Improved Mahalanobis Distance Research Articles

Short-term electricity price forecasting based on similarity day screening, two-layer decomposition technique and Bi-LSTM neural network

Electricity price forecasting (EPF) has been challenged by the widespread grid integration of renewable energy (RE), so it is critical to develop a highly accurate and reliable EPF model. In this study, novel considerations of RE generation factors are made, and a quantitative model for the impact of RE on electricity prices is built using random forests (RF) and improved Mahalanobis Distance (IMD). To reduce data duplication, similar days of EPF are first selected. Then it is suggested to decompose the electricity price series into multiple intrinsic mode functions (IMF) and residuals with different frequencies using a two-layer decomposition model based on improved comprehensive ensemble empirical mode decomposition (ICEEMD) and variational mode decomposition (VMD), in order to reduce data noise and volatility. Finally, EPF model based on Bi-directional long short-term memory (Bi-LSTM) is established to forecast multiple subsequences, and the final price forecasting result is obtained after integrated processing. Experimental results show that the RF-IMD-ICEEMD-VMD-Bi-LSTM hybrid model can significantly improve the forecasting performance, reduce the prediction error of EPF, and has the best performance among the comparison models.

Point Clouds Outlier Removal Method Based on Improved Mahalanobis and Completion

Point clouds have been regarded as a representative format for 3D visualization of real-world objects or scenes. However, point clouds acquired from depth cameras or laser scanning devices commonly contain outliers. Outlier removal performance will directly affect the downstream applications. Existing methods mainly perform outlier removal directly on the raw data, which are designed without finding a proper balance between outlier removal precision and detail feature retention. In this letter, we propose a novel outlier removal method called CIMD that can obtain outlier-free data by conducting the improved Mahalanobis distance and point clouds completion. A layered statistical outlier removal approach is introduced as preprocessing strategy to obtain feature points and incomplete point clouds. We filter feature points to fill the incomplete point clouds according to the improved Mahalanobis distances. The theoretical analysis proves that the improved Mahalanobis distance can magnify the difference between outliers and ground truth compared with the original way. Using publicly available dataset PointCleanNet and real-world scanned data, the experimental results show that the proposed method has superior performance compared with state-of-the-art methods.

Research on deterioration evolution trend of primary loop piping in nuclear power plant based on fusion health index

The perennial operation of nuclear power primary loop piping leads to deterioration of the piping and the potential risk of leakage. At present, there is no reliable technology to detect the leakage directly. This paper proposes a method to obtain the piping deterioration evolution trend based on the analysis of sets of data that can reflect real state of the piping. The specific method is to quantitatively construct a fusion health index model based on the improved Mahalanobis distance, which integrates the analytic hierarchy process and the entropy weight method. A prediction model combining convolutional neural networks and long short-term memory neural networks is established to make the trend analysis and prediction. The experimental results show that the method can better reflect the actual health state of the piping and effectively predict the deterioration evolution trend, which provides a specific reference value for ensuring the safe and stable operation of equipment.

Improved Mahalanobis Distance Based JITL-LSTM Soft Sensor for Multiphase Batch Processes

To predict key variables of complicated batch processes, the long short-term memory (LSTM) soft sensor is developed to deal with both data nonlinearity and dynamics. To extract proper historical samples and implement the real-time modeling scheme with model updating strategy, the just-in-time learning (JITL) algorithm is widely used at the data selection stage of LSTM soft sensor. However, the multiphase issue of batch processes are not considered for the conventional JITL-LSTM soft sensor. In this paper, a multiphase Mahalanobis distance based JITL framework is developed to integrate the phase recognition strategy into the similarity measurement and data selection scheme, by which an extra step of phase identification can be avoided and the accuracy of JITL can be significantly improved. Thus, batch samples from different operating phases can be recognized without an additional phase identification step. By the use of the Mahalanobis Distance based JITL-LSTM Soft Sensor, the probability of data mismatch can be significantly reduced so that the accuracy of quality prediction can be promoted. Two simulation cases are provided to verify the effectiveness of the proposed method consisting of a fed-batch reactor process and the penicillin fermentation process.

An adaptive parameters adjustment and planning method for robotic belt grinding using modified quality model

As a kind of flexible manufacturing system, the machining quality of a robotic belt grinding system is related to a variety of factors with strong time variation, which easily leads to process fluctuations and affects the final quality. Therefore, it is a great challenge to control the quality precisely during the whole grinding procedure. Focusing on this problem, an adaptive parameters adjustment and planning method for robotic belt grinding using the modified quality model is proposed in this paper. Firstly, the correlation analysis method of grinding parameters in time domain is presented based on an improved-Mahalanobis distance. The response surface methodology (RSM) is utilized to construct the quality prediction model, and furtherly the parameter sensitivity function, which can characterize the influence degree of different parameters on the grinding quality, is introduced to calculate the Mahalanobis distance for improving the accuracy of the correlation analysis method. Secondly, based on the correlation analysis, a conversion method from the old samples into the new samples space is presented using vector field smoothing algorithm (VFS), then the modified grinding quality model can be re-established adopting the new samples. Furtherly, taking the problem of poor robotic response rate into consideration, a multi-parameters collaborative planning method under the smoothness constraint is developed using particle swarm optimization (PSO) algorithm, which can avoid the parameter mutation and improve the process stability. Finally, belt grinding experiments on a curved surface were carried out based on the robotic grinding platform. The results show that the approach can improve the grinding shape accuracy, which verify the effectiveness of the proposed methods.

Multi-information fusion-based belt condition monitoring in grinding process using the improved-Mahalanobis distance and convolutional neural networks

Due to the advantages of flexibility, high efficiency, and low processing heat, belt grinding has been widely applied in manufacturing industries. As belt wear will cause deterioration of the removal capacity, increasing the surface irregularity and adversely affecting the grinding quality, interests in belt condition monitoring have signficantly augmented in recent years, which not only secures the surface quality, but also helps to optimize the utilization of the belt’s life cycle. A multi-information fusion-based belt condition monitoring method in grinding process using the improved-Mahalanobis distance and Convolutional Neural Networks (CNN) is proposed in this paper. Firstly, a time-domain mapping relationship between belt wear and material removal rate is put forward and a factor k t is derived to characterize the wear status. Furtherly, the evolution of abrasive grains degradation as well as the wear effect on grinding quality is analyzed. Secondly, a parallel multi-sensor integration grinding system including force, vibration, sound and acoustic emission sensors is established, based on which the single-factor and multi-factor sensitivity experiments are conducted to determine the optimal combination of characteristic signals. Finally, a multi-layer model including the grinding conditions classification and belt stages identification is established adopting the methods of improved-Mahalanobis distance and CNN. On one hand the model is not limited to a fixed condition and has a wider application scope, on the other hand avoids the impact of human experience on the features extraction and improves the model accuracy from the theoretical perspective. The experimental results show that the identification accuracy of the belt wear stage adopting the method in this paper is no less than 94 % for the 16 sampling conditions and more than 86 % for other grinding conditions. Furtherly, the contrast experiments indicate that the method in this paper is of a higher accuracy than the single-layer CNN model, which proves the effectiveness of the proposed method.

Missing data filling algorithm in OA (office automation) cooperative office system

The current algorithms can only effectively fill in fewer missing data when filling out missing data in office automation (OA) cooperative office system, and the filling accuracy is worse when the missing data is large. In order to solve this problem, a missing data filling algorithm based on improved Mahalanobis distance is proposed. It used a top-down recursion to process raw data, got the training set of the observation sample and the associated class labels, and generated readable rules and decision trees. It optimizes the generated decision tree by using the genetic algorithm of the elite strategy, and classifies the missing data through the optimized decision tree. According to the classification results combined with the entropy calculation method in the improved Mahalanobis distance algorithm, there is a single decreasing relationship between entropy size and Mahalanobis distance of missing data, which can reduce the impact on the predictive value of related data, and achieve the missing data filling in OA cooperative office system. The experimental results showed that the proposed algorithm can effectively improve the accuracy of data filling, and the accuracy of data filling can be stable over 80% under different missing rates.

SAR Image Change Detection via Spatial Metric Learning With an Improved Mahalanobis Distance

The log-ratio (LR) operator has been widely employed to generate the difference image for synthetic aperture radar (SAR) image change detection. However, the difference image generated by this pixel-wise operator can be subject to SAR images speckle and unavoidable registration errors between bitemporal SAR images. In this letter, we proposed a spatial metric learning method to obtain a difference image more robust to the speckle by learning a metric from a set of constraint pairs. In the proposed method, spatial context is considered in constructing constraint pairs, each of which consists of patches in the same location of bitemporal SAR images. Then, a semi-definite positive metric matrix $\bf M$ can be obtained by the optimization with the max-margin criterion. Finally, we verify our proposed method on four challenging datasets of bitemporal SAR images. Experimental results demonstrate that the difference map obtained by our proposed method outperforms than other state-of-art methods.

Risk Assessment and Decision-Making Based on Mean-CVaR-Entropy for Flood Control Operation of Large Scale Reservoirs

The risk inevitably exists in the process of flood control operation and decision-making of reservoir group, due to the hydrologic and hydraulic uncertain factors. In this study different stochastic simulation methods were applied to simulate these uncertainties in multi-reservoir flood control operation, and the risk caused by different uncertainties was evaluated from the mean value, extreme value and discrete degree of reservoir occupied storage capacity under uncertain conditions. In order to solve the conflict between risk assessment indexes and evaluate the comprehensive risk of different reservoirs in flood control operation schemes, the subjective weight and objective weight were used to construct the comprehensive risk assessment index, and the improved Mahalanobis distance TOPSIS method was used to select the optimal flood control operation scheme. The proposed method was applied to the flood control operation system in the mainstream and its tributaries of upper reaches of the Yangtze River basin, and 14 cascade reservoirs were selected as a case study. The results indicate that proposed method can evaluate the risk of multi-reservoir flood control operation from all perspectives and provide a new method for multi-criteria decision-making of reservoir flood control operation, and it breaks the limitation of the traditional risk analysis method which only evaluated by risk rate and cannot evaluate the risk of the multi-reservoir flood control operation system.

Outlier Detection Based on Robust Mahalanobis Distance and Its Application

Classical Mahalanobis distance is used as a method of detecting outliers, and is affected by outliers. Some robust Mahalanobis distance is proposed via the fast MCD estimator. However, the bias of the MCD estimator increases significantly as the dimension increases. In this paper, we propose the improved Mahalanobis distance based on a more robust Rocke estimator under high-dimensional data. The results of numerical simulation and empirical analysis show that our proposed method can better detect the outliers in the data than the above two methods when there are outliers in the data and the dimensions of data are very high.

An improved Mahalanobis distance-based colour segmentation method for rural building recognition

Aiming at the rapid identification of rural buildings in complex environments from high-spatialresolution images, an improved Mahalanobis distance colour segmentation method (IMDCSM) is proposed and realised in Red, Green and Blue (RGB) space. Vector sets of a lower discrete degree are obtained by filtering the colour vector sets of the building samples, and a standard ellipsoid equation can be constructed based on these vector sets. The threshold of interested colour range can be flexibly and intuitively selected by changing the shape and size of this ellipsoid. Then, according to the relationship between the location of the image pixel colour vector and the ellipsoid, all building information can be extracted quickly. To verify the effectiveness of the proposed method, unmanned aerial vehicle (UAV) images of two areas in the suburbs of Chengdu city and Deyang city were utilised as experimental data for image segmentation, and the existing colour segmentation method based on the Mahalanobis distance was selected as an indicator to assess the effectiveness of this method. The experimental results demonstrate that the completeness and correctness of this method reached 95% and 83.0%, respectively, values that are higher than those of the Mahalanobis distance colour segmentation method (MDCSM). In general, this method is suitable for the rapid extraction of rural building information, and provides a new threshold selection method for classification.

A New Analog Circuit Fault Diagnosis Method Based on Improved Mahalanobis Distance

This paper presents a new analog circuit fault diagnosis method based on improved Mahalanobis Distance. The Mahalanobis Distance is improved according to the characteristics of analog circuit, and then introduced into analog circuit fault detection. First, the circuit testability was analyzed, and the relation of ambiguity groups was determined on the basis of the test matrix, and then the separable potential faulty components under the assumption of single fault were also determined. Finally, the suspicious components could be classified using the improved Mahalanobis Distance according to the feature values of the test points, so as to reduce the number of classes and enhance the speed when classifying faults. The experiment shows that the method can achieve fast analog circuit fault diagnosis and better results of analog circuit diagnosis detection.