/datathon2017

Optimising power using weather predictive models

Primary LanguageJupyter Notebook

datathon2017

Concatenate Datasets

  • Inputs: downloaded data from folder data/ with naming convention: "Chiller1_ConFlow_201706.csv" for June 2017
  • Change the second cell accordingly for the chiller number and datatype.
  • Combines data from May to November 2017 and puts into one file
  • Outputs: "data/Chiller1_ConFlow_full.csv"

Merge Datatypes

  • Inputs: full datasets generated by Concatenate Datasets
  • Keeps only important columns and concatenates by each minute values from Power/Temp/Conflow/Evaflow
  • Outputs: "data/Chiller1_full.csv"
  • Note: NEW Inputs includes external temperature data, saves as "data/Chiller1_full_ext.csv"

Prediction-METHOD.ipynb

  • Prediction and optimization for chiller (step 1 and 2)
  • Inputs: "Chiller1_full.csv"
  • Try different methods, evaluate best learning model on test set
  • Perform optimization, get optimized chiller variables

Prediction-METHOD-w cooling tower.ipynb

  • Prediction and optimization for cooling tower (step 3 and 4)
  • Inputs: "Chiller1_full_ext.csv"
  • Try different methods, evaluate best learning model on test set
  • Use optimized chiller variables before to predict power and optimize cooling tower variables