Benchmark

Инструменты для объективного сравнения систем управления по стандартным метрикам.

Что оцениваем

• Перерегулирование
• Время переходного процесса
• Степень затухания
• Статическую ошибку

API

ControlBenchmark

Class for control system evaluation and beautiful plot generation.

Provides tools for analyzing transient process quality in automatic control systems with results visualization.

becnchmarking_one_step = benchmarking_one_step class-attribute instance-attribute

benchmarking_one_step(control_signal, system_signal, signal_val, dt)

Evaluate control system on one step and return extended set of results as dictionary.

Parameters:

Name	Type	Description	Default
control_signal	ndarray	System control signal.	required
system_signal	ndarray	System signal that is affected by control.	required
signal_val	float	Signal value from which the step function begins.	required
dt	float	Discretization step.	required

Returns:

Name	Type	Description
dict	dict	Dictionary with control system evaluation results: - "overshoot" (float): overshoot (%), - "settling_time" (float): settling time (s), - "damping_degree" (float): damping degree, - "static_error" (float): static error, - "rise_time" (float): rise time (s), - "peak_time" (float): peak time (s), - "maximum_deviation" (float): maximum deviation, - "iae" (float): integral absolute error, - "ise" (float): integral squared error, - "itae" (float): integral time absolute error, - "oscillation_count" (int): oscillation count, - "steady_state_value" (float): steady state value, - "performance_index" (float): comprehensive quality index.

plot(control_signal, system_signal, signal_val, dt, tps, figsize=(1600, 1000), title='Control System Analysis')

Build interactive control system analysis plot using Plotly.

Parameters:

Name	Type	Description	Default
control_signal	ndarray	System control signal.	required
system_signal	ndarray	System signal that is affected by control.	required
signal_val	float	Signal value from which the step function begins.	required
dt	float	Discretization step.	required
tps	ndarray	Array of time stamps.	required
figsize	tuple	Plot size in pixels, defaults to (1600, 1000).	(1600, 1000)
title	str	Plot title.	'Control System Analysis'

compare_systems(systems_data, signal_val, dt, figsize=(1800, 1200))

Compare multiple control systems on a single interactive plot.

Parameters:

Name	Type	Description	Default
systems_data	Dict[str, Dict]	Mapping from system name to a dictionary with keys: - control_signal: reference/command signal (np.ndarray) - system_signal: system response (np.ndarray) - time: time stamps for plotting (np.ndarray)	required
signal_val	float	Threshold value used to detect the step start.	required
dt	float	Discretization time step used for metric conversions.	required
figsize	tuple	Plot size in pixels. Defaults to (1800, 1200).	(1800, 1200)

Returns:

Type	Description
Dict[str, Dict]	Dict[str, Dict]: Mapping from system name to computed metrics.

generate_report(control_signal, system_signal, signal_val, dt, system_name='Система управления')

Generate a formatted text report with control quality metrics.

Parameters:

Name	Type	Description	Default
control_signal	ndarray	Reference/command signal.	required
system_signal	ndarray	System response signal.	required
signal_val	float	Threshold value used to detect the step start.	required
dt	float	Discretization time step.	required
system_name	str	Label used in the report header.	'Система управления'

Returns:

Name	Type	Description
str	str	Formatted report string.

Пример использования

from tensoraerospace.benchmark import ControlBenchmark

bench = ControlBenchmark()
metrics = bench.benchmarking_one_step(control_signal, system_signal, 1.0, dt)

print("Статическая ошибка:", metrics['static_error'])
print("Время переходного процесса:", metrics['settling_time'])
print("Степень затухания:", metrics['damping_degree'])
print("Перерегулирование:", metrics['overshoot'])

# Визуализация сравнения сигналов и метрик
bench.plot(control_signal, system_signal, 1.0, dt, tps, figsize=(15, 5))

Единицы и входные данные

• control_signal, system_signal — массивы одинаковой длины
• 1.0 — желаемое установившееся значение (пример)
• dt — шаг дискретизации; tps — временная ось

Обратная совместимость

Старое имя метода becnchmarking_one_step по-прежнему работает как псевдоним для benchmarking_one_step для обеспечения обратной совместимости.