Skip to content

Commit

Permalink
Big changes to performance._cleanDataFrameResults and performance.cal…
Browse files Browse the repository at this point in the history 
…culatePerformanceGencumsky.

Cumulativesky module scans and ground scans are now working!
  • Loading branch information
@cdeline
cdeline committed Sep 11, 2024
1 parent 78d2352 commit 013fa7c
Show file tree
Hide file tree
Showing 5 changed files with 426 additions and 299 deletions.
4 changes: 2 additions & 2 deletions bifacial_radiance/main.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3126,10 +3126,10 @@ def _printRow(analysisobj, key):
module_local = module
self.compiledResults = performance.calculatePerformanceGencumsky(results=self.results,
bifacialityfactor=module_local.bifi,
fillcleanedSensors=True)
fillcleanedSensors=False)

self.compiledResults.to_csv(os.path.join('results', 'Cumulative_Results.csv'),
float_format='%0.3f')
float_format='%0.3f', index=False)

self.trackerdict = trackerdict
return self.compiledResults
Expand Down
120 changes: 68 additions & 52 deletions bifacial_radiance/performance.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -144,36 +144,46 @@ def RMSE_abs(meas, model):
def _cleanDataFrameResults(mattype, rearMat, Wm2Front, Wm2Back,
fillcleanedSensors=False):

if len(rearMat) == 0:
# 2D array of values. if a row of rearMat is nan then it's single-sided and
# agriPV is true for that row
_matchAgriPV = ['sky', 'pole', 'tube', 'bar', '3267', '1540', '1540']
_match = ['sky', 'pole', 'tube', 'bar', 'ground', '3267', '1540']
matcharray = np.row_stack([_matchAgriPV if row.isna().all() else _match
for (n,row) in rearMat.iterrows()])
"""
if Wm2Front.size != Wm2Back.size:
agriPV = True
else:
agriPV = False

if agriPV:
matchers = ['sky', 'pole', 'tube', 'bar', '3267', '1540']
else:
agriPV = False
matchers = ['sky', 'pole', 'tube', 'bar', 'ground', '3267', '1540']
"""

maskfront = np.column_stack([mattype[col].str.contains('|'.join(matchers),
na=False) for col in
mattype])
maskfront = np.row_stack([row.str.contains('|'.join(matcharray[index]),
na=False) for (index, row) in
mattype.iterrows()])

Wm2Front[maskfront] = np.nan

if not agriPV:
maskback = np.column_stack([rearMat[col].str.contains('|'.join(matchers),
na=False) for col in
rearMat])
try:
maskback = np.row_stack([row.str.contains('|'.join(matcharray[index]),
na=False) for (index, row) in
rearMat.iterrows()])
Wm2Back[maskback] = np.nan
except AttributeError: # rearMat is empty
pass

# Filling Nans...
filledFront = Wm2Front.mean(axis=1)

if fillcleanedSensors:
filledBack = Wm2Back.copy().interpolate()
frontcopy = Wm2Front.copy().interpolate(axis=1)
filledBack = Wm2Back.copy().interpolate(axis=1)
else:
frontcopy = Wm2Front.copy()
filledBack = Wm2Back.copy() # interpolate()

return filledFront, filledBack
return filledFront, filledBack, frontcopy


def calculatePerformance(module, csvfile=None, results=None,
Expand Down Expand Up @@ -244,7 +254,7 @@ def calculatePerformance(module, csvfile=None, results=None,
if 'rearMat' in data:
rearMat = data['rearMat'].str.strip('[]').str.split(',', expand=True)
else:
rearMat = pd.DataFrame(None)
rearMat = pd.DataFrame([np.nan])
if 'timestamp' in data:
dfst['timestamp'] = data['timestamp']
if 'modNum' in data:
Expand All @@ -269,7 +279,7 @@ def calculatePerformance(module, csvfile=None, results=None,
rearMat = pd.DataFrame.from_dict(dict(zip(
results.index, results['rearMat']))).T
else:
rearMat = pd.DataFrame(None)
rearMat = pd.DataFrame([np.nan])

if 'timestamp' in results:
dfst['timestamp'] = results['timestamp']
Expand All @@ -284,7 +294,7 @@ def calculatePerformance(module, csvfile=None, results=None,
print("Data or file not passed. Ending arrayResults")
return

filledFront, filledBack = _cleanDataFrameResults(
filledFront, filledBack, frontcopy = _cleanDataFrameResults(
mattype, rearMat, Wm2Front, Wm2Back,
fillcleanedSensors=fillcleanedSensors)

Expand Down Expand Up @@ -324,7 +334,7 @@ def calculatePerformance(module, csvfile=None, results=None,

def calculatePerformanceGencumsky(csvfile=None, results=None,
bifacialityfactor=1.0,
fillcleanedSensors=True):
fillcleanedSensors=False):
'''
Compile calculate results for cumulative 1 axis tracking routine
Expand Down Expand Up @@ -359,26 +369,26 @@ def _dict2DF(df, key):
dfst = pd.DataFrame()

if csvfile is not None:
data = pd.read_csv(csvfile)
Wm2Front = data['Wm2Front'
results = pd.read_csv(csvfile)
Wm2Front = results['Wm2Front'
].str.strip('[]').str.split(',',
expand=True).astype(float)
Wm2Back = data['Wm2Back'
Wm2Back = results['Wm2Back'
].str.strip('[]').str.split(',',
expand=True).astype(float)
mattype = data['mattype'
mattype = results['mattype'
].str.strip('[]').str.split(',',
expand=True)
rearMat = data['rearMat'
rearMat = results['rearMat'
].str.strip('[]').str.split(',',
expand=True)

if 'modNum' in data:
dfst['module'] = data['modNum']
if 'rowNum' in data:
dfst['row'] = data['rowNum']
if 'sceneNum' in data:
dfst['sceneNum'] = data['sceneNum']
if 'modNum' in results:
dfst['module'] = results['modNum']
if 'rowNum' in results:
dfst['row'] = results['rowNum']
if 'sceneNum' in results:
dfst['sceneNum'] = results['sceneNum']
else:
if results is not None:
Wm2Front = _dict2DF(results, 'Wm2Front')
Expand All @@ -405,46 +415,52 @@ def _dict2DF(df, key):

# Data gets cleaned but need to maintain same number of sensors
# due to adding for the various tracker angles.
filledFront, filledBack = _cleanDataFrameResults(
filledFront, filledBack, frontcopy = _cleanDataFrameResults(
mattype, rearMat, Wm2Front, Wm2Back,
fillcleanedSensors=fillcleanedSensors)
cumFront = []
cumWM2 = []
cumBack = []
cumRow = []
cumMod = []
cumScene = []
Grear_mean = []
# Gfront_mean = []
POA_eff = []

# NOTE change 26.07.22 'row' -> 'rowNum' and 'mod' -> 'ModNumber
# NOTE change March 13 2024 ModNumber -> modNum
for rownum in results['rowNum'].unique():
for modnum in results['modNum'].unique():
mask = (results['rowNum'] == rownum) & (
results['modNum'] == modnum)
cumBack.append(list(filledBack[mask].sum(axis=0)))
cumFront.append(filledFront[mask].sum(axis=0))
cumRow.append(rownum)
cumMod.append(modnum)

# Maybe this would be faster by first doing the DF with the above,
# exploding the column and calculating.
POA_eff.append(list(
(filledBack[mask].apply(lambda x: x*bifacialityfactor +
filledFront[mask])).sum(axis=0)))
Grear_mean.append(filledBack[mask].sum(axis=0).mean())
# Gfront_mean.append(filledFront[mask].sum(axis=0).mean())

dfst = pd.DataFrame(zip(cumRow, cumMod, cumFront, cumBack, Grear_mean,
POA_eff), columns=('row', 'module', 'Gfront_mean',
'Wm2Back', 'Grear_mean',
#for rownum in results['rowNum'].unique():
# for modnum in results['modNum'].unique():
for i, df in results.groupby(['rowNum','modNum','sceneNum']):
#mask = (results['rowNum'] == rownum) & (
# results['modNum'] == modnum)
cumBack.append(list(filledBack.loc[df.index].sum(axis=0, min_count=1)))
cumFront.append(filledFront.loc[df.index].sum(axis=0, min_count=1))
cumWM2.append(list(frontcopy.loc[df.index].sum(axis=0, min_count=1)))
cumRow.append(i[0])
cumMod.append(i[1])
cumScene.append(i[2])


# Maybe this would be faster by first doing the DF with the above,
# exploding the column and calculating.
POA_eff.append(list(
(filledBack.loc[df.index].apply(lambda x: x*bifacialityfactor +
filledFront.loc[df.index])).sum(axis=0)))
Grear_mean.append(filledBack.loc[df.index].sum(axis=0, min_count=1).mean())
# Gfront_mean.append(filledFront[mask].sum(axis=0).mean())

dfst = pd.DataFrame(zip(cumRow, cumMod, cumScene, cumFront, cumWM2, cumBack, Grear_mean,
POA_eff), columns=('rowNum', 'modNum','sceneNum', 'Gfront_mean',
'Wm2Front', 'Wm2Back', 'Grear_mean',
'POA_eff'))

dfst['BGG'] = dfst['Grear_mean']*100*bifacialityfactor/dfst['Gfront_mean']

# Reordering columns
cols = ['row', 'module', 'BGG', 'Gfront_mean', 'Grear_mean', 'POA_eff',
'Wm2Back']
cols = ['rowNum', 'modNum','sceneNum', 'BGG', 'Gfront_mean', 'Grear_mean', 'POA_eff',
'Wm2Front','Wm2Back']
dfst = dfst[cols]

return dfst
Loading

0 comments on commit 013fa7c

Please sign in to comment.