基于极限学习机的采煤机功率预测模型构建

Construction of Power Prediction Model of Shearer based on Limit Learning Machine

为提高采煤机概念设计的效率和质量,充分继承专家知识经验,面向采煤机概念设计过程,分析了采煤机内部参数的耦合关系,结合粗糙集属性约简和规则提取,提出基于极限学习机的采煤机电机功率预测模型。采用遗传算法优化隐含层神经元个数,并通过分析不同激励函数对决定系数的影响,优选激励函数。最后选用设备制造商的数据进行实验分析,并与支持向量机模型进行对比。结果表明,该模型决定系数保持在0.95以上,平均相对误差在5%以内,精度优于支持向量机模型,并且ELM模型在计算速度方面具有显著优势,能够满足应用需求。

In order to improve the efficiency and quality of the conceptual design of shearer, by fully inheriting the knowledge and experience of experts, the coupling relationship of the internal parameters of shearer was analyzed on account of the conceptual design process of shearer. Based on the rough set attribute reduction and rule extraction, a prediction model of shearer motor power based on limit learning machine was proposed. The genetic algorithm was used to optimize the number of neurons in the hidden layer, and the incentive function was optimized by analyzing the influences of different incentive functions on the decision coefficient. Finally, the manufacturers- data was chose to carry out experimental analysis and to compare with the support vector machine model. The results showed that the decision coefficient of the model maintained above 0.95, the average relative error was less than 5%, and the accuracy was better than that of the support vector machine model. Moreover, the ELM model had a significant advantage in computing speed, and could meet the application needs.