Abstract
With the gradually stringent emission regulation, it is in needs for new methods to optimize conventional combustion engines in terms of exhaust emission, working performance, and abnormal combustion. In response, cheaper, more reliable, and more responsive engine control schemes based on the ion current detection method appear, and a challenge in the method is to retrieve combustion information involving max-pressure and knock condition from ion current signals. To cope with the challenge, we develop an image understanding based FusionNet model that transforms ion current signals to spectrograms and takes the spectrograms to predict max-pressure and knock condition simultaneously. As a result, FusionNet can predict the crank angle and the numerical value of the max-pressure of samples in the test set, with an average of the Mean Squared Error valuing 6.802 and 7.142 respectively. Moreover, FusionNet can predict pressure oscillation related to the knock condition with an average of the Cosine Similarity valuing 0.00932, and apply the detection result of the oscillation to predict the knock condition of samples in the test set, with an average of the F1-Score valuing 0.92684.
| Original language | English |
|---|---|
| Pages (from-to) | 1080-1089 |
| Number of pages | 10 |
| Journal | Fuel |
| Volume | 253 |
| DOIs | |
| Publication status | Published - Oct 1 2019 |
Fingerprint
Keywords
- Cylinder pressure
- Deep neural network
- Image understanding
- Ion current detection
- Knock detection
- Signal processing
ASJC Scopus subject areas
- Chemical Engineering(all)
- Fuel Technology
- Energy Engineering and Power Technology
- Organic Chemistry
Cite this
An image understanding based model with ion current signals for predicting combustion information. / Deng, Yu; Gao, Zhongquan; Tomita, Eiji; Wen, Yang.
In: Fuel, Vol. 253, 01.10.2019, p. 1080-1089.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - An image understanding based model with ion current signals for predicting combustion information
AU - Deng, Yu
AU - Gao, Zhongquan
AU - Tomita, Eiji
AU - Wen, Yang
PY - 2019/10/1
Y1 - 2019/10/1
N2 - With the gradually stringent emission regulation, it is in needs for new methods to optimize conventional combustion engines in terms of exhaust emission, working performance, and abnormal combustion. In response, cheaper, more reliable, and more responsive engine control schemes based on the ion current detection method appear, and a challenge in the method is to retrieve combustion information involving max-pressure and knock condition from ion current signals. To cope with the challenge, we develop an image understanding based FusionNet model that transforms ion current signals to spectrograms and takes the spectrograms to predict max-pressure and knock condition simultaneously. As a result, FusionNet can predict the crank angle and the numerical value of the max-pressure of samples in the test set, with an average of the Mean Squared Error valuing 6.802 and 7.142 respectively. Moreover, FusionNet can predict pressure oscillation related to the knock condition with an average of the Cosine Similarity valuing 0.00932, and apply the detection result of the oscillation to predict the knock condition of samples in the test set, with an average of the F1-Score valuing 0.92684.
AB - With the gradually stringent emission regulation, it is in needs for new methods to optimize conventional combustion engines in terms of exhaust emission, working performance, and abnormal combustion. In response, cheaper, more reliable, and more responsive engine control schemes based on the ion current detection method appear, and a challenge in the method is to retrieve combustion information involving max-pressure and knock condition from ion current signals. To cope with the challenge, we develop an image understanding based FusionNet model that transforms ion current signals to spectrograms and takes the spectrograms to predict max-pressure and knock condition simultaneously. As a result, FusionNet can predict the crank angle and the numerical value of the max-pressure of samples in the test set, with an average of the Mean Squared Error valuing 6.802 and 7.142 respectively. Moreover, FusionNet can predict pressure oscillation related to the knock condition with an average of the Cosine Similarity valuing 0.00932, and apply the detection result of the oscillation to predict the knock condition of samples in the test set, with an average of the F1-Score valuing 0.92684.
KW - Cylinder pressure
KW - Deep neural network
KW - Image understanding
KW - Ion current detection
KW - Knock detection
KW - Signal processing
UR - http://www.scopus.com/inward/record.url?scp=85066074028&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85066074028&partnerID=8YFLogxK
U2 - 10.1016/j.fuel.2019.04.162
DO - 10.1016/j.fuel.2019.04.162
M3 - Article
AN - SCOPUS:85066074028
VL - 253
SP - 1080
EP - 1089
JO - Fuel
JF - Fuel
SN - 0016-2361
ER -