A new method based on the combination of a neural network and a genetic algorithm was proposed to rank the order of exploitation priority of coalbed methane reservoirs. The neural network was used to acquire the weigh...A new method based on the combination of a neural network and a genetic algorithm was proposed to rank the order of exploitation priority of coalbed methane reservoirs. The neural network was used to acquire the weights of reservoir parameters through sample training and genetic algorithm was used to optimize the initial connection weights of nerve cells in case the neural network fell into a local minimum. Additionally, subordinate functions of each parameter were established to normalize the actual values of parameters of coalbed methane reservoirs in the range between zero and unity. Eventually, evaluation values of all coalbed methane reservoirs could be obtained by using the comprehensive evaluation method, which is the basis to rank the coalbed methane reservoirs in the order of exploitation priority. The greater the evaluation value, the higher the exploitation priority. The ranking method was verified in this paper by ten exploited coalbed methane reservoirs in China. The evaluation results are in agreement with the actual exploitation cases. The method can ensure the truthfulness and credibility of the weights of parameters and avoid the subjectivity caused by experts. Furthermore, the probability of falling into local minima is reduced, because genetic the algorithm is used to optimize the neural network system.展开更多
外涵静子叶片是大涵道比涡扇发动机气路的核心部件之一,外涵静子脱出是一种较为严重的故障模式,此故障可能会导致飞机或其他发动机部件损伤,进而造成灾难性事故。对外涵静子叶片脱出故障的预警是一项重要的工作。但因其早期特征不明显,...外涵静子叶片是大涵道比涡扇发动机气路的核心部件之一,外涵静子脱出是一种较为严重的故障模式,此故障可能会导致飞机或其他发动机部件损伤,进而造成灾难性事故。对外涵静子叶片脱出故障的预警是一项重要的工作。但因其早期特征不明显,现有的方法较难对此类故障进行有效的预警。因此,针对该问题,基于监控数据提出一种深度特征提取的支持向量数据域描述(Support vector data description,SVDD)的故障预警方法,以实现对外涵静子叶片脱出故障的早期预警。首先,采用基于发动机气路性能辨识的建模方法,建立发动机特定性能参数的观测模型对气路参数进行深度特征提取,以真实状态量与模型观测量的差值作为航空发动机是否发生故障的特征;然后利用SVDD算法建立决策边界,实现故障数据的自动划分,决策边界生成的阈值可在故障发生之前的一定时间之内给出告警;最后,经过多次计算,结果表明,在故障早期直至故障发生的区间内,表征其健康状态的性能参数都与观测量有较大的偏移,表明了所选特征的有效性。使用数据增强方法生成故障仿真数据与真实数据进行对比验证,预警时间比故障真实发生时间预警模型平均提前3.14 h。展开更多
基金EU-China Energy and Environment Programme(Europe Aid/120723/D/SV/CN)Research Fund for the Doctoral Program of Higher Education of China(20030425001)
文摘A new method based on the combination of a neural network and a genetic algorithm was proposed to rank the order of exploitation priority of coalbed methane reservoirs. The neural network was used to acquire the weights of reservoir parameters through sample training and genetic algorithm was used to optimize the initial connection weights of nerve cells in case the neural network fell into a local minimum. Additionally, subordinate functions of each parameter were established to normalize the actual values of parameters of coalbed methane reservoirs in the range between zero and unity. Eventually, evaluation values of all coalbed methane reservoirs could be obtained by using the comprehensive evaluation method, which is the basis to rank the coalbed methane reservoirs in the order of exploitation priority. The greater the evaluation value, the higher the exploitation priority. The ranking method was verified in this paper by ten exploited coalbed methane reservoirs in China. The evaluation results are in agreement with the actual exploitation cases. The method can ensure the truthfulness and credibility of the weights of parameters and avoid the subjectivity caused by experts. Furthermore, the probability of falling into local minima is reduced, because genetic the algorithm is used to optimize the neural network system.
文摘外涵静子叶片是大涵道比涡扇发动机气路的核心部件之一,外涵静子脱出是一种较为严重的故障模式,此故障可能会导致飞机或其他发动机部件损伤,进而造成灾难性事故。对外涵静子叶片脱出故障的预警是一项重要的工作。但因其早期特征不明显,现有的方法较难对此类故障进行有效的预警。因此,针对该问题,基于监控数据提出一种深度特征提取的支持向量数据域描述(Support vector data description,SVDD)的故障预警方法,以实现对外涵静子叶片脱出故障的早期预警。首先,采用基于发动机气路性能辨识的建模方法,建立发动机特定性能参数的观测模型对气路参数进行深度特征提取,以真实状态量与模型观测量的差值作为航空发动机是否发生故障的特征;然后利用SVDD算法建立决策边界,实现故障数据的自动划分,决策边界生成的阈值可在故障发生之前的一定时间之内给出告警;最后,经过多次计算,结果表明,在故障早期直至故障发生的区间内,表征其健康状态的性能参数都与观测量有较大的偏移,表明了所选特征的有效性。使用数据增强方法生成故障仿真数据与真实数据进行对比验证,预警时间比故障真实发生时间预警模型平均提前3.14 h。
