TY  - BOOK
AU  - Vaishnav, Shubham
AU  - Desai, Kaushal A.
TI  - Machine Learning Based Instantaneous Cutting Force Models for End Milling Operation
U1  - 621 
PY  - 2019///
CY  - IIT Jodhpur
PB  - Department of Mechanical Engineering
KW  - Instantaneous Cutting Force Models
KW  - MTech Theses
KW  - Department of Mechanical Engineering
N2  - Cutting  force  is  a  fundamental  parameter  determining  productivity and  quality  ofthe  end  milling  operation. The  development  of  a  generic  cutting  force  model  forend  milling  operation  requires  a  large  number  of machining  experimentsfollowed  by  determining  empirical  relationships  between  process  parameterswith cutting force. This thesis presents a novel approach combining Mechanisticmodel and supervised ANN model to  predict  instantaneous  cutting  force  variationduring  the  end  milling  operation.  The  approach  proposes training  of  an  ANNmodel  using  datasets  generated  from the  Mechanistic  force  model  instead  ofusing experimental data. The proposed approach can generate a large number ofdatasets for the training of an ANN model without conducting rigorousexperimentation. The thesis work explores different ANN architecturesanddetermines  an  appropriate  network  by  comparing  a  set  of  performanceparameters.  A  series  of  end  milling experiments are conducted to examine theefficacy of the proposed approach in predicting cutting forces over a wide  rangeof  cutting  conditions.  Subsequently,  the  application  of  modern  machine  learningmodels  such  as Long Short Term Memory (LSTM) is explored to detect processparameters from the experimentally measured cutting force data. The studyuses a multi-layered LSTM to map measured cutting force data (input sequence)to a vector of end milling process parameters (fixed dimensionality labelledoutput vector). The dataset for the training of an LSTM model is generated usingthe Mechanistic force model. The testing of trained LSTM model is  carried  out  toidentify  process  parameters  from  experimentally  measured  cutting  force  data.  Ithas  been observed that the proposed model can detect cutting conditions quiteeffectively for CNC end milling.KeywordsMachining; End milling; Cutting forces; Mechanistic Model; Machine Learning

ER  -