现有负荷敏感度随机估计法以电压暂降的物理特征为基础，用主观概率模型描述负荷电压耐受曲线(voltage tolerance curve，VTC)的随机分布规律，所需样本量大，在实际中难以实现且可能引入主观误差。将电压暂降特征转换为电压暂降严重性指标，在负荷VTC曲线分布规律未知和样本数较少的情况下，根据最大熵原理确定VTC曲线的概率密度函数，用累计求和法计算负荷故障率，提出一种适合于小样本的随机评估方法。对评估原理、最大熵模型、约束条件、求解算法与评估过程等进行详细研究。对个人计算机(personal computers，PC)进行仿真并与现有4种评估方法比较，结果证明，该方法对样本量依赖性小，无需主观假设，在未知VTC曲线随机分布规律时，评估结果准确，适应性强。

Based on physical characteristics, existing stochastic estimation methods of equipment sensitivity to voltage sag use subjective probability models to express the probability distribution of voltage tolerance curve (VTC) of equipment in the uncertain region. But the parameter estimation needs vast sample data. These methods may result in man-made errors. In order to investigate the universal rule of sensitivity estimation method, the concept of severity index was introduced and a new stochastic assessment method was proposed based on maximum entropy principle in this paper. In this method, the probability density function of VTC was determined by the maximum entropy model under limited sample data. The accumulative summing was used to calculate the failure rate of equipment during voltage sag. The estimation principle, the maximum entropy model, its constraints and the solution were investigated in detail. The approaches were also presented. As a case study, the personal computer was simulated. The simulation results compared with existing methods show that the method needs no subjective assumption under the condition of small samples and the results accord with the practical situation when the probability distribution of VTC is unknown. And this method is with good adaptability.

国家自然科学基金项目(50877049，50677041)；四川省应用基础研究项目(2008JY0043-2)。