a small script for predicting solar output based on predicted clouds cover/ air visablity from various weather API's
OpenWeather,Climacel,Weather-API, Weatherbit, VisualCrossing, and Aeris. it fairly accurate depending on the accuracy of the weather API
you will need to install pysolar , lynx and bc
pysolar can be install vis pip install pysolar and lynx and bc through your basic apt install method.
it configured to work with collectd-mqtt that sends it to influxdb to generate graphs or graphs viable in collectd interface . it also generates a CSV so you can import it in other databases
you need to configure which ever API with your local corordinates and your weather API key/s . the lat/lon has two functions it configures pysolar with your local postion to calculate probable clear sky solar radation. and get the weather api cloud cover estimate to generate a probable masking effects..
the second file, to configure, is the panelconf. this helps calculates the power production. it supports up to 3 orientations and predicts solar panel potential production depending on the sun's position in relations to the panel groupings.. in there you enter your panel orientation , panel grouping wattage size, inverter effieciency and albedo effect ( ie: reflected light that still generates power output - to calculater your general albedo at midday measure you maxmium out put for a panel then point the panel away from the sun and see what your power output is. then divide the maxmium from the reflective to get a percentage ( whole number ie 0.05 = 5%)
it also has a simple tool called graph using gnuplot to generate a simple graph usage is : graph W_API.csv to see predicted power production from the weather api csv
example below
Graphing in Chronograf - you notice that different apis gave different cloudcover prediction as with air visablity hence why the graph is slightly different than the actual predicted cloud cover
for me weather API and weatherbit are the most consistanly accurate. but I live in a fairly remote regions that are not generally covered that well by most weather API services . you can generate higher accuracy by combining several api out puts into one or taking sample from several different gps location close to you IE 50km away from you and combining into one graph out put
the CSV are sent to /dev/shm to save on writing to and SD card since many will likely use in pi devices it also creates a 24 hr period the Maxmium Clear sky output, the estimated solar production and measured solar production this can vary by 0-5 % also the percentage of normal clear sky production and number of hrs of power production ( rounded to 1/2 hr)