基于IBAS-BP算法的热电厂 负荷预测及工程应用

Load Prediction and Engineering Application of Thermal Power Plant Based on IBAS-BP Algorithm

摘要:

针对热电厂负荷随机性强、预测精度差、计算时间长等问题，提出一种结合改进天牛须搜索算法IBAS和BP神经网络的组合预测方法。模型以热电厂的历史有功负荷、季节、日期类型和气象数据为输入因子，通过引入精英策略，将单个天牛寻优扩充为群体寻优，同时改进天牛搜索步长，使BP参数在IBAS搜索范围内有效寻优，从而优化BP神经网络的权值，增强其搜索和寻优能力，提高预测网络的性能和精度。采用4个标准测试函数，将改进模型与标准天牛须算法对比。引入均方根误差RMSE、平均绝对百分比误差MAPE精度评价指标对PSO-BP网络、BAS-BP模型、IBAS-BP模型预测结果进行评估。实验结果表明，与其他模型的算例结果相比，IBAS-BP模型具有更好的预测性能。将热电厂负荷预测的结果，作为其厂级负荷优化分配系统(厂级AGC)的输入，通过负荷优化分配系统，得出单台机组未来负荷的预测值，最大限度地降低供电煤耗量，提高热电厂机组运行的经济性。

Abstract:

In order to solve the thermal power plant load randomness, short-term power load prediction accuracy poor, long calculation time and other problems, the paper proposes a prediction method combined with Improved Beetle Antennae Search（IBAS）algorithm and BP neural network. Model in the history of the thermal power plant active load, season, date type and weather data as input factors, by introducing elite strategy, the optimization of single beetle is extended to group optimization, and the search step length of beetle is improved. So that BP parameters could be effectively optimized within the search range of IBAS, to optimize the weight of BP neural network, enhance its search and optimization ability, and improve the performance and accuracy of the prediction network. Using four standard test functions, the improved model compared with standard BAS algorithm, the introduction of Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE), the precision evaluation index of PSO-BP network, BAS-BP model, IBAS-BP prediction model for evaluation. The experimental results show that compared with other kinds of model calculation results, the IBAS-BP model has better prediction performance.The load forecast result of thermal power plant is taken as the input of the plant level load optimization distribution system (plant level AGC), and the predicted value of the future load of a single unit is obtained through the load optimization distribution system, so as to minimize the coal consumption of power supply and improve the operation economy of thermal power plant units.