误差理论与数据处理(费业泰)最全课后答案.doc
精选优质文档-倾情为你奉上专心-专注-专业误差理论习题答案误差理论习题答案1-4在测量某一长度时,读数值为2.31m,其最大绝对误差为 20um,试求其最大相对误差。解:最大相对误差(最大绝对误差)/测得值,所以6420 10 100%=8.66 10 %2.31最大相对误差1-5使 用 凯 特 摆 时 , 由 公 式21224hhgT()给 定 。 今 测 出 长 度12()hh为(1.042300.00005)m, 振动时间 T为(2.04800.0005)s,试求g 及最大相对误差。如果12()hh测出为(1.042200.0005)m,为了使g的误差能小于20.001/m s,T 的测量必须精确到多少?解:由21224()hhgT得2241.042309.81053/2.0480gm s对21224()hhgT进行全微分,令12hhh并 令ghT,代 替dddghT,得222348hh TgTT 从 而2ghTghT的最大相对误差为:4maxmaxmax0.000050.0005225.3625 10 %1.042302.0480ghTghT由21224()hhgT,得24ghT,所以243.141591.042202.047909.81053T1-7 为什么在使用微安表时,总希望指针在全量程的2/3范围内使用?解:设微安表的量程为0 nX,测量时指针的指示值为X,微安表的精度等级为S,最大误差%nX S,相对误差%nX SX,一般nXX,故当X越接近nX相对误差就越小,故在使用微安表时,希望指针在全量程的2/3范围内使用。1-9多级弹导火箭的射程为10000km时, 其射击偏离预定点不超过0.1km,优秀选手能在距离50m远处准确射中直径为2cm的靶心,试评述哪一个射击精度高?解:火箭射击的相对误差:30.1100%10 %10000选手射击的相对误差:20.02100%4 10 %50所以,相比较可见火箭的射击精度高。1-10 若用两种测量方法测量某零件的长度L1=100mm,其测量误差分别为11 m和9 m,而用第三种方法测量另一零件的长度L2 =150mm ,其测量误差为12 m,试比较三种测量方法精度的高低.解:第一种方法测量的相对误差为:11100%0.01%110mmm 第二种方法测量的相对误差为:精选优质文档-倾情为你奉上专心-专注-专业9100%0.0082%110mmm 第三种方法测量的相对误差为:12100%0.008%150mmm 相比较可知:第三种方法测量的精度最高,第一种方法测量的精度最低。第二章:误差基本原理1知识点:21.算术平均值32.标准差及算术平均值的标准差43.测量结果表达方式54.粗大误差判断及剔除2-2 测量某物体共8次,测得数据(单位为g)为236.45,236.37,23.51,236.34,236.39,236.48,236.47,236.40。试求算术平均值及其标准差.解:算术平均值为:111(236.45236.37236.51236.34236.39236.48236.47236.40)8236.43niixxng82211222222221810.02( 0.06)0.08( 0.09)( 0.04)0.050.04( 0.03)70.02510.05997niiiivvng 算术平均值的标准差是:0.05990.02128xgn2-3用别捷尔斯法、极差法和最大误差法计算习题2-2的标准差,并比较之。解: 别捷尔斯法:1|1.253(1)0.020.060.080.090.040.050.040.031.253871.253 0.410.513730.068655656niivn ng0.068650.024278xgn 极差法:精选优质文档-倾情为你奉上专心-专注-专业maxmin236.51236.340.17nxxg查表得82.85ndd所以0.170.05965 ,2.85nngd0.059650.021098xgn 最大误差法:查表得8110.61nKK所以max8|0.090.610.0549ivgK,0.05490.019418xgn综上所述,用贝塞尔公式得到的标准差是0.0212g,别捷尔斯法计算得到的标准是0.02427g、 极差法是0.02109g和最大误差法是0.01941g, 故最大误差法计算的得到的标准差最小,别捷尔斯法最大。2-9. 已知某仪器测量的标准差为0.5m。 若在该仪器上,对某一轴径测量一次,测得值为26.2025mm, 试写出测量结果。 若重复测量 10次, 测得值 (单位为mm) 为26.2025,26.2028,26.2028,26.2025,26.2026,26.2022,26.2023,26.2025,26.2026,26.2022,试写出测量结果。 若手头无该仪器测量的标准差值的资料,试由中10次重复测量的测量值,写出上述、的测量结果。解:lim33 0.51.50.0015xmmm 测量结果:(26.20250.0015)xmm1126.2025+26.2028+26.2028+26.2025+26.2023+26.2022+26.2023+26.2025+26.2026+26.20221026.2025niixxnmm44lim0.00051.58 10100.00053334.74 1010 xxxmmnmmn 测量结果:lim(26.20250.0005)xxxmm 可由测得数据计算得:1027414.2 102.16 1019iivmmn452.16 106.83 1010 xmmn所以 对,测量结果为:13(26.20250.0006)xxmm精选优质文档-倾情为你奉上专心-专注-专业对 ,测量结果为:3(26.20250.0002)xxxmm2-12 甲、乙两测量者用正弦尺对一锥体的锥角各测量五次,测得值如下:甲:7 220,7 30,7 235,7 220,7 215乙:7 225,7 225,7 220,7 250,7 245试求其测量结果。解:对于甲来说5157 220 7 30 7 235 7 220 7 21557 2307.04116ii 甲甲52122222(1)( 0.00276)0.008340.00134( 0.00276)( 0.00416 )5(51)0.00228iivn n 甲对于乙来说5157 225 7 225 7 220 7 250 7 24557 2337.0425ii 乙乙510.00167(1)iivn n乙所以两个测量者的权是:22221111:0.5360.002280.00167PP乙甲乙甲:不妨取0.536,1PP乙甲,所以,1.536PP乙甲。1 7.041660.5367.04257.04227 2321.536PPPP 乙乙甲甲乙甲精选优质文档-倾情为你奉上专心-专注-专业22221212221)(1)1 0.002280.5360.001670.00041.441 1.536iixxiiiPvPPPPmP 乙甲乙甲x乙甲()(33 1.444.32 即为所求。2-16对某一线圈电感测量10次,前4次是和一个标准线圈比较得到的,后6次是和另一个标准线圈比较得到的,测得结果如下(单位为mH) :50.82,50.83,50.87,50.89;50.78,50.78,50.75,50.85,50.82,50.81。试判断两组数据间有无系统误差。解:用秩和检验法有:将两组数据混合排列,得124,6nn,5.5791031.5T 因为14,30TT,TT,所以有根据怀疑存在系统误差。2-17 等精度测量某一电压10次,测得结果(单位为V)为25.94,25.97,25.98,26.01,26.04,26.02,26.04,25.98,25.96,26.07。测量完毕后,发现测量装置有接触松动现象,为判断是否接触不良而引入系统误差,将接触改善后,又重新作了10次等精度测量,测得结果(单位为V)为25.93,25.94,25.98,26.02,26.01,25.90,25.93,26.04,25.94,26.02。试用t检验法(取为 0.05)判断两测量值之间是否有系统误差。解:用t检验法判断:第一次测量的数据26.001x 2223111()1.549 101.549*1010nxiixxxn第二次测量数据:25.97y 10223111()0.02152.15 1010yiiyyyn所以2233(2)()()()10 10(10102)(26.00125.97)(1010)(10 1.549 10102.15 10 )1.53xyxyxyxxyyn n nntxynnnn因为1010218v ,取0.05,查 t 分布表,得2.10t精选优质文档-倾情为你奉上专心-专注-专业| | 1.532.10tt所以,无根据怀疑测量列中存在系统误差。2-19 对某量进行两组测量,测得数据如下:xi0.620.861.131.131.161.181.201.211.221.261.301.341.391.411.57yi0.991.121.211.251.311.311.381.411.481.501.591.601.601.841.95试用秩和检验法判断两组测量值之间是否有系统误差。解:将两组混合排列成下表:i12345678910ix0.620.861.131.131.161.181.201.21iy0.991.12i11121314151617181920ix1.221.261.301.341.39iy1.211.251.3113.11.38i21222324252627282930ix1.411.57iy1.411.481.501.591.601.601.841.95所以,数学期望为,112(1)15(15151)232.522n nna标准差,1212(1)15 15(15151)24.11212n n nn所以,174232.52.42724.1Tat 故,当置信概率98.36%P ,此时2.40t,| | 2.4272.40tt精选优质文档-倾情为你奉上专心-专注-专业此时有根据怀疑两组测量值之间存在系统误差。而当置信概率98.76%p 时,2.50t| | 2.4272.50tt此时无根据怀疑两组测量值之间存在系统误差。2-20 对某量进行 15 次测量, 测得数据为28.53,28.52,28.50,28.52,28.53,28.53,28.50,28.49,28.49,28.51,28.53,28.52,28.49,28.40,28.50,若这些测得值以消除系统误差, 试用莱以特准则、 格罗布斯准则和狄克松准则分别判断该测量列中是否含有系统误差的测量值。思路:1.莱以特准则:计算得2211.496 100.03271151niivn33 0.03270.0981根据莱以特准则,第14次测量值的残余误差14| 0.1040.0981v所以它含有粗大误差,故将它剔除。再根据剩下的14个测量值重复上述步骤。2.格罗布斯准则:28.504x ,0.0327,按照测量值的大小,顺序排列得,(1)(15)28.40,28.53xx现在有 2 个测量值(1)(15),xx可怀疑,由于(1)28.50428.400.104xx(15)28.5328.5040.026xx故应该先怀疑(1)x是否含有粗大误差,计算,(1)(1)28.50428.403.18040.0327xxg取0.05,查表得,0(15,0.05)2.41g,则103.1804(15,0.05)2.41gg故第 14 个测量值(1)x含有粗大误差,应剔除。注意:此时不能直接对 x(15)进行判断,一次只能剔除一个粗差。精选优质文档-倾情为你奉上专心-专注-专业3.重复上述步骤,判断是否还含有粗差。4.狄克松准则同理,判断后每次剔除一个粗差后重复。第三章:误差的合成与分解知识点:1.系统误差合成2.随机误差合成3.相关系数4.微小误差取舍原则5.误差的分解及等作用原则3-2为求长方体体积 V,直接测量其各边长 a=161.6mm,b=44.5mm,c=11.2mm,已知测量的系统误差为1.2amm ,0.8bmm ,0.5cmm ,测量的极限误差为0.8amm ,0.5bmm ,0.5cmm ,试求立方体的体积及其体积的极限误差。思路:1. 按测得值计算得 V;2. 根据系统误差的合成原理求得 V 的系统误差;3. 计算长方体的体积;4. 根据极限误差的合成原理求得极限误差;此时可写出测量结果表达式。解:因为Vabc30161.644.5 11.280541.44Vabcmm 44.5 11.2498.4161.6 11.21809.92161.644.57191.2VbcaVacbVabc体积的系统误差:3498.4 1.21809.92( 0.8)7191.20.52745.744VVVVabcabcmm 所以,长方体的体积是:3080541.442745.74477795.696VVVmm 极限误差为(局部误差方和根) :精选优质文档-倾情为你奉上专心-专注-专业2222222222222222223()()()()()()498.40.81809.920.57191.20.53729.11VabcabcVVVabcbcacabmm 所以,立方体的体积是377795.696mm,体积的极限误差是33729.11mm。3-4测 量 某 电 路 的 电 流22.5ImA, 电 压12.6UV, 测 量 的 标 准 差 分 别 为0.5,0.1IUmAV,求所耗功率PUI及其标准差p。解:先求所耗功率:312.622.5 100.2835PUIW因为,322.5 1012.6PIUPUI且 U,I 完全线形相关,故1,所以,22223222323366663()()2(22.5 10 )0.112.6(0.5 10 )2 1 22.5 1012.60.1 0.5105.0625 1039.69 1028.35 1073.1025 108.55 10puIuIPPPPUIUIW 所以,该电路所耗功率为0.2835W,其标准差为38.55 10 W。3-6已知x与y的相关系数1xy ,试求2uxay的方差2u。解:因为2 ,uuxaxy1xy 所以,222222222222()()242( 1)244(2)uxyxyxyxyxyxyxyxyuuuuxyxyxaxaxaxaxa 所以,2uxay的方差2u为2(2)xyxa精选优质文档-倾情为你奉上专心-专注-专业3-8如图 3-6 所示,用双球法测量孔的直径 D,其钢球直径分别为12,d d,测出的距离分别为12HH, 试求被测孔径 D 与各直接测量量的函数关系1212(,)Df d dH H及其误差传递函数。解:如图所示,图 3-6由勾股定理得22212121212()()()222ddddddDHH22212121212()()()222ddddddDHH即,22121212121212121212121212112212()()222()()22222()()2ddddddDHHddddddddddHHHHdddHHdHH然后对 d1,d2,H1,H2 分别求偏导,即得出误差传递系数。3-10假定从支点到重心的长度为L的单摆振动周期为T,重力加速度可由公式2LTg给出。 若要求测量g的相对标准差0.1%gg, 试问按等作用原则分配误差时, 测量L和T的相对标准差应该是多少?解:由重力加速度公式,2LTg得,222244 LLTggT因为,222348 ggLLTTT 因为测量项目有两个,所以2n 。按等作用原理分配误差,得2222411442222gggggLLTLgLgggnL110.1%0.07072%22gLLg同理,精选优质文档-倾情为你奉上专心-专注-专业232222411888222222 2ggggggTLTTTTTgTgLLLggnT 11|0.1%0.03536%2 22 2gTTg综上所述,测量L和T的相对标准差分别是0.07072%和0.03536%。第第五章:最小二乘法原理理知识点:1.最小二乘法原理2.正规方程3.两种参数估计的方法4.精度估计推荐掌握:基于矩阵的的最小二乘法参数估计参数最小二乘法估计矩阵形式的简单推导及回顾:由误差方程VLAX且要求 VTV 最小,则:() ()TTV VLAXLAX()()TTTLX ALAXTTTTTTL LL AXX A LX A AXXX令其等于f( ),要f( )最小,需其对应偏导为0:()0TTTTTTdd XL AL AA AXX A AXf( ), TTTTTL AX A AA LA AX所以:1()TTXA AA L理论基础:TTddddXXXXf( )=f( )ddddddXXXXXXXXXf( )g( )=g( )f( ) f( )g( )5-1 由测量方程32.9xy20.9xy231.9xy试求x、y的最小二乘法处理及其相应精度。解:方法一(常规) :1.列出误差方程组:精选优质文档-倾情为你奉上专心-专注-专业1232.9(32 )0.9(2 )1.9(23 )vxyvxyvxy32222ii1(2.9(3)(0.9(2 )(1.9(23 )Vxyxyxy分别对xy,求偏导,并令它们的结果为 0,2(3)2.9)32(2 )0.9)2(23 )1.9)20 xyxyxy 2(3)2.9)2(2 )0.9)22(23 )1.9)30 xyxyxy即,14513.45144.6xyxy 由上式可解得结果:0.9626x 0.0152y 2. 直接列表计算给出正规方程常数项和系数i1 ia2ia21 ia22ia12iia ail1 iia l2iia l1319132.98.72.921-214-20.90.9-1.832-349-61.93.8-5.7-1414-5-13.4-4.6可得正规方程14513.45144.6xyxy 将xy,的结果代入分别求得:1232.9(3 0.96260.0152)0.0030.9(0.962620.0152)0.03221.9(20.96263 0.0152)0.0204vvv得,322221231222( 0.003)( 0.0322)(0.0204)0.00146iivvvv 由题已知,32nt,得32311.46 100.038232ivnt得,由不定乘数的方程组1112111214515140dddd2122212214505141dddd精选优质文档-倾情为你奉上专心-专注-专业得110.0819d220.0819d得110.0382 0.08190.0109xd220.0382 0.08190.0109yd方法二(按矩阵形式计算) :由误差方程VLAX1232.9(32 )0.9(2 )1.9(23 )vxyvxyvxy上式可以表示为1122333 11223vlxvlyvl 即123vvvV;1232.90.91.9lll L;3 11223A;xy X可得11()TTTxy XC A LA AA L式中1111()3 13 12145121235142314514511145 145145171514TCA A所以精选优质文档-倾情为你奉上专心-专注-专业12.91453 1210.91451231711.92.94741310.92923321711.9164.612.61710.96260.0152Txy XC A L即解得,0.96260.0152xy将最佳估计值代入误差方程可得,1231233 11223vvvlxlyl VLAX3 12.90.96260.9120.01521.9230.00300.03220.0204 将计算得到的数据代入式中32311.46 100.03823-2ivnt为求出估计量xy,的标准差,首先求出不定常数ijd(12)ij ,。由已知,不定常数ijd的系数与正规方程的系数相同,因而ijd是矩阵1C中各元素,即1112121221451145171ddCdd精选优质文档-倾情为你奉上专心-专注-专业则11140.0819171d22140.0819171d可得估计量的标准差为110.0382 0.08190.0109xd220.0382 0.08190.0109yd5-3 测力计示值与测量时的温度t的对应值独立测得如下表所示。t/C151821242730/FN43.6143.6343.6843.7143.7443.78设t无误差,t值随t的变化呈线性关系0Fkkt,试给出线性方程中系数0k和k的最小二乘估计及其相应精度。解:利用矩阵求解,误差方程VLAX可写成1122330445566115118121124127130vlvlvlkvlkvlvl 即123456vvvvvvV;12345643.6143.6343.6843.7143.7443.78llllll L;115118121124127130A;0kkX可得011()TTTkkXC A LA AA L式中精选优质文档-倾情为你奉上专心-专注-专业1111()11511811111112115182124273012412713061351353165319513561351356135316531951351356945TCA A=所以0143.6143.63319513511111143.68135615182124273043.7194543.7443.7843.43240.01152TkkXC A L将最佳估计值代入误差方程VLAX得11223344556611511812143.43241240.011521271300.00480.009760.005680.001120.003440.002vlvlvlvlvlvl V精选优质文档-倾情为你奉上专心-专注-专业为求出估计量0kk,的标准差,需要求出不定乘数ijd的系数,而不定乘数ijd的系数与正规方程的系数相同,因而ijd是矩阵1C中各元素,即11121212231951351356945ddddC则1131953.38095945d2260.00635945d可得估计量的标准差为0110.00647 3.380950.00119kd1220.00647 0.006350.000516kd55 不等精度测量的方程组如下:35.6xy ,11P ;48.1xy,22P ;230.5xy,33P 试求x、y的最小二乘法处理及其相应精度。解:利用矩阵计算134121A;1235.68.10.5lll L;xy X;100020003P由1()TTxy XA PAA PL另精选优质文档-倾情为你奉上专心-专注-专业1001314202041311003211318641323214511 14TTCA AA PA得1141141114511451456291 14C则11()1005.614114210208.11453116290030.51.4352.352TTTxy XA PAA PLCA PL将最佳估计值代入误差方程VLAX,得1235.6131.4358.1412.3520.5210.0230.00950.0165vvvV可计算2221 0.023020.00953 ( 0.0165)0.039232Tnt V PV又知11140.022260.0223629d22450.071540.0715629d可得估计量的标准差为110.0392 0.02230.0059xd220.0392 0.07150.0105yd精选优质文档-倾情为你奉上专心-专注-专业57 将下面的非线性误差方程组化成线性的形式,并给出未知参数12xx,的二乘法处理及其相应精度。115.13vx31213.21()vxx228.26vx124123.01x xvxx解:1. 由前面三个线性的误差方程VLAX可解得12xx,的近似估计值1020 xx,利用矩阵形式求解:1235.138.2613.21lll L;123vvvV;10011 1A;12xxX可得1112()TTTxxXC A LA AA L式中1111()10101010111 12112211123TCA A所以,精选优质文档-倾情为你奉上专心-专注-专业1125.132110118.2612011313.215.1321 118.26121313.2115.2100124.600035.07008.2000TxxXC A L2. 取12xx,得近似值10205.07008.200 xx,令11012202xxxx可将误差方程线性化,现分别对测量方程求偏导11011111xxfax22011220 xxfax11022110 xxfax22022221xxfax11033111xxfax22033221xxfax110101022022024212122412211212()0.3818()()xxxxxxxxxxfxxxx xxaxxxxx220101022022024112121422221212()0.1460()()xxxxxxxxxxfx xxx xxaxxxxx则误差方程化成线性方程组VLA,1234vvvvV;1110201221020233102034410204(,)0.06(,)0.04(,)0.06(,)0.12lf xxllfxxllfxxllfxxl L;1211122122313241421001110.38180.1460aaaaaaaaA可得精选优质文档-倾情为你奉上专心-专注-专业1112()TTTC A LA AA L式中1111()101010.3818010110.1460110.38180.14602.14581.05571.05572.02130.62720.32760.32760.6658TCA A所以1120.060.62720.32761010.38180.040.32760.66580110.14600.060.120.060.62720.32760.29960.19160.040.32760.66580.33820.02790.060.120.0TC A L1640.0100解得,120.01640.0100 则11015.07000.0164 5.0536xx22028.20000.0100 8.1900 xx解得,120.01640.0100 则11015.07000.0164 5.0536xx精选优质文档-倾情为你奉上专心-专注-专业22028.20000.0100 8.1900 xx可得,0.02510.112042TntV V再由,10.62720.32760.32760.6658C则,110.6272d220.6658d可得估计量的标准差为,1110.1120 0.62720.0887xd2220.1120 0.66580.0914xd第六章回归分析知识点:1.一元线性回归2.多元线性回归3.方差分析及显著性检验6-1 材料的抗剪强度与材料承受的正应力有关。对某种材料试验的数据如下:正应力/x Pa26.825.428.923.627.723.9抗剪强度/y Pa26.527.324.227.123.625.9/x Pa24.728.126.927.422.625.6/y Pa26.322.521.721.425.824.9假设正应力的数值是精确的,求抗剪强度与正应力之间的线性回归方程。当正应力为24.5Pa时,抗剪强度的估计值是多少?解:序号/x Pa/y Pa2x2yxy126.826.5718.24702.25710.2225.427.3645.16745.29693.42328.924.2935.21585.64699.38精选优质文档-倾情为你奉上专心-专注-专业423.627.1556.96734.41639.56527.723.6767.29556.96653.72623.925.9571.21670.81619.01724.726.3610.09691.69649.61828.122.5789.61506.25632.25926.921.7723.61470.89583.731027.421.4750.76457.96586.361122.625.8510.76665.64583.081225.624.9655.36620.01637.44和311.6297.28134.267407.87687.761211311.625.9712iixxn,1211297.224.7712iiyyn22211()311.68134.2643.0512ninixxiixlxn,22211()297.27407.847.1512niniyyiiylyn111311.6297.27687.7629.5312nniiniixyiiixylx yn 29.530.686143.05xyxxlbl ,0297.2311.60.686142.58181212bybx所以,综上所述,42.58180.686142.580.69yxx。 当正应力x为 24.5Pa时,抗剪强度的估计值是:42.580.6861 24.525.77525.8yPaPa6-7在 4 种不同温度下观测某化学反应生成物含量的百分数,每种在同一温度下重复观测三次,数据如下:温度x/C150200250300生成物含量的百分数y77.476.778.284.184.583.788.989.289.794.894.795.9求y对x的先行回归方程,并进行方差分析和显著性检验精选优质文档-倾情为你奉上专心-专注-专业解:ty为同一温度三次下观测生成物含量的百分数的平均值,177.476.778.277.4333 3y,284.1 84.5 83.784.1 3y,388.9 89.2 89.789.2673y,494.8 94.795.995.13333yN1N21N2212N211900900225421500900202500N4 215000202500 12500iiiiiiNiixxiixxxxxlxN()()1212212N2110345.9333345.933386.4833430087.614()345.93334 29917.4620() 30087.61429917.4620 170.1521456.66750.116512500NiiNiiNiiNiiyyiixyyyyyyyNylyNlblbyb 86.48330.1165225 60.270860.27080.1165xyxi1i11i1i1i1479291.66()()900345.9333477834.9925()()79291.6677834.992541456.6675NiiNNiiiNNNiiiixyNx yxyNx yxylN现将计算结果写入方差分析表中由于1531.14F 0.01(1,8)11.26F,回归高度显著12.022F 0.01(1,10)10.04F故回归方程拟合的很好。解:将表中xy,画图得曲线如图所示,从曲线上按 x0.5 读取xy,y,2y列入下表。因表中2y极接近常数,此组观测数据可用2yabxcx表示观测值自图上读数值顺序差值xyxyy2y0.200.4004.4-0.20.504.320.44.20.60.40.704.450.84.60.40.81.205.331.25.40.41.21.606.881.66.60.61.82.108.912.08.40.42.22.5011.222.410.60.52.72.8013.392.813.30.53.23.2016.533.216.50.53.73.7021.203.620.26-12炼焦炉的焦化时间y与炉宽1x及烟道管相对温度2x的数据如下:精选优质文档-倾情为你奉上专心-专注-专业y/min6.4015.0518.7530.2544.8548.9451.5561.50100.44111.421x/m1.322.693.564.415.356.207.128.879.8010.652x1.153.404.108.7514.8215.1515.3218.1835.1940.40求回归方程0122ybb xb x,检验显著性,并讨论1x,2x对y的影响。解:利用最小二乘法:所求的回归方程为:0122ybb xb x式中的012bbb, ,是回归方程的回归系数。对每一组的1x,2x可以确定一个回归值ty,实际测得值ty与回归值ty之差就是残余误差tv,利用矩阵形式:令6.4011.321.1515.0512.693.4018.7513.564.1030.2514.418.7544.8515.3514.82 48.9416.2015.1551.5517.1215.3261.5018.8718.18100.4419.8835.19111.421 10.6540.40YX12340516278910 vvvvbvbvbvvvv bV则误差方程化成矩阵形式为V = Y - Xb根据最小二乘原理,回归系数的矩阵解为T-1TbX XX Y = CB()精选优质文档-倾情为你奉上专心-专注-专业1211211121112212211111111111111.322.693.564.415.356.207.128.879.8010.6 51.153.404.108.7514.8215.1515.3218.1835.1940.40NNttttNNNtttttttNNNtttttttNxxxxx xxx xxTA = X X11.321.1512.693.4013.564.1014.418.7515.3514.8216.2015.1517.1215.3218.8718.1819.8835.191 10.6540.401059.97156.4659.97446.99651282.5215156.461282.52153991.12083t131t132t16.4015.0518.7530.25111111111144.85 1.322.693.564.415.356.207.128.879.8010 .6548.941.153.404.108.7514.8215.1515.3218.1835.1 940.4051.5561.50tttttyx yx yTBX Y100.44111.42489.15 3875.936511749.8781精选优质文档-倾情为你奉上专心-专注-专业555443331.3916 101.3915 106.9757 101 3.8648 101.5431 103.4423 101059.97156.466.9757 103.4423 10873.564159.97446.99651282.5215156.461282.52153991.12080.85320.23720.0428 0.237-1C = A20.09460.02110.04280.02110.00540.85320.23720.0428489.150.23720.09460.02113875.93650.04280.02110.005411749.87810.58002.71222.0497 b = CB所求回归方程为:120.58002.71222.0497yxx检验回归方程的显著性:021 12 212110217() 2.6997942.50402.04534096.6372 10923.330110956.9961 10923.3301 33.6660/10923.3301/ 21135.6162/ (1)33.6660 / 7QMjyyytjyyMNMUyybLblb lQLUUMFQNM 由于0.011135.6162(2.7)12.25FF因此回