基于特征参数估计的关键元器件健康状态评估是实现电路级故障预测与健康管理的关键技术.然而,工作温度的变化会导致元器件特征参数估计值与失效标准(25℃条件下)不统一,致使状态评估结果不准确.针对该关键问题,以非理想 Boost 变换器为例,首先,采用改进粒子群优化算法辨识关键元器件特征参数值;然后,分别建立电解电容、功率MOSFET器件特征参数与工作温度的数学模型,并定义器件特征参数归一化因子(NF)及健康指标(HI);最后,提出基于无迹粒子滤波(UPF)动态更新模型参数的方法估计归一化后的特征参数值,从而实现变温度下关键元器件健康状态评估.搭建硬件实验平台,通过研究结果表明参数辨识误差小于5%,健康评估准确度可达90%以上,验证了所提方法的有效性.%The health condition assessment for key components based on characteristic parameter estimation is one of the most significant technologies for circuit-level prognostics and health management (PHM). However, changes in operating temperature can lead to inconsistencies in the estimated values of characteristic parameters and their failure criteria (at 25 degrees Celsius), resulting in inaccurate condition evaluation. Therefore, for the non-ideal Boost converter, firstly, this paper provides the values of characteristic parameters by particle swarm optimization (PSO) algorithm. Then, the mathematical models of the characteristic parameters and operating temperature for electrolytic capacitor and power MOSFET are established, respectively. Meanwhile, the normalized factor (NF) and health index (HI) are defined in this paper. Finally, to achieve health condition assessment under various temperature conditions, a novel approach of updating the model parameters dynamically based on unscented particle filter is proposed. In addition, experimental results show that the parameter identification error is less than 5% and the health evaluation accuracy is over 90%.
展开▼
