In this work,synchronous cutting of concave and convex surfaces was achieved using the duplex helical method for the hypoid gear,and the problem of tooth surface error correction was studied.First,the mathematical mod...In this work,synchronous cutting of concave and convex surfaces was achieved using the duplex helical method for the hypoid gear,and the problem of tooth surface error correction was studied.First,the mathematical model of the hypoid gears machined by the duplex helical method was established.Second,the coordinates of discrete points on the tooth surface were obtained by measurement center,and the normal errors of the discrete points were calculated.Third,a tooth surface error correction model is established,and the tooth surface error was corrected using the Levenberg-Marquard algorithm with trust region strategy and least square method.Finally,grinding experiments were carried out on the machining parameters obtained by Levenberg-Marquard algorithm with trust region strategy,which had a better effect on tooth surface error correction than the least square method.After the tooth surface error is corrected,the maximum absolute error is reduced from 30.9μm before correction to 6.8μm,the root mean square of the concave error is reduced from 15.1 to 2.1μm,the root mean square of the convex error is reduced from 10.8 to 1.8μm,and the sum of squared errors of the concave and convex surfaces was reduced from 15471 to 358μm^(2).It is verified that the Levenberg-Marquard algorithm with trust region strategy has a good accuracy for the tooth surface error correction of hypoid gear machined by duplex helical method.展开更多
针对近端策略优化(PPO)算法难以严格约束新旧策略的差异和探索与利用效率较低这2个问题,提出一种基于裁剪优化和策略指导的PPO(COAPG-PPO)算法。首先,通过分析PPO的裁剪机制,设计基于Wasserstein距离的信任域裁剪方案,加强对新旧策略差...针对近端策略优化(PPO)算法难以严格约束新旧策略的差异和探索与利用效率较低这2个问题,提出一种基于裁剪优化和策略指导的PPO(COAPG-PPO)算法。首先,通过分析PPO的裁剪机制,设计基于Wasserstein距离的信任域裁剪方案,加强对新旧策略差异的约束;其次,在策略更新过程中,融入模拟退火和贪心算法的思想,提升算法的探索效率和学习速度。为了验证所提算法的有效性,使用MuJoCo测试基准对COAPG-PPO与CO-PPO(PPO based on Clipping Optimization)、PPO-CMA(PPO with Covariance Matrix Adaptation)、TR-PPO-RB(Trust Region-based PPO with RollBack)和PPO算法进行对比实验。实验结果表明,COAPG-PPO算法在大多数环境中具有更严格的约束能力、更高的探索和利用效率,以及更高的奖励值。展开更多
基金Projects(52075552,51575533,51805555,11662004)supported by the National Natural Science Foundation of China。
文摘In this work,synchronous cutting of concave and convex surfaces was achieved using the duplex helical method for the hypoid gear,and the problem of tooth surface error correction was studied.First,the mathematical model of the hypoid gears machined by the duplex helical method was established.Second,the coordinates of discrete points on the tooth surface were obtained by measurement center,and the normal errors of the discrete points were calculated.Third,a tooth surface error correction model is established,and the tooth surface error was corrected using the Levenberg-Marquard algorithm with trust region strategy and least square method.Finally,grinding experiments were carried out on the machining parameters obtained by Levenberg-Marquard algorithm with trust region strategy,which had a better effect on tooth surface error correction than the least square method.After the tooth surface error is corrected,the maximum absolute error is reduced from 30.9μm before correction to 6.8μm,the root mean square of the concave error is reduced from 15.1 to 2.1μm,the root mean square of the convex error is reduced from 10.8 to 1.8μm,and the sum of squared errors of the concave and convex surfaces was reduced from 15471 to 358μm^(2).It is verified that the Levenberg-Marquard algorithm with trust region strategy has a good accuracy for the tooth surface error correction of hypoid gear machined by duplex helical method.
文摘针对近端策略优化(PPO)算法难以严格约束新旧策略的差异和探索与利用效率较低这2个问题,提出一种基于裁剪优化和策略指导的PPO(COAPG-PPO)算法。首先,通过分析PPO的裁剪机制,设计基于Wasserstein距离的信任域裁剪方案,加强对新旧策略差异的约束;其次,在策略更新过程中,融入模拟退火和贪心算法的思想,提升算法的探索效率和学习速度。为了验证所提算法的有效性,使用MuJoCo测试基准对COAPG-PPO与CO-PPO(PPO based on Clipping Optimization)、PPO-CMA(PPO with Covariance Matrix Adaptation)、TR-PPO-RB(Trust Region-based PPO with RollBack)和PPO算法进行对比实验。实验结果表明,COAPG-PPO算法在大多数环境中具有更严格的约束能力、更高的探索和利用效率,以及更高的奖励值。
