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

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

357

358

359

360

361

362

363

364

365

366

367

368

369

370

371

372

373

374

375

376

377

378

379

380

381

382

383

384

385

386

387

388

389

390

391

392

393

394

395

396

397

398

399

400

401

402

403

404

405

406

407

408

409

410

411

412

413

414

415

416

417

418

419

420

421

422

423

424

425

426

427

428

429

430

431

432

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

456

457

458

459

460

461

462

463

464

465

""" 

ltisys -- a collection of functions to convert linear time invariant systems 

from one representation to another. 

""" 

from __future__ import division, print_function, absolute_import 

 

import numpy 

import numpy as np 

from numpy import (r_, eye, atleast_2d, poly, dot, 

asarray, product, zeros, array, outer) 

from scipy import linalg 

 

from .filter_design import tf2zpk, zpk2tf, normalize 

 

 

__all__ = ['tf2ss', 'abcd_normalize', 'ss2tf', 'zpk2ss', 'ss2zpk', 

'cont2discrete'] 

 

 

def tf2ss(num, den): 

r"""Transfer function to state-space representation. 

 

Parameters 

---------- 

num, den : array_like 

Sequences representing the coefficients of the numerator and 

denominator polynomials, in order of descending degree. The 

denominator needs to be at least as long as the numerator. 

 

Returns 

------- 

A, B, C, D : ndarray 

State space representation of the system, in controller canonical 

form. 

 

Examples 

-------- 

Convert the transfer function: 

 

.. math:: H(s) = \frac{s^2 + 3s + 3}{s^2 + 2s + 1} 

 

>>> num = [1, 3, 3] 

>>> den = [1, 2, 1] 

 

to the state-space representation: 

 

.. math:: 

 

\dot{\textbf{x}}(t) = 

\begin{bmatrix} -2 & -1 \\ 1 & 0 \end{bmatrix} \textbf{x}(t) + 

\begin{bmatrix} 1 \\ 0 \end{bmatrix} \textbf{u}(t) \\ 

 

\textbf{y}(t) = \begin{bmatrix} 1 & 2 \end{bmatrix} \textbf{x}(t) + 

\begin{bmatrix} 1 \end{bmatrix} \textbf{u}(t) 

 

>>> from scipy.signal import tf2ss 

>>> A, B, C, D = tf2ss(num, den) 

>>> A 

array([[-2., -1.], 

[ 1., 0.]]) 

>>> B 

array([[ 1.], 

[ 0.]]) 

>>> C 

array([[ 1., 2.]]) 

>>> D 

array([[ 1.]]) 

""" 

# Controller canonical state-space representation. 

# if M+1 = len(num) and K+1 = len(den) then we must have M <= K 

# states are found by asserting that X(s) = U(s) / D(s) 

# then Y(s) = N(s) * X(s) 

# 

# A, B, C, and D follow quite naturally. 

# 

num, den = normalize(num, den) # Strips zeros, checks arrays 

nn = len(num.shape) 

if nn == 1: 

num = asarray([num], num.dtype) 

M = num.shape[1] 

K = len(den) 

if M > K: 

msg = "Improper transfer function. `num` is longer than `den`." 

raise ValueError(msg) 

if M == 0 or K == 0: # Null system 

return (array([], float), array([], float), array([], float), 

array([], float)) 

 

# pad numerator to have same number of columns has denominator 

num = r_['-1', zeros((num.shape[0], K - M), num.dtype), num] 

 

if num.shape[-1] > 0: 

D = atleast_2d(num[:, 0]) 

 

else: 

# We don't assign it an empty array because this system 

# is not 'null'. It just doesn't have a non-zero D 

# matrix. Thus, it should have a non-zero shape so that 

# it can be operated on by functions like 'ss2tf' 

D = array([[0]], float) 

 

if K == 1: 

D = D.reshape(num.shape) 

 

return (zeros((1, 1)), zeros((1, D.shape[1])), 

zeros((D.shape[0], 1)), D) 

 

frow = -array([den[1:]]) 

A = r_[frow, eye(K - 2, K - 1)] 

B = eye(K - 1, 1) 

C = num[:, 1:] - outer(num[:, 0], den[1:]) 

D = D.reshape((C.shape[0], B.shape[1])) 

 

return A, B, C, D 

 

 

def _none_to_empty_2d(arg): 

if arg is None: 

return zeros((0, 0)) 

else: 

return arg 

 

 

def _atleast_2d_or_none(arg): 

if arg is not None: 

return atleast_2d(arg) 

 

 

def _shape_or_none(M): 

if M is not None: 

return M.shape 

else: 

return (None,) * 2 

 

 

def _choice_not_none(*args): 

for arg in args: 

if arg is not None: 

return arg 

 

 

def _restore(M, shape): 

if M.shape == (0, 0): 

return zeros(shape) 

else: 

if M.shape != shape: 

raise ValueError("The input arrays have incompatible shapes.") 

return M 

 

 

def abcd_normalize(A=None, B=None, C=None, D=None): 

"""Check state-space matrices and ensure they are two-dimensional. 

 

If enough information on the system is provided, that is, enough 

properly-shaped arrays are passed to the function, the missing ones 

are built from this information, ensuring the correct number of 

rows and columns. Otherwise a ValueError is raised. 

 

Parameters 

---------- 

A, B, C, D : array_like, optional 

State-space matrices. All of them are None (missing) by default. 

See `ss2tf` for format. 

 

Returns 

------- 

A, B, C, D : array 

Properly shaped state-space matrices. 

 

Raises 

------ 

ValueError 

If not enough information on the system was provided. 

 

""" 

A, B, C, D = map(_atleast_2d_or_none, (A, B, C, D)) 

 

MA, NA = _shape_or_none(A) 

MB, NB = _shape_or_none(B) 

MC, NC = _shape_or_none(C) 

MD, ND = _shape_or_none(D) 

 

p = _choice_not_none(MA, MB, NC) 

q = _choice_not_none(NB, ND) 

r = _choice_not_none(MC, MD) 

if p is None or q is None or r is None: 

raise ValueError("Not enough information on the system.") 

 

A, B, C, D = map(_none_to_empty_2d, (A, B, C, D)) 

A = _restore(A, (p, p)) 

B = _restore(B, (p, q)) 

C = _restore(C, (r, p)) 

D = _restore(D, (r, q)) 

 

return A, B, C, D 

 

 

def ss2tf(A, B, C, D, input=0): 

r"""State-space to transfer function. 

 

A, B, C, D defines a linear state-space system with `p` inputs, 

`q` outputs, and `n` state variables. 

 

Parameters 

---------- 

A : array_like 

State (or system) matrix of shape ``(n, n)`` 

B : array_like 

Input matrix of shape ``(n, p)`` 

C : array_like 

Output matrix of shape ``(q, n)`` 

D : array_like 

Feedthrough (or feedforward) matrix of shape ``(q, p)`` 

input : int, optional 

For multiple-input systems, the index of the input to use. 

 

Returns 

------- 

num : 2-D ndarray 

Numerator(s) of the resulting transfer function(s). `num` has one row 

for each of the system's outputs. Each row is a sequence representation 

of the numerator polynomial. 

den : 1-D ndarray 

Denominator of the resulting transfer function(s). `den` is a sequence 

representation of the denominator polynomial. 

 

Examples 

-------- 

Convert the state-space representation: 

 

.. math:: 

 

\dot{\textbf{x}}(t) = 

\begin{bmatrix} -2 & -1 \\ 1 & 0 \end{bmatrix} \textbf{x}(t) + 

\begin{bmatrix} 1 \\ 0 \end{bmatrix} \textbf{u}(t) \\ 

 

\textbf{y}(t) = \begin{bmatrix} 1 & 2 \end{bmatrix} \textbf{x}(t) + 

\begin{bmatrix} 1 \end{bmatrix} \textbf{u}(t) 

 

>>> A = [[-2, -1], [1, 0]] 

>>> B = [[1], [0]] # 2-dimensional column vector 

>>> C = [[1, 2]] # 2-dimensional row vector 

>>> D = 1 

 

to the transfer function: 

 

.. math:: H(s) = \frac{s^2 + 3s + 3}{s^2 + 2s + 1} 

 

>>> from scipy.signal import ss2tf 

>>> ss2tf(A, B, C, D) 

(array([[1, 3, 3]]), array([ 1., 2., 1.])) 

""" 

# transfer function is C (sI - A)**(-1) B + D 

 

# Check consistency and make them all rank-2 arrays 

A, B, C, D = abcd_normalize(A, B, C, D) 

 

nout, nin = D.shape 

if input >= nin: 

raise ValueError("System does not have the input specified.") 

 

# make SIMO from possibly MIMO system. 

B = B[:, input:input + 1] 

D = D[:, input:input + 1] 

 

try: 

den = poly(A) 

except ValueError: 

den = 1 

 

if (product(B.shape, axis=0) == 0) and (product(C.shape, axis=0) == 0): 

num = numpy.ravel(D) 

if (product(D.shape, axis=0) == 0) and (product(A.shape, axis=0) == 0): 

den = [] 

return num, den 

 

num_states = A.shape[0] 

type_test = A[:, 0] + B[:, 0] + C[0, :] + D 

num = numpy.zeros((nout, num_states + 1), type_test.dtype) 

for k in range(nout): 

Ck = atleast_2d(C[k, :]) 

num[k] = poly(A - dot(B, Ck)) + (D[k] - 1) * den 

 

return num, den 

 

 

def zpk2ss(z, p, k): 

"""Zero-pole-gain representation to state-space representation 

 

Parameters 

---------- 

z, p : sequence 

Zeros and poles. 

k : float 

System gain. 

 

Returns 

------- 

A, B, C, D : ndarray 

State space representation of the system, in controller canonical 

form. 

 

""" 

return tf2ss(*zpk2tf(z, p, k)) 

 

 

def ss2zpk(A, B, C, D, input=0): 

"""State-space representation to zero-pole-gain representation. 

 

A, B, C, D defines a linear state-space system with `p` inputs, 

`q` outputs, and `n` state variables. 

 

Parameters 

---------- 

A : array_like 

State (or system) matrix of shape ``(n, n)`` 

B : array_like 

Input matrix of shape ``(n, p)`` 

C : array_like 

Output matrix of shape ``(q, n)`` 

D : array_like 

Feedthrough (or feedforward) matrix of shape ``(q, p)`` 

input : int, optional 

For multiple-input systems, the index of the input to use. 

 

Returns 

------- 

z, p : sequence 

Zeros and poles. 

k : float 

System gain. 

 

""" 

return tf2zpk(*ss2tf(A, B, C, D, input=input)) 

 

 

def cont2discrete(system, dt, method="zoh", alpha=None): 

""" 

Transform a continuous to a discrete state-space system. 

 

Parameters 

---------- 

system : a tuple describing the system or an instance of `lti` 

The following gives the number of elements in the tuple and 

the interpretation: 

 

* 1: (instance of `lti`) 

* 2: (num, den) 

* 3: (zeros, poles, gain) 

* 4: (A, B, C, D) 

 

dt : float 

The discretization time step. 

method : {"gbt", "bilinear", "euler", "backward_diff", "zoh"}, optional 

Which method to use: 

 

* gbt: generalized bilinear transformation 

* bilinear: Tustin's approximation ("gbt" with alpha=0.5) 

* euler: Euler (or forward differencing) method ("gbt" with alpha=0) 

* backward_diff: Backwards differencing ("gbt" with alpha=1.0) 

* zoh: zero-order hold (default) 

 

alpha : float within [0, 1], optional 

The generalized bilinear transformation weighting parameter, which 

should only be specified with method="gbt", and is ignored otherwise 

 

Returns 

------- 

sysd : tuple containing the discrete system 

Based on the input type, the output will be of the form 

 

* (num, den, dt) for transfer function input 

* (zeros, poles, gain, dt) for zeros-poles-gain input 

* (A, B, C, D, dt) for state-space system input 

 

Notes 

----- 

By default, the routine uses a Zero-Order Hold (zoh) method to perform 

the transformation. Alternatively, a generalized bilinear transformation 

may be used, which includes the common Tustin's bilinear approximation, 

an Euler's method technique, or a backwards differencing technique. 

 

The Zero-Order Hold (zoh) method is based on [1]_, the generalized bilinear 

approximation is based on [2]_ and [3]_. 

 

References 

---------- 

.. [1] http://en.wikipedia.org/wiki/Discretization#Discretization_of_linear_state_space_models 

 

.. [2] http://techteach.no/publications/discretetime_signals_systems/discrete.pdf 

 

.. [3] G. Zhang, X. Chen, and T. Chen, Digital redesign via the generalized 

bilinear transformation, Int. J. Control, vol. 82, no. 4, pp. 741-754, 

2009. 

(http://www.mypolyuweb.hk/~magzhang/Research/ZCC09_IJC.pdf) 

 

""" 

if len(system) == 1: 

return system.to_discrete() 

if len(system) == 2: 

sysd = cont2discrete(tf2ss(system[0], system[1]), dt, method=method, 

alpha=alpha) 

return ss2tf(sysd[0], sysd[1], sysd[2], sysd[3]) + (dt,) 

elif len(system) == 3: 

sysd = cont2discrete(zpk2ss(system[0], system[1], system[2]), dt, 

method=method, alpha=alpha) 

return ss2zpk(sysd[0], sysd[1], sysd[2], sysd[3]) + (dt,) 

elif len(system) == 4: 

a, b, c, d = system 

else: 

raise ValueError("First argument must either be a tuple of 2 (tf), " 

"3 (zpk), or 4 (ss) arrays.") 

 

if method == 'gbt': 

if alpha is None: 

raise ValueError("Alpha parameter must be specified for the " 

"generalized bilinear transform (gbt) method") 

elif alpha < 0 or alpha > 1: 

raise ValueError("Alpha parameter must be within the interval " 

"[0,1] for the gbt method") 

 

if method == 'gbt': 

# This parameter is used repeatedly - compute once here 

ima = np.eye(a.shape[0]) - alpha*dt*a 

ad = linalg.solve(ima, np.eye(a.shape[0]) + (1.0-alpha)*dt*a) 

bd = linalg.solve(ima, dt*b) 

 

# Similarly solve for the output equation matrices 

cd = linalg.solve(ima.transpose(), c.transpose()) 

cd = cd.transpose() 

dd = d + alpha*np.dot(c, bd) 

 

elif method == 'bilinear' or method == 'tustin': 

return cont2discrete(system, dt, method="gbt", alpha=0.5) 

 

elif method == 'euler' or method == 'forward_diff': 

return cont2discrete(system, dt, method="gbt", alpha=0.0) 

 

elif method == 'backward_diff': 

return cont2discrete(system, dt, method="gbt", alpha=1.0) 

 

elif method == 'zoh': 

# Build an exponential matrix 

em_upper = np.hstack((a, b)) 

 

# Need to stack zeros under the a and b matrices 

em_lower = np.hstack((np.zeros((b.shape[1], a.shape[0])), 

np.zeros((b.shape[1], b.shape[1])))) 

 

em = np.vstack((em_upper, em_lower)) 

ms = linalg.expm(dt * em) 

 

# Dispose of the lower rows 

ms = ms[:a.shape[0], :] 

 

ad = ms[:, 0:a.shape[1]] 

bd = ms[:, a.shape[1]:] 

 

cd = c 

dd = d 

 

else: 

raise ValueError("Unknown transformation method '%s'" % method) 

 

return ad, bd, cd, dd, dt