enphase_solar_plot.R

library(httr)
library(jsonlite)
library(lubridate)
library(tidyverse)



########### config
#
# my info - replace this with your enphase api key, app id, user_id
api_key <- '80b9...9f87'
auth_url <- 'https://enlighten.enphaseenergy.com/app_user_auth/new?app_id=140...001'

user_id <- '4d...0a'

# enphase info: api url v2
api_url <- 'https://api.enphaseenergy.com/api/v2/systems'


# set from and to dates (365 days default, you can hardcode dates, too)

from_date <- format.Date(as.POSIXct(Sys.Date()-365), '%Y-%m-%d')
to_date <- format.Date(as.POSIXct(Sys.Date()), '%Y-%m-%d')


###################

# get the response, containing the system ID, and extract it
response <- GET(paste0(
  api_url,'?key=',api_key,'&user_id=',user_id
))

json_r <- fromJSON(  content(response, as = "text")   )

system_id <- json_r[[1]]$system_id[1]


# now, get the lifetime usage info, to find all day totals

energy_lifetime <- GET(paste0(
  api_url, '/', system_id,
  '/energy_lifetime?start_date=', from_date,
  '&end_date=', to_date,
  '&key=',api_key,'&user_id=',user_id
  
))

elr <- fromJSON(  content(energy_lifetime, as = "text")   )

# grab the first date and last date available, make a list

first_day <- format.Date( as.POSIXct( elr$start_date ), '%Y-%m-%d') 
last_day  <- format.Date( as.POSIXct( elr$meta$last_report_at, origin="1970-01-01" ), '%Y-%m-%d') 
date_list <- seq(as.Date(first_day), as.Date(last_day), by = "day")



# loop over the list of dates, with 7 second pause to prevent throttling

all_stats_single <- date_list[1:364] %>% 
  map_dfr(
    .f = function(d) {
      start <- as.numeric(as.POSIXct( d ))
      end   <- start + 24 * 3600 -1
      
      stats <- GET(paste0(
        api_url, '/', system_id,
        '/stats?start_at=', start,
        '&end_at=', end,
        '&key=',api_key,'&user_id=',user_id
      ))
      
      allstats <- fromJSON(  content(stats, as = "text")   )
      
      if(stats$status_code == 200 & class(allstats$intervals) == "data.frame") {
        message(d)
        ret <- allstats$intervals %>% mutate( date_string = d )
      } else {
        message(stats$status_code)
        message(class(allstats$intervals))
        ret <- NULL
      }
      
      Sys.sleep(7)
      ret
    }
  )

# done!

# now, since historic data does not change, no need to fetch it again, we store it:

write_csv(all_stats_single, "./panel_history.csv")

# Now reformat the pulled data for plotting, and make the plot

history <- all_stats_single %>%
  mutate(
    ts = as.POSIXct( end_at, origin="1970-01-01" ),
    day = wday(ts, label = TRUE, abbr = FALSE),
    week = paste(year(ts), 'week', str_pad(week(ts), 2, side = "left", pad = "0")),
    month = month(ts),
    actualday = as.Date(as.character(ts)),
    intraday = hour(ts) * 60 + minute(ts)
  ) %>% 
  group_by(actualday) %>% mutate(`Daily kWh` = sum(enwh)/1000) %>% 
  ungroup()


ggplot(history,
       aes(x = intraday, y = enwh))  +
  geom_line(aes(color = `Daily kWh`, group = actualday), size = 1, alpha = .7) +
  theme_void() +
  scale_color_viridis_c(option = "magma") +
  facet_wrap(. ~ week, nrow=6, strip.position="bottom") +
  theme(plot.margin=unit(c(1,1,1.5,1.2),"cm"), legend.position = "left") +
  theme(plot.title = element_text(margin=margin(0,0,30,0))) +
  theme(plot.background = element_rect(fill = "lightgrey")) +
  ggtitle("  Solar panel power production during the day")
  

The plot