Accelerometer Output Research Articles

Calibration and Cross-Validation of the ActiGraph wGT3X+ Accelerometer for the Estimation of Physical Activity Intensity in Children with Intellectual Disabilities.

BackgroundValid objective measurement is integral to increasing our understanding of physical activity and sedentary behaviours. However, no population-specific cut points have been calibrated for children with intellectual disabilities. Therefore, this study aimed to calibrate and cross-validate the first population-specific accelerometer intensity cut points for children with intellectual disabilities.MethodsFifty children with intellectual disabilities were randomly assigned to the calibration (n = 36; boys = 28, 9.53±1.08yrs) or cross-validation (n = 14; boys = 9, 9.57±1.16yrs) group. Participants completed a semi-structured school-based activity session, which included various activities ranging from sedentary to vigorous intensity. Direct observation (SOFIT tool) was used to calibrate the ActiGraph wGT3X+, which participants wore on the right hip. Receiver Operating Characteristic curve analyses determined the optimal cut points for sedentary, moderate, and vigorous intensity activity for the vertical axis and vector magnitude. Classification agreement was investigated using sensitivity, specificity, total agreement, and Cohen’s kappa scores against the criterion measure of SOFIT.ResultsThe optimal (AUC = .87−.94) vertical axis cut points (cpm) were ≤507 (sedentary), 1008−2300 (moderate), and ≥2301 (vigorous), which demonstrated high sensitivity (81−88%) and specificity (81−85%). The optimal (AUC = .86−.92) vector magnitude cut points (cpm) of ≤1863 (sedentary), 2610−4214 (moderate), and ≥4215 (vigorous) demonstrated comparable, albeit marginally lower, accuracy than the vertical axis cut points (sensitivity = 80−86%; specificity = 77−82%). Classification agreement ranged from moderate to almost perfect (κ = .51−.85) with high sensitivity and specificity, and confirmed the trend that accuracy increased with intensity, and vertical axis cut points provide higher classification agreement than vector magnitude cut points.ConclusionsThis study provides the first valid methods of interpreting accelerometer output in children with intellectual disabilities. The calibrated physical activity cut points are notably higher than existing cut points, thus raising questions on the validity of previous low physical activity estimates in children with intellectual disabilities that were based on typically developing cut points.

Strain Modal Analysis of Small and Light Pipes Using Distributed Fibre Bragg Grating Sensors.

Vibration fatigue failure is a critical problem of hydraulic pipes under severe working conditions. Strain modal testing of small and light pipes is a good option for dynamic characteristic evaluation, structural health monitoring and damage identification. Unique features such as small size, light weight, and high multiplexing capability enable Fibre Bragg Grating (FBG) sensors to measure structural dynamic responses where sensor size and placement are critical. In this paper, experimental strain modal analysis of pipes using distributed FBG sensors ispresented. Strain modal analysis and parameter identification methods are introduced. Experimental strain modal testing and finite element analysis for a cantilever pipe have been carried out. The analysis results indicate that the natural frequencies and strain mode shapes of the tested pipe acquired by FBG sensors are in good agreement with the results obtained by a reference accelerometer and simulation outputs. The strain modal parameters of a hydraulic pipe were obtained by the proposed strain modal testing method. FBG sensors have been shown to be useful in the experimental strain modal analysis of small and light pipes in mechanical, aeronautic and aerospace applications.

An Analysis System for Attitude Angles of Four-rotor Aircrafts

accurate attitude angles can make the control of four- rotor aircrafts more precise. Various filters have been adopted to get more precise attitude angles. However, so far there is no suitable analysis system to evaluate the performance of filters. In this paper, an analysis system has been designed which can get the realistic attitude angles directly and compare the processing results of filters so as to analyze the performance of filters. The hardware of the analysis system contains the main control module, data acquisition module and wireless communication module. And the software of the system which is designed by LabVIEW mainly includes displaying and saving module, analysis module and the front panel. An extended Kalman filter and an innovation-based extended Kalman filter are compared by the system and results show that the attitude angles got by the innovation-based extended Kalman filter are more accurate than the extended Kalman filter. better fusion effect than Kalman filter for the nonlinear characteristic (5). In the paper (6), Kalman filter combines 3- axis gyroscope and computed quaternion to determine pitch and roll angles. This algorithm is adequate for the real-time estimation of the orientation of a quadrotor. A nonlinear complementary filter that combines accelerometer output for low frequency was proposed by Euston (7). The results were provided with a real-world data set and the performance of the filter was evaluated against the output from a full GPS/INS (Global Position System/ Inertial Navigation System) that was available for the data set. The Four-rotor aircraft system is nonlinear system. At present, the most widely estimator used in the four-rotor aircraft is extended Kalman filter. In this paper, a simple and practical analysis system is designed of hardware and software. The system implements a hardware platform of real-time data measurement and transmission. The hardware contains the main controller design, the data acquisition module design and the wireless communication module design. The programs of the lower computer are written in the C++ language and the upper computer is designed by LabVIEW (8, 9). First, the attitude angles are measured by sensor MPU6050, and transmitted through the wireless communication module to the computer. Then these data are estimated accurately by the extended Kalman filter based on innovation and shown by the virtual oscilloscope. The system can provide more real and accurate data for the four-rotor aircrafts' control system and the interface of the system is real-time, high speed and intuitive.

Contactless on-bed activity sensing using first-reflection echolocation.

A novel system capable of detecting on-bed activity during sleep using first-reflection ultrasonic echolocation is described herein. As is employed in many existing solutions, such activity detection may be utilized in the assessment of sleep quality. Compared to current approaches using either wearable devices or sensors collocated on the surface of the bed, the proposed architecture greatly enhances convenience for the end-user by providing minimal disruptions to his or her standard sleep routine. A series of experiments were conducted in order to investigate the capacity of the system to detect activity during sleep. System performance was benchmarked against both a wrist-worn accelerometer as well as a smartphone application placed adjacent to the subject on the bed. Analysis demonstrates a statistically significant correlation between features computed from the system's output and the filtered activity data produced by the application, with maximum p values on the order of 10-3. Comparison with activity estimates formulated from the wrist-worn accelerometer output suggests stronger agreement, as indicated by increased correlation coefficient values.

An enhanced primary shock calibration procedure to reduce the zero shift effect of piezoelectric transducers by using a virtual amplifier

The low-frequency response of a charge amplifier induces an accelerometer (a combination of a piezoelectric transducer and a charge amplifier) output voltage with zero shift. Hence, a virtual amplifier with same input–output characteristics as the charge amplifier was designed to reduce the zero shift effect. The charge shock sensitivities of a piezoelectric transducer were evaluated by applying the input acceleration to the virtual amplifier. The results of the study indicated that the charge shock sensitivities were comparable to the vibration calibration results.

Statistical approaches to account for missing values in accelerometer data: Applications to modeling physical activity.

Physical inactivity is a recognized risk factor for many chronic diseases. Accelerometers are increasingly used as an objective means to measure daily physical activity. One challenge in using these devices is missing data due to device nonwear. We used a well-characterized cohort of 333 overweight postmenopausal breast cancer survivors to examine missing data patterns of accelerometer outputs over the day. Based on these observed missingness patterns, we created psuedo-simulated datasets with realistic missing data patterns. We developed statistical methods to design imputation and variance weighting algorithms to account for missing data effects when fitting regression models. Bias and precision of each method were evaluated and compared. Our results indicated that not accounting for missing data in the analysis yielded unstable estimates in the regression analysis. Incorporating variance weights and/or subject-level imputation improved precision by >50%, compared to ignoring missing data. We recommend that these simple easy-to-implement statistical tools be used to improve analysis of accelerometer data.

Agreement between activPAL3c accelerometers placed at different thigh positions

ActivPAL3c triaxial accelerometers are increasingly used to assess physical activity and sedentary behaviours. However, little is known how attachment site (left vs. right or upper vs. lower thigh) affects activPAL3c accelerometer outcomes. Twenty-eight adults wore four activPAL3c accelerometers attached 2cm above and below the midpoint of the left and right thigh during five minutes of laboratory-based walking at a self-selected pace, treadmill walking at 0.89 and 1.56m/s, and treadmill running at 2.22m/s, and during approximately seven hours of free-living conditions. One-way ANOVA revealed no statistically significant differences in accelerometer output between ActivPAL3c accelerometers at these locations during both laboratory-based, and free-living conditions. Intraclass correlation coefficients showed a high level of agreement between activPAL3c accelerometers during laboratory and free living activities (ICC(2,1) 0.76–1.00). The attachment sites used in this study do not appear to significantly influence outcomes for step count, sitting or walking time, or number of postural transitions from triaxial activPAL3c accelerometers during laboratory-based walking or free-living conditions. Therefore if necessary, to enhance participant compliance during monitoring periods, these sites could be used interchangeably. Studies of longer duration incorporating additional laboratory and recreational activities are needed to confirm these findings.

Effect of Raw Acceleration Filtering Methods on the Relationship Between Accelerometer Outputs and Energy Expenditure

Despite the ubiquity of accelerometers in physical activity research, there remains no consensus method for processing raw acceleration data into end-user appropriate metrics. One aspect of accelerometer data processing that has received limited attention is the manner in which raw acceleration data are filtered to remove non-physical noise prior to computing summary outputs. PURPOSE: To evaluate how raw acceleration filtering methods affect the relationship between accelerometer outputs and measured energy expenditure in children and adolescents. METHODS: 122 children and adolescents (61 boys, 13.0 ± 4.3 years, 22.4 ± 6.2 kg/m2) wore an ActiGraph GT3X+ triaxial accelerometer (80 Hz) on their non-dominant wrist while performing jumping jacks (JJ), dribbling a basketball (DB), and ambulating on a treadmill at 2.4 (T1), 4.8 (T2), and 7.2 km/h (T3). Raw data for each axis were filtered nine times using 4th order Butterworth bandpass filters defined by all combinations of three lower (0.15, 0.25, and 0.35 Hz) and upper (2.5, 4.5, and 8.5 Hz) cutoff frequencies. Average signal vector magnitude (SVM) was calculated each second for all nine filters and the average 1 s value for the last 2 min of each 5 min activity trial was retained. Oxygen uptake (VO2) was measured by the COSMED K4b2 system concurrent with accelerometry measures. METs were derived by dividing mass relative VO2 by resting energy expenditure (Schofield equation estimate). Pearson correlations were used to quantify associations between SVM and METs for each activity/filter combination. Meng’s Z-test was used to evaluate differences between dependent correlations within each activity. RESULTS: Correlations between SVM and METs across the evaluated filters for JJ, DB, T1, T2, and T3 ranged from 0.43–0.55, 0.10–0.25, 0.11–0.15, 0.19–0.32, and 0.11–0.22, respectively. Significant between-filter differences (p < 0.05) in the magnitude of association between SVM and METs were noted for JJ (n = 20), DB (n = 20), T2 (n = 9), and T3 (n = 8). CONCLUSION: Filter choice can significantly influence the relationship between an accelerometer’s output and measured energy expenditure. Further study is needed to identify filtering methods which produce accelerometer outputs that best reflect the energy cost of human movement. Support: NIH NICHD 1R21HD073807-01A1

Sedentary Sphere

Access to raw acceleration data should facilitate comparisons between accelerometer outputs regardless of monitor brand. To evaluate the accuracy of posture classification using the Sedentary Sphere in data from two widely used wrist-worn triaxial accelerometers. Laboratory: Thirty-four adults wore a GENEActiv and an ActiGraph GT3X+ on their nondominant wrist while performing four lying, seven sitting, and five upright activities. Free-living: The same participants wore both accelerometers on their nondominant wrist and an activPAL3 on their right thigh during waking hours for 2 d. Laboratory: Using the Sedentary Sphere with 15-s epoch GENEActiv data, sedentary and upright postures were correctly identified 74% and 91% of the time, respectively. Corresponding values for the ActiGraph data were 75% and 90%. Free-living: Total sedentary time was estimated at 534 ± 144, 523 ± 143, and 528 ± 137 min by the activPAL, the Sedentary Sphere with GENEActiv data and with ActiGraph data, respectively. The mean bias, relative to the activPAL, was small with moderate limits of agreement (LoA) for both the GENEActiv (mean bias = -12.5 min, LoA = -117 to 92 min) and ActiGraph (mean bias = -8 min, LoA = -103 to 88 min). Strong intraclass correlations (ICC) were evident for the activPAL with the GENEActiv (0.93, 0.84-0.97 (95% confidence interval) and the ActiGraph (0.94, 0.86-0.97). Agreement between the GENEActiv and ActiGraph posture classifications was very high (ICC = 0.98 (0.94-0.99), mean bias = +3 min, LoA = -58 to 63 min). These data support the efficacy of the Sedentary Sphere for classification of posture from a wrist-worn accelerometer in adults. The approach is equally valid with data from both the GENEActiv and ActiGraph accelerometers.

The development and validation of using inertial sensors to monitor postural change in resistance exercise

This research presented and validated a method of assessing postural changes during resistance exercise using inertial sensors. A simple lifting task was broken down to a series of well-defined tasks, which could be examined and measured in a controlled environment. The purpose of this research was to determine whether timing measures obtained from inertial sensor accelerometer outputs are able to provide accurate, quantifiable information of resistance exercise movement patterns. The aim was to complete a timing measure validation of inertial sensor outputs. Eleven participants completed five repetitions of 15 different deadlift variations. Participants were monitored with inertial sensors and an infrared three dimensional motion capture system. Validation was undertaken using a Will Hopkins Typical Error of the Estimate, with a Pearson׳s correlation and a Bland Altman Limits of Agreement analysis. Statistical validation measured the timing agreement during deadlifts, from inertial sensor outputs and the motion capture system. Timing validation results demonstrated a Pearson׳s correlation of 0.9997, with trivial standardised error (0.026) and standardised bias (0.002). Inertial sensors can now be used in practical settings with as much confidence as motion capture systems, for accelerometer timing measurements of resistance exercise. This research provides foundations for inertial sensors to be applied for qualitative activity recognition of resistance exercise and safe lifting practices.

Feasibility of a Chest-worn accelerometer for physical activity measurement.

A "proof-of-concept" study to examine the feasibility of wearing an Actigraph GT3X+ at the Chest (resembling a neck pendant) for physical activity measurement. A convenience sample of 45 healthy adults (23 male, mean age 20.0±4.5 years) was included in data analysis. Participants simultaneously wore three GT3X+ accelerometers, on the Waist, Wrist, and Chest and completed 8 bouts of slow (.67, .89, 1.11m/s), average (1.33, 1.56, 1.78m/s) and fast (2.00, 2.22m/s) walking on a treadmill. Paired t-test, correlations and absolute percentage errors (APE) of accelerometer output (vector magnitude, VM) were computed for the key pairs: Waist-Wrist; and Waist-Chest. The Wrist-site overestimated VM to a greater extent at all speeds in comparison to the Chest. Pearson's r correlations were weaker for Waist-Wrist (<.80) in comparison to the Waist-Chest (>.85). The APE's were much lower (i.e. higher agreement) for the Chest (9.23-15.5%) compared to the Wrist (19.7-54.9%). Participants also felt the Chest-site was more acceptable than the Waist-site. PA measurements recorded by a Chest worn GT3X+ more closely resembled PA measurements recorded at the traditional Waist site than when compared to the Wrist site. When combined with high Chest site preference, the findings of our study suggest that the Chest is a feasible site for accelerometer wear.

Calibration and correction method of the output signals of the triaxial accelerometer

Background. Calibration is one of the most important stages of work for putting into operation such navigation sensors as accelerometers. Objective. The aim of this study is to investigate the possibility of using the calibration and correction method of the output signals of the triaxial accelerometer. Methods. To work with the model outputs matrix methods of linear algebra are used. In particular, determining the unknown parameters of the model is based on the method of least squares. Correction algorithm is given in the form of a matrix notation for solving system of linear equations. Such methods of mathematical statistics as the standard deviation and the mathematical expectation were used for the output signals correction estimation. Results. Formulas for calculations of calibration parameters were obtained, and correction link for output accelerometer signals was developed. Experimental calibration was organized and the data obtained by the algorithm of calibration and correction link was processed. The results of studies for several test samples of triaxial accelerometers are presented. Conclusions. Using the calibration coefficients, which combine error scale factors and non-orthogonal axes errors can significantly reduce the computational load on the electronic signals processing unit and make output signals with a satisfactory accuracy.

Acceleration Measurement Drift Rejection in Motion Control Systems by Augmented-State Kinematic Kalman Filter

This paper deals with the use of MEMS accelerometers to improve the performances of positioning control systems equipped with low-resolution positioning sensors. A kinematic Kalman filter (KKF) is used to combine the position and acceleration measurements and get a smooth estimate of the kinematic variables, even in the presence of a coarse position quantization. Compared to similar schemes existing in literature, the state of the proposed KKF is augmented, to include the accelerometer output bias/drift among the variables estimated by the filter. In this way, the intrinsic robustness of the KKF scheme is further improved, by making the estimation process of the kinematic variables practically insensitive to the variation of the sensor bias/drift. The proposed KKF is used to provide a smooth and robust estimate of the kinematic variables to a positioning control system consisting of a two degrees-of-freedom (DOF) proportional–derivative (PD) position control combined with an acceleration-based disturbance observer (ADOB). Compared to a solution based on a conventional KKF, not accounting for the accelerometer output disturbance, the proposed solution exhibits better positioning performances, and insensitivity to the accelerometer output bias/drift. This feature is validated through several experimental tests on a positioning system based on a linear motor.

Prediction of activity-related energy expenditure using accelerometer-derived physical activity under free-living conditions: a systematic review.

Activity-related energy expenditure (AEE) might be an important factor in the etiology of chronic diseases. However, measurement of free-living AEE is usually not feasible in large-scale epidemiological studies but instead has traditionally been estimated based on self-reported physical activity. Recently, accelerometry has been proposed for objective assessment of physical activity, but it is unclear to what extent this methods explains the variance in AEE. We conducted a systematic review searching MEDLINE database (until 2014) on studies that estimated AEE based on accelerometry-assessed physical activity in adults under free-living conditions (using doubly labeled water method). Extracted study characteristics were sample size, accelerometer (type (uniaxial, triaxial), metrics (for example, activity counts, steps, acceleration), recording period, body position, wear time), explained variance of AEE (R(2)) and number of additional predictors. The relation of univariate and multivariate R(2) with study characteristics was analyzed using nonparametric tests. Nineteen articles were identified. Examination of various accelerometers or subpopulations in one article was treated separately, resulting in 28 studies. Sample sizes ranged from 10 to 149. In most studies the accelerometer was triaxial, worn at the trunk, during waking hours and reported activity counts as output metric. Recording periods ranged from 5 to 15 days. The variance of AEE explained by accelerometer-assessed physical activity ranged from 4 to 80% (median crude R(2)=26%). Sample size was inversely related to the explained variance. Inclusion of 1 to 3 other predictors in addition to accelerometer output significantly increased the explained variance to a range of 12.5-86% (median total R(2)=41%). The increase did not depend on the number of added predictors. We conclude that there is large heterogeneity across studies in the explained variance of AEE when estimated based on accelerometry. Thus, data on predicted AEE based on accelerometry-assessed physical activity need to be interpreted cautiously.

Standardized Physical Activities Measured with Actical Accelerometers in Patients after Transient Ischemic Attack

Objective: The purpose of the study was to quantitatively measure the amount of physical activity during standardized activities using accelerometers in patients with transient ischemic attack (TIA) in order to compare different physical activities based on objective measurements. Methods: Thirteen hospitalized TIA patients performed 11 different standardized activities with 5 Actical accelerometers attached to both wrists and ankles and the hip. The output was registered as activity counts (AC). Total AC were the sum of AC from all five accelerometers. Results: Higher total AC/minute were found in more intensive activities, e.g. the median total AC/minute were 11216 (IQR 8447-14153) during climbing stairs and 22 (IQR 12-42) during sleep. Conclusion: The study demonstrates that it is possible to compare different physical activities based on amounts of AC/minute obtained during standardized activities in TIA patients. Accelerometer outputs recorded during standardized physical activities enable future tailoring of specific rehabilitation programs for individual stroke patients aiming at a certain amount of AC by combining different specific physical activities.

Visual Inspection for Determining Days When Accelerometer Is Worn: Is This Valid?

Logs have been traditionally used for ascertaining accelerometer wear days in mail study designs, but not all participants complete logs. Visual inspection of accelerometer output may supplement missing logs; however, no data on the validity of this method are available. We compared visual inspection with participant logs in 197 women (mean age, 71.0 yr). Women were mailed an accelerometer to be worn during waking hours for 7 d, marking each wear day on a log before returning the accelerometer by mail. For every participant, we created a series of graphs of accelerometer counts by time of day (one chart for each day with accelerometer output, including mail days). Two raters, masked to log wear status, independently inspected these graphs and scored each day as "worn" or "not worn." The median (interquartile range) number of valid wear days using either visual inspection or log was 7 (7-7). For rater 1, the sensitivity and specificity of visual inspection was 99.7% (95% confidence interval, 99.2%-99.9%) and 97.2% (95.2%-98.6%), respectively; for rater 2, the sensitivity and specificity of visual inspection was 99.7% (99.2%-99.9%) and 97.0% (94.9%-98.4%), respectively. Interrater agreement was 99.5%. Visual inspection of accelerometer data is a valid alternative to missing participant wear logs when determining wear days in mail study designs.

Calibration of low-cost triaxial inertial sensors

Accelerometers (ACCs) and gyroscopes (gyros) are commonly known as inertial sensors and their orthogonal triads generally form an inertial measurement unit (IMU) used as a core means of a navigation system. Before the navigation system is to be used, it is necessary to perform its calibration. A typical process of the IMU calibration usually estimates scale-factors, orthogonality or misalignment errors, and offsets of both triads. These parameters compose the so-called sensor error model (SEM). The process of obtaining accurate information that describes the motion performed within the calibration generally requires a costly and specialized means [1], [2]. Therefore, much effort has been put into cost-effective calibration using an optical motion tracking system [3]–[6], or transferring the calibration into a state estimation problem [7]. In the ACC case, most of the current calibration methods utilize the fact that ACCs are affected by gravity when they are under static conditions. Therefore, we proceed with calibration performed under static conditions, which utilizes the knowledge of the gravity magnitude and ACC output measurements collected at predetermined orientations and performs ACC SEM estimation using nonlinear optimization [6], [8]–[10]. In the gyro case, calibration based on the Earth's rate might be inapplicable, for instance due to the fact that Earth's rate is under or around the resolution of the gyro, and thus other means to apply and measure angular rates need to be used. This situation commonly arises in the case of low-cost MEMS (Micro-Electro-Mechanical System) based gyros. Thus, expensive mechanical platforms are often inevitable [11]–[14].

An Adaptive Dual-Window Step Detection Method for a Waist-Worn Inertial Navigation System

In waist-worn pedestrian navigation systems, the periodic vertical acceleration peak signal at body centre of gravity is widely used for detecting steps. Due to vibration and waist shaking interference, accelerometer output signals contain false peaks and thus reduce step detection accuracy. This paper analyses the relationship between periodic acceleration at pedestrian centre of gravity and walking stance during walking. An adaptive dual-window step detection method is proposed based on this analysis. The peak signal is detected by a dual-window and the window length is adjusted according to the change in step frequency. The adaptive dual window approach is shown to successfully suppress the effects of vibration and waist shaking, thereby improving the step detection accuracy. The effectiveness of this method is demonstrated through step detection experiments and pedestrian navigation positioning experiments respectively. The step detection error rate was found to be less than 0·15% in repeated experiments consisting of 345 steps, while the longer (about 1·3 km) pedestrian navigation experiments demonstrated typical positioning error was around 0·67% of the distance travelled.

Wearable Goniometer and Accelerometer Sensory Fusion for Knee Joint Angle Measurement in Daily Life.

Human motion analysis is crucial for a wide range of applications and disciplines. The development and validation of low cost and unobtrusive sensing systems for ambulatory motion detection is still an open issue. Inertial measurement systems and e-textile sensors are emerging as potential technologies for daily life situations. We developed and conducted a preliminary evaluation of an innovative sensing concept that combines e-textiles and tri-axial accelerometers for ambulatory human motion analysis. Our sensory fusion method is based on a Kalman filter technique and combines the outputs of textile electrogoniometers and accelerometers without making any assumptions regarding the initial accelerometer position and orientation. We used our technique to measure the flexion-extension angle of the knee in different motion tasks (monopodalic flexions and walking at different velocities). The estimation technique was benchmarked against a commercial measurement system based on inertial measurement units and performed reliably for all of the various tasks (mean and standard deviation of the root mean square error of and , respectively). In addition, the method showed a notable improvement in angular estimation compared to the estimation derived by the textile goniometer and accelerometer considered separately. In future work, we will extend this method to more complex and multi-degree of freedom joints.

Temperature characteristics and error compensation for quartz flexible accelerometer

Reduction of error due to the influence of temperature on the quartz flexible accelerometer without any heating device is a difficult task, and is also a tendency for research and application. In this paper, static and dynamic temperature compensation models are established in order to reduce the temperature influence on accelerometer measurement accuracy. Combined with the experiment data, the relationship between the accelerometer output accuracy, temperature and the magnitude of acceleration is analyzed. The data collected from the temperature experiment show that output value of the accelerometer varies with temperature. The method of uniaxial quadrature experiment is adopted and the accelerometer output value is gauged at temperature ranging from ?20°C to 50°C. Having used the static and the dynamic temperature compensation models, the accelerometer temperature error compensation experiment is conducted and the compensated errors by the two models are analyzed. The result shows that the compensated value meets the technical requirements. Two technical indicators, the zero bias K0 and the scaling factor K1, which are used to measure the degree of accelerometers, are both improved and their fluctuation ranges are reduced.