Infer thresholds for detect_clearsky

docs/sphinx/source/whatsnew/v0.10.2.rst

@@ -15,6 +15,8 @@ Enhancements
   :py:func:`pvlib.iotools.get_pvgis_hourly`, :py:func:`pvlib.iotools.get_cams`,
   :py:func:`pvlib.iotools.get_bsrn`, and :py:func:`pvlib.iotools.read_midc_raw_data_from_nrel`.
   (:pull:`1800`)
+* Added option to infer threshold values for 
+  :py:func:`pvlib.clearsky.detect_clearsky` (:issue:`1808`, :pull:`1784`)
 
 
 Bug fixes
@@ -39,4 +41,5 @@ Requirements
 Contributors
 ~~~~~~~~~~~~
 * Adam R. Jensen (:ghuser:`AdamRJensen`)
+* Abigail Jones (:ghuser:`ajonesr`)
 * Taos Transue (:ghuser:`reepoi`)

pvlib/clearsky.py

@@ -644,8 +644,38 @@ def _clear_sample_index(clear_windows, samples_per_window, align, H):
     return clear_samples
 
 
-def detect_clearsky(measured, clearsky, times=None, window_length=10,
-                    mean_diff=75, max_diff=75,
+def _clearsky_get_threshold(sample_interval):
+    """
+    Returns threshold values for kwargs in detect_clearsky. See
+    Table 1 in [1].
+
+    References
+    ----------
+    .. [1] Jordan, D.C. and C. Hansen, "Clear-sky detection for PV
+       degradation analysis using multiple regression", Renewable Energy,
+       v209, p. 393-400, 2023.
+    """
+
+    if (sample_interval < 1 or sample_interval > 30):
+        raise ValueError("`infer_limits=True` can only be used for inputs \
+                          with time step from 1 to 30 minutes")
+
+    data_freq = np.array([1, 5, 15, 30])
+
+    window_length = np.interp(sample_interval, data_freq, [50, 60, 90, 120])
+    mean_diff = np.interp(sample_interval, data_freq, [75, 75, 75, 75])
+    max_diff = np.interp(sample_interval, data_freq, [60, 65, 75, 90])
+    lower_line_length = np.interp(sample_interval, data_freq, [-45,-45,-45,-45])
+    upper_line_length = np.interp(sample_interval, data_freq, [80, 80, 80, 80])
+    var_diff = np.interp(sample_interval, data_freq, [0.005, 0.01, 0.032, 0.07])
+    slope_dev = np.interp(sample_interval, data_freq, [50, 60, 75, 96])
+
+    return (window_length, mean_diff, max_diff, lower_line_length,
+            upper_line_length, var_diff, slope_dev)
+
+
+def detect_clearsky(measured, clearsky, times=None, infer_limits=False,
+                    window_length=10, mean_diff=75, max_diff=75,
                     lower_line_length=-5, upper_line_length=10,
                     var_diff=0.005, slope_dev=8, max_iterations=20,
                     return_components=False):
@@ -676,6 +706,9 @@ def detect_clearsky(measured, clearsky, times=None, window_length=10,
     times : DatetimeIndex or None, default None.
         Times of measured and clearsky values. If None the index of measured
         will be used.
+    infer_limits : bool, default False
+        If True, does not use passed in kwargs (or defaults), but instead
+        interpolates these values from Table 1 in [2]_.
     window_length : int, default 10
         Length of sliding time window in minutes. Must be greater than 2
         periods.
@@ -731,6 +764,9 @@ def detect_clearsky(measured, clearsky, times=None, window_length=10,
     .. [1] Reno, M.J. and C.W. Hansen, "Identification of periods of clear
        sky irradiance in time series of GHI measurements" Renewable Energy,
        v90, p. 520-531, 2016.
+    .. [2] Jordan, D.C. and C. Hansen, "Clear-sky detection for PV
+       degradation analysis using multiple regression", Renewable Energy,
+       v209, p. 393-400, 2023. :doi:`10.1016/j.renene.2023.04.035`
 
     Notes
     -----
@@ -773,6 +809,21 @@ def detect_clearsky(measured, clearsky, times=None, window_length=10,
     sample_interval, samples_per_window = \
         tools._get_sample_intervals(times, window_length)
 
+    # if infer_limits, find threshold values using the sample interval
+    if infer_limits:
+        window_length, mean_diff, max_diff, lower_line_length, \
+            upper_line_length, var_diff, slope_dev = \
+                _clearsky_get_threshold(sample_interval)
+
+        # recalculate samples_per_window using returned window_length
+        _, samples_per_window = \
+            tools._get_sample_intervals(times, window_length)
+
+    # check that we have enough data to produce a nonempty hankel matrix
+    if len(times) < samples_per_window:
+        raise ValueError(f"times has only {len(times)} entries, but it must \
+                           have at least {samples_per_window} entries")
+
     # generate matrix of integers for creating windows with indexing
     H = hankel(np.arange(samples_per_window),
                np.arange(samples_per_window-1, len(times)))
@@ -826,7 +877,22 @@ def detect_clearsky(measured, clearsky, times=None, window_length=10,
         def rmse(alpha):
             return np.sqrt(np.mean((clear_meas - alpha*clear_clear)**2))
 
-        alpha = minimize_scalar(rmse).x
+        optimize_result = minimize_scalar(rmse)
+        if not optimize_result.success:
+            try:
+                message = "Optimizer exited unsuccessfully: " \
+                           + optimize_result.message
+            except AttributeError:
+                message = "Optimizer exited unsuccessfully: \
+                           No message explaining the failure was returned. \
+                           If you would like to see this message, please \
+                           update your scipy version (try version 1.8.0 \
+                           or beyond)."
+            raise RuntimeError(message)
+
+        else:
+            alpha = optimize_result.x
+
         if round(alpha*10000) == round(previous_alpha*10000):
             break
     else:

pvlib/data/detect_clearsky_threshold_data.csv

@@ -0,0 +1,126 @@
+# latitude:35.04
+# longitude:-106.62
+# elevation:1619
+# window_length:10
+Time (UTC),GHI,Clear or not
+4/1/2012 17:30,862.0935267144818,1
+4/1/2012 17:31,863.9568654037691,1
+4/1/2012 17:32,865.802767695237,1
+4/1/2012 17:33,867.6311939211416,1
+4/1/2012 17:34,869.4421084958154,1
+4/1/2012 17:35,871.2354749580768,1
+4/1/2012 17:36,873.0112572200139,1
+4/1/2012 17:37,874.7694195555663,1
+4/1/2012 17:38,876.5099266138993,1
+4/1/2012 17:39,878.2327434081394,1
+4/1/2012 17:40,879.9378353205582,1
+4/1/2012 17:41,881.6251681074431,1
+4/1/2012 17:42,500.0,0
+4/1/2012 17:43,300.0,0
+4/1/2012 17:44,400.0,0
+4/1/2012 17:45,888.1962370575133,1
+4/1/2012 17:46,889.7942734332919,1
+4/1/2012 17:47,891.3743529660121,1
+4/1/2012 17:48,892.9364440027008,1
+4/1/2012 17:49,894.4805152569894,1
+4/1/2012 17:50,896.0065358138269,1
+4/1/2012 17:51,897.514475133874,1
+4/1/2012 17:52,899.0043030438522,1
+4/1/2012 17:53,900.4759897479845,1
+4/1/2012 17:54,901.9295058184756,1
+4/1/2012 17:55,903.3648231563722,1
+4/1/2012 17:56,950.0,0
+4/1/2012 17:57,906.1807424805852,1
+4/1/2012 17:58,907.5612891914101,1
+4/1/2012 17:59,908.9235237269177,1
+4/1/2012 18:00,910.2674188971979,1
+4/1/2012 18:01,911.592947889838,1
+4/1/2012 18:02,912.9000842614388,1
+4/1/2012 18:03,914.1888019477013,1
+4/1/2012 18:04,915.4590752551169,1
+4/1/2012 18:05,916.7108788648916,1
+4/1/2012 18:06,917.9441886574208,1
+4/1/2012 18:07,919.1589784086842,1
+4/1/2012 18:08,920.3552247618425,1
+4/1/2012 18:09,921.5329038989687,1
+4/1/2012 18:10,922.691992377965,1
+4/1/2012 18:11,923.8324671359268,1
+4/1/2012 18:12,924.9543054818693,1
+4/1/2012 18:13,926.057485105408,1
+4/1/2012 18:14,927.141984069634,1
+4/1/2012 18:15,928.2077808145282,1
+4/1/2012 18:16,929.254854160055,1
+4/1/2012 18:17,930.2831839827653,1
+4/1/2012 18:18,931.2927484780441,1
+4/1/2012 18:19,932.2835282910601,1
+4/1/2012 18:20,933.2555037490038,1
+4/1/2012 18:21,934.2086555597849,1
+4/1/2012 18:22,935.1429648059466,1
+4/1/2012 18:23,936.058412951919,1
+4/1/2012 18:24,936.9549818381174,1
+4/1/2012 18:25,937.8326536837798,1
+4/1/2012 18:26,938.6914110895166,1
+4/1/2012 18:27,939.5312370318452,1
+4/1/2012 18:28,940.3521148697101,1
+4/1/2012 18:29,941.1540288705581,1
+4/1/2012 18:30,941.9369620746523,1
+4/1/2012 18:31,942.7008995238247,1
+4/1/2012 18:32,943.4458260928685,1
+4/1/2012 18:33,944.1717270394992,1
+4/1/2012 18:34,944.8785879996012,1
+4/1/2012 18:35,945.5663949895419,1
+4/1/2012 18:36,946.2351344081491,1
+4/1/2012 18:37,946.8847930330305,1
+4/1/2012 18:38,947.5153580226918,1
+4/1/2012 18:39,948.126816918376,1
+4/1/2012 18:40,948.7191580309192,1
+4/1/2012 18:41,949.2923688693852,1
+4/1/2012 18:42,949.8464385206038,1
+4/1/2012 18:43,950.3813560493355,1
+4/1/2012 18:44,950.8971109033968,1
+4/1/2012 18:45,951.3936929103834,1
+4/1/2012 18:46,951.8710922815745,1
+4/1/2012 18:47,952.3292996087865,1
+4/1/2012 18:48,952.7683058659376,1
+4/1/2012 18:49,953.1881024102506,1
+4/1/2012 18:50,953.5886809795984,1
+4/1/2012 18:51,953.9700339452911,1
+4/1/2012 18:52,954.3321532981175,1
+4/1/2012 18:53,954.6750321861093,1
+4/1/2012 18:54,954.9986638782349,1
+4/1/2012 18:55,955.3030420248847,1
+4/1/2012 18:56,955.5881606602495,1
+4/1/2012 18:57,955.8540142004646,1
+4/1/2012 18:58,956.1005974445337,1
+4/1/2012 18:59,956.3279055749699,1
+4/1/2012 19:00,956.5359341563872,1
+4/1/2012 19:01,956.7246791371283,1
+4/1/2012 19:02,956.8941369551687,1
+4/1/2012 19:03,957.0443040971436,1
+4/1/2012 19:04,957.175177780536,1
+4/1/2012 19:05,957.2867554853715,1
+4/1/2012 19:06,957.3790350752402,1
+4/1/2012 19:07,957.4520147966049,1
+4/1/2012 19:08,957.5056932791584,1
+4/1/2012 19:09,957.5400695359402,1
+4/1/2012 19:10,957.5551429630466,1
+4/1/2012 19:11,957.5509133398784,1
+4/1/2012 19:12,957.527380828979,1
+4/1/2012 19:13,957.4845459762313,1
+4/1/2012 19:14,957.4224096624041,1
+4/1/2012 19:15,957.3409732831614,1
+4/1/2012 19:16,957.2402384985319,1
+4/1/2012 19:17,957.1202073871192,1
+4/1/2012 19:18,956.9808824107419,1
+4/1/2012 19:19,956.8222664139458,1
+4/1/2012 19:20,956.6443626248492,1
+4/1/2012 19:21,956.4471746540574,1
+4/1/2012 19:22,956.2307064955613,1
+4/1/2012 19:23,955.9949625264852,1
+4/1/2012 19:24,955.7399475061342,1
+4/1/2012 19:25,955.4656663871413,1
+4/1/2012 19:26,955.1721250622959,1
+4/1/2012 19:27,954.8593292624192,1
+4/1/2012 19:28,954.5272852786308,1
+4/1/2012 19:29,954.175999783701,1
+4/1/2012 19:30,953.8054798341012,1

pvlib/tests/test_clearsky.py

@@ -533,6 +533,49 @@ def detect_clearsky_data():
     return expected, cs
 
 
+@pytest.fixture
+def detect_clearsky_threshold_data():
+    # this is (roughly) just a 2 hour period of the same data in
+    # detect_clearsky_data (which only spans 30 minutes)
+    data_file = DATA_DIR / 'detect_clearsky_threshold_data.csv'
+    expected = pd.read_csv(
+        data_file, index_col=0, parse_dates=True, comment='#')
+    expected = expected.tz_localize('UTC').tz_convert('Etc/GMT+7')
+    metadata = {}
+    with data_file.open() as f:
+        for line in f:
+            if line.startswith('#'):
+                key, value = line.strip('# \n').split(':')
+                metadata[key] = float(value)
+            else:
+                break
+    metadata['window_length'] = int(metadata['window_length'])
+    loc = Location(metadata['latitude'], metadata['longitude'],
+                   altitude=metadata['elevation'])
+    # specify turbidity to guard against future lookup changes
+    cs = loc.get_clearsky(expected.index, linke_turbidity=2.658197)
+    return expected, cs
+
+
+def test_clearsky_get_threshold():
+    out = clearsky._clearsky_get_threshold(4.5)
+    expected = (58.75, 75, 64.375, -45, 80.0, 0.009375, 58.75)
+    assert np.allclose(out, expected)
+
+
+def test_clearsky_get_threshold_raises_error():
+    with pytest.raises(ValueError, match='can only be used for inputs'):
+        clearsky._clearsky_get_threshold(0.5)
+
+
+def test_detect_clearsky_calls_threshold(mocker, detect_clearsky_threshold_data):
+    threshold_spy = mocker.spy(clearsky, '_clearsky_get_threshold')
+    expected, cs = detect_clearsky_threshold_data
+    threshold_actual = clearsky.detect_clearsky(expected['GHI'], cs['ghi'],
+                       infer_limits=True)
+    assert threshold_spy.call_count == 1
+
+
 def test_detect_clearsky(detect_clearsky_data):
     expected, cs = detect_clearsky_data
     clear_samples = clearsky.detect_clearsky(
@@ -629,6 +672,18 @@ def test_detect_clearsky_missing_index(detect_clearsky_data):
         clearsky.detect_clearsky(expected['GHI'].values, cs['ghi'].values)
 
 
+def test_detect_clearsky_not_enough_data(detect_clearsky_data):
+    expected, cs = detect_clearsky_data
+    with pytest.raises(ValueError, match='have at least'):
+       clearsky.detect_clearsky(expected['GHI'], cs['ghi'], window_length=60)
+
+
+def test_detect_clearsky_optimizer_failed(detect_clearsky_data):
+    expected, cs = detect_clearsky_data
+    with pytest.raises(RuntimeError, match='Optimizer exited unsuccessfully'):
+        clearsky.detect_clearsky(expected['GHI'], cs['ghi'], window_length=15)
+
+
 @pytest.fixture
 def detect_clearsky_helper_data():
     samples_per_window = 3