asim 26.1.6

Automatic simulation system powered by neural networks

Automatic simulation system powered by neural networks

Installation

pip install asim

What is asim

• Physical field modeling with automatic constraint enforcement
• Flexible data grouping and normalization
• Built-in support for recurrent architectures
• Export/import of trained models

Using asim

import matplotlib.pyplot as plt
import pandas as pd
from asim.dataset import PhysicalDataManage, Vt, Vu, Vx, DummyDatasets
from asim.model import PhysicalFieldModel
from asim.optimizer import ContinuousOptimizer
from asim.simulator import PhysicalSimulator

# 1. Define data and structure
df = DummyDatasets.boiler_minimum(size=2000)  # pd.read_csv("demo.csv")
cols = [
    Vt(label="ts"),
    Vx(group="boiler", label="heat_temp", minmax=(0.0, None), unit="kj", enable=True),
    Vu(group="boiler", label="heat_power1", minmax=(0.0, None), unit="kw", enable=True),
    Vu(group="boiler", label="heat_power2", minmax=(0.0, None), unit="kw", enable=True),
]
dm = PhysicalDataManage(df, columns=cols, batch_size=64)

# 2A. Select the model, define the parameters, train and save
fm = PhysicalFieldModel(dm, lr=0.003)
fm.fit(epochs=100)
fm.export("demo.sim.onnx")

# 2B. Use a simulator to simulate the operation
sim = PhysicalSimulator("demo.sim.onnx", dm=dm)
sim_df = df[0:1440].copy()
sim_df.index = pd.to_datetime(sim_df["ts"], unit="s", utc=True).dt.tz_convert("Etc/GMT-8")
sim_df = sim.steps(sim_df, x0={"heat_temp": 150.0})
sim.plots(sim_df)
plt.show()

# 3A. Build the optimizer
lTerm = lambda x, u, p: (x - p) ** 2 + 0.5 * (u[0] ** 2 + u[1] ** 2)
mTerm = lambda x, u, p: (x - p) ** 2
opt = ContinuousOptimizer("demo.sim.onnx", dm=dm, lTerm=lTerm, mTerm=mTerm)
opt.fit(epochs=100)
opt.export("demo.opt.onnx")

# 3B. Use the optimizer
opt_df = opt.steps(df[0:360].copy())
opt.plots(opt_df)
plt.show()