摘要
为提高岩性识别的精度,引入SSA算法解决BP神经网络的性能受连接权值cj、ωij和连接阈值ε、θj的影响较大的问题,对网络连接权值以及阈值进行选择性优化,提出一种基于SSA-BP的岩性识别方法。将声波、补偿中子、微电极2 m梯度、井径、4 m梯度、2.5 m梯度、感应电导、浅侧向和微电极差等9项指标输入SSA-BP算法,评价岩性识别效果,使用岩性类别作为SSA-BP算法的输出,得到SSA-BP的岩性识别模型。与ELM、SVM和BPNN对比发现,SSA-BP可以有效提高岩性识别的精度,识别准确率达到96.41%,为岩性识别研究和应用提供了新的方法和途径。
In order to improve the accuracy of lithology recognition,the SSA algorithm is introduced to solve the problem that the performance of BP neural is greatly affected by the connection weights cj and ωij,and the connection thresholds ε and θj.The network connection weights and thresholds are selectively optimized,with the proposal of SSA-BP’s lithology identification method.Input 9 indicators such as acoustic wave,compensated neutron,microelectrode 2 m gradient,borehole diameter,4 m gradient,2.5 m gradient,induced conductance,shallow lateral and microelectrode difference,etc.into the SSA&algorithm to evaluate the lithology recognition effect,and use the lithology category as the output of the SSA&BP algorithm,the lithology recognition model of SSA&BP is obtained.Compared with ELM,SVM and BPNN,it is found that SSA-BP can effectively improve the accuracy of lithology recognition,and the recognition accuracy rate reaches 96.41%,which provides new methods and approaches for the research and application of lithology recognition.
作者
杨远宏
YANG Yuan-hong(Guizhou University,Guiyang 550025,China)
出处
《长春工程学院学报(自然科学版)》
2021年第1期87-91,共5页
Journal of Changchun Institute of Technology:Natural Sciences Edition
关键词
樽海鞘算法
BP神经网络
岩性识别
极限学习机
支持向量机
测井
salp swarm algorithm
BP neural network
lithology recognition
extreme learning machine
support vector machine
logging