郑州大学学报(工学版) /oa 松辽盆地干热岩热储水力压裂参数敏感性分析 /oa/darticle.aspx?type=view&id=202404067 水力压裂是提取地热能的关键技术,通过提高储层岩石的渗透性来增加产热量。本文以松辽盆地干热岩热储层为研究对象,基于Abaqus软件建立水力压裂二维数值模型,结合正交试验法,对影响水力裂缝特性的参数进行了敏感性分析。结果表明:(1)数值模拟结果与室内试验结果的误差为2.6%,说明该模型用来研究水力压裂是准确可靠的。(2)通过极差分析法对各参数进行敏感性分析可知,对裂缝宽度影响最大的因素是岩石弹性模量,压裂液排量的影响程度最小;对起裂压力影响最大的因素为水平应力差异系数,岩石弹性模量的影响程度最小。该研究结果可以为松辽盆地干热岩储层水力压裂施工提供一定的指导。 2026年01月30 00:00 2024年pre1 1 7 1833208 张延军<sup>1,2</sup>,孙铭浩<sup>1</sup>,于子望<sup>1</sup>,刘玉龙<sup>1</sup> 基于深度强化学习的无人机边缘计算任务卸载策略 /oa/darticle.aspx?type=view&id=202404070 <span style=""> <p style="text-indent: 18pt;"> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">针对地理条件较为复杂的环境中存在的缺乏基础设施</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">、</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">任务延时高和带宽需求量大等问题</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">提出一种联合任务卸载和功率分配的多级移动边缘计算</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">(</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">MEC</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">)</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">系统模型</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">。</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">所提模型考虑将配备</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">MEC </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">的服务器部署在无人机附近提供计算服务</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">综合分析无人机的任务卸载</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">、</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">功耗和计算资源分配等问题并给出度量方法</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">同时考虑无人机可执行的任务类型以及任务对无人机的</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">CPU </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">和</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">GPU </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">要求</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">将该问题表述为混合整数非线性问题</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">。针对该问题提出一种基于深度强化学习的计算任务卸载算法</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">该算法基于改进双深度</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">Q </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">学习算法</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">在深度强化学习中利用深度神经网络找到无人机之间的映射</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">从状态空间中找到潜在的模式并估计最优动作</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">并使用无模型的</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">DRL </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">方法</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">使每个无人机根据局部观察快速做出卸载决策</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">。仿真结果表明</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">:</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">所提算法相比</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">LCGP </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">算法</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">平均卸载成本降低了</span>42. 8%</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">;</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">相比</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">DDPG </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">算法</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">能耗减少了</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">16%</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">;</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">相比</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">DDQN </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">算法</span> </span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;">,</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">任务执行延迟减少了</span>12. 9%</span> <span style="font-size: 9pt; font-family: &quot;Times New Roman&quot;;"> <span style="font-family: 宋体;">。</span> </span> </p> </span> <br /> 2026年01月30 00:00 2024年pre1 10 2097588 王 峰1,马星宇2,孟鹏帅2,赵 薇2,翟伟光2