solar-data-tools 0.4.0

Tools for performing common tasks on solar PV data signals

solar-data-tools

Tools for performing common tasks on solar PV data signals. These tasks include finding clear days in a data set, common data transforms, and fixing time stamp issues. These tools are designed to be automatic and require little if any input from the user. Libraries are included to help with data IO and plotting as well.

There is close integration between this repository and the Statistical Clear Sky repository, which provides a "clear sky model" of system output, given only measured power as an input.

See notebooks folder for examples.

Setup

Recommended: Set up conda environment with provided .yml file

Updated September 2020

We recommend seting up a fresh Python virutal environment in which to use solar-data-tools. We recommend using the Conda package management system, and creating an environment with the environment configuration file named pvi-user.yml, provided in the top level of this repository. This will install the statistical-clear-sky package as well.

Please see the Conda documentation page, "Creating an environment from an environment.yml file" for more information.

Installing this project as PIP package

$ pip install solar-data-tools

As of March 6, 2019, it fails because scs package installed as a dependency of cxvpy expects numpy to be already installed. scs issue 85 says, it is fixed. However, it doesn't seem to be reflected in its pip package. Also, cvxpy doesn't work with numpy version less than 1.16. As a work around, install numpy separatly first and then install this package. i.e.

$ pip install 'numpy>=1.16'
$ pip install statistical-clear-sky

Solvers

By default, ECOS solver is used, which is supported by cvxpy because it is Open Source.

However, it is found that Mosek solver is more stable. Thus, we encourage you to install it separately as below and obtain the license on your own.

• mosek - For using MOSEK solver.

  $ pip install -f https://download.mosek.com/stable/wheel/index.html Mosek
  

Installing this project as Anaconda package

$ conda install -c slacgismo solar-data-tools

If you are using Anaconda, the problem described in the section for PIP package above doesn't occur since numpy is already installed. And during solar-data-tools installation, numpy is upgraded above 1.16.

Solvers

By default, ECOS solver is used, which is supported by cvxpy because it is Open Source.

However, it is found that Mosek solver is more stable. Thus, we encourage you to install it separately as below and obtain the license on your own.

• mosek - For using MOSEK solver.

  $ conda install -c mosek mosek
  

Using this project by cloning this GIT repository

From a fresh python environment, run the following from the base project folder:

$ pip install -r requirements.txt

As of March 6, 2019, it fails because scs package installed as a dependency of cxvpy expects numpy to be already installed. scs issue 85 says, it is fixed. However, it doesn't seem to be reflected in its pip package. Also, cvxpy doesn't work with numpy version less than 1.16. As a work around, install numpy separatly first and then install this package. i.e.

$ pip install 'numpy>=1.16'
$ pip install -r requirements.txt

To test that everything is working correctly, launch

$ jupyter notebook

and run the two notebooks in the notebooks/ folder.

Usage

Clear Day Detection

This algorithm estimates the clear days in a data set two ways and then combines the estimates for the final estimations. The first estimate is based on the "smoothness" of each daily power signal. The second estimate is based on the seasonally adjusted daily energy output of the system.

import numpy as np
from solardatatools.clear_day_detection import find_clear_days
from solardatatools.data_transforms import make_2d
from solardatatools.dataio import get_pvdaq_data

pv_system_data = get_pvdaq_data(sysid=35, api_key='DEMO_KEY', year=[2011, 2012, 2013])

power_signals_d = make_2d(pv_system_data, key='dc_power')

clear_days = find_clear_days(power_signals_d)

Time Shift Detection and Fixing

This algorithm determines if the time stamps provided with the data have "shifted" at any point and then corrects the shift if found. These shifts can often be caused by incorrect handling of daylight savings time, but can come from other sources as well.

from solardatatools.data_transforms import fix_time_shifts, make_2d
from solardatatools.dataio import get_pvdaq_data
from solardatatools.plotting import plot_2d

pv_system_data = get_pvdaq_data(sysid=1199, year=[2015, 2016, 2017], api_key='DEMO_KEY')

power_signals_d = make_2d(pv_system_data, key='dc_power')

fixed_power_signals_d, time_shift_days_indices_ixs = fix_time_shifts(
    power_signals_d, return_ixs=True)

Versioning

We use Semantic Versioning for versioning. For the versions available, see the tags on this repository.

Authors

• Bennet Meyers - Initial work and Main research work - Bennet Meyers GitHub

See also the list of contributors who participated in this project.

License

This project is licensed under the BSD 2-Clause License - see the LICENSE file for details