Abstract

Failure of safety-critical systems (SCSs) devastates human life and the environment and involves huge costs. So, quality assurance is essential for the SCSs. Performability is a combined study of performance and reliability. The performability parameters should be precise for the critical systems since each model has some limitations that cause uncertainty. This paper proposes a model for uncertainty Prediction and optimization for the performability of the SCSs. A wireless sensor network fire alarm system is taken as a case study to illustrate the concepts and demonstrate their applicability. The performability of the fire alarm system is modelled using the hyper-exponential distribution, and synthetic data are generated using this distribution. The output uncertainty is calibrated using the probabilistic neural network (PNN). Failure of safety-critical systems (SCSs) has catastrophic effects on both the environment and human life and is extremely expensive. Quality control is, therefore, crucial for SCSs. The study of performability combines dependability with performance. Given that each model has some limits that lead to uncertainty, the performability parameters should be accurate for the key systems. This article suggests a model for uncertainty predictions and performance improvement for the SCSs. A wireless sensor network fire alarm system is used as a case study to clarify the ideas and show how they apply. The hyper-exponential distribution is used to model the fire alarm system’s performance, and it is also used to create synthetic data. The probabilistic neural network (PNN) is used to calibrate the output uncertainty.

中文翻译：

---

安全关键系统性能预测中的不确定性建模

摘要

安全关键系统 (SCS) 的故障会破坏人类生命和环境，并造成巨大损失。因此，质量保证对于 SCS 至关重要。性能是性能和可靠性的综合研究。对于关键系统来说，性能参数应该是精确的，因为每个模型都有一些导致不确定性的限制。本文提出了一种用于 SCS 性能的不确定性预测和优化的模型。以无线传感器网络火灾报警系统为例来说明这些概念并证明其适用性。使用超指数分布对火灾报警系统的性能进行建模，并使用该分布生成综合数据。使用概率神经网络 (PNN) 校准输出不确定性。安全关键系统 (SCS) 的故障会对环境和人类生命造成灾难性影响，而且代价极其高昂。因此，质量控制对于 SCS 至关重要。对性能的研究将可靠性与性能结合起来。鉴于每个模型都有一些导致不确定性的限制，因此关键系统的性能参数应该是准确的。本文提出了一种用于 SCS 的不确定性预测和性能改进的模型。以无线传感器网络火灾报警系统作为案例研究来阐明这些想法并展示它们如何应用。超指数分布用于对火灾报警系统的性能进行建模，并且还用于创建合成数据。概率神经网络（PNN）用于校准输出不确定性。