def testGoodVehicle(self, plot_results=False):
logging.getLogger().setLevel(logging.DEBUG)
mdl = goodVehicle()
experiment = Experiment(mdl)
mdl = experiment.run()
self.assertTrue("cycle" in mdl, 'No result "cycle" in Model: %s' % mdl)
print("DRIVEABILITY: \n%s" % experiment.driveability_report())
cycle = mdl["cycle"]
gears = cycle["gears"]
print(
"G1: %s, G2: %s"
% (np.count_nonzero(gears == 1), np.count_nonzero(gears == 2))
)
self._compare_exp_results(cycle, "goodveh", self.run_comparison)
if plot_results:
print(mdl["cycle"])
# print([get_wltc_data()['classes']['class3b']['cycle'][k] for k in mdl['cycle']['driveability_issues'].keys()])
self._plotResults(mdl)
np.set_printoptions(edgeitems=16)
# print(driveability_issues)
# print(v_max)
# results['target'] = []; print(results)
plt.show()
def test_two_ramps_smoke_test(self):
mdl = goodVehicle()
V = np.hstack((np.r_[0:100:2], np.r_[100:0:-2]))
mdl['cycle_run'] = {'v_target': V}
experiment = Experiment(mdl)
mdl = experiment.run()
def test_badCycle(self):
mdl = goodVehicle()
mdl["cycle"] = 1
with self.assertRaisesRegex(
PandasError, "DataFrame constructor not properly called"):
experiment = Experiment(mdl)
mdl = experiment.run()
def test_two_ramps_smoke_test(self):
mdl = goodVehicle()
V = np.hstack((np.r_[0:100:2], np.r_[100:0:-2]))
mdl['cycle_run'] = {'v_target': V}
experiment = Experiment(mdl)
mdl = experiment.run()
def test_badCycle(self):
mdl = goodVehicle()
mdl['params'] = params = {}
mdl['cycle_run'] = 1
with assertRaisesRegex(self, PandasError, 'DataFrame constructor not properly called'):
experiment = Experiment(mdl)
mdl = experiment.run()
def test_two_ramps_smoke_test(self):
mdl = goodVehicle()
mdl = datamodel.upd_resistance_coeffs_regression_curves(mdl)
V = np.hstack((np.r_[0:100:2], np.r_[98:0:-2]))
mdl["cycle"] = {"v_target": V}
experiment = Experiment(mdl)
mdl = experiment.run()
def test_badCycle(self):
mdl = goodVehicle()
mdl['params'] = params = {}
mdl['cycle_run'] = 1
with assertRaisesRegex(self, PandasError,
'DataFrame constructor not properly called'):
experiment = Experiment(mdl)
mdl = experiment.run()
def _run_the_experiments(
transplant_original_gears=False,
plot_results=False,
compare_results=False,
encoding="ISO-8859-1",
):
# rated_power,kerb_mass,rated_speed,idling_speed,test_mass,no_of_gears,ndv_1,ndv_2,ndv_3,ndv_4,ndv_5,ndv_6,ndv_7,ID_cat,user_def_driv_res_coeff,user_def_power_curve,f0,f1,f2,Comment
# 0 5 10 15 19
df = read_vehicle_data()
for (ix, row) in df.iterrows():
veh_num = ix
model = goodVehicle()
veh = model["vehicle"]
veh["test_mass"] = row["test_mass"]
veh["unladen_mass"] = veh["test_mass"] - driver_weight
veh["f0"] = row["f0"]
veh["f1"] = row["f1"]
veh["f2"] = row["f2"]
veh["p_rated"] = row["rated_power"]
veh["n_rated"] = row["rated_speed"]
veh["n_idle"] = int(row["idling_speed"])
ngears = int(row["no_of_gears"])
veh["gear_ratios"] = list(row[6:6 + ngears]) #'ndv_1'
if transplant_original_gears:
log.warning(">>> Transplanting gears from Heinz's!")
df_h = read_heinz_file(veh_num)
forced_cycle = df_h["g_max"]
forced_cycle.columns = ["gears_orig"]
model["forced_cycle"] = forced_cycle
experiment = Experiment(model)
model = experiment.run()
params = model["params"]
f_downscale = params["f_downscale"]
if f_downscale > 0:
log.warning(">> DOWNSCALE %s", f_downscale)
# ankostis_mdb: 't', "v in km/h","v_orig","a in m/s²","gear","g_min","g_max","gear_modification","error_description"
# heinz: 't', 'km_h', 'stg', 'gear'
(root, ext) = os.path.splitext(vehs_data_inp_fname)
outfname = os.path.join(mydir, samples_dir,
"{}-{:05}{}".format(root, veh_num, ext))
df = pd.DataFrame(model["cycle"])
_compare_exp_results(df, outfname, compare_results)
df.to_csv(outfname, index_label="time")
def testUnderPowered(self, plot_results=False):
mdl = goodVehicle()
mdl["vehicle"]["p_rated"] = 50
experiment = Experiment(mdl)
mdl = experiment.run()
print("DRIVEABILITY: \n%s" % experiment.driveability_report())
self._compare_exp_results(mdl["cycle"], "unpower1", self.run_comparison)
mdl = goodVehicle()
veh = mdl["vehicle"]
veh["test_mass"] = 1000
veh["unladen_mass"] = veh["test_mass"] - driver_weight
veh["p_rated"] = 80
veh["v_max"] = 120
veh["gear_ratios"] = [120.5, 95, 72, 52]
experiment = Experiment(mdl)
mdl = experiment.run()
print("DRIVEABILITY: \n%s" % experiment.driveability_report())
self._compare_exp_results(mdl["cycle"], "unpower2", self.run_comparison)
if plot_results:
self._plotResults(mdl)
plt.show()
def run_pyalgo_on_accdb_vehicle(
h5,
vehnum,
additional_properties=False,
props_group_suffix="prop",
pwot_group_suffix="wot",
) -> Tuple[dict, pd.DataFrame, pd.DataFrame]:
"""
Quick 'n dirty way to invoke python-algo (bc model will change).
:param h5:
the `WltpGs-msaccess.h5` file (path or h5db) to read input from
:return:
the *out-props* key-values, the *cycle* data-frame,
and the grid-wots constructed to solve v_max.
"""
from wltp import io as wio, engine, utils
from wltp.experiment import Experiment
props, wot, n2vs = load_vehicle_accdb(h5, vehnum)
mdl = mdl_from_accdb(props, wot, n2vs)
datamodel.validate_model(mdl, additional_properties=additional_properties)
exp = Experiment(mdl, skip_model_validation=True)
mdl = exp.run()
## Keep only *output* key-values, not to burden HDF data-model
# (excluding `driveability`, which is a list, and f0,f1,f2, addume were input).
#
# oprops = {k: v for k, v in veh if np.isscalar(v)}
out_mdl = {
"pmr": mdl["pmr"],
"n95_low": mdl["n95_low"],
"n95_high": mdl["n95_high"],
"v_max": mdl["v_max"],
"n_vmax": mdl["n_vmax"],
"g_vmax": mdl["g_vmax"],
"is_n_lim_vmax": mdl["is_n_lim_vmax"],
"n_max1": mdl["n_max1"],
"n_max2": mdl["n_max2"],
"n_max3": mdl["n_max3"],
"n_max": mdl["n_max"],
"wltc_class": mdl["wltc_class"],
"f_dsc_raw": mdl["f_dsc_raw"],
"f_dsc": mdl["f_dsc"],
}
cycle = mdl["cycle"]
return out_mdl, cycle, mdl["wots_vmax"]
def testPerf(self):
logging.getLogger().setLevel(logging.WARNING)
nexp = 100
start = time.time()
for _ in range(nexp):
mdl = goodVehicle()
experiment = Experiment(mdl)
experiment.run()
elapsed = time.time() - start
print(">> ELAPSED: %.2fsec, RUN/EXP: %.4fsec" %
(elapsed, elapsed / nexp))
def testOverlayOnInit(self):
mdl = goodVehicle()
nval = 6000
mdl2 = {"n_rated": nval}
exp = Experiment(mdl, mdl2)
mdl = exp._model
self.assertEqual(mdl["n_rated"], nval)
def testGoodVehicle():
mdl = goodVehicle()
exp = Experiment(mdl)
mdl = exp._model
defwot = datamodel.upd_default_load_curve({})["wot"]
assert pd.DataFrame(mdl["wot"][["n_norm",
"p_norm"]]).equals(pd.DataFrame(defwot))
def testOverlayOnInit():
mdl = goodVehicle()
n_rated = 6000
mdl2 = {"n_rated": n_rated}
exp = Experiment(datamodel.merge(mdl, mdl2))
mdl = exp._model
assert mdl["n_rated"] == n_rated
def test_two_ramps_with_map(self):
V = np.array([0, 2, 5, 10, 20, 30, 40, 50, 30, 20, 10, 5, 1, 0])
v_columns = ('v_class', 'v_target')
slope = [0, 0, 0.1, 0.1, 0.2, 0.2, 0.1, 0.1, 0.1, 0.1, 0.1, 0, 0, 0]
for col in v_columns:
results = []
for s in (slope, None):
mdl = {
"vehicle": {
"unladen_mass": 1430,
"test_mass": 1500,
"v_max": None,
"p_rated": 100,
"n_rated": 5450,
"n_idle": 950,
"gear_ratios": [120.5, 75, 50, 43, 37, 32],
"resistance_coeffs": [100, 0.5, 0.04],
},
'cycle_run': {
col: V,
}
}
if s:
mdl['cycle_run']['slope'] = s
proc = Experiment(mdl)
mdl = proc.run()
#print(mdl)
cycle_run = mdl['cycle_run']
if s:
self.assertIn('slope', cycle_run)
self.assertIn('gears', cycle_run)
for vcol in v_columns:
npt.assert_array_equal(
V, cycle_run[vcol],
"V_Column(%s) not overriden!" % vcol)
results.append(cycle_run['gears'])
npt.assert_array_equal(V, cycle_run[vcol],
"V_Column(%s) not overriden!" % vcol)
def build_models(vehs_df, **locals_kws):
"""
Builds all input-dataframes as Experiment classes and returns them in a list of (veh_id, exp) pairs.
:param vehs_df: A dataframe indexed by veh_id, and with columns *json-pointer* paths into the model
:return: a list of (veh_id, :class:`wltp.experiment.Experiment`) tuples
"""
experiment_pairs = []
for veh_id, row in vehs_df.iterrows():
try:
mdl_in = {}
for colname, colval in row.items():
log.debug("veh_id(%s): Column(%s): %s", veh_id, colname,
colval)
if not colval or (isinstance(colval, str)
and not colval.strip()):
continue
if isinstance(colval, str):
## Is it an excel-ref like:
# @<sheet_name>!A1[:R10:c1].table
#
try:
colval = resolve_excel_ref(colval)
except ValueError:
## Try to parse value as python-code.
#
try:
old_v = colval
colval = eval(
colval, globals(), locals_kws
) ## NOTE: Total insecure, but we're scientists, aren't we?
except Exception:
log.info(
"Failed parsing value(%s) as python code due to: %s\n Assuming plain string.",
old_v,
traceback.format_exc(),
)
else:
log.info(
"Parsed value(%s) as python code into: %s",
old_v,
colval,
)
pdl.set_jsonpointer(mdl_in, colname, colval)
exp = Experiment(mdl_in)
experiment_pairs.append((veh_id, exp))
except Exception as ex:
raise Exception("Invalid model for vehicle(%s): %s" %
(veh_id, ex)) from ex
return experiment_pairs
def test_two_ramps_with_slope(self):
V = np.hstack((np.r_[0:100:2], np.r_[100:0:-2]))
SLOPE = np.random.rand(V.shape[0])
v_columns = ('v_class', 'v_target')
for col in v_columns:
mdl = goodVehicle()
mdl['cycle_run'] = pd.DataFrame({
col: V,
'slope': SLOPE,
})
proc = Experiment(mdl)
mdl = proc.run()
#print(pd.DataFrame(mdl['cycle_run']))
cycle_run = mdl['cycle_run']
self.assertIn('slope', cycle_run)
self.assertIn('gears', cycle_run)
for vcol in v_columns:
npt.assert_array_equal(V, cycle_run[vcol], "V_Column(%s) not overriden!"%vcol)
def test_two_ramps_with_slope(self):
V = np.hstack((np.r_[0:100:2], np.r_[100:0:-2]))
SLOPE = np.random.rand(V.shape[0])
v_columns = ('v_class', 'v_target')
for col in v_columns:
mdl = goodVehicle()
mdl['cycle_run'] = pd.DataFrame({
col: V,
'slope': SLOPE,
})
proc = Experiment(mdl)
mdl = proc.run()
#print(pd.DataFrame(mdl['cycle_run']))
cycle_run = mdl['cycle_run']
self.assertIn('slope', cycle_run)
self.assertIn('gears', cycle_run)
for vcol in v_columns:
npt.assert_array_equal(V, cycle_run[vcol],
"V_Column(%s) not overriden!" % vcol)
def test_two_ramps_with_map(self):
V = np.array([0,2,5,10,20,30,40,50,30,20,10,5,1,0])
v_columns = ('v_class', 'v_target')
slope = [0, 0, 0.1, 0.1, 0.2, 0.2, 0.1, 0.1, 0.1, 0.1, 0.1,0,0,0]
for col in v_columns:
results = []
for s in (slope, None):
mdl = {
"vehicle": {
"unladen_mass": 1430,
"test_mass": 1500,
"v_max": None,
"p_rated": 100,
"n_rated": 5450,
"n_idle": 950,
"gear_ratios": [120.5, 75, 50, 43, 37, 32],
"resistance_coeffs":[100, 0.5, 0.04],
}, 'cycle_run': {
col: V,
}
}
if s:
mdl['cycle_run']['slope'] = s
proc = Experiment(mdl)
mdl = proc.run()
#print(mdl)
cycle_run = mdl['cycle_run']
if s:
self.assertIn('slope', cycle_run)
self.assertIn('gears', cycle_run)
for vcol in v_columns:
npt.assert_array_equal(V, cycle_run[vcol], "V_Column(%s) not overriden!"%vcol)
results.append(cycle_run['gears'])
npt.assert_array_equal(V, cycle_run[vcol], "V_Column(%s) not overriden!"%vcol)
def _run_the_experiments(
transplant_original_gears=False,
plot_results=False,
compare_results=False,
encoding="UTF-8",
):
## If file existent, it contains also calculated fields
# from the previous experiment run.
#
out_df = _read_vehicles_out()
inp_df = _read_vehicles_inp()
## Reconstruct the columns only presetn in the out_df.
#
inp_df["pmr"] = np.NAN
inp_df["wltc_class"] = ""
inp_df["f_downscale"] = np.NAN
wots = _read_wots()
failed_vehicles = 0
for (ix, row) in inp_df.iterrows():
veh_num = ix
heinz_fname = _make_heinz_fname(veh_num)
outfname = _make_gened_fname(transplant_original_gears, veh_num)
if not out_df is None and _is_file_up_to_date(outfname, [heinz_fname]):
inp_df.loc[ix] = out_df.loc[ix]
continue
mdl = goodVehicle()
veh = mdl["vehicle"]
veh["test_mass"] = row["test_mass"]
veh["unladen_mass"] = row["kerb_mass"]
veh["f0"] = row["f0_real"]
veh["f1"] = row["f1_real"]
veh["f2"] = row["f2_real"]
veh["p_rated"] = row["rated_power"]
veh["n_rated"] = row["rated_speed"]
veh["n_idle"] = int(row["idling_speed"])
veh["v_max"] = row["v_max"]
ngears = int(row["no_of_gears"])
veh["gear_ratios"] = list(row["ndv_1" : "ndv_%s" % ngears]) #'ndv_1'
veh["wot"] = _select_wot(wots, row["IDcat"] == 2)
if transplant_original_gears:
log.warning(">>> Transplanting gears from Heinz's!")
df_h = _read_heinz_file(veh_num)
mdl["cycle"] = {"gears_orig": df_h["g_max"].values}
try:
experiment = Experiment(mdl)
mdl = experiment.run()
except Exception as ex:
log.warning("VEHICLE_FAILED(%s): %s", veh_num, str(ex))
failed_vehicles += 1
continue
else:
params = mdl["params"]
veh = mdl["vehicle"]
inp_df.loc[ix, "pmr"] = veh["pmr"]
inp_df.loc[ix, "wltc_class"] = veh["wltc_class"]
inp_df.loc[ix, "f_downscale"] = params["f_downscale"]
# ankostis_mdb: 't', "v in km/h","v_orig","a in m/s²","gear","g_min","g_max","gear_modification","error_description"
# heinz: 't', 'km_h', 'stg', 'gear'
cycle_df = pd.DataFrame(mdl["cycle"])
_compare_exp_results(cycle_df, outfname, compare_results)
cycle_df.to_csv(outfname, index_label="time")
fail_limit_prcnt = 0.1
assert failed_vehicles < fail_limit_prcnt * inp_df.shape[0], (
"TOO MANY(>%f) vehicles have Failed(%i out of %i)!"
% (fail_limit_prcnt, failed_vehicles, inp_df.shape[0])
)
if not transplant_original_gears:
_write_vehicle_data(inp_df)
return inp_df
3000 93.00
3250 94.15
3500 95.30
3750 95.00
4000 94.70
4250 92.15
4500 89.60
4750 86.95
5000 84.30
"""),
sep="\t",
header=0,
)
datamodel.validate_model(mdl, additional_properties=True)
exp = Experiment(mdl, skip_model_validation=True)
mdl = exp.run()
oprops = {
"pmr": mdl["pmr"],
"n95_low": mdl["n95_low"],
"n95_high": mdl["n95_high"],
"v_max": mdl["v_max"],
"n_vmax": mdl["n_vmax"],
"g_vmax": mdl["g_vmax"],
"n_max1": mdl["n_max1"],
"n_max2": mdl["n_max2"],
"n_max3": mdl["n_max3"],
"n_max": mdl["n_max"],
"wltc_class": mdl["wltc_class"],
