Citation
Ismail, Nazarul Abidin
(2021)
Artificial intelligence-based tool condition monitoring in robotic incremental sheet forming through vibration, tool wear and surface roughness analyses.
Doctoral thesis, Universiti Putra Malaysia.
Abstract
Sheet metal forming is a fabrication process that allows sheet metal to be formed
in 3D shapes with the use of a specific tool and die. However, the conventional
sheet metal forming has disadvantages in terms of quality and low flexibility, and
it also prolongs the time-to-market in producing low costs prototype products.
Robot-based incremental sheet forming (ISF) is a new prospect and one of the
relatively new sheet metal forming processes to fabricate a product with 3D
complex shapes. Interests in new techniques with a variety approach for forming
processes have created more studies by researchers on the robot-based ISF
process. However, tool wear always makes the difficulty of the ISF process for
sustaining the process performance. In the present study, the development of a
comprehensive predictive model for tool wear in robot-based ISF using artificial
intelligence (AI) has been conducted. The model would predict the critical
degradation of tool wear and simultaneously the relationship with quality of the
formed workpiece surface. The robot-based ISF experiments were carried out
using a forming tool of AISI D2 tool steel with a 10 mm diameter that attached to
the ABB IRB 4400/60 IRC5 industrial robotic arm. Three different materials of
SUS316 stainless steel, Cu60Zn40 copper alloy and AA3003 aluminum alloy
with 0.5 mm thickness were used as workpieces. As preliminary experiments, a
parametric optimization was carried out to determine optimum processing
parameters in robot-based ISF using L18 orthogonal array design of
experiments. The vibration signals of the ISF process were recorded by the
accelerometer sensors, which are located on the forming tool and workpiece.
Subsequently, after the vibration signals through signal processing, pattern
recognition was conducted to identify and categorize the tool condition by two
clusters, which are a tool in good condition and worn out. The increasing of
surface roughness on the workpieces can also be seen noticeably with the
increasing of vibration on the forming process due to tool wear. This proving that
vibration signals can provide the tool wear identification for the ISF process. The
predictive models were developed and compared between three different AI
models, which are artificial neural network (ANN), fuzzy logic (FL) and adaptive
network-based fuzzy inference system (ANFIS). The prediction using ANN
model with two hidden layers showed that it has an excellent prediction accuracy
of 99.94 % for tool wear (architecture 2-4-4-1) and 91.77 % (architecture 2-5-3-
1) for surface roughness. The use of the ANN with two hidden layers is the best
model to predict the tool wear in robot-based ISF. The successful development
of prediction of tool wear in robot-based incremental sheet forming can provide
a significant way in tool condition monitoring system to minimize downtime
related with tool damaged and affected the quality of the workpieces.
Download File
Additional Metadata
Actions (login required)
![View Item View Item](/style/images/action_view.png) |
View Item |