A study on the GA-BP neural network model for surface roughness of basswood-veneered medium-density fiberboard

Holzforschung ◽  
2020 ◽  
Vol 74 (10) ◽  
pp. 979-988
Author(s):  
Xizhi Wu ◽  
Han Niu ◽  
Xian-Jun Li ◽  
Yiqiang Wu
Keyword(s):  
Neural Network ◽  
Surface Roughness ◽  
Bp Neural Network ◽  
Medium Density Fiberboard ◽  
Research Work ◽  
Feed Speed ◽  
Medium Density ◽  
Belt Speed ◽  
Interface Bonding Strength ◽  
Grain Roughness

AbstractRoughness is an important property of wood surface and has a significant influence on the interface bonding strength and surface coating quality. However, there are no theoretical models for basswood-veneered medium-density fiberboard (MDF) by fine sanding from existing research work. In this paper, the basswood-veneered MDF was fine sanded with an air drum. Orthogonal experiment was implemented to study the effects of abrasive granularity, feed rate, belt speed, air drum deformation and air drum pressure on the surface roughness of basswood-veneered MDF. The simulation models of the parallel-grain roughness and the vertical-grain roughness of the sanded surface were conducted based on the BP (error back propagation) neural network, which was optimized by a genetic algorithm (GA) (GA-BP neural network), and these models were verified by extensive experimental data. The results showed that the influence of sanding parameters on parallel-grain roughness was similar to that on vertical-grain roughness. The order of influence was that: abrasive granularity > belt speed > feed speed > air drum deformation and air drum pressure. Based on the work, the parallel-grain roughness and vertical-grain roughness of basswood-veneered MDF could be well predicted by the GA-BP neural network. The average relative errors on parallel-grain roughness and vertical-grain roughness were 3.4% and 1.9%, respectively.

Effect of species and grinding disc distance on the surface roughness parameters of medium density fiberboard

2016 ◽  
Vol 75 (3) ◽  
pp. 335-346 ◽  
Author(s):  
Lidia Gurau ◽  
Nadir Ayrilmis ◽  
Jan Thore Benthien ◽  
Martin Ohlmeyer ◽  
Manja Kitek Kuzman ◽  
...  
Keyword(s):  
Surface Roughness ◽  
Medium Density Fiberboard ◽  
Medium Density ◽  
Roughness Parameters ◽  
Surface Roughness Parameters

Surface Roughness Prediction Model of Cylinder Grinding

2012 ◽  
Vol 157-158 ◽  
pp. 123-126 ◽  
Author(s):  
Ning Ding ◽  
Yi Chen Wang ◽  
Ding Tong Zhang ◽  
Yu Xiang Shi ◽  
Jian Shi
Keyword(s):  
Neural Network ◽  
Surface Roughness ◽  
Prediction Model ◽  
Fuzzy Neural Network ◽  
Vibration Signal ◽  
Feed Speed ◽  
Vibration Data ◽  
Proposed Model ◽  
Fuzzy Neural ◽  
Roughness Prediction

Based on the theory of roughness during cylinder grinding and the theory of fuzzy-neural network, a surface roughness intelligent prediction model is developed in this paper. The feed, speed, and the vibration data are the inputs for the model. An accelerometer is used to gather the vibration signal in real time. The model is used in the grinding experiment, and verifies the feasibility of the proposed model.

Modeling for prediction of surface roughness in milling medium density fiberboard with a parallel robot

Sensor Review ◽  
2019 ◽  
Vol 39 (5) ◽  
pp. 716-723 ◽  
Author(s):  
Mustafa Ayyildiz
Keyword(s):  
Surface Roughness ◽  
Learning Algorithm ◽  
Medium Density Fiberboard ◽  
Back Propagation ◽  
Parallel Robot ◽  
Medium Density ◽  
Mean Square ◽  
Ann Model ◽  
Content Type ◽  
Milling Medium

Purpose This paper aims to discuss the utilization of artificial neural networks (ANNs) and multiple regression method for estimating surface roughness in milling medium density fiberboard (MDF) material with a parallel robot. Design/methodology/approach In ANN modeling, performance parameters such as root mean square error, mean error percentage, mean square error and correlation coefficients (R2) for the experimental data were determined based on conjugate gradient back propagation, Levenberg–Marquardt (LM), resilient back propagation, scaled conjugate gradient and quasi-Newton back propagation feed forward back propagation training algorithm with logistic transfer function. Findings In the ANN architecture established for the surface roughness (Ra), three neurons [cutting speed (V), feed rate (f) and depth of cut (a)] were contained in the input layer, five neurons were included in its hidden layer and one neuron was contained in the output layer (3-5-1).Trials showed that LM learning algorithm was the best learning algorithm for the surface roughness. The ANN model obtained with the LM learning algorithm yielded estimation training values R2 (97.5 per cent) and testing values R2 (99 per cent). The R2 for multiple regressions was obtained as 96.1 per cent. Originality/value The result of the surface roughness estimation model showed that the equation obtained from the multiple regressions with quadratic model had an acceptable estimation capacity. The ANN model showed a more dependable estimation when compared with the multiple regression models. Hereby, these models can be used to effectively control the milling process to reach a satisfactory surface quality.

Effects of sanding parameters on sanding force and normal force in sanding wood-based panels

Holzforschung ◽  
2015 ◽  
Vol 69 (2) ◽  
pp. 241-245 ◽  
Author(s):  
Bin Luo ◽  
Li Li ◽  
Hongguang Liu ◽  
Mingzhi Wang ◽  
Meijun Xu ◽  
...  
Keyword(s):  
Energy Consumption ◽  
Normal Force ◽  
Medium Density Fiberboard ◽  
Linear Increase ◽  
Feed Speed ◽  
Medium Density ◽  
Particle Board ◽  
Processing Efficiency ◽  
Reduce Energy Consumption ◽  
Increasing Trends

Abstract The proper parameters of sanding with abrasive sanding machine are significant to reduce energy consumption and to improve processing efficiency and quality. The parameters sanding speed, feed speed, and granularity have been investigated in terms of the sanding force (sF) and normal force (nF) for particle board (PB) and medium-density fiberboard (MDF). For PB, the sF and nF show decreasing trends of second power with increasing sanding speed and linear increase when feed speed increases. The sF and nF are almost constant when granularity increases from 40 to 80, but these forces show increasing trends of second power when granularity increases from 80 to 150. For MDF, the sF and nF change as trends of second power with increasing sanding speed and increase in trends of second power with increasing feed speed. The sF and nF force decrease when granularity increases from 40 to 80, but these forces present linear increasing trends when granularity increases from 80 to 150.

Sign in / Sign up

Export Citation Format

Share Document