 redundant structure is analyzed by the model of the model and the Markov model. The meta-evaluation and analysis of the evaluation results; and the application of the improved model to a complete safety instrument system for verification The results of the experiment show that the evaluation of the DS-Markov model is higher than the point value obtained by the Markov model evaluation, and the accuracy of the DS-Markov model is higher.
【学位授予单位】：西南大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：X924.4;X913.4

【参考文献】