Proceedings of the
9th International Conference of Asian Society for Precision Engineering and Nanotechnology (ASPEN2022)
15 – 18 November 2022, Singapore
doi:10.3850/978-981-18-6021-8_OR-06-0260
Energy Consumption Estimation of Machine Tool Using Machine Learning
1School of Mechanical Engineering, Yonsei University, Republic of Korea
2Digital Transformation R&D Department, Korea Institute of Industrial Technology (KITECH), Republic of Korea
ABSTRACT
In the manufacturing industry, interest in energy consumption reduction of machine tools is increasing for cost reduction and sustainable production. To improve the energy efficiency of machine tools, energy estimation models have been studied. In particular, physics-based models that reflect machine tool dynamics are widely used to enhance energy estimation accuracy. However, such conventional models require an understanding of complex physical behaviors and many computational loads for the various components of machine tools. In this paper, we studied machine learning algorithms to estimate the energy consumption of machine tools. Also, we proposed energy estimation models based on such algorithms without the physical model. The proposed models were built using different algorithms such as LSTM (long short-term memory), GRU (gated recurrent unit), and 1D CNN (convolution neural network), respectively. These algorithms, which are mainly applied to time-series data, were used to estimate time-varying energy consumption. Training and test datasets were collected by performing machining experiments on the commercial machine tool. These datasets include power consumption and CNC (computerized numerical control) data measured by power meter and CNC controller, respectively. The proposed models, which were learned using the training datasets, estimate the energy consumption of each machine tool component from the input NC code. To calculate the estimation accuracy of the proposed models, estimated energy was compared with the measured energy. Through these comparison results, we showed the most suitable machine learning algorithm for energy estimation of machine tools.
Keywords: Energy efficiency, Machining process, Machine learning algorithm
1School of Mechanical Engineering, Yonsei University, Republic of Korea
2Digital Transformation R&D Department, Korea Institute of Industrial Technology (KITECH), Republic of Korea
ABSTRACT
In the manufacturing industry, interest in energy consumption reduction of machine tools is increasing for cost reduction and sustainable production. To improve the energy efficiency of machine tools, energy estimation models have been studied. In particular, physics-based models that reflect machine tool dynamics are widely used to enhance energy estimation accuracy. However, such conventional models require an understanding of complex physical behaviors and many computational loads for the various components of machine tools. In this paper, we studied machine learning algorithms to estimate the energy consumption of machine tools. Also, we proposed energy estimation models based on such algorithms without the physical model. The proposed models were built using different algorithms such as LSTM (long short-term memory), GRU (gated recurrent unit), and 1D CNN (convolution neural network), respectively. These algorithms, which are mainly applied to time-series data, were used to estimate time-varying energy consumption. Training and test datasets were collected by performing machining experiments on the commercial machine tool. These datasets include power consumption and CNC (computerized numerical control) data measured by power meter and CNC controller, respectively. The proposed models, which were learned using the training datasets, estimate the energy consumption of each machine tool component from the input NC code. To calculate the estimation accuracy of the proposed models, estimated energy was compared with the measured energy. Through these comparison results, we showed the most suitable machine learning algorithm for energy estimation of machine tools.
Keywords: Energy efficiency, Machining process, Machine learning algorithm
