mirror of
https://git.roussel.pro/telecom-paris/GIN206.git
synced 2026-02-09 02:30:17 +01:00
53 lines
2.1 KiB
Python
53 lines
2.1 KiB
Python
import pandas as pd
|
|
import numpy as np
|
|
from opensimplex import OpenSimplex
|
|
import datetime
|
|
|
|
def generate_greenhouse_data(filepath):
|
|
# Read the CSV file into a DataFrame, parsing 'time' as datetime
|
|
df = pd.read_csv(filepath, parse_dates=["time"], dtype={"id": str, "value": float})
|
|
|
|
# Compute the absolute differences between consecutive temperature readings
|
|
df['diff'] = df['value'].diff().abs()
|
|
|
|
# Initial value for 'diff' will be NaN; we can fill it with 0 or a small number
|
|
df['diff'] = df['diff'].fillna(0)
|
|
|
|
# Filter the DataFrame:
|
|
# 1. Exclude temperature values that are too high (>50) or too low (<-10)
|
|
# 2. Exclude rows where the difference from the previous reading is greater than 6
|
|
filtered_df = df[(df['value'] > 0) & (df['value'] < 50) & (df['diff'] <= 6)]
|
|
|
|
# Drop the 'diff' column as it's no longer needed after filtering
|
|
filtered_df = filtered_df.drop(columns=['diff'])
|
|
|
|
return filtered_df
|
|
|
|
|
|
def generate_simplex(start_time=None, end_time=None, interval=600, max_temp=30, min_temp=10, frequency=10):
|
|
# Default time settings if none provided
|
|
if end_time is None:
|
|
end_time = datetime.datetime.now()
|
|
if start_time is None:
|
|
start_time = end_time - datetime.timedelta(days=1)
|
|
|
|
# Calculate the number of samples needed based on the interval
|
|
total_seconds = int((end_time - start_time).total_seconds())
|
|
steps = total_seconds // interval
|
|
|
|
# Time array
|
|
times = [start_time + datetime.timedelta(seconds=i * interval) for i in range(steps + 1)]
|
|
|
|
# Simplex noise generator
|
|
simplex = OpenSimplex(seed=np.random.randint(0, 1000))
|
|
|
|
# Generate noise values and scale them
|
|
temperatures = [simplex.noise2(x=i / frequency, y=0) for i in range(steps + 1)]
|
|
|
|
# Map Simplex noise output (usually in range [-1, 1]) to the [min_temp, max_temp]
|
|
scaled_temperatures = min_temp + (np.array(temperatures) + 1) / 2 * (max_temp - min_temp)
|
|
|
|
# Create DataFrame
|
|
df = pd.DataFrame({'time': times, 'value': scaled_temperatures})
|
|
return df
|