Mahalanobis Distance Algorithm Research Articles

Robust adaptive non‐linear alignment algorithm for SINS/DVL integrated navigation system based on variational Bayesian

AbstractThe problem of dynamic attitude alignment for strapdown inertial navigation system (SINS) and Doppler velocity log (DVL) integrated navigation has become a research hotspot. Underwater complex environment makes DVL output vulnerable to non‐Gaussian noise pollution, which makes it difficult to converge the SINS misalignment angle to within 1° based on analytical coarse alignment. This is to say, the performance of SINS fine alignment via Kalman filter will be degraded because the model of initial alignment exhibits non‐linearity. The inaccurate and/or unknown prior information of measurement noise statistical characteristics will also degrade the filtering performance. To ensure the SINS/DVL alignment accuracy under non‐linear, non‐Gaussian and uncertain conditions, this paper proposed a robust adaptive unscented Kalman filter (UKF) based on Variational Bayesian (VB) method (VBRAUKF). The proposed VBRAUKF improves the adaptability and robustness of UKF from the following two main aspects. Firstly, an adaptive estimation strategy for the measurement noise covariance is designed based on the VB algorithm, which can restrain the effect of inaccurate observation model and improve the adaptive ability of the filtering method. Secondly, an expansion factor is designed based on Mahalanobis distance algorithm, which can restrain the effect of non‐Gaussian noise and improve the robustness of the filtering method. The experimental results for the problem of the SINS/DVL dynamic alignment under mixed Gaussian noise and/or outlier conditions demonstrate the superiority of the proposed VBRAUKF over the traditional ones.

Design and validation of novel maize grain cleaning loss detection system based on classification models of particle time-domain signals

Quantitative evaluation of grain cleaning loss is a central concern in maize intelligent direct harvesting. This study integrates particle time-domain signals with machine learning models to develop a novel grain cleaning loss detection system and create modular embedded companion software. Specifically, collisional time-domain signals of grains and impurities with varying mass gradients are captured, and abnormal data are eliminated utilizing the Mahalanobis Distance (MD) algorithm. Signal classification models incorporating diverse preprocessing methods are trained using Naive Bayes (NB) and Support Vector Machine (SVM) algorithms. Model performance is evaluated through internal leave-one-out-cross-validation (LOOCV) and external validation sets. The best classification model is integrated into the grain loss detection software, and bench tests verify real-time cleaning loss rates for detection systems. Results demonstrate that the SVM model with Savitzky-Golay (S-G) smoothing + Second-Order Derivative (SOD) preprocessing achieves the highest predictive accuracy, with 97.44 % and 96.15 % in training and validation sets, respectively. When the feeding rate of maize mixtures is 5 kg/s, absolute grain loss rates for three measurements are 1.62 %, 1.72 % and 1.86 %, respectively. Compared to traditional loss detection systems with multifunctional circuits, the novel loss detection system enhances the average accuracy of grain loss rate by 6.54, 6.82 and 8.93 percentage points. Hence, the proposed method and apparatus can be applied to detect grain cleaning loss quantitatively.

Health Condition Identification and Quantitative Assessment of Emergency Diesel Generator Sets Based on Back Propagation Neural Network and Mahalanobis Distance

To address the problem of a comprehensive assessment of emergency diesel generator set health status, a quantitative assessment method of emergency diesel generator set health status based on data-driven and distance measurement is proposed. BP neural network was used to extract the feature sequences of emergency diesel generator sets under different health states. An emergency diesel generator set health state identification model was established. To further quantitatively evaluate the health level of the emergency diesel generator set, the distance between each state of the emergency diesel generator set and the health state is measured by introducing the Mahalanobis distance algorithm, which realizes the unification of both qualitative and quantitative evaluation of the health state of the emergency diesel generator set. The operating states of diesel generator sets were subdivided into four states: healthy, sub-healthy, abnormal, and fault, and a simulation test study was conducted using the actual measurement data set of emergency diesel generator set operation. It is verified that the proposed method can classify the operating status of diesel generator sets more accurately and obtain the health assessment score index. The integration of qualitative and quantitative assessment of the health status of emergency diesel generator sets is realized, and the status warning is carried out accordingly, which provides an accurate basis for taking different maintenance and repair measures in a targeted manner.

A near‐infrared spectroscopy method for the detection of texture profile analysis of Litopeneo vannamei based on partial least squares regression

AbstractAmong aquatic products freshness testing methods, spectral detection method is the most commonly used rapid nondestructive testing method, and the quality structure index can be used as the judgment index of aquatic products freshness, so studying the near‐infrared (NIR) spectroscopy detection method of aquatic products based on texture index is helpful to improve the detection accuracy of freshness of products. In order to find a method for NIR spectroscopy based on texture indicators, we took shrimp as the research target, collected total 216 NIR spectral data of 680–2600 nm and texture index of sample for 8 consecutive days, and established a quantitative prediction model. In this study, we compared the effects of six spectral pretreatments on partial least squares regression (PLSR) models, including multiple scattering correction, standard normal variate transformation, mean centering, Savitzky–Golay (S‐G) smoothing, moving average method, and normalization. The results show that the PLSR model based on S‐G smoothing has the best model accuracy in prediction after the S‐G smoothing algorithm has preprocessed the spectral data. In addition, principal component analysis and the Mahalanobis distance algorithm were used in the study to remove abnormal samples to improve the accuracy of the model. Finally, the quantitative prediction model for qualitative construct indicators constructed based on the S‐G smoothing algorithm and the PLSR algorithm had better prediction results than the other models algorithm we considered, with a correlation coefficient (R) of 0.868 for the cohesive validation set and 0.781 for the elastic validation set of qualitative construct indicators. Finally, based on the above study, this study developed a prediction graphical user interface system for qualitative metrics of texture.Practical ApplicationsMost of the existing aquatic meat freshness detection methods are offline detection based on physical and chemical tests, and some use spectral technology to detect the chemical parameters of aquatic meat freshness. This project innovatively applies NIR spectroscopy technology to aquatic products. The physical parameters of meat freshness can be quickly detected online. By preprocessing the data samples and selecting the response characteristic wavelengths, a NIR spectroscopy prediction model of texture indicators with self‐learning ability and high precision is constructed, and the texture is established based on this. Indicator NIR spectroscopy prediction system. This whole process provides new research methods and analytical tools for rapid nondestructive testing of aquatic product freshness.

Development and experimental analysis of a small integrated edge navigation sensor based on principle of circular arc array.

Signal, accuracy, and real-time performance of satellite, radar, and machine vision is a subject of concern in various complex agricultural environments. Therefore, the demand for a robust navigation sensor for indoor and vertical agricultural environment remains crucial, and it is a significant subject. In view of this, the relative edge pose detection method based on the ideal target band principle of the lateral center arc array, in this research, a small integrated arc array navigation sensor module based on adaptive detection arc technology, is developed, which costs about $100, autonomous edge navigation position, and attitude detection is realized in facility agriculture environment with continuous structured corridor or roadside features. In this research, a coupling method of reducing the radius of distance sensor arrangement, adjusting the unequal center angle, and increasing the detection distance is used to realize the miniaturization of the arc array arrangement. A semicircular modular rocket was designed to slide and adjust the center angle of the distance sensor, and the longitudinal installation position of the modularized sensor was adjusted by translating the circular arc of the detection; the convenient moving arrangement under different vehicle width and detection arc characteristics is realized. An adaptive construction method of detecting a circular arc based on self-calibrating detection distance of a distance sensor is proposed, which effectively reduces the difficulty of arranging the lateral central circular arc array; the fast construction of lateral detection arc is realized. In addition, in order to improve the accuracy and stability of the pose detection, the Mahalanobis distance algorithm and the standard Kalman filter are used to optimize the estimation of the ranging information and the relative pose of the edge. The experimental results show that the small integrated arc array navigation sensor module can independently construct photoelectric detection arcs with different characteristics to detect the position and attitude of the relative edge. When the road surface is concave and convex, the small integrated arc array navigation sensor module can still maintain the stable position and attitude detection of the relative edge for more than 30 s. In addition, when the walking speed of the autonomous navigation platform is 0.15 m/s to 0.35 m/s, the detection errors of lateral deviation and heading deviation relative to the road edge are less than 40 mm and 4.5°, respectively. The small integrated arc array navigation sensor module is less affected by the change of operating speed, and still has good accuracy and stability. The results show that the modularized edge navigation sensor has the advantages of fast and convenient use, high accuracy, and low cost; it can be applied to autonomous edge navigation control in greenhouse and plant and animal factories.

Selection of Myrciaria dubia clones under conditions of the savanna/forest transition of Roraima through multivariate analysis

Camu-camu (Myrciaria dubia), a fruit from the Amazon region, has received attention due to its high content of antioxidant compounds such as ascorbic acid and polyphenols. This study’s objective was to select camu-camu clones according to their morphological characteristics using multivariate analyses. We evaluated 56 camu-camu clones distributed in three blocks with two plants per experimental plot. The variables measured were plant height (H), number of basal branches (NBB), number of terminal buds (NTB), basal stem diameter (BSD), chlorophylls a (Chl a) and b (Chl b), and total chlorophyll (Chl a + b). Principal component (PCA) and multivariate clustering analyses were performed using the average linkage mean and Mahalanobis distance algorithms. After 24 months, results showed that the characteristic that least contributed to clone selection was NBB. The plant height (H), number of terminal buds (NTB), basal stem diameter (BSD), chlorophylls a (Chl a) and b (Chl b), and total chlorophyll (Chl a + b) had the most positive contribution towards the initial selection of camu-camu clones at two years of age. The 22 camu-camu clones showed the highest rate of vegetative development or vegetative quality index (VQI) in the transitional savanna/forest area of the northern Amazon, Brazil, after two years of transplanting.

Development and application of a modularized geometry optimizer for future supercritical CO2 turbomachinery optimization

During the operation of supercritical CO (sCO) turbomachineries, large pressure ratios, supersonic conditions and non-ideal gas fluid dynamic phenomenon may happen, which will decrease the whole cycle efficiencies. Hence non-standard designs for the turbomachineries geometry are needed. Successfully simulations in capturing non-ideal gas fluid dynamics and coupled with optimization algorithm are hard and currently not available in any commercial CFD software. Therefore, the problem of optimizing sCO turbomachinery is difficult to solve, which has become a major obstacle to obtain compact and high-efficiency sCO turbomachineries. In this study, a modularized geometry optimizer is developed to obtain the non-standard geometric designs for sCO turbomachineries. Multiple techniques are applied to this optimizer, include the Nelder–Mead algorithm, Mahalanobis distance and stochastic algorithm. The newly developed optimizer can successfully find the optimum satisfying the objective function under given weighting factors. The computational cost can be reduced through a stochastic algorithm. To validate the optimizer, a convergent–divergent nozzle for air with a target Mach number equal to 2.4 is optimized. Different starting points and combinations of weighting factors are used to create a Pareto front. Adjust the weighting factors for different terms of the objective function leading the optimizer to go to different directions in n-dimensional spaces. Three optimized cases, one is Mach number optimized, one is outlet flow uniformity optimized and the other is compromised case, are picked out and analyzed. The results show that the optimizer can successfully find optimized geometry than the reference case and potentially save computational cost. Due to the modularized characteristics, the components of this optimizer can be replaced with any available techniques, which mean the optimizer can be applied to solve different types of optimization problems.

An incipient fault diagnosis methodology using local Mahalanobis distance: Detection process based on empirical probability density estimation

Incipient fault detection is growing as a challenging and hot topic in industrial and academic areas. It is essential to avoid slight unpermitted changes of a system state that can be aggravated and lead to severe security issues. The main challenge of this problem lies in the fact that tiny changes in the early stage can be blurred with noise and create confusion leading to poor detection performance of typical fault detection methods. To detect subtle deviations buried in noise and cope with the non-Gaussian distributed data condition while keeping with the time series information, a sensitive fault detection methodology combining a specifically tuned Local Mahalanobis Distance (LMD) algorithm and an Empirical Probability Density (EPD) estimation technique is proposed. More specifically, first, a healthy domain estimation is proposed to compute the local Mahalanobis distance with optimally tuned characteristics. To approximate a healthy domain, this work proposes a down-sampling algorithm for anchors generation and a parameter estimation method optimally tuned and based on Generalized Extreme Value distribution (GEV) for the domain margin selection. Subsequently, the EPD cumulative sum technique is applied to the LMD result for improving the detection sensitivity further. The performance analysis based on simulation data shows that our proposal is effective to non-Gaussian data and sensitive for incipient fault detection. A case study based on the Continuous-flow Stirred Tank Reactor (CSTR) further validates the effectiveness of our proposal and highlights its benefit by comparing it with state-of-the-art-based solutions in terms of detection delay, detection probability, false alarm probability, and area under the receiver operating characteristic curve (AUC).

Effectiveness of two different at-line instruments for the assessment of cheese composition, major minerals and fatty acids content

The at-line performance of two different NIRS instruments to predict major and minor cheese nutritional traits was evaluated. For this purpose, 158 samples from dairy products were collected and analysed by reference methods. Spectra were acquired using a transmittance and a reflectance instrument. Predictive equations were developed on the whole dataset or dividing samples in groups. Samples clustering was performed using pairwise Mahalanobis distance and centroid linkage algorithm. Prediction models for protein, fat, saturated fatty acids and minerals showed good prediction performances (R2 > 0.80). Instrument configuration had a limited impact on prediction accuracy. Overall, clustering approach reduced prediction error but coefficient of determination also decreased. Prediction of minor compounds with models built from a large variety of cheeses could be useful for process control. Cluster approach is recommended for specific traits and cheese type, for the fine tuning of final product characteristics.

Robust adaptive Kalman filter for strapdown inertial navigation system dynamic alignment

The measurement noise covariance R plays a vital role in the Kalman filter (KF) algorithm. Traditionally, a constant R is obtained by experience or experiments. However, the KF cannot achieve optimal estimation using constant R under non-Gaussian conditions. A robust strategy for adaptive estimation of R is proposed to suppress the influence of non-Gaussian noise on the in-motion alignment performance of the Doppler velocity log (DVL) velocity-aided strapdown inertial navigation system (SINS). Furthermore, an improved Sage–Husa robust adaptive KF algorithm (SHRAKF) based on the Mahalanobis distance (MD) algorithm is proposed to handle the outliers that frequently occur within the complicated underwater environment. The contributions of this work are twofold—the SHRAKF (1) designs a robust strategy to adaptively estimate R in real time and (2) further improves filtering robustness and adaptability with the MD algorithm, conditional on the DVL outputs being contaminated by non-Gaussian noise. A semi-physical simulation experiment for SINS/DVL in-motion alignment based on the test data is carried out, and the experimental results show that the SHRAKF adaptively estimates R in real time and effectively suppresses observational outliers. For non-Gaussian noise pollution, including outliers and heavy-tailed noise, the SHRAKF performs better than traditional methods.

Method and system for nondestructive detection of freshness in Penaeus vannamei based on hyperspectral technology

To test the freshness of Penaeus vannamei rapidly and non-destructively, we constructed a quantitative prediction model for total volatile basic nitrogen (TVB-N) content by collecting 860–1700 nm hyperspectral data and TVB-N content of samples of P. vannamei during eight consecutive days. In this study, we compared the influence of six kinds of spectral pretreatment on the partial least-squares regression modeling (PLSR), which included multiple scattering correction (MSC), standard normal variate transformation (SNV), normalized mean centering, 11-point moving-average smoothing, and Savitzky–Golay smoothing. The results demonstrate that the prediction effect of the MSC-PLSR model has the best relative model accuracy, under the pretreatment of the MSC algorithm for spectral data. In addition, principal component analysis (PCA) and the Mahalanobis distance algorithm were used to remove anomalous samples. The results reveal that the prediction effect of the MSC-PLSR model is better than other models that we considered, and the correlation coefficient (R) of the validation set reached 0.8147. Finally, this study developed a graphical user interface system to predict freshness, based on our research, which provides a method and an analysis tool to rapidly detect the freshness of P. vannamei.

Rapid determination of active components in Ginkgo biloba leaves by near infrared spectroscopy combined with genetic algorithm joint extreme learning machine

Near-infrared spectroscopy(NIRS) combined with band screening method and modeling algorithm can be used to achieve the rapid and non-destructive detection of the traditional Chinese medicine(TCM) production process. This paper focused on the ginkgo leaf macroporous resin purification process, which is the key technology of Yinshen Tongluo Capsules, in order to achieve the rapid determination of quercetin, kaempferol and isorhamnetin in effluent. The abnormal spectrum was eliminated by Mahalanobis distance algorithm, and the data set was divided by the sample set partitioning method based on joint X-Y distances(SPXY). The key information bands were selected by synergy interval partial least squares(siPLS); based on that, competitive adaptive reweighted sampling(CARS), successive projections algorithm(SPA) and Monte Carlo uninformative variable(MC-UVE) were used to select wavelengths to obtain less but more critical variable data. With selected key variables as input, the quantitative analysis model was established by genetic algorithm joint extreme learning machine(GA-ELM) algorithm. The performance of the model was compared with that of partial least squares regression(PLSR). The results showed that the combination with siPLS-CARS-GA-ELM could achieve the optimal model performance with the minimum number of variables. The calibration set correlation coefficient R_c and the validation set correlation coefficient R_p of quercetin, kaempferol and isorhamnetin were all above 0.98. The root mean square error of calibration(RMSEC), the root mean square error of prediction(RMSEP) and the relative standard errors of prediction(RSEP) were 0.030 0, 0.029 2 and 8.88%, 0.041 4, 0.034 8 and 8.46%, 0.029 3, 0.027 1 and 10.10%, respectively. Compared with the PLSR me-thod, the performance of the GA-ELM model was greatly improved, which proved that NIRS combined with GA-ELM method has a great potential for rapid determination of effective components of TCM.

An Underwater Integrated Navigation Algorithm to Deal With DVL Malfunctions Based on Deep Learning

In underwater navigation systems, Global Navigation Satellite System (GNSS) information cannot be used for navigation. The mainstream method of autonomous underwater vehicles (AUV) underwater navigation system is Doppler Velocity Log (DVL) aided strapdown inertial navigation system (SINS). However, because the DVL is an instrument based on Doppler frequency shift to measure velocity, it is easily affected by the external environment. In a complex underwater environment, DVL output is easily polluted by outliers or even interrupted. In this paper, A new integrated navigation algorithm based on deep learning model is proposed to deal with DVL malfunctions. First, use RKF based on Mahalanobis distance algorithm to eliminate outliers, and then train the Nonlinear AutoRegressive with eXogenous input (NARX) model when DVL is available. When DVL is interrupted, use the NARX model to predict the output of DVL and continue integrated navigation. The proposed NARX-RKF scheme’s effectiveness verification was performed on the data set collected by the SINS/DVL ship-mounted experimental system. For comparison, different methods are also compared in the experiment. Experimental results show that NARX-RKF can effectively predict the output of DVL and is significantly better than other methods.

Field identification of weed species and glyphosate-resistant weeds using high resolution imagery in early growing season

Accurate weed mapping in early growing season is an essential step in the site-specific weed management system. This study focuses on validating the potential application of high resolution multispectral and thermal UAS images in classification of weed species and glyphosate-resistant weeds at early phenological stages. A field experiment was conducted to evaluate supervised classification methods to identify three-weed species including waterhemp (Amaranthus rudis), kochia (Kochia scoparia), and ragweed (Ambrosia artemisiifolia L.). The accuracy of six classification algorithms namely Parallelepiped, Mahalanobis Distance, Maximum Likelihood, Spectral Angle Mapper, Support Vector Machine and Decision Tree implemented at pixel and object-based levels in weed species classification were evaluated. Thermal infrared imagery was also used to assess the canopy temperature variance within the weed species to identify the glyphosate-resistance status in detected weeds. The object-based algorithms developed with mosaicked imagery effectively classified weed species with the overall accuracy and Kappa coefficient values greater than 86% and 0.77, respectively. The lowest accuracy and Kappa coefficient (67% and 0.58) were observed for pixel-based Mahalanobis Distance algorithm. The canopy temperature-based classification of susceptible and resistant weeds resulted in the discrimination accuracies of 88%, 93% and 92% in glyphosate-resistant kochia, waterhemp and ragweed, respectively.

Online DDoS attack detection using Mahalanobis distance and Kernel-based learning algorithm

Distributed denial-of-service (DDoS) attacks are constantly evolving as the computer and networking technologies and attackers' motivations are changing. In recent years, several supervised DDoS detection algorithms have been proposed. However, these algorithms require a priori knowledge of the classes and cannot automatically adapt to frequently changing network traffic trends. This emphasizes the need for the development of new DDoS detection mechanisms that target zero-day and sophisticated DDoS attacks. In this paper, we propose an online, sequential, DDoS detection scheme that is suitable for use with multivariate data. The proposed algorithm utilizes a kernel-based learning algorithm, the Mahalanobis distance, and a chi-square test. Initially, we extract four entropy-based and four statistical features from network flows per minute as detection metrics. Then, we employ the kernel-based learning algorithm using the entropy features to detect input vectors that were suspected to be DDoS. This algorithm assumes no model for network traffic or DDoS. It constructs and adapts a dictionary of features that approximately span the subspace of normal behavior. Every T minutes, the Mahalanobis distance between suspicious vectors and the distribution of dictionary members is measured. Subsequently, the chi-square test is used to evaluate the Mahalanobis distance. The proposed DDoS detection scheme was applied to the CICIDS2017 dataset, and we compared the results with those given by existing algorithms. It was demonstrated that the proposed online detection scheme outperforms almost all available DDoS classification algorithms with an offline learning process.

Integrated navigation for doppler velocity log aided strapdown inertial navigation system based on robust IMM algorithm

Global navigation satellite system (GNSS) is the most widely used position, navigation and time service system. However, GNSS such as BeiDou Navigation Satellite System (BDS) cannot provide the external assistance information for the autonomous underwater vehicle (AUV) because of the blocking effect of water on electromagnetic wave. The uncertain and non-Gaussian characteristics of the measurement noise may degrade the performance of SINS/DVL integrated navigation under the standard Kalman filter (KF) framework. To solve this problem, this paper proposes a robust interacting multiple model algorithm (IMM-RKF) based on the Mahalanobis distance algorithm. Furthermore, two switching sub-models will depict the uncertain parameters. The contributions of the work presented here are twofold. First, this paper robustifies the IMM algorithm under non Gaussian condition, and second the research method is verified in the problem of SINS/DVL integrated navigation. Simulation experimental results for the problem of SINS/DVL integrated navigation under non-Gaussian and uncertain conditions demonstrate the validity and superiority of the proposed method.

Introducing the coupled stepwise areal constraining and Mahalanobis distance: a promising MCDM-based probabilistic model for landfill site selection.

This study sets out to propose a new ensemble of probabilistic spatial modeling and multi-criteria decision-making comprised of stepwise areal constraining and Mahalanobis distance algorithms in order to assess areal suitability for landfilling. The Ardak watershed was selected as the study area due to encountering several cases of open garbage dumps and uncontrolled landfills which are one of the main sources of river water pollution in the upstream of the Ardak dam. The results revealed that the proposed algorithm successfully assists in inventory-irrespective probabilistic modeling of landfill siting which is mainly indebted to the role of areal constraining in providing training and validation samples for the Mahalanobis distance model. The latter also showed a robust pattern recognition results from which a discernible differentiation of the area was attained while the spatial dependencies between the environmental factors were taken into account. Mahalanobis distance also gave an outstanding performance in terms of goodness of fit (area under the success rate 89.367) and prediction power (area under the success rate 89.252). Based on a five-point scale classification scheme, about 2.7% and 2.6% of the study area, respectively, have high and very high suitability for landfilling, while the remaining area is shared between very low-to-moderate suitability classes. According to the current trail of literature regarding landfill site selection which mostly relies on mere areal filtering, a probabilistic model would give invaluable inferences regarding the pattern of suitability/susceptibility of the area of interest and causative role of the influential factors. Graphical Abstract.

Nondestructive detection for egg freshness grade based on hyperspectral imaging technology

AbstractIn order to identify the freshness grade of eggs nondestructively and rapidly, hyperspectral imaging technology was used in this article. The hyperspectral data of 200 samples of three freshness grades was acquired by using hyperspectral image acquisition system (400.68–1,001.612 nm), and then the freshness grade of egg samples was measured by stoichiometry. First, Mahalanobis distance algorithm was used to remove abnormal sample data. Second, savitzky–golay and wavelet threshold denoising combined with standard normalized variable were used to pretreat the spectral data, respectively. Third, iteratively retains informative variables (IRIV), variable iterative space shrinkage approach, and competitive adaptive reweighted sampling were used for feature wavelength selection. Since the classification accuracy of support vector machine (SVM) model was affected by the selection of parameters, genetic algorithm (GA) was introduced to search the optimal parameters in SVM and compared with grid search algorithm. Finally, the result indicated that the classification accuracy of training set and test set of the optimal classification model (IRIV‐GA‐SVM) reached 99.29% and 97.87%, respectively. Thus, it is feasible to use hyperspectral image technology to detect egg freshness grade.Practical ApplicationsThe freshness grade is one of the most important indexes to measure the quality of eggs. Traditional methods of detecting egg freshness grade are time‐consuming and destructive, which cannot meet the test needs of modern agriculture. Hyperspectral imaging, an emerging technology, can provide both spectral and spatial information simultaneously, and has the advantages of nondestructive, fast and nonpollution. The result indicated that hyperspectral imaging technology for the detection of freshness grade of eggs is feasible.

Study on Establishing Degradation Model of Chip Solder Joint Material under Coupled Stress.

The chip is the core component of the integrated circuit. Degradation and failure of chip solder joints can directly lead to function loss of the integrated circuit. In order to establish the degradation model of chip solder joints under coupled stress, this paper takes quad flat package (QFP) chip solder joints as the study object. First, solder joint degradation data and failure samples were obtained through fatigue tests under coupled stress. Three types of micro failure modes of solder joints were obtained by scanning electron microscope (SEM) analysis and finite element model (FEM) simulation results. Second, the characterization of degradation data was obtained by the principal component of Mahalanobis distance (PCMD) algorithm. It is found that solder joint degradation is divided into three stages: strain accumulation stage, crack propagation stage, and failure stage. Later, Coffin–Manson model and Paris model were modified based on the PCMD health index and strain simulation. The function relationship between strain accumulation time, crack propagation time, and strain was determined, respectively. Solder joint degradation models at different degradation stage were established. Finally, through strain simulation, the models can predict the strain accumulation time and failure time effectively under each failure mode, and their prediction accuracy is above 85%.

Land-Use Land-Cover Classification by Machine Learning Classifiers for Satellite Observations—A Review

Rapid and uncontrolled population growth along with economic and industrial development, especially in developing countries during the late twentieth and early twenty-first centuries, have increased the rate of land-use/land-cover (LULC) change many times. Since quantitative assessment of changes in LULC is one of the most efficient means to understand and manage the land transformation, there is a need to examine the accuracy of different algorithms for LULC mapping in order to identify the best classifier for further applications of earth observations. In this article, six machine-learning algorithms, namely random forest (RF), support vector machine (SVM), artificial neural network (ANN), fuzzy adaptive resonance theory-supervised predictive mapping (Fuzzy ARTMAP), spectral angle mapper (SAM) and Mahalanobis distance (MD) were examined. Accuracy assessment was performed by using Kappa coefficient, receiver operational curve (RoC), index-based validation and root mean square error (RMSE). Results of Kappa coefficient show that all the classifiers have a similar accuracy level with minor variation, but the RF algorithm has the highest accuracy of 0.89 and the MD algorithm (parametric classifier) has the least accuracy of 0.82. In addition, the index-based LULC and visual cross-validation show that the RF algorithm (correlations between RF and normalised differentiation water index, normalised differentiation vegetation index and normalised differentiation built-up index are 0.96, 0.99 and 1, respectively, at 0.05 level of significance) has the highest accuracy level in comparison to the other classifiers adopted. Findings from the literature also proved that ANN and RF algorithms are the best LULC classifiers, although a non-parametric classifier like SAM (Kappa coefficient 0.84; area under curve (AUC) 0.85) has a better and consistent accuracy level than the other machine-learning algorithms. Finally, this review concludes that the RF algorithm is the best machine-learning LULC classifier, among the six examined algorithms although it is necessary to further test the RF algorithm in different morphoclimatic conditions in the future.