|
265 | 265 | \hat{\boldsymbol L}=\boldsymbol L+\boldsymbol V |
266 | 266 | $$ |
267 | 267 |
|
| 268 | +::: example |
| 269 | + |
| 270 | +为确定某一抛物线方程 $y^2=ax$,观测了 4 组数据,且 $x_i$ 无误差,$y_i$ 为相互独立的等精度观测值,试用间接平差求: |
| 271 | + |
| 272 | +1. 该抛物线方程; |
| 273 | +2. 待定系数 $\hat a$ 的中误差。 |
| 274 | + |
| 275 | +| 点号 | $x$/cm | $y$/cm | |
| 276 | +| ---- | ------ | ------ | |
| 277 | +| 1 | 1 | 1.90 | |
| 278 | +| 2 | 2 | 2.70 | |
| 279 | +| 3 | 3 | 3.35 | |
| 280 | +| 4 | 4 | 4.32 | |
| 281 | + |
| 282 | +--- |
| 283 | + |
| 284 | +依题意,有 $n=4,t=1,r=n-t=3$。 |
| 285 | + |
| 286 | +令 $\varGamma=\sqrt a$,则 |
| 287 | + |
| 288 | +$$ |
| 289 | +\hat y_i=\hat\varGamma\sqrt{x_i} |
| 290 | +$$ |
| 291 | + |
| 292 | +有 |
| 293 | + |
| 294 | +$$ |
| 295 | +\hat{\boldsymbol L}=\boldsymbol B\hat\varGamma |
| 296 | +$$ |
| 297 | + |
| 298 | +其中 |
| 299 | + |
| 300 | +$$ |
| 301 | +\boldsymbol L= |
| 302 | +\begin{bmatrix} |
| 303 | +1.90 \\ |
| 304 | +2.70 \\ |
| 305 | +3.35 \\ |
| 306 | +4.32 |
| 307 | +\end{bmatrix}, |
| 308 | +\quad |
| 309 | +\boldsymbol B= |
| 310 | +\begin{bmatrix} |
| 311 | +1 \\ |
| 312 | +\sqrt2 \\ |
| 313 | +\sqrt3 \\ |
| 314 | +2 |
| 315 | +\end{bmatrix} |
| 316 | +$$ |
| 317 | + |
| 318 | +取近似值 |
| 319 | + |
| 320 | +$$ |
| 321 | +\varGamma_0=\frac14\sum\frac{y_i}{\sqrt{x_i}}=1.97583 |
| 322 | +$$ |
| 323 | + |
| 324 | +设 |
| 325 | + |
| 326 | +$$ |
| 327 | +\hat\varGamma=\varGamma_0+\hat\lambda |
| 328 | +$$ |
| 329 | + |
| 330 | +因观测值等精度,故 |
| 331 | + |
| 332 | +$$ |
| 333 | +\boldsymbol P=\boldsymbol I |
| 334 | +$$ |
| 335 | + |
| 336 | +误差方程为 |
| 337 | + |
| 338 | +$$ |
| 339 | +\boldsymbol V=\boldsymbol B\hat\lambda-\boldsymbol l |
| 340 | +$$ |
| 341 | + |
| 342 | +其中 |
| 343 | + |
| 344 | +$$ |
| 345 | +\boldsymbol l=\boldsymbol L-\boldsymbol B\varGamma_0= |
| 346 | +\begin{bmatrix} |
| 347 | +-0.07583 \\ |
| 348 | +-0.09455 \\ |
| 349 | +-0.07224 \\ |
| 350 | +0.36834 |
| 351 | +\end{bmatrix} |
| 352 | +$$ |
| 353 | + |
| 354 | +法方程为 |
| 355 | + |
| 356 | +$$ |
| 357 | +\boldsymbol N_{bb}\hat\lambda-\boldsymbol U=0 |
| 358 | +$$ |
| 359 | + |
| 360 | +其中 |
| 361 | + |
| 362 | +$$ |
| 363 | +\begin{gathered} |
| 364 | +\boldsymbol N_{bb}=\boldsymbol B^{\rm T}\boldsymbol P\boldsymbol B |
| 365 | +=\boldsymbol B^{\rm T}\boldsymbol B=10 \\ |
| 366 | +\boldsymbol U=\boldsymbol B^{\rm T}\boldsymbol P\boldsymbol l |
| 367 | +=\boldsymbol B^{\rm T}\boldsymbol l=0.40245 |
| 368 | +\end{gathered} |
| 369 | +$$ |
| 370 | + |
| 371 | +故 |
| 372 | + |
| 373 | +$$ |
| 374 | +\hat\lambda=\boldsymbol N_{bb}^{-1}\boldsymbol U |
| 375 | +=\frac1{10}\times0.40245=0.04025 |
| 376 | +$$ |
| 377 | + |
| 378 | +于是 |
| 379 | + |
| 380 | +$$ |
| 381 | +\begin{gathered} |
| 382 | +\hat a=\hat\varGamma^2=4.06458 \\ |
| 383 | +\hat\varGamma=\varGamma_0+\hat\lambda=2.01608 |
| 384 | +\end{gathered} |
| 385 | +$$ |
| 386 | + |
| 387 | +又有 |
| 388 | + |
| 389 | +$$ |
| 390 | +\boldsymbol Q_{\hat\lambda\hat\lambda}=\boldsymbol N_{bb}^{-1}=\frac1{10} |
| 391 | +$$ |
| 392 | + |
| 393 | +改正数为 |
| 394 | + |
| 395 | +$$ |
| 396 | +\boldsymbol V=\boldsymbol B\hat\lambda-\boldsymbol l= |
| 397 | +\begin{bmatrix} |
| 398 | +0.11608 \\ |
| 399 | +0.15117 \\ |
| 400 | +0.14195 \\ |
| 401 | +-0.28784 |
| 402 | +\end{bmatrix} |
| 403 | +$$ |
| 404 | + |
| 405 | +单位权方差估值为 |
| 406 | + |
| 407 | +$$ |
| 408 | +\hat\sigma_0^2 |
| 409 | +=\frac{\boldsymbol V^{\rm T}\boldsymbol P\boldsymbol V}{r} |
| 410 | +=\frac{\boldsymbol V^{\rm T}\boldsymbol V}{3} |
| 411 | +=0.04644 |
| 412 | +$$ |
| 413 | + |
| 414 | +由误差传播定律, |
| 415 | + |
| 416 | +$$ |
| 417 | +\boldsymbol Q_{\hat a\hat a} |
| 418 | +=(2\hat\varGamma)^2\boldsymbol Q_{\hat\lambda\hat\lambda} |
| 419 | +=1.62583 |
| 420 | +$$ |
| 421 | + |
| 422 | +故 |
| 423 | + |
| 424 | +$$ |
| 425 | +D_{\hat a\hat a} |
| 426 | +=\hat\sigma_0^2\boldsymbol Q_{\hat a\hat a} |
| 427 | +=0.07550 |
| 428 | +=\sigma_{\hat a}^2 |
| 429 | +$$ |
| 430 | + |
| 431 | +因此 |
| 432 | + |
| 433 | +$$ |
| 434 | +\sigma_{\hat a}=\sqrt{\sigma_{\hat a}^2}=0.27478 |
| 435 | +$$ |
| 436 | + |
| 437 | +综上,有 |
| 438 | + |
| 439 | +$$ |
| 440 | +\hat a=4.065,\quad \sigma_{\hat a}=0.275 |
| 441 | +$$ |
| 442 | + |
| 443 | +抛物线方程近似为 $y^2=4.065x$。 |
| 444 | + |
| 445 | +::: |
| 446 | + |
268 | 447 | ## 应用要点 |
269 | 448 |
|
270 | 449 | - 测边网:参数取待定点坐标,观测方程为边长函数,对坐标线性化后列误差方程 |
|
0 commit comments