摘要
管道密封性能对工业流体传输系统的可靠性至关重要。文章介绍了一种改进的粒子群优化算法(IPSO),用于预测管道密封性能。IPSO算法结合了粒子群算法和新的非线性递减惯性权重策略,以提高速度和位置更新速度,同时引入了遗传交叉算子以增加多样性。IPSO算法优化了神经网络的权重和阈值,创建了IPSO-BPNN模型。该模型在两个管道数据集上的测试结果显示出色,MAE为0.5154、0.7085,MAPE为3.76%、2.67%,RMSE为0.6726、0.9472。与LR、FEA、BPNN以及PSO-BPNN模型相比,IPSO-BPNN模型的预测性能更佳。这种改进的粒子群算法显著提高了管道密封性能的预测准确性,为管道检查提供更精确的依据。
Pipe seal performance is critical to the reliability of industrial fluid transfer systems.This paper introduces an improved particle swarm optimization(IPSO)algorithm to predict the sealing performance of pipelines.IPSO algorithm combines particle swarm optimization(PSO)and a new nonlinear descending inertia weight strategy to improve speed and position update speed,while introducing genetic crossover operators to increase diversity.The IPSO algorithm optimizes the weights and thresholds of the neural network and creates the IPSO-BPNN model.The model has excellent test results on two pipeline datasets,with MAE of 0.5154 and 0.7085,MAPE of 3.76%and 2.67%,and RMSE of 0.6726 and 0.9472.Compared with LR,FEA,BPNN and PSO-BPNN models,IPSO-BPNN model has better predictive performance.The improved PSO significantly improves the prediction accuracy of pipeline sealing performance and provides a more accurate basis for pipeline inspection.
作者
张静文
Zhang Jingwen(Chongqing Creation Vocational College,Chongqing,Sichuan 402160,China)
出处
《化工设备与管道》
CAS
北大核心
2023年第6期106-113,共8页
Process Equipment & Piping
基金
2021年重庆市教育科学十四五规划课题(基于核心素养的高职通识教育改革研究与实践)(课题编号2021-GX-184)。
关键词
管道密封性能
预测模型
粒子群算法
神经网络
pipeline sealing performance
predictive model
particle swarm optimization
neural network
