High-accurate Tool Wear Prediction With Convolutional Neural Network


Key Lab of Contemporary Design and Integrated Manufacturing Technology, Northwestern Polytechnical University

We develop a model which can monitor single-machining tool wear, also known as TCM, from 7-dimensional signals better than any other traditional method.

Key to exceeding expert and traditional performance is a deep convolutional network which can map a sequence of signal to tool wear label along with PHM 2010 dataset similar with previous dataset of its kind.

READ OUR PAPER Read Patent (CN201810977274) Download Database SOURCE CODE

Whole structure of ResNet

We train a 32-layer convolutional neural network (CNN) to predict tool wear in 7-dimensional sensor signals

The network takes as input a time-series of secondary-sampled sensor signal, and outputs a sequence of responding tool wear. The original signal is sampled at 50KHz. We arrive at an architecture which is 32 layers of convolution followed by a fully connected layer.

To make the optimization of such a deep model tractable, we use residual connections and batch-normalization. The depth increases both the non-linearity of the computation as well as the size of the context window for each regression task.

Each channel of signal is corresponding to:
1: Force (N) in X dimension
2: Force (N) in Y dimension
3: Force (N) in Z dimension
4: Vibration (g) in X dimension
5: Vibration (g) in Y dimension
6: Vibration (g) in Z dimension
7: AE-RMS (V)

The data is collected a dataset of 3 tools under the same machining circumstance

The PHM data is sampled at a frequency of 50000Hz and have 8GB size. In this machining condition, the spindle speed of the cutter was 10400 RPM; feed rate was 1555 mm/min; Y depth of cut (radial) was 0.125 mm; Z depth of cut (axial) was 0.2 mm.

You may find more details about this experiment at HERE

The model outperforms all of competitors in PHM 2010 Contest

Although we have no data about test data, according to the score on known dataset, loss of this model is ~1000 times less than the 1st prize winner, which shows a great accuracy of CNN.

Loss compared with all of competitors

The smaller, the better.

kidozh393 student
gtl-phm55000 professional
UNO-PHM57000 professional
PathFinder64000 student
Projector120000 professional
Shirazi150000 student
Maverick250000 student
westlake320000 professional
DCO1300000 professional
COSMI3700000 student
scr5200000 professional
Fall2010NYTX5700000 student
myc210000000 student
Tardec2300000000 professional
hcbdy16000000000 student
ITRI18000000000 professional
sniknam19000000000 professional
EchoLake33000000000 professional
Luckbox110000000000 student
phm4152200000000000 professional
Meliksah2900000000000 professional
gtisyeprog4100000000000 student
rsm45000000000000 professional

Given that more than 900 run are recorded in PHM data, high-accuracy prediction by CNN can monitor tool wear and decrease the need of prior experience. Furthermore, we hope that this technology dealt with sensor signal enables more widespread use of deep learning in places where access to experts is difficult.

If you have questions about our work, contact us at: kidozh@gmail.com