论文部分内容阅读
论述了人工神经网络方法和非线性混合离散变量的优化设计方法应用于配煤中的必要性和可行性．对实际应用中所用到的神经网络ＢＰ算法及其一些改进和非线性混合离散变量的优化设计方法———ＭＤＯＤ法作了简要介绍．针对用加权平均或某些经验公式对配煤过程中所遇到的一些指标进行预测时效果不很理想这一现状，采用了神经网络技术对发热量、挥发分、水分、灰分及煤灰软化温度等指标进行建模以实现非线性映射，同时根据现有配煤系统的实际生产情况，把每种煤的配比取为离散变量，从而使ＭＤＯＤ算法在动力配煤中得以应用．
The necessity and feasibility of applying artificial neural network method and optimal design method of nonlinear mixed discrete variables to coal blending are discussed. The BP neural network algorithm used in practical applications and some improvements and optimization design methods of nonlinear mixed discrete variables - MDOD method are briefly introduced. Aiming at the current situation that the weighted average or some empirical formulas are not ideal for predicting some indexes encountered during coal blending, neural network technology is used to simulate the calorific value, volatile matter, moisture content, ash content and coal ash softening Temperature and other indicators to achieve the nonlinear mapping. At the same time, according to the actual production of the existing coal blending system, the ratio of each coal is taken as discrete variables, so that the MDOD algorithm can be applied in power blending.