基于广义极值分布的流化床压力波动风险预警

为预警气泡运动所引起的流化床粉煤气化压力波动风险,提出预测压力波动极值以及压力波动重现水平的方法;首先采用自相关分析法将压力波动母样本数据合理分段,再用区间极值法统计子样本的压力波动极值数据,以广义极值(GEV)分布方法建立GEV分布模型和Gumbel分布模型,并经过模型诊断选择最优模型;然后通过子样本与母样本的参数关...

Risk warning of pressure fluctuation in fluidized beds based on generalized extreme value distribution

In order to forewarn pressure fluctuation risks of pulverized coal gasification in fluidized beds caused by bubble movement,a method to predict extreme value and return level of pressure fluctuation was proposed.Firstly,pressure fluctuation data of parent sample were segmented reasonably by autocorrelation analysis method,and interval extreme value method was employed to calculate extreme value data for sub-samples.Secondly,GEV distribution model and Gumbel distribution model were established through GEV distribution method,and an optimal model was selected by model diagnosis.Then,extreme distribution parameters of parent sample were obtained from parametric relationship between it and sub-samples,and expected value of extreme pressure fluctuation was acquired based on extreme value distribution parameters of parent sample,from which,return level of pressure fluctuation was also obtained by Bayesian inference.The results show that the error rate of extreme value obtained by utilizing GEV distribution is controlled within±10%,with its average error rate within-1.7%-3.7%.It can be believed with 95%confidence that return level of pressure fluctuation will be maintained within confidence interval.