def make_week_predict(opt, tstart, tstop, bs_cmds, tlm, db):
print "In make_week_predict"
# Try to make initial state0 from cmd line options
state0 = dict((x, getattr(opt, x))
for x in ('pitch', 'simpos', 'ccd_count', 'fep_count',
'vid_board', 'clocking', 'T_dea'))
print state0
state0.update({'tstart': tstart - 30,
'tstop': tstart,
'datestart': DateTime(tstart - 30).date,
'datestop': DateTime(tstart).date,
'q1': 0.0, 'q2': 0.0, 'q3': 0.0, 'q4': 1.0,
}
)
print state0
# If cmd lines options were not fully specified then get state0 as last
# cmd_state that starts within available telemetry. Update with the
# mean temperatures at the start of state0.
if None in state0.values():
state0 = cmd_states.get_state0(tlm['date'][-5], db,
datepar='datestart')
ok = ((tlm['date'] >= state0['tstart'] - 700) &
(tlm['date'] <= state0['tstart'] + 700))
state0.update({'T_dea': np.mean(tlm['1deamzt'][ok])})
# TEMPORARY HACK: core model doesn't actually support predictive
# active heater yet. Initial temperature determines active heater
# state for predictions now.
if state0['T_dea'] < 15:
state0['T_dea'] = 15.0
logger.debug('state0 at %s is\n%s' % (DateTime(state0['tstart']).date,
pformat(state0)))
# Get commands after end of state0 through first backstop command time
cmds_datestart = state0['datestop']
cmds_datestop = bs_cmds[0]['date']
# Get timeline load segments including state0 and beyond.
timeline_loads = db.fetchall("""SELECT * from timeline_loads
WHERE datestop > '%s'
and datestart < '%s'"""
% (cmds_datestart, cmds_datestop))
logger.info('Found {} timeline_loads after {}'.format(
len(timeline_loads), cmds_datestart))
# Get cmds since datestart within timeline_loads
db_cmds = cmd_states.get_cmds(cmds_datestart, db=db, update_db=False,
timeline_loads=timeline_loads)
# Delete non-load cmds that are within the backstop time span
# => Keep if timeline_id is not None or date < bs_cmds[0]['time']
db_cmds = [x for x in db_cmds if (x['timeline_id'] is not None or
x['time'] < bs_cmds[0]['time'])]
logger.info('Got %d cmds from database between %s and %s' %
(len(db_cmds), cmds_datestart, cmds_datestop))
# Get the commanded states from state0 through the end of backstop commands
states = cmd_states.get_states(state0, db_cmds + bs_cmds)
states[-1].datestop = bs_cmds[-1]['date']
states[-1].tstop = bs_cmds[-1]['time']
logger.info('Found %d commanded states from %s to %s' %
(len(states), states[0]['datestart'], states[-1]['datestop']))
# Create array of times at which to calculate DEA temps, then do it.
logger.info('Calculating DEA thermal model')
print state0
model = calc_model(opt.model_spec, states, state0['tstart'], tstop,
state0['T_dea'])
# Make the DEA limit check plots and data files
plt.rc("axes", labelsize=10, titlesize=12)
plt.rc("xtick", labelsize=10)
plt.rc("ytick", labelsize=10)
temps = {'dea': model.comp['1deamzt'].mvals}
plots = make_check_plots(opt, states, model.times, temps, tstart)
viols = make_viols(opt, states, model.times, temps)
write_states(opt, states)
write_temps(opt, model.times, temps)
return dict(opt=opt, states=states, times=model.times, temps=temps,
plots=plots, viols=viols)
def get_prediction_states(self, tbegin):
"""
Get the states used for the prediction.
Parameters
----------
tbegin : string
The starting date/time from which to obtain states for
prediction.
"""
"""
Get state0 as last cmd_state that starts within available telemetry.
The original logic in get_state0() is to return a state that
is absolutely, positively reliable by insisting that the
returned state is at least ``date_margin`` days old, where the
default is 10 days. That is too conservative (given the way
commanded states are actually managed) and not what is desired
here, which is a recent state from which to start thermal propagation.
Instead we supply ``date_margin=None`` so that get_state0 will
find the newest state consistent with the ``date`` criterion
and pcad_mode == 'NPNT'.
"""
state0 = cmd_states.get_state0(tbegin,
self.db,
datepar='datestart',
date_margin=None)
self.logger.debug('state0 at %s is\n%s' %
(DateTime(state0['tstart']).date, pformat(state0)))
# Get commands after end of state0 through first backstop command time
cmds_datestart = state0['datestop']
cmds_datestop = self.bs_cmds[0]['date']
# Get timeline load segments including state0 and beyond.
timeline_loads = self.db.fetchall("""SELECT * from timeline_loads
WHERE datestop >= '%s'
and datestart < '%s'""" %
(cmds_datestart, cmds_datestop))
self.logger.info('Found {} timeline_loads after {}'.format(
len(timeline_loads), cmds_datestart))
# Get cmds since datestart within timeline_loads
db_cmds = cmd_states.get_cmds(cmds_datestart,
db=self.db,
update_db=False,
timeline_loads=timeline_loads)
# Delete non-load cmds that are within the backstop time span
# => Keep if timeline_id is not None (if a normal load)
# or date < bs_cmds[0]['time']
# If this is an interrupt load, we don't want to include the end
# commands from the continuity load since not all of them will be valid,
# and we could end up evolving on states which would not be present in
# the load under review. However, once the load has been approved and is
# running / has run on the spacecraft, the states in the database will
# be correct, and we will want to include all relevant commands from the
# continuity load. To check for this, we find the current time and see
# the load under review is still in the future. If it is, we then treat
# this as an interrupt if requested, otherwise, we don't.
current_time = DateTime().secs
interrupt = self.interrupt and self.bs_cmds[0]["time"] > current_time
db_cmds = [
x for x in db_cmds
if ((x['timeline_id'] is not None and not interrupt)
or x['time'] < self.bs_cmds[0]['time'])
]
self.logger.info('Got %d cmds from database between %s and %s' %
(len(db_cmds), cmds_datestart, cmds_datestop))
# Get the commanded states from state0 through the end of backstop commands
states = cmd_states.get_states(state0, db_cmds + self.bs_cmds)
states[-1].datestop = self.bs_cmds[-1]['date']
states[-1].tstop = self.bs_cmds[-1]['time']
self.logger.info(
'Found %d commanded states from %s to %s' %
(len(states), states[0]['datestart'], states[-1]['datestop']))
return states, state0
print '%s not found' % table
for table in tables:
sqldef = file(table + '_def.sql').read()
db.execute(sqldef, commit=True)
datestart = state0['datestart']
timeline_loads = db.fetchall("""SELECT * from timeline_loads
WHERE datestop > '%s'""" % datestart)
timeline_loads_mod = timeline_loads.copy()[:-2]
timeline_loads_mod[-1].datestop = '2009:053:00:00:00.000'
print '=' * 40
print 'Processing with timeline_loads'
cmds = cmd_states.get_cmds(datestart, db=db, update_db=True, timeline_loads=timeline_loads)
states = cmd_states.get_states(state0, cmds)
print 'len(cmds) =',len(cmds)
cmd_states.update_states_db(states, db)
if 0:
print '=' * 40
print 'Processing with timeline_loads_mod'
cmds = cmd_states.get_cmds(datestart, db=db, update_db=True, timeline_loads=timeline_loads_mod)
states = cmd_states.get_states(state0, cmds)
print 'len(cmds) =',len(cmds)
print states[0]
print states[-1]
cmd_states.update_states_db(states, db)
print '=' * 40
def make_week_predict(opt, tstart, tstop, bs_cmds, tlm, db):
logger.debug("In make_week_predict")
# Try to make initial state0 from cmd line options
state0 = dict((x, getattr(opt, x))
for x in ('pitch', 'simpos', 'ccd_count', 'fep_count',
'vid_board', 'clocking', 'T_psmc','T_pin1at',
'dh_heater'))
state0.update({'tstart': tstart - 30,
'tstop': tstart,
'datestart': DateTime(tstart - 30).date,
'datestop': DateTime(tstart).date,
'q1': 0.0, 'q2': 0.0, 'q3': 0.0, 'q4': 1.0,
}
)
logger.debug("Completed state0 update")
# If cmd lines options were not fully specified then get state0 as last
# cmd_state that starts within available telemetry. Update with the
# mean temperatures at the start of state0.
if None in state0.values():
state0 = cmd_states.get_state0(tlm['date'][-5], db,
datepar='datestart')
ok = ((tlm['date'] >= state0['tstart'] - 700) &
(tlm['date'] <= state0['tstart'] + 700))
state0.update({'T_psmc': np.mean(tlm['1pdeaat'][ok])})
# state0.update({'T_pin1at': np.mean(tlm['1pin1at'][ok]) + 3.0 })
state0.update({'T_pin1at': np.mean(tlm['1pdeaat'][ok]) - 10.0 })
# TEMPORARY HACK: core model doesn't actually support predictive
# active heater yet. Initial temperature determines active heater
# state for predictions now.
if state0['T_psmc'] < 15:
state0['T_psmc'] = 15.0
logger.info('state0 at %s is\n%s' % (DateTime(state0['tstart']).date,
pformat(state0)))
# Get commands after end of state0 through first backstop command time
cmds_datestart = state0['datestop']
cmds_datestop = bs_cmds[0]['date']
# Get timeline load segments including state0 and beyond.
timeline_loads = db.fetchall("""SELECT * from timeline_loads
WHERE datestop > '%s'
and datestart < '%s'"""
% (cmds_datestart, cmds_datestop))
logger.info('Found {} timeline_loads after {}'.format(
len(timeline_loads), cmds_datestart))
# Get cmds since datestart within timeline_loads
db_cmds = cmd_states.get_cmds(cmds_datestart, db=db, update_db=False,
timeline_loads=timeline_loads)
# Delete non-load cmds that are within the backstop time span
# => Keep if timeline_id is not None or date < bs_cmds[0]['time']
db_cmds = [x for x in db_cmds if (x['timeline_id'] is not None or
x['time'] < bs_cmds[0]['time'])]
logger.info('Got %d cmds from database between %s and %s' %
(len(db_cmds), cmds_datestart, cmds_datestop))
# Get the commanded states from state0 through the end of backstop commands
states = cmd_states.get_states(state0, db_cmds + bs_cmds)
states[-1].datestop = bs_cmds[-1]['date']
states[-1].tstop = bs_cmds[-1]['time']
logger.info('Found %d commanded states from %s to %s' %
(len(states), states[0]['datestart'], states[-1]['datestop']))
# htrbfn='/home/edgar/acis/thermal_models/dhheater_history/dahtbon_history.rdb'
htrbfn='dahtbon_history.rdb'
logger.info('Reading file of dahtrb commands from file %s' % htrbfn)
htrb=Ska.Table.read_ascii_table(htrbfn,headerrow=2,headertype='rdb')
dh_heater_times=Chandra.Time.date2secs(htrb['time'])
dh_heater=htrb['dahtbon'].astype(bool)
# Create array of times at which to calculate PSMC temps, then do it.
logger.info('Calculating PSMC thermal model')
logger.info('state0 at start of calc is\n%s' % (pformat(state0)))
model = calc_model(opt.model_spec, states, state0['tstart'], tstop,
state0['T_psmc'],None,state0['T_pin1at'], None,
dh_heater,dh_heater_times)
# Make the PSMC limit check plots and data files
plt.rc("axes", labelsize=10, titlesize=12)
plt.rc("xtick", labelsize=10)
plt.rc("ytick", labelsize=10)
temps = dict(psmc=model.comp['1pdeaat'].mvals,pin=model.comp['pin1at'].mvals)
plots = make_check_plots(opt, states, model.times, temps, tstart)
viols = make_viols(opt, states, model.times, temps)
write_states(opt, states)
write_temps(opt, model.times, temps)
return dict(opt=opt, states=states, times=model.times, temps=temps,
plots=plots, viols=viols)
def get_prediction_states(self, tbegin):
"""
Get the states used for the prediction.
Parameters
----------
tbegin : string
The starting date/time from which to obtain states for
prediction.
"""
"""
Get state0 as last cmd_state that starts within available telemetry.
The original logic in get_state0() is to return a state that
is absolutely, positively reliable by insisting that the
returned state is at least ``date_margin`` days old, where the
default is 10 days. That is too conservative (given the way
commanded states are actually managed) and not what is desired
here, which is a recent state from which to start thermal propagation.
Instead we supply ``date_margin=None`` so that get_state0 will
find the newest state consistent with the ``date`` criterion
and pcad_mode == 'NPNT'.
"""
state0 = cmd_states.get_state0(tbegin, self.db, datepar='datestart',
date_margin=None)
self.logger.debug('state0 at %s is\n%s' % (DateTime(state0['tstart']).date,
pformat(state0)))
# Get commands after end of state0 through first backstop command time
cmds_datestart = state0['datestop']
cmds_datestop = self.bs_cmds[0]['date']
# Get timeline load segments including state0 and beyond.
timeline_loads = self.db.fetchall("""SELECT * from timeline_loads
WHERE datestop >= '%s'
and datestart < '%s'"""
% (cmds_datestart, cmds_datestop))
self.logger.info('Found {} timeline_loads after {}'.format(
len(timeline_loads), cmds_datestart))
# Get cmds since datestart within timeline_loads
db_cmds = cmd_states.get_cmds(cmds_datestart, db=self.db, update_db=False,
timeline_loads=timeline_loads)
# Delete non-load cmds that are within the backstop time span
# => Keep if timeline_id is not None (if a normal load)
# or date < bs_cmds[0]['time']
# If this is an interrupt load, we don't want to include the end
# commands from the continuity load since not all of them will be valid,
# and we could end up evolving on states which would not be present in
# the load under review. However, once the load has been approved and is
# running / has run on the spacecraft, the states in the database will
# be correct, and we will want to include all relevant commands from the
# continuity load. To check for this, we find the current time and see
# the load under review is still in the future. If it is, we then treat
# this as an interrupt if requested, otherwise, we don't.
current_time = DateTime().secs
interrupt = self.interrupt and self.bs_cmds[0]["time"] > current_time
db_cmds = [x for x in db_cmds
if ((x['timeline_id'] is not None and not interrupt) or
x['time'] < self.bs_cmds[0]['time'])]
self.logger.info('Got %d cmds from database between %s and %s' %
(len(db_cmds), cmds_datestart, cmds_datestop))
# Get the commanded states from state0 through the end of backstop commands
states = cmd_states.get_states(state0, db_cmds + self.bs_cmds)
states[-1].datestop = self.bs_cmds[-1]['date']
states[-1].tstop = self.bs_cmds[-1]['time']
self.logger.info('Found %d commanded states from %s to %s' %
(len(states), states[0]['datestart'],
states[-1]['datestop']))
return states, state0
def make_week_predict(opt, tstart, tstop, bs_cmds, tlm, db):
# Try to make initial state0 from cmd line options
state0 = dict((x, getattr(opt, x)) for x in ("pitch", "simpos", "power", "T_dea", "T_pin"))
state0.update(
{
"tstart": tstart - 30,
"tstop": tstart,
"datestart": DateTime(tstart - 30).date,
"datestop": DateTime(tstart).date,
}
)
# If cmd lines options were not fully specified then get state0 as last
# cmd_state that starts within available telemetry. Update with the
# mean temperatures at the start of state0.
if None in state0.values():
state0 = cmd_states.get_state0(tlm[-5].date, db, datepar="datestart")
ok = (tlm.date >= state0["tstart"] - 150) & (tlm.date <= state0["tstart"] + 150)
state0.update({"T_dea": np.mean(tlm["1pdeaat"][ok]), "T_pin": np.mean(tlm["1pin1at"][ok])})
logger.debug("state0 at %s is\n%s" % (DateTime(state0["tstart"]).date, pformat(state0)))
if opt.old_cmds:
cmds_datestart = DateTime(state0["tstop"]).date
cmds_datestop = DateTime(bs_cmds[0]["time"]).date
db_cmds = cmd_states.get_cmds(cmds_datestart, cmds_datestop, db)
else:
# Get the commands after end of state0 through first backstop command time
cmds_datestart = state0["datestop"]
cmds_datestop = bs_cmds[0]["date"] # *was* DateTime(bs_cmds[0]['time']).date
# Get timeline load segments including state0 and beyond.
timeline_loads = db.fetchall(
"""SELECT * from timeline_loads
WHERE datestop > '%s' and datestart < '%s'"""
% (cmds_datestart, cmds_datestop)
)
logger.info("Found %s timeline_loads after %s" % (len(timeline_loads), cmds_datestart))
# Get cmds since datestart within timeline_loads
db_cmds = cmd_states.get_cmds(cmds_datestart, db=db, update_db=False, timeline_loads=timeline_loads)
# Delete non-load cmds that are within the backstop time span
# => Keep if timeline_id is not None or date < bs_cmds[0]['time']
db_cmds = [x for x in db_cmds if (x["timeline_id"] is not None or x["time"] < bs_cmds[0]["time"])]
logger.info("Got %d cmds from database between %s and %s" % (len(db_cmds), cmds_datestart, cmds_datestop))
# Get the commanded states from state0 through the end of the backstop commands
states = cmd_states.get_states(state0, db_cmds + bs_cmds)
states[-1].datestop = bs_cmds[-1]["date"]
states[-1].tstop = bs_cmds[-1]["time"]
logger.info(
"Found %d commanded states from %s to %s" % (len(states), states[0]["datestart"], states[-1]["datestop"])
)
# Add power column based on ACIS commanding in states
states = Ska.Numpy.add_column(states, "power", get_power(states))
# Create array of times at which to calculate PSMC temperatures, then do it.
times = np.arange(state0["tstart"], tstop, opt.dt)
logger.info("Calculating PSMC thermal model")
T_pin, T_dea = twodof.calc_twodof_model(states, state0["T_pin"], state0["T_dea"], times, characteristics.model_par)
# Make the PSMC limit check plots and data files
plt.rc("axes", labelsize=10, titlesize=12)
plt.rc("xtick", labelsize=10)
plt.rc("ytick", labelsize=10)
temps = dict(dea=T_dea, pin=T_pin)
plots = make_check_plots(opt, states, times, temps, tstart)
viols = make_viols(opt, states, times, temps)
write_states(opt, states)
write_temps(opt, times, temps)
return dict(opt=opt, states=states, times=times, temps=temps, plots=plots, viols=viols)