@article{oai:mie-u.repo.nii.ac.jp:00006328,
 author = {李, 明珠 and Li, Mingzhu and 田尻, 功郎 and Tajiri, Isao and 佐藤, 邦夫 and Sato, Kunio and 伊佐, 務 and Isa, Tsutomu and 端, 俊一 and Hata, Shunichi},
 journal = {三重大學農學部學術報告 = The bulletin of the Faculty of Agriculture, Mie University},
 month = {Mar},
 note = {application/pdf, L-shaped six component force transducer was developed to measure six component forces on the basis of bending strains being interfered with the components. The component transducer has three sections for measuring. Two bending strains at each section are measured independently in axial directions.
To continue, it is necessary to define the general theme; The bending strain on each side of the beam consists of three portions. It is expressed by a equation summing these portions. Each portion is the strain that is brought about by a moment with a coefficient of sensitivity in each axial direction. 
Consequently, the one equation is concerned in three coefficients. When the measured strains are equal to the compound strains in consideration of interference, six equations bring about six component forces. They will be obtained by a solution of these equations. 
Then,it is most important that calibration is performed with accuracy as to the coefficients of sensitivity. There are two methods in calibration. One of them is linear regression that is used to determine the coefficients on the basis of relation of the strain and moment. The other is multiple regression. It depends on relation of the strain and two or three moments. We obtained 18 coefficients of sensitivity by both methods, and investigated the coefficients to compare the two cases. 
On the other hand, the known forces can be restored to the original state by the calculation using equations with coefficients. So, we explored restorable possibility as to six known forces in three axial directions. The following matters were made clear as a result. 
1. Main coefficient of sensitivity is nearly equal to 1/EZ among three coefficients in each equation, and the difference to the calculated values is 5.5%. E: modulus of elasticity, Z: section modulus.  
2. The values of coefficients on interfering strains among three portions are 3.7% of the main on an average. It becomes clear that twisting moment on L-shaped beam interferes remarkably in bending strains of measuring section 3 in comparison with other sections. Preceding ratios to the main are at most 7%.  
3. There is a shade of difference between coefficients by linear regression and ones by multiple regression without α₁,α₆.
4. We librated the three known forces in each axial direction at the point being apart from L-shaped beam end. Then, we calculated to reconstruct these forces on the basis of measured bending strains and simultaneous equation. The average of error in unit load is 5.3% on all reconstructions with both methods. In these cases, the reconstructions depending on coefficients by multiple regression are better than ones of linear regression. Its difference is about 0.8% on an average.  
5. The reconstruction of moments is more difficult than that of axial forces. The difference of error is about 1%.  
6. The measuring limit is about 300kgf on this transducer, and we can obtain six component forces under 5-7% error as to accurate values. It will be better for the user to attach this transducer to measuring apparatus in consideration of the axial directions. It will be selected on the basis of greater flexural rigidity to estimated large force. Then, main specifications are as follows: Cross section of the beam is 70 × 31 mm. Material is SS41P. Approximate dimension is 530×388 mm. Distance of gage setting is 100mm and 150mm.},
 pages = {69--79},
 title = {Ｌ型６分力計の試作に関する研究},
 volume = {64},
 year = {1982}
}