基于改进GWO-LightGBM的磨煤机故障预警方法研究

Research on Fault Early Warning Method Based on Improved GWO-LightGBM Coal Mills

为提高燃煤电厂磨煤机运维效率、降低运维成本,对磨煤机故障预警进行了研究。创新性地提出一种基于改进灰狼优化(GWO)算法的轻量级梯度提升机(LightGBM)故障预警方法。通过建立LightGBM轴承温度预测模型获取磨煤机轴承温度阈值,并引入改进GWO算法优化模型超参数,以提高算法效率和性能。试验结果表明,改进GWO-LightGBM算法相比支持向量机(SVM)等传统算法具有更高的精度和更优的泛化能力。通过实际故障案例证明,该方法能够提前2 h对磨煤机进行早期故障预警。该方法对燃煤电厂磨煤机安全运维具有指导意义。

To improve the efficiency and reduce the cost of coal mill operation and maintenance in coal-fired power plants,coal mill fault early warning is studied.A light-gradient boosting machine(LightGBM)fault warning method based on the grey wolf optimization(GWO)algorithm is innovatively proposed.The coal mill bearing temperature threshold is obtained by establishing a LightGBM bearing temperature prediction model and introduces the improved GWO algorithm to optimise the model hyperparameters to improve the algorithm efficiency and performance.The experimental results show that the improved GWO-LightGBM algorithm has higher accuracy and better generalisation capability compared with traditional algorithms such as support vector machine(SVM).The method can provide early fault warning for coal mills up to 2 h in advance,as demonstrated by actual fault cases.The method is of guiding significance for the safe operation and maintenance of coal mills in coal-fired power plants.