【摘要】 由于放射性统计涨落以及仪器自身稳定性的影响,碳氧比测井伽马能谱中会包含一定量的噪声,影响碳氧比值的计算精度,在资料解释时需要对能谱进行滤波。传统小波阈值算法进行信号滤波处理时采用的软硬阈值函数存在均方根误差偏大与伪吉布斯效应等问题。提出了一种具有调节参数的改进阈值函数,根据不同的参数取值实现不同的滤波效果。为了获得改进阈值函数中的最佳参数,提出了基于反向学习策略的遗传算法,并用于对阈值函数中的参数进行优化。与经典遗传算法相比,改进遗传算法具有更好的收敛性与寻优精度。采用优化后的阈值函数对能谱进行滤波处理得到的数据和软硬阈值函数的处理结果相比具有更小的均方根误差,能谱中的噪声也得到了有效压制,表明所提算法具有更好的适用性。更多还原

【Abstract】 Due to the fluctuation of radioactivity statistics and the influence of the stability of the instrument, the gamma energy spectrum measured by the C/O logging instrument would contain a certain amount of noise, which can affect the calculation accuracy of the C/O, so it is necessary to filter the energy spectrum during the process of data interpretation. The traditional wavelet threshold algorithm used the soft and hard threshold function for signal filtering, but the function has the problems of large mean square root error and pseudo-Gibbs effect. In order to solve these problems, an improved threshold function which can adjust parameters was proposed in this paper, which can be used according to different parameters to realize different filtering effects. Meanwhile, to obtain the best parameters in the improved threshold function, this paper also proposed a genetic algorithm based on backward learning strategy, which was used to optimize the parameters in the threshold function. Compared with the classical genetic algorithm, the improved genetic algorithm has better convergence and optimization accuracy. By comparison, it was found that compared with the processing results of the soft and hard threshold function, the data obtained by the optimized threshold function had a smaller mean square root error, and the noise in the energy spectrum was effectively suppressed, which indicated that the proposed algorithm has better applicability.更多还原