The Modeling Experimental Research Of Walking And Running Energy Expenditure Based On Pressure Sensors And Accelerometer

Objective:Building recognition regression models of walking and running energy expenditure based onpressure sensors and accelerometer. Finding the higher correlation parameters between the energyexpenditure and parameters obtained from the pressure and acceleration sensors. Taking intoaccount the height, age, weight and gait patterns of human affect the accuracy to improving themodeling accuracy. Model of physical activity energy expenditure to estimate the daily activitiesof the human body on the public for pedestrian movement, interference-free, comprehensiveconsideration of the evaluation environment and the object of individual factors applies, includingthe China Youth. The model has high practical significance and value of health guidance forChinese youthResearch Methods:Part of an experiment carried out in high school. Adolescents as subjects, including juniorhigh school the total number of68, Age from11years to14years (33boys,35girls). High schoolstudents aged from14to18years, Before the experiment，Collecting personal information，Measuring height, weight, leg length, percentage of body fat of subjects, resting heart rate andblood pressure, Wear apparatus, the subjects familiar with the treadmill for six minutes(Matsas etal.2000), Treadmill speed is3,4,5,6,7and8km/h, the speed of3,4,5and6km/h was walkingspeed;7km/h and8km/h were running speed. Each speed lasted5minutes; K4b2was used tocollect energy consumption data in experiment.Another part of an experiment carried out in Shanghai Sports Performance Laboratory.Subjects were20University student volunteers. Subjects were physical health, whose Lowerlimbs had not medical history. The subjects aged19.54±5.36. The walking protocol consisted oftreadmill walking for five min at each of the following speeds:0.6，0.8，1.0，1.2，1.4，1.6，1.8,and2.0m/s. The running protocol consisted of treadmill running for five min at each of thefollowing speeds:1.4,1.6,1.8,2.0,2.2,2.4,2.6, and2.8m/s. The subjects should rest to quiet statefrom walking to running mode conversion. VO2000was used to test quiet and exercise metabolicgas values. POLAR heart rate watch was used in experiment. Two-axis accelerometer was placedon the first sacral vertebra and sampling frequency was1000Hz.Mathematical statistics method: The data was processed using Matlab7.0, graphics using theOrigin8.0, stepwise regression, logistic regression, ANOVA, covariance analysis.Research Results:1) Established linear regression model between walking speed and step frequency, the linear regression model between step frequency and energy expenditure. The multiple correlationcoefficient of the regression equations were higher than0.75. Improve the accuracy of themeasures regression model.2) There was significant linear relationship between net energy expenditure and the verticalityacceleration RMS (R~2=0.92), AP acceleration RMS (R~2=0.81), Integral value of the verticalityacceleration (R~2=0.94) in walking pattern. If uniaxial accelerometer was used, the estimateaccuracy of energy expenditure would improve using the vertical axis of accelerometer. Multiplecorrelation coefficients of the velocity and RMS of AP acceleration, integral value of APacceleration were respectively R~2=0.82and R~2=0.81. The net energy expenditure and RMS ofverticality acceleration (R~2=0.61), integral value of verticality acceleration (R~2=0.61) had a linearrelationship, but multiple correlation coefficients was lower for RMS of AP acceleration (R~2=0.33),integral value of AP acceleration (R~2=0.33) in running pattern. Multiple correlation coefficient oflinear correlation larger difference about between running velocity and RMS of verticalityacceleration (R~2=0.58), integral value of verticality acceleration (R~2=0.55), between runningvelocity and RMS of AP acceleration (R~2=0.35), integral value of AP acceleration (R~2=0.39).3) There was a quadratic curve relationship between walking velocity and vertical reaction forcepeak (R~2=0.86), a linear relationship between vertical reaction force peak and energy expenditure(R~2=0.85); There was a quadratic curve relationship between walking velocity and WeightAcceptance Rate (R~2=0.85), a linear relationship between Weight Acceptance Rate and energyexpenditure (R~2=0.80); Peak plantar pressure slope and velocity, and the net energy expenditurehave a high linear relationship. There was a linear relationship between the walking velocity andvertical reaction force peak (R~2=0.79), Weight Acceptance Rate (R~2=0.87), the third mask WeightAcceptance Rate (R~2=0.75). Multiple correlation coefficient of linear correlation was lowerbetween running velocity and Weight Acceptance Rate (R~2=0.69). The walking energyexpenditure and Weight Acceptance Rate had a high linear relationship. Multiple correlationcoefficient of linear correlation between running energy expenditure and Weight Acceptance Ratewas R~2=0.73, but for stance time was R~2=0.64.4) We established fitting equations between energy expenditure (energy consumed per kilogram ofbody weight for a given walking time) and velocity of walking and running. This analysis revealedthat a quadratic curve relationship between walking velocity and energy expenditure (R~2=0.88), alinear relationship between running velocity and energy expenditure (R~2=0.72). Coordinates of theintersection of two fitted curves is (2.35m/s,141.7cal/kg/min). In the test velocity range, at thesame velocity, the energy expenditure of walking and running is significant difference (P<0.01),running burns more calories per kilogram bodyweight per unit minute than walking; We alsoestablished fitting equations between energy cost (energy consumed per kilogram of body weight per unit distance) and velocity of walking. This analysis revealed that a quadratic curverelationship between walking velocity and energy cost(R~2=0.98), The fitted curve lowest pointcoordinates is (1.14m/s，0.553cal/kg/m), a linear relationship between running velocity and energycost(R~2=0.68).Conclusions：1) Movement velocity and step frequency has a high linear correlation. Height affects the stepfrequency; height increase has led to decreases in step frequency at the same speed. Height shouldtake into account affect the correlation between step velocity and step frequency. Accuracy of theregression equation in step frequency and step velocity will increase when grouped according toheight. Age factor should be considered in the establishment the regression model of stepfrequency and energy expenditure. Within the error tolerance, the model as a fitness activityevaluation is reliable.2) Using the waist accelerometer can better estimate the energy expenditure and velocity of humanwalking running; RMS value, integral value and energy expenditure as well as the velocity has ahigh linear relationship after the acceleration signals detrended. Even if the acceleration integral isequal, the energy expenditure is not equal in walking and running pattern; Acceleration RMS ofthe horizontal direction and vertical direction of the integral value can effectively distinguishbetween walking and running pattern.3) The peak vertical reaction force and travel speed was high quadratic relationship; there is higherlinear relationship between peak vertical reaction force and energy expenditure; WeightAcceptance Rate and velocity have a high quadratic relationship; Weight Acceptance Rate andenergy expenditure is high linear relationship.4) Quadratic curve relationship between walking velocity and energy expenditure，a linearrelationship between running velocity and energy expenditure; In a certain range of velocity,running burns more calories per kilogram bodyweight per unit minute than walking; Therelationship of the energy cost and the velocity is U-shaped curve for walking; the relationship ofthe energy cost and the velocity is sloping line for running. The higher running velocity the lowerthe energy cost in the measuring range.