Abstract: High-temperature ceramic filter tube is composed of a support body with a large pore size and a membrane structure with a small pore size. In practical applications, it has a lot of dust with particle size smaller than 1 ?m, which can move through the membrane structure and deposit the support body finally. It can't be removed by pulse jet cleaning effectively. Particle deposition within the support body and the compression of residual dust cake are responsible for the decreasing permeability and increasing residual pressure drop of high temperature ceramic filter tubes. In this work, on the basis of particle deposition within the filter medium and the compression of residual dust cake can't be measured directly, Bayesian estimation theory was used to establish a state-space model to describe the ceramic filter tubes degradation process using the residual pressure drop measured in the filtration system. This method can incorporate the latest residual pressure drop data and update the model parameter timely, the remaining life of the ceramic filter tube was predicted in real time. At the same time the failure probability density distribution of the remaining life of the ceramic filter tube and the change rate of degeneration status of the ceramic filter tube were predicted. By analyzing the actual data of the ceramic filter tube residual pressure drop from high-temperature experiment device and shell coal gasification process respectively, the prediction accuracy of the remaining life increased gradually with the increase of the residual pressure drop data, the accuracy of the prediction at the later stage was higher than 95%, and the ceramic filter tubes change rate of degradation status gradually decreased. This was consistent with the conclusion that the residual pressure drop of ceramic filter tube increases fast at the early stage and slow at the late stage.

Key words: ceramic filter tube, residual pressure drop, state-space model, degradation process, remaining life prediction

摘要： 高温陶瓷过滤管由内部孔径较大的支撑体和外部孔径较小的过滤膜双层结构构成，在实际应用中，存在大量粒径较小的粉尘颗粒，会穿过过滤膜沉积到支撑体内部，脉冲反吹无法有效清除. 因支撑体内颗粒沉积及管壁外残余粉尘层不断压缩，使陶瓷过滤管渗透率不断下降，残余压降逐渐增加。本工作基于高温陶瓷过滤管壁内颗粒沉积特性及残余粉尘层压缩不可直接观测的特点，结合贝叶斯估计理论，利用过滤管运行期间采集的残余压降数据，提出一种基于状态空间模型的过滤管性能退化建模方法。该方法能融入最新采集到的残余压降数据，实时对模型参数进行更新，可对陶瓷过滤管的剩余寿命进行实时预测，同时对陶瓷过滤管剩余寿命的失效概率密度分布及陶瓷过滤管的退化状态变化率进行预测。对某高温试验装置及壳牌煤气化装置中的陶瓷过滤管残余压降分析表明，预测剩余寿命准确率随残余压降数据增加而逐渐增加，后期预测准确率高于95%，且陶瓷过滤管退化状态变化率逐渐变小，与陶瓷过滤管残余压降前期增加快后期增加慢的现象一致。

关键词: 陶瓷过滤管, 残余压降, 状态空间模型, 性能退化, 寿命预测