-
Notifications
You must be signed in to change notification settings - Fork 3
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
1 changed file
with
125 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,125 @@ | ||
import matplotlib as mpl | ||
import matplotlib.pyplot as plt | ||
import numpy as np | ||
import pandas as pd | ||
|
||
from wind_up.models import PlotConfig | ||
|
||
mpl.use("Agg") | ||
|
||
|
||
def plot_ws_est_gain_xs_one_ttype( | ||
pc_low_high: pd.DataFrame, | ||
ttype: str, | ||
rated_power_kw: float, | ||
x0: float, | ||
x1: float, | ||
x2: float, | ||
x3: float, | ||
plot_cfg: PlotConfig, | ||
) -> None: | ||
plt.figure() | ||
plt.plot(pc_low_high["x_mean"], pc_low_high["y_low"], label="q low") | ||
plt.plot(pc_low_high["x_mean"], pc_low_high["y_high"], label="q high") | ||
plt.vlines(x0, 0, rated_power_kw, label="ws0", color="#C30") | ||
plt.vlines(x1, 0, rated_power_kw, label="ws1", color="#D20") | ||
plt.vlines(x2, 0, rated_power_kw, label="ws2", color="#E10") | ||
plt.vlines(x3, 0, rated_power_kw, label="ws3", color="#F00") | ||
plot_title = f"{ttype} ws_est_gain xs" | ||
plt.title(plot_title) | ||
plt.xlabel("mean wind speed [m/s]") | ||
plt.ylabel("mean PositiveActivePower [kW]") | ||
plt.grid() | ||
plt.legend() | ||
if plot_cfg.show_plots: | ||
plt.show() | ||
if plot_cfg.save_plots: | ||
plt.savefig(plot_cfg.plots_dir / ttype / f"{plot_title}.png") | ||
plt.close() | ||
|
||
|
||
def plot_ws_est_one_ttype_or_wtg( | ||
df: pd.DataFrame, | ||
ttype_or_wtg: str, | ||
pc_transposed: pd.DataFrame, | ||
plot_cfg: PlotConfig, | ||
) -> None: | ||
plt.figure() | ||
plt.plot(pc_transposed.index, pc_transposed["ws_cp_corrected"]) | ||
plot_title = f"{ttype_or_wtg} transposed power curve" | ||
plt.title(plot_title) | ||
plt.xlabel("pw_clipped bin centre [kW]") | ||
plt.ylabel("ws_cp_corrected [m/s]") | ||
plt.grid() | ||
if plot_cfg.show_plots: | ||
plt.show() | ||
if plot_cfg.save_plots: | ||
tdir = plot_cfg.plots_dir / ttype_or_wtg | ||
tdir.mkdir(exist_ok=True, parents=True) | ||
plt.savefig(tdir / f"{plot_title}.png") | ||
plt.close() | ||
|
||
plt.figure() | ||
plt.scatter(df["ws_cp_corrected"], df["pw_clipped"], s=1, c=df["ws_est_gain"]) | ||
plt.xlabel("ws_cp_corrected [m/s]") | ||
plt.ylabel("pw_clipped [kW]") | ||
plot_title = f"{ttype_or_wtg} power curve colored by ws_est_gain" | ||
plt.title(plot_title) | ||
plt.colorbar() | ||
plt.grid() | ||
if plot_cfg.show_plots: | ||
plt.show() | ||
if plot_cfg.save_plots: | ||
plt.savefig(plot_cfg.plots_dir / ttype_or_wtg / f"{plot_title}.png") | ||
plt.close() | ||
|
||
plt.figure() | ||
plt.scatter(df["ws_est_blend"], df["pw_clipped"], s=1) | ||
plt.xlabel("ws_est_blend [m/s]") | ||
plt.ylabel("pw_clipped [kW]") | ||
plot_title = f"{ttype_or_wtg} wind speed estimated from power" | ||
plt.title(plot_title) | ||
plt.grid() | ||
if plot_cfg.show_plots: | ||
plt.show() | ||
if plot_cfg.save_plots: | ||
plt.savefig(plot_cfg.plots_dir / ttype_or_wtg / f"{plot_title}.png") | ||
plt.close() | ||
|
||
plt.figure() | ||
plt.scatter(df["WindSpeedMean"], df["ws_est_blend"], s=1, c=df["ws_est_gain"]) | ||
plt.xlabel("WindSpeedMean [m/s]") | ||
plt.ylabel("ws_est_blend [m/s]") | ||
plot_title = f"{ttype_or_wtg} wind speed estimated vs original colored by ws_est_gain" | ||
plt.title(plot_title) | ||
plt.grid() | ||
if plot_cfg.show_plots: | ||
plt.show() | ||
if plot_cfg.save_plots: | ||
plt.savefig(plot_cfg.plots_dir / ttype_or_wtg / f"{plot_title}.png") | ||
plt.close() | ||
|
||
plt.figure() | ||
wsbins = np.arange(0, max(df["WindSpeedMean"].max(), df["ws_est_blend"].max()) + 0.5, 0.5) | ||
plt.hist( | ||
df["WindSpeedMean"], | ||
bins=wsbins.tolist(), | ||
label="WindSpeedMean", | ||
) | ||
plt.hist( | ||
df["ws_est_blend"], | ||
bins=wsbins.tolist(), | ||
alpha=0.5, | ||
label="ws_est_blend", | ||
) | ||
plt.xlabel("WindSpeedMean, ws_est_blend [m/s]") | ||
plt.ylabel("count") | ||
plot_title = f"{ttype_or_wtg} turbines wind speed histograms" | ||
plt.title(plot_title) | ||
plt.legend() | ||
plt.grid() | ||
if plot_cfg.show_plots: | ||
plt.show() | ||
if plot_cfg.save_plots: | ||
plt.savefig(plot_cfg.plots_dir / ttype_or_wtg / f"{plot_title}.png") | ||
plt.close() |