def dayrun(start, cluster):
'''
Simulation of a single day according to start state 'start'
and the stochastics associated with 'cluster'.
'''
# script ##########################################################
# we set the default daycheck at False for the first run and loop
# creating days while False, meaning while the simulated day does
# not correspond to the agreed-on rules in check().
# ----->>> The check is currently not effectively implemented.
daycheck = False
end = datetime.datetime.utcnow() + datetime.timedelta(seconds=10)
# define the corresponding MCSA object from stats.py depicting a
# Monte Carlo Survival Analysis object.
SA = stats.MCSA(cluster)
# and then keep simulating a day until True
while daycheck == False:
# set start state conditions
tbin = 0
occs = np.zeros(144, dtype=int)
occs[0] = start
# occupancy data from survey are given per 30 min, so we need to know in which of 48 bins to look for data:
t48 = np.array(sorted(list(range(1, 49)) * 3))
dt = SA.duration(
start, t48[0]
) # get duration of current state at start time (4am)
# and loop sequentially transition and duration functions
while tbin < 143:
tbin += 1
if dt == 0: # previous state duration has ended
occs[tbin] = SA.transition(
occs[tbin - 1],
t48[tbin]) # find most probable next state
dt = SA.duration(
occs[tbin], t48[tbin]
) - 1 # restart duration counter for new state
# -1 is necessary, as the occupancy state already started
else:
occs[tbin] = occs[tbin - 1] # maintain current state
dt += -1 # count down duration of current state
# whereafer we control if this day is ok
daycheck = check(
occs
) # ----->>> The check is currently not effectively implemented, always TRUE!
# and we include a break if the while-loop takes to long until
# check()-conditions are fulfilled.
if datetime.datetime.utcnow() > end:
break
# ouput ###########################################################
# return occupants array if daycheck is ok
return occs
def dayrun(start, cluster):
'''
Simulation of a single day according to start state 'start'
and the stochastics stored in cluster 'BxDict'and daytype 'Bx'.
'''
# script ##########################################################
# we set the default daycheck at False for the first run and loop
# creating days while False, meaning while the simulated day does
# not correspond to the agreed-on rules in check().
daycheck = False
end = datetime.datetime.utcnow() + datetime.timedelta(seconds=10)
# defin the corresponding MCSA object from stats.py depicting a
# Monte Carlo Survival Analysis object.
SA = stats.MCSA(cluster)
# and then keep simulating a day until True
while daycheck == False:
# set start state conditions
tbin = 0
occs = np.zeros(144, dtype=int)
occs[0] = start
t48 = np.array(sorted(list(range(1, 49)) * 3))
dt = SA.duration(start, t48[0])
# and loop sequentially transition and duration functions
while tbin < 143:
tbin += 1
if dt == 0:
occs[tbin] = SA.transition(occs[tbin - 1], t48[tbin])
dt = SA.duration(occs[tbin], t48[tbin]) - 1
# -1 is necessary, as the occupancy state already started
else:
occs[tbin] = occs[tbin - 1]
dt += -1
# whereafer we control if this day is ok
daycheck = check(occs)
# and we include a break if the while-loop takes to long until
# check()-conditions are fulfilled.
if datetime.datetime.utcnow() > end:
break
# ouput ###########################################################
# return occupants array if daycheck is ok according to Bx
return occs
def __occupancy__(self, min_form=True, min_time=False):
'''
Simulation of occupancy profiles for each occupant based on their clusters.
- A weekday, saturday and sunday are simulated, after which a typical week is created and repeated for the entire simulation period.
- The starting day of the year is taken into account.
- The occupancy profiles start at 4:00 AM, in accordance with the used survey data. (later shifted, see roundUp())
'''
def check(occday,
min_form=True,
min_time=False): # -->> this check is not effective !!!
'''
We set a check which becomes True if the simulated day behaves
according to the cluster, as a safety measure for impossible
solutions.
---->>> This check is not effectively implemented for the moment !!!
'''
# script 1 ########################################################
# First we check if the simulated occ-chain has the same shape
shape = True # --->>> Check not effective, shape=True always !!!!!!!!!!!
if min_form:
location = np.zeros(1, dtype=int)
reduction = occday[0] * np.ones(1, dtype=int)
for i in range(len(occday) - 1):
if occday[i + 1] != occday[i]:
location = np.append(location, i + 1)
reduction = np.append(reduction, occday[i + 1])
# shape = np.array_equal(reduction, RED) --->>> RED not available any more!!
# script 2 ########################################################
# And second we see if the chain has no sub-30 min differences
length = True # --->>> Check not effective, length=True always !!!!!!!!!!!
if min_time: # ------>>> min_time=False by default !!
minlength = 99
for i in location:
j = 0
while occday[i +
j] == occday[i] and i + j < len(occday) - 1:
j = j + 1
if j < minlength:
minlength = j
# and we neglect the very short presences of 20 min or less
length = not minlength < 3
# output ##########################################################
# both have to be true to allow continuation, and we return boolean
return shape and length # --->>> Check not effective, always True !!!!!!!!!!!
def dayrun(start, cluster):
'''
Simulation of a single day according to start state 'start'
and the stochastics associated with 'cluster'.
'''
# script ##########################################################
# we set the default daycheck at False for the first run and loop
# creating days while False, meaning while the simulated day does
# not correspond to the agreed-on rules in check().
# ----->>> The check is currently not effectively implemented.
daycheck = False
end = datetime.datetime.utcnow() + datetime.timedelta(seconds=10)
# define the corresponding MCSA object from stats.py depicting a
# Monte Carlo Survival Analysis object.
SA = stats.MCSA(cluster)
# and then keep simulating a day until True
while daycheck == False:
# set start state conditions
tbin = 0
occs = np.zeros(144, dtype=int)
occs[0] = start
# occupancy data from survey are given per 30 min, so we need to know in which of 48 bins to look for data:
t48 = np.array(sorted(list(range(1, 49)) * 3))
dt = SA.duration(
start, t48[0]
) # get duration of current state at start time (4am)
# and loop sequentially transition and duration functions
while tbin < 143:
tbin += 1
if dt == 0: # previous state duration has ended
occs[tbin] = SA.transition(
occs[tbin - 1],
t48[tbin]) # find most probable next state
dt = SA.duration(
occs[tbin], t48[tbin]
) - 1 # restart duration counter for new state
# -1 is necessary, as the occupancy state already started
else:
occs[tbin] = occs[tbin - 1] # maintain current state
dt += -1 # count down duration of current state
# whereafer we control if this day is ok
daycheck = check(
occs
) # ----->>> The check is currently not effectively implemented, always TRUE!
# and we include a break if the while-loop takes to long until
# check()-conditions are fulfilled.
if datetime.datetime.utcnow() > end:
break
# ouput ###########################################################
# return occupants array if daycheck is ok
return occs
def merge(occ):
'''
Merge the occupancy profiles of all household members to a single
profile denoting the most active state of all members.
'''
# scirpt ##########################################################
# We start defining an array of correct length filled with the
# least active state and loop to see if more-active people are
# present at the depicted moment.
occs = int(3) * np.ones(len(occ[0]))
for member in occ:
for to in range(len(member)):
if member[to] < occs[to]:
occs[to] = member[to]
# ouput ###########################################################
# return the merge occupancy states
return occs
# script ##############################################################
# We change the directory to to location where the data is stored,
# and create a typical week.
cdir = os.getcwd()
occ_week = [] #typical week
for member in self.clustersList:
week = [] # initiate empty week
SA = stats.MCSA(member['wkdy'])
startstate = SA.startstate(
) # random starting state of week, depending on cluster
for i in range(5):
week = np.append(week, dayrun(startstate, member['wkdy']))
startstate = week[-1]
week = np.append(week, dayrun(week[-1], member['sat']))
week = np.append(week, dayrun(week[-1], member['son']))
occ_week.append(week)
# A merge occupancy is created depicted the most active state of all
# household members, later-on used for set-point temperatures and hot water tappings.
occ_merg = merge(occ_week)
# and combine the weekly occupancy states for the entire year by
# repeating them every week and correcting for the first day of year and stop time,
# including for the merged occupancy.
bins = 144 # number of datapoints in one day
tstart = bins * (self.dow[0]) # need to sart on correct day of week
tstop = tstart + bins * (
self.nday
) + 1 # need nday in total, plus one step (for IDEAS simulations)
occ_year = []
for line in range(len(occ_week)): # per separate member
#repeat week more than enough times (54), take days needed to have correct nday and start day
occ_year.append(np.tile(occ_week, 54)[line][tstart:tstop])
occ_merged = []
occ_merged.append(np.tile(occ_merg, 54)[tstart:tstop])
# output ##############################################################
# chdir back to original and return the occupancy states to the class
# object.
os.chdir(cdir)
self.occ = occ_year
self.occ_m = occ_merged
# and print statements
presence = [to for to in self.occ_m[0] if to < 2]
hours = len(presence) / 6.
print(' - Total presence time is {0:.1f} out of {1} hours'.format(
hours, self.nday * 24))
print('\tbeing {:.1f} percent)'.format(hours * 100 / (self.nday * 24)))
return None