基于模拟退火-粒子群算法优化支持向量机参数的连铸漏钢预报

发布时间：2019-05-22 01:25

【摘要】：针对连铸漏钢预报神经网络模型在小样本训练数据情况下难以获得较高预报准确率的问题,提出了一种基于模拟退火-粒子群(SA-PSO)算法优化支持向量机(SVM)参数的连铸漏钢预报算法。将粒子群优化算法引入支持向量机的训练过程中,利用其调整参数少、寻优速度快的优点,有效地提高了漏钢预报模型的寻优速度;利用模拟退火算法对粒子群算法迭代更新后粒子的新位置加以评价,来决定新位置是否被接受,避免了粒子群算法在迭代寻优过程中陷入局部极值的问题。结合某钢厂连铸现场历史数据对提出的连铸漏钢预报算法进行了测试,测试结果表明,所提算法的连铸漏钢预报准确率可达98.8%。
[Abstract]:In order to solve the problem that the neural network model of continuous casting breakout prediction is difficult to obtain high prediction accuracy under the condition of small sample training data, A continuous casting breakout prediction algorithm based on simulated annealing particle swarm optimization (SA-PSO) algorithm is proposed to optimize the (SVM) parameters of support vector machine (SVM). In the training process of support vector machine, particle swarm optimization algorithm is introduced into the training process of support vector machine, and the optimization speed of breakout prediction model is effectively improved by using the advantages of less adjustment parameters and fast optimization speed. The simulated annealing algorithm is used to evaluate the new position of particles after iterative updating of particle swarm optimization algorithm to determine whether the new position is accepted or not, which avoids the problem that particle swarm optimization algorithm falls into local extremum in the process of iterative optimization. Based on the historical data of continuous casting site in a steel plant, the proposed prediction algorithm of continuous casting breakout is tested. the test results show that the accuracy of the proposed algorithm can reach 98.8%.
【作者单位】：燕山大学工业计算机控制工程河北省重点实验室;国家冷轧板带装备及工艺工程技术研究中心;中船航海科技有限责任公司;
【基金】：国家自然科学基金委员会与宝钢集团有限公司联合资助项目(U1260203) 国家自然科学基金资助项目(61403332) 河北省高等学校科学技术研究青年基金资助项目(QN2016122) 河北省高等学校创新团队领军人才培育计划资助项目(LJRC013)
【分类号】：TF777;TP18

【共引文献】