File size: 16,899 Bytes
ac2f8e9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 |
from math import isclose
from sympy.core.numbers import I, all_close
from sympy.core.symbol import Dummy
from sympy.functions.elementary.complexes import (Abs, arg)
from sympy.functions.elementary.exponential import log
from sympy.functions.elementary.miscellaneous import sqrt
from sympy.abc import s, p, a
from sympy import pi
from sympy.external import import_module
from sympy.physics.control.control_plots import \
(pole_zero_numerical_data, pole_zero_plot, step_response_numerical_data,
step_response_plot, impulse_response_numerical_data,
impulse_response_plot, ramp_response_numerical_data,
ramp_response_plot, bode_magnitude_numerical_data,
bode_phase_numerical_data, bode_plot, nyquist_plot_expr,
nichols_plot_expr)
from sympy.physics.control.lti import (TransferFunction,
Series, Parallel, TransferFunctionMatrix)
from sympy.testing.pytest import raises, skip
matplotlib = import_module(
'matplotlib', import_kwargs={'fromlist': ['pyplot']},
catch=(RuntimeError,))
numpy = import_module('numpy')
tf1 = TransferFunction(1, p**2 + 0.5*p + 2, p)
tf2 = TransferFunction(p, 6*p**2 + 3*p + 1, p)
tf3 = TransferFunction(p, p**3 - 1, p)
tf4 = TransferFunction(10, p**3, p)
tf5 = TransferFunction(5, s**2 + 2*s + 10, s)
tf6 = TransferFunction(1, 1, s)
tf7 = TransferFunction(4*s*3 + 9*s**2 + 0.1*s + 11, 8*s**6 + 9*s**4 + 11, s)
tf8 = TransferFunction(5, s**2 + (2+I)*s + 10, s)
ser1 = Series(tf4, TransferFunction(1, p - 5, p))
ser2 = Series(tf3, TransferFunction(p, p + 2, p))
par1 = Parallel(tf1, tf2)
def _to_tuple(a, b):
return tuple(a), tuple(b)
def _trim_tuple(a, b):
a, b = _to_tuple(a, b)
return tuple(a[0: 2] + a[len(a)//2 : len(a)//2 + 1] + a[-2:]), \
tuple(b[0: 2] + b[len(b)//2 : len(b)//2 + 1] + b[-2:])
def y_coordinate_equality(plot_data_func, evalf_func, system):
"""Checks whether the y-coordinate value of the plotted
data point is equal to the value of the function at a
particular x."""
x, y = plot_data_func(system)
x, y = _trim_tuple(x, y)
y_exp = tuple(evalf_func(system, x_i) for x_i in x)
return all(Abs(y_exp_i - y_i) < 1e-8 for y_exp_i, y_i in zip(y_exp, y))
def test_errors():
if not matplotlib:
skip("Matplotlib not the default backend")
# Invalid `system` check
tfm = TransferFunctionMatrix([[tf6, tf5], [tf5, tf6]])
expr = 1/(s**2 - 1)
raises(NotImplementedError, lambda: pole_zero_plot(tfm))
raises(NotImplementedError, lambda: pole_zero_numerical_data(expr))
raises(NotImplementedError, lambda: impulse_response_plot(expr))
raises(NotImplementedError, lambda: impulse_response_numerical_data(tfm))
raises(NotImplementedError, lambda: step_response_plot(tfm))
raises(NotImplementedError, lambda: step_response_numerical_data(expr))
raises(NotImplementedError, lambda: ramp_response_plot(expr))
raises(NotImplementedError, lambda: ramp_response_numerical_data(tfm))
raises(NotImplementedError, lambda: bode_plot(tfm))
# More than 1 variables
tf_a = TransferFunction(a, s + 1, s)
raises(ValueError, lambda: pole_zero_plot(tf_a))
raises(ValueError, lambda: pole_zero_numerical_data(tf_a))
raises(ValueError, lambda: impulse_response_plot(tf_a))
raises(ValueError, lambda: impulse_response_numerical_data(tf_a))
raises(ValueError, lambda: step_response_plot(tf_a))
raises(ValueError, lambda: step_response_numerical_data(tf_a))
raises(ValueError, lambda: ramp_response_plot(tf_a))
raises(ValueError, lambda: ramp_response_numerical_data(tf_a))
raises(ValueError, lambda: bode_plot(tf_a))
# lower_limit > 0 for response plots
raises(ValueError, lambda: impulse_response_plot(tf1, lower_limit=-1))
raises(ValueError, lambda: step_response_plot(tf1, lower_limit=-0.1))
raises(ValueError, lambda: ramp_response_plot(tf1, lower_limit=-4/3))
# slope in ramp_response_plot() is negative
raises(ValueError, lambda: ramp_response_plot(tf1, slope=-0.1))
# incorrect frequency or phase unit
raises(ValueError, lambda: bode_plot(tf1,freq_unit = 'hz'))
raises(ValueError, lambda: bode_plot(tf1,phase_unit = 'degree'))
def test_pole_zero():
def pz_tester(sys, expected_value):
_z, _p = pole_zero_numerical_data(sys)
z_check = all_close(_z, expected_value[0])
p_check = all_close(_p, expected_value[1])
return p_check and z_check
exp1 = [[], [-0.24999999999999994-1.3919410907075054j, -0.24999999999999994+1.3919410907075054j]]
exp2 = [[0.0], [-0.25-0.3227486121839514j, -0.25+0.3227486121839514j]]
exp3 = [[0.0], [0.9999999999999998+0j, -0.5000000000000004-0.8660254037844395j,
-0.5000000000000004+0.8660254037844395j]]
exp4 = [[], [0.0, 0.0, 0.0, 5.0]]
exp5 = [[-5.645751311064592, -0.5000000000000008, -0.3542486889354093],
[-0.24999999999999986-0.322748612183951348j,
-0.2499999999999998+0.32274861218395134j,
-0.24999999999999986-1.3919410907075052j,
-0.2499999999999998+1.3919410907075052j]]
exp6 = [[], [-1.1641600331447917-3.545808351896439j,
-0.8358399668552097+2.5458083518964383j]]
assert pz_tester(tf1, exp1)
assert pz_tester(tf2, exp2)
assert pz_tester(tf3, exp3)
assert pz_tester(ser1, exp4)
assert pz_tester(par1, exp5)
assert pz_tester(tf8, exp6)
def test_bode():
if not numpy:
skip("NumPy is required for this test")
def bode_phase_evalf(system, point):
expr = system.to_expr()
_w = Dummy("w", real=True)
w_expr = expr.subs({system.var: I*_w})
return arg(w_expr).subs({_w: point}).evalf()
def bode_mag_evalf(system, point):
expr = system.to_expr()
_w = Dummy("w", real=True)
w_expr = expr.subs({system.var: I*_w})
return 20*log(Abs(w_expr), 10).subs({_w: point}).evalf()
def test_bode_data(sys):
return y_coordinate_equality(bode_magnitude_numerical_data, bode_mag_evalf, sys) \
and y_coordinate_equality(bode_phase_numerical_data, bode_phase_evalf, sys)
assert test_bode_data(tf1)
assert test_bode_data(tf2)
assert test_bode_data(tf3)
assert test_bode_data(tf4)
assert test_bode_data(tf5)
def check_point_accuracy(a, b):
return all(isclose(*_, rel_tol=1e-1, abs_tol=1e-6
) for _ in zip(a, b))
def test_impulse_response():
if not numpy:
skip("NumPy is required for this test")
def impulse_res_tester(sys, expected_value):
x, y = _to_tuple(*impulse_response_numerical_data(sys,
adaptive=False, n=10))
x_check = check_point_accuracy(x, expected_value[0])
y_check = check_point_accuracy(y, expected_value[1])
return x_check and y_check
exp1 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445,
5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0),
(0.0, 0.544019738507865, 0.01993849743234938, -0.31140243360893216, -0.022852779906491996, 0.1778306498155759,
0.01962941084328499, -0.1013115194573652, -0.014975541213105696, 0.0575789724730714))
exp2 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445, 5.555555555555555,
6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0), (0.1666666675, 0.08389223412935855,
0.02338051973475047, -0.014966807776379383, -0.034645954223054234, -0.040560075735512804,
-0.037658628907103885, -0.030149507719590022, -0.021162090730736834, -0.012721292737437523))
exp3 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445, 5.555555555555555,
6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0), (4.369893391586999e-09, 1.1750333000630964,
3.2922404058312473, 9.432290008148343, 28.37098083007151, 86.18577464367974, 261.90356653762115,
795.6538758627842, 2416.9920942096983, 7342.159505206647))
exp4 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445, 5.555555555555555,
6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0), (0.0, 6.17283950617284, 24.69135802469136,
55.555555555555564, 98.76543209876544, 154.320987654321, 222.22222222222226, 302.46913580246917,
395.0617283950618, 500.0))
exp5 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445, 5.555555555555555,
6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0), (0.0, -0.10455606138085417,
0.06757671513476461, -0.03234567568833768, 0.013582514927757873, -0.005273419510705473,
0.0019364083003354075, -0.000680070134067832, 0.00022969845960406913, -7.476094359583917e-05))
exp6 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445,
5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0),
(-6.016699583000218e-09, 0.35039802056107394, 3.3728423827689884, 12.119846079276684,
25.86101014293389, 29.352480635282088, -30.49475907497664, -273.8717189554019, -863.2381702029659,
-1747.0262164682233))
exp7 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335,
4.444444444444445, 5.555555555555555, 6.666666666666667, 7.777777777777779,
8.88888888888889, 10.0), (0.0, 18.934638095560974, 5346.93244680907, 1384609.8718249386,
358161126.65801865, 92645770015.70108, 23964739753087.42, 6198974342083139.0, 1.603492601616059e+18,
4.147764422869658e+20))
assert impulse_res_tester(tf1, exp1)
assert impulse_res_tester(tf2, exp2)
assert impulse_res_tester(tf3, exp3)
assert impulse_res_tester(tf4, exp4)
assert impulse_res_tester(tf5, exp5)
assert impulse_res_tester(tf7, exp6)
assert impulse_res_tester(ser1, exp7)
def test_step_response():
if not numpy:
skip("NumPy is required for this test")
def step_res_tester(sys, expected_value):
x, y = _to_tuple(*step_response_numerical_data(sys,
adaptive=False, n=10))
x_check = check_point_accuracy(x, expected_value[0])
y_check = check_point_accuracy(y, expected_value[1])
return x_check and y_check
exp1 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445,
5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0),
(-1.9193285738516863e-08, 0.42283495488246126, 0.7840485977945262, 0.5546841805655717,
0.33903033806932087, 0.4627251747410237, 0.5909907598988051, 0.5247213989553071,
0.4486997874319281, 0.4839358435839171))
exp2 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445,
5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0),
(0.0, 0.13728409095645816, 0.19474559355325086, 0.1974909129243011, 0.16841657696573073,
0.12559777736159378, 0.08153828016664713, 0.04360471317348958, 0.015072994568868221,
-0.003636420058445484))
exp3 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445,
5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0),
(0.0, 0.6314542141914303, 2.9356520038101035, 9.37731009663807, 28.452300356688376,
86.25721933273988, 261.9236645044672, 795.6435410577224, 2416.9786984578764, 7342.154119725917))
exp4 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445,
5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0),
(0.0, 2.286236899862826, 18.28989519890261, 61.72839629629631, 146.31916159122088, 285.7796124828532,
493.8271703703705, 784.1792566529494, 1170.553292729767, 1666.6667))
exp5 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445,
5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0),
(-3.999999997894577e-09, 0.6720357068882895, 0.4429938256137113, 0.5182010838004518,
0.4944139147159695, 0.5016379853883338, 0.4995466896527733, 0.5001154784851325,
0.49997448824584123, 0.5000039745919259))
exp6 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445,
5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0),
(-1.5433688493882158e-09, 0.3428705539937336, 1.1253619102202777, 3.1849962651016517,
9.47532757182671, 28.727231099148135, 87.29426924860557, 265.2138681048606, 805.6636260007757,
2447.387582370878))
assert step_res_tester(tf1, exp1)
assert step_res_tester(tf2, exp2)
assert step_res_tester(tf3, exp3)
assert step_res_tester(tf4, exp4)
assert step_res_tester(tf5, exp5)
assert step_res_tester(ser2, exp6)
def test_ramp_response():
if not numpy:
skip("NumPy is required for this test")
def ramp_res_tester(sys, num_points, expected_value, slope=1):
x, y = _to_tuple(*ramp_response_numerical_data(sys,
slope=slope, adaptive=False, n=num_points))
x_check = check_point_accuracy(x, expected_value[0])
y_check = check_point_accuracy(y, expected_value[1])
return x_check and y_check
exp1 = ((0.0, 2.0, 4.0, 6.0, 8.0, 10.0), (0.0, 0.7324667795033895, 1.9909720978650398,
2.7956587704217783, 3.9224897567931514, 4.85022655284895))
exp2 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445,
5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0),
(2.4360213402019326e-08, 0.10175320182493253, 0.33057612497658406, 0.5967937263298935,
0.8431511866718248, 1.0398805391471613, 1.1776043125035738, 1.2600994825747305, 1.2981042689274653,
1.304684417610106))
exp3 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445, 5.555555555555555,
6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0), (-3.9329040468771836e-08,
0.34686634635794555, 2.9998828170537903, 12.33303690737476, 40.993913948137795, 127.84145222317912,
391.41713691996, 1192.0006858708389, 3623.9808672503405, 11011.728034546572))
exp4 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445, 5.555555555555555,
6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0), (0.0, 1.9051973784484078, 30.483158055174524,
154.32098765432104, 487.7305288827924, 1190.7483615302544, 2469.1358024691367, 4574.3789056546275,
7803.688462124678, 12500.0))
exp5 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445, 5.555555555555555,
6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0), (0.0, 3.8844361856975635, 9.141792069209865,
14.096349157657231, 19.09783068994694, 24.10179770390321, 29.09907319114121, 34.10040420185154,
39.09983919254265, 44.10006013058409))
exp6 = ((0.0, 1.1111111111111112, 2.2222222222222223, 3.3333333333333335, 4.444444444444445, 5.555555555555555,
6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0), (0.0, 1.1111111111111112, 2.2222222222222223,
3.3333333333333335, 4.444444444444445, 5.555555555555555, 6.666666666666667, 7.777777777777779, 8.88888888888889, 10.0))
assert ramp_res_tester(tf1, 6, exp1)
assert ramp_res_tester(tf2, 10, exp2, 1.2)
assert ramp_res_tester(tf3, 10, exp3, 1.5)
assert ramp_res_tester(tf4, 10, exp4, 3)
assert ramp_res_tester(tf5, 10, exp5, 9)
assert ramp_res_tester(tf6, 10, exp6)
def test_nyquist_plot_expr():
r1, i1, w1 = nyquist_plot_expr(tf1)
r2, i2, w2 = nyquist_plot_expr(tf2)
r3, i3, w3 = nyquist_plot_expr(tf3)
r4, i4, w4 = nyquist_plot_expr(tf4)
assert r1 == (2 - w1**2)/(0.25*w1**2 + (2 - w1**2)**2)
assert i1 == -0.5*w1/(0.25*w1**2 + (2 - w1**2)**2)
assert r2 == 3*w2**2/(9*w2**2 + (1 - 6*w2**2)**2)
assert i2 == w2*(1 - 6*w2**2)/(9*w2**2 + (1 - 6*w2**2)**2)
assert r3 == -w3**4/(w3**6 + 1)
assert i3 == -w3/(w3**6 + 1)
assert r4 == 0
assert i4 == 10/w4**3
def test_nichols_expr():
m1, p1, w1 = nichols_plot_expr(tf1)
m2, p2, w2 = nichols_plot_expr(tf2)
m3, p3, w3 = nichols_plot_expr(tf3)
m4, p4, w4 = nichols_plot_expr(tf4)
assert m1 == 20*log(1/sqrt(w1**4 - 3.75*w1**2 + 4))/log(10)
assert p1 == 180*arg(1/(-w1**2 + 0.5*w1*I + 2))/pi
assert m2 == 20*log(Abs(w2)/sqrt(36*w2**4 - 3*w2**2 + 1))/log(10)
assert p2 == 180*arg(w2*I/(-6*w2**2 + 3*w2*I + 1))/pi
assert m3 == 20*log(Abs(w3)/sqrt(w3**6 + 1))/log(10)
assert p3 == 180*arg(-w3*I/(w3**3*I + 1))/pi
assert m4 == 20*log(10/(w4**2*Abs(w4)))/log(10)
assert p4 == 180*arg(I/w4**3)/pi
|