基于动态神经网络的水泥熟料堆积体建模方法

Modeling Method of Cement Clinker Accumulation Body Based on Dynamic Neural Network

工程中存在许多颗粒堆积问题。为了构建堆积体离散元数值模型,首先设计了堆积体结构参数采集试验,并以工程中篦冷机内的熟料堆积为背景,使用水泥熟料组建堆积体,采用切片法获取切片图像,随后进行一系列处理,获得了堆积体内颗粒的相关结构参数。同时,为了能够重建堆积体模型,使用神经网络训练了一个预测模型,并进行了相关验证。模型运行良好,表明在一定程度上能够满足堆积体建模的要求。

The particle packing is an important issue in engineering.In order to construct the discrete element numerical model of the particle accumulation body,an experiment for collecting its structural parameters such as diameter and coordinate was designed firstly.With the cement clinker packing of the grate cooler in the background,cement clinker was used to form the accumulation body.And a series of images were collected by cutting the particle accumulation body into slices.Then a battery of treatments was carried out to obtain its structural parameters.At the same time,a prediction model was trained based on neural network in order to rebuild accumulation body model and its statistical accuracy was verified.The good precision of the model indicates that it meets the modeling requirements to a certain extent.