A study on improving bone conduction speaker performance by electromagnetic prediction and performance distribution by statistical analysis method

• Published in: Journal of mechanical science and technology Vol. 31; no. 4; pp. 1673 - 1681
• Main Authors: Ko, Dong Shin, Hur, Deog Jae, Park, Tae Won, Lee, Jai Hyuk
• Format: Journal Article
• Language: English
• Published: Seoul Korean Society of Mechanical Engineers 01.04.2017; Springer Nature B.V; 대한기계학회
• Subjects: Bones; Computer simulation; Control; Design engineering; Design factors; Design improvements; Design parameters; Dynamical Systems; Electromagnetic properties; Engineering; Fractional factorial design; Industrial and Production Engineering; Interaction parameters; Mathematical models; Mechanical Engineering; Monte Carlo simulation; Nonlinearity; Parameters; Performance prediction; Regression analysis; Regression models; Reliability; Response surface methodology; Statistical analysis; Statistical prediction; Test procedures; Vibration; 기계공학; Electromagnetic analysis; Bone conduction speaker; Design of experiments; Monte Carlo simulation
• ISSN: 1738-494X; 1976-3824
• DOI: 10.1007/s12206-017-0315-x

The present paper is focused on the stochastic characteristics of the electromagnetic force, one of the performance parameters for the bone conduction speaker which is one of the devices in the “smart glass”. The design parameters were taken as significant, affecting the electromagnetic force. Characteristic analysis of significant parameters was considered by using the factorial design method. Significant factor of main effect was selected via fractional factorial design method. Main effect and interaction of selected factor were analyzed applying the full factorial design method. The independency of the selected parameter and their significant interaction were examined by using the F-test method. Linear and non-linear characteristics for the selected parameters and performance were examined by creation of the median point within the analysis results for significance analysis. Therefore, prediction model derived non-linear regression model from the central composite design of response surface method. For probability distribution of the electromagnetic characteristics, related prediction model and Monte Carlo simulation method were applied. Electromagnetic performance prediction result showed 98.5 % level and improved maximum 99.8 % reliability level under 3σ level of dimensional management. In view of this, such stochastic design approach could improve design efficiency via verification of individual design parameters’ effect on the performance levels, thereby proving design reliability based on the object levels.