A Forecast-Driven Deep Learning Framework Using ICEEMDAN and PatchTST for Stock Prediction and Dynamic Portfolio Optimization

Doppler radar for monitoring vital signals is an emerging tool, and how to remove the noise during the detection process and reconstruct the accurate respiration and heartbeat signals are hot issues in current research. In this paper, a novel radar vital signal separation and de-noising technique based on improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), sample entropy (SampEn), and wavelet threshold is proposed. First, the noisy radar signal was decomposed into a series of intrinsic mode functions (IMFs) using ICEEMDAN. Then, each IMF was analyzed using SampEn to find out the first few IMFs containing noise, and these IMFs were de-noised using the wavelet threshold. Finally, in order to extract accurate vital signals, spectrum analysis and Kullback–Leible (KL) divergence calculations were performed on all IMFs, and appropriate IMFs were selected to reconstruct respiration and heartbeat signals. Moreover, as far as we know, there is almost no previous research on radar vital signal de-noising based on the proposed technique. The effectiveness of the algorithm was verified using simulated and measured experiments. The results show that the proposed algorithm could effectively reduce the noise and was superior to the existing de-noising technologies, which is beneficial for extracting more accurate vital signals.

Ferromagnetic materials are extensively utilized in industrial settings where the early detection and repair of defects is paramount for ensuring industrial safety. During the enhanced magnetic memory detection of micro-defects, many interference signals appear in the detection signal, which makes it difficult to accurately extract the characteristics of the micro-defect signals, significantly affecting detection effectiveness. When improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) is employed independently for signal denoising, the noise and feature signals of the transition components are retained or removed. When variational mode decomposition (VMD) is employed independently for signal denoising, the denoising effect is restricted because of the difficulty in determining the penalty factor α and the number of decomposition layers m. To solve these problems, a denoising method for enhanced magnetic memory detection signals based on ICEEMDAN and VMD, called ICEEMDAN–VMD, is proposed in this paper. First, a comprehensive index (CI) combining information entropy (IE) and the correlation coefficient R is proposed, then the signal components obtained by performing decomposition with the ICEEMDAN method are divided into noise-dominant components, transition components, and useful signal components based on the CI. Subsequently, VMD is employed to perform secondary decomposition on the transition components obtained from the ICEEMDAN method and calculate the correlation coefficients. Ultimately, the optimal VMD components and useful signal components obtained by the ICEEMDAN method are selected for signal reconstruction to obtain a denoised signal. To validate the effectiveness of the proposed method, the denoising effects of the ICEEMDAN–VMD, ICEEMDAN, and VMD methods were compared based on the signal-to-noise ratio (SNR) and fuzzy entropy (FE). The comparison indicated that the ICEEMDAN–VMD denoising method significantly enhanced the denoising effect, and the SNRs of the components of the magnetic field signal could be increased by up to 69.426%. The SNR of each gradient component of the magnetic field signal could be improved by up to ten times, and the FEs of the signal components and their corresponding gradient components could be reduced by 24.198%–81.011%, respectively.

To improve the accuracy of rock failure monitoring, this article addresses the optimization of denoising acoustic emission (AE) signals. Combining laboratory experiments on rock AE and theoretical research on signal denoising, a denoising method based on improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) is proposed for rock fracture AE signals. The ICEEMDAN algorithm is used to decompose the original noisy signal into multiple intrinsic mode functions (IMFs) and employs cluster analysis to determine data thresholds based on their characteristics. Subsequently, using multiple criteria such as permutation entropy, correlation coefficient, and variance contribution rate, the IMFs are categorized into two groups. The low-correlation IMFs are partially removed based on the combination of indicators, while the high-correlation portion is denoised using wavelet thresholding (WT). Finally, a wavelet analysis is performed to reconstruct the signal, effectively achieving an optimized denoising of the original signal. Quantitative analysis of denoising effects on typical rock uniaxial compression fracture AE signals reveals that the optimized method has a positive impact on high-frequency noise reduction. The peak frequency range is unaffected before and after optimization, while the main amplitude reduction is concentrated in the high-frequency range. Compared to traditional wavelet denoising methods, the proposed method exhibits higher signal-to-noise ratio (SNR) improvement, as well as varying degrees of reduction in mean squared error (MSE) and total harmonic distortion (THD). This research introduces a novel approach for optimizing the application of rock AE signals.

Based on improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) algorithm and kernel online sequential extreme learning machine (KOSELM) algorithm, a new hybrid short-term traffic flow prediction model (ICEEMDAN-KOSELM-ARIMA) for signalized intersections is proposed according to the current and historical traffic flow data. First, traffic flow historical time series are decomposed by ICEEMDAN algorithm for the purpose of improving the prediction accuracy. Several intrinsic mode functions could be obtained by the decomposition process. Then, permutation entropy algorithm is employed to analyze the random properties of intrinsic mode function components. According to the different random properties of intrinsic mode functions, different prediction models can be built. On this basis, KOSELM prediction models are established for the intrinsic mode function components with big randomness. And auto-regressive integrated moving average (ARIMA) prediction models are built for the intrinsic mode function components with small randomness. Finally, an actual signalized intersection is selected to verify the effect and performance of the hybrid prediction model proposed in this article. Results show that compared with other models, the new proposed hybrid prediction model can effectively improve prediction accuracy, of which prediction errors are the lowest and fitting effect with actual values is the best.

Accurate prediction of Global Navigation Satellite System-derived precipitable water vapor (GNSS-PWV), which is a crucial indicator for climate change monitoring, holds significant scientific value for climate disaster prevention and mitigation. In the study of GNSS-PWV prediction, the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) algorithm within a decomposition–integration framework effectively addresses the non-stationarity and complexity of PWV sequences, enhancing prediction accuracy. However, residual noise and pseudo-modes from decomposition can distort signals, reducing the predictor system’s reliability. Additionally, independent modeling of all decomposed components decreases computational efficiency. To address these challenges, this paper proposes a hybrid model combining the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), autoregressive integrated moving average (ARIMA), and long short-term memory (LSTM) networks. Enhanced by local mean optimization and adaptive noise regulation, the ICEEMDAN algorithm effectively suppresses pseudo-modes and minimizes residual noise, enabling its decomposed intrinsic mode functions (IMFs) to more accurately capture the multi-scale features of GNSS-PWV. Sample entropy (SE) is used to quantify the complexity of IMFs, and components with similar entropy values are reconstructed into the following three sub-sequences: high-frequency, low-frequency, and trend. This process significantly reduces modeling complexity and improves computational efficiency. We propose different modeling strategies tailored to the dynamics of various subsequences. For the nonlinear and non-stationary high-frequency components, the LSTM network is used to effectively capture their complex patterns. The LSTM’s gating mechanism and memory cell design proficiently address the long-term dependency issue. For the stationary and weakly nonlinear low-frequency and trend components, linear patterns are extracted using ARIMA. Differencing eliminates trends and moving average operations capture random fluctuations, effectively addressing periodicity and trends in the time series. Finally, the prediction results of the three components are linearly combined to obtain the final prediction value. To validate the model performance, experiments were conducted using measured GNSS-PWV data from several stations in Hong Kong. The results demonstrate that the proposed model reduces the root mean square error by 56.81%, 37.91%, and 13.58% at the 1 h scale compared to the LSTM, EMD-LSTM, and ICEEMDAN-SE-LSTM benchmark models, respectively. Furthermore, it exhibits strong robustness in cross-month forecasts (accounting for seasonal influences) and multi-step predictions over the 1–6 h period. By improving the accuracy and efficiency of PWV predictions, this model provides reliable technical support for the real-time monitoring and early warning of extreme weather events in Hong Kong while offering a universal methodological reference for multi-scale modeling of geophysical parameters.

This paper presents a fault diagnosis method for rolling bearings working in non-stationary running conditions. The proposed approach is based on an improved version of the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), the multivariate denoising using wavelet analysis and principal component analysis (PCA), the spectral kurtosis, and the order tracking analysis (OTA). The results show that the improved CEEMDAN has completely decomposed the raw signal into different intrinsic mode functions (IMFs) representing the natural oscillatory modes embedded into the signal. The most relevant IMF from which the defect was extracted is selected by the kurtogram plot which allows locating the optimal frequency band having the highest kurtosis value. Multivariate denoising based on wavelet analysis and PCA is used to increase the signal-to-noise ratio (SNR) of the selected IMF. The results show the great contribution of the denoising approach when comparing the selected denoised IMF with the original one. Finally, order tracking analysis is applied on the denoised IMF’s envelope to remove the effect of speed variation, and an envelope order spectrum is obtained. The proposed approach is first applied on theoretical signal simulating rolling bearing defect in variable regime including three different phases. The final order spectrum shows exactly the simulated defect order and several of its harmonics. For the experimental validation, several signals of defective rolling bearings have been measured on the Machine Fault Simulator test rig in variable regime. Despite the combined variable regime including acceleration-constant regime-deceleration, at the same time, the obtained results indicate the efficiency of the proposed method to extract the fault order with high accuracy. The maximum error between the theoretical order and the experimentally obtained one was 1.3% for outer race defect and 1% for inner race defect. Finally, the performances of the proposed method are compared to those of another diagnosis method designed for variable regime conditions. Both outer race and inner race defects are considered in acceleration regime. The results show the superiority of the proposed method to highlight the defect order with highest clarity.

Power quality events caused by renewable-energy integration are usually associated with complex disturbances; therefore, their type identification is the primary task of subsequent pollution control. This study proposes a novel three-step classification approach based on time-dependent spectral features (TDSFs) for the classification of complex power quality disturbances (PQDs). First, the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) is adopted to decompose the PQDs into several intrinsic mode functions (IMFs). The related IMFs are selected by correlation coefficient and kurtosis. Then the eight eigenvalues of each IMF are extracted, including TDSFs. Finally, the eigenvalue dimension of each IMF is reduced by linear discriminant analysis (LDA). Moreover, the classifier of the adaptive k-nearest neighbor with excluding outliers (AdaKNNEO) can confirm the PQDs type. To verify the effectiveness of the proposed approach, a series of simulations and hardware experiments are conducted. The overall result shows robustness and high accuracy of the proposed method, and especially for the complex PQDs, it possesses the highest entirely correct of 96% compared to other advanced methods.

A hybrid air quality early-warning framework: An hourly forecasting model with online sequential extreme learning machines and empirical mode decomposition algorithms

Aiming to address the nonlinear and non-stationary characteristics of rolling bearing fault vibration signals, traditional diagnostic methods face challenges such as mode mixing, insufficient feature discrimination, and sensitivity to classifier parameters, which limit further improvements in diagnostic accuracy. This paper proposes a comprehensive optimization framework for fault diagnosis, covering signal decomposition and fault feature classification. First, improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) is applied to decompose the signal into intrinsic mode functions (IMF), effectively suppressing mode mixing. Next, the correlation coefficient and multiscale permutation entropy (MPE) are employed to quantify the dynamic complexity of each IMF across multiple scales, enabling the extraction of highly discriminative multidimensional feature vectors. Finally, the black-winged kite algorithm (BKA) is used to optimize the penalty factor and kernel parameters of the least squares support vector machine (LSSVM). BKA achieves a dynamic balance between local exploitation and global exploration, and its integration with Cauchy mutation improves robustness to noise, thereby enhancing convergence speed and stability. Experimental results from publicly available datasets and a custom experimental platform show that the proposed approach effectively distinguishes among various fault types, offering superior performance, improved interpretability, and enhanced robustness.

Photovoltaic power generation prediction constitutes a significant research area within the realm of power system artificial intelligence. Accurate prediction of future photovoltaic output is imperative for the optimal dispatchment and secure operation of the power grid. This study introduces a photovoltaic prediction model, termed ICEEMDAN-Bagging-XGBoost, aimed at enhancing the accuracy of photovoltaic power generation predictions. In this paper, the original photovoltaic power data initially undergo decomposition utilizing the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) algorithm, with each intrinsic mode function (IMF) derived from this decomposition subsequently reconstructed into high-frequency, medium-frequency, and low-frequency components. Targeting the high-frequency and medium-frequency components of photovoltaic power, a limiting gradient boosting tree (XGBoost) is employed as the foundational learner in the Bagging parallel ensemble learning method, with the incorporation of a sparrow search algorithm (SSA) to refine the hyperparameters of XGBoost, thereby facilitating more nuanced tracking of the changes in the photovoltaic power’s high-frequency and medium-frequency components. Regarding the low-frequency components, XGBoost-Linear is utilized to enable rapid and precise prediction. In contrast with the conventional superposition reconstruction approach, this study employs XGBoost for the reconstruction of the prediction output’s high-frequency, intermediate-frequency, and low-frequency components. Ultimately, the efficacy of the proposed methodology is substantiated by the empirical operation data from a photovoltaic power station in Hebei Province, China. Relative to integrated and traditional single models, this paper’s model exhibits a markedly enhanced prediction accuracy, thereby offering greater applicational value in scenarios involving short-term photovoltaic power prediction.

DC–DC converters are important components that affect the reliability of power electronic systems. Predicting the health status of converters helps to obtain performance degradation information in advance before shutdown failures occur in the system, and formulate reasonable maintenance strategies accordingly, improving the reliability and safety of the system. This paper proposes a data-driven method combining Health indicator (HI) extraction and deep learning technology. Firstly, the output voltage of the converter is taken as the health status sensitive information, and the permutation entropy is extracted as the HI. Secondly, the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise method is adopted to decompose the HI and extract the monotonic degradation trend of the health status. Finally, the bidirectional gated recurrent unit method with the introduced attention mechanism is used to conduct single-step predictions on each intrinsic mode function component and the trend term. The results of each predicted component are reconstructed to realize circuit-level health status prediction. Experiments are carried out using the High-temperature degradation experimental data set with the Boost converter as the research object for verification. The results show that compared with traditional methods, the prediction error of the proposed method is reduced by 53.16%, demonstrating higher accuracy and reliability.

Accurate wind power prediction can effectively alleviate the pressure of the power system peak frequency regulation, and is more conducive to the economic dispatch of the power system. To enhance wind power forecasting accuracy, a hybrid approach for wind power interval prediction is proposes in this study. Firstly, an Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) is applied to decompose the initial wind power sequence into multiple modes, and Variational Mode Decomposition is used to further decompose the high-frequency non-stationary components. Next, Fuzzy Entropy (FE) is utilized to assess the complexity of the post-decomposed Intrinsic Mode Functions (IMFs), and different forecasting methods are employed accordingly, the point predictions were obtained by linearly summing the component predictions.Additionally, an improved sparrow search algorithm (ISSA) is used to seek the optimal hyperparameters of the prediction algorithm. Finally, the prediction intervals are constructed using the point prediction results based on kernel density estimation (KDE). The root mean square errors (RMSE) of deterministic predictions are 2.8458 MW and 1.8605 MW, with uncertainty coverage rates of 95.83% and 97.92% at a 95% confidence level.

Hybrid wind speed forecasting using ICEEMDAN and transformer model with novel loss function

To enhance the accuracy of short-term photovoltaic (PV) power forecasting, this study proposes a novel hybrid model that integrates Random Forest (RF), Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN), Sample Entropy (SE), the Random Walk with Compulsory Evolution (RWCE) algorithm, and the Gated Recurrent Unit (GRU) network. Initially, RF is applied to select relevant meteorological features, minimizing redundancy and improving both training efficiency and predictive robustness under complex operating conditions. ICEEMDAN is then employed to decompose the PV power series into multiple quasi-stationary components, mitigating the adverse effects of non-stationarity on forecasting accuracy. Following this, SE is applied to quantify the complexity of each component and reconstruct the decomposed signals into high-, mid-, and low-frequency bands, simplifying the inputs to the forecasting model. To further improve performance, the RWCE algorithm optimizes GRU network hyperparameters through global exploration, individual evolution, and enforced evolution strategies. The optimized GRU network then predicts each reconstructed component, and the component-wise forecasts are aggregated to yield the final PV power output. Simulation results from several representative months indicate that the proposed approach reduces RMSE by an average of 9.02% compared to comparison model and by 43.41% relative to the baseline model, demonstrating its superior forecasting capability. Additionally, the model demonstrated scalability across varying climate conditions, confirming its applicability in real-world scenarios.

Climate change is of great significance to vegetation coverage. However, the long time series of climatic factors and vegetation index is non-stationary and non-linear, containing different information in frequency and time scales. The study innovatively integrated Support Vector Machine (SVM) and Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN), and analyzed the relationship between climatic factors and the Normalized Difference Vegetation Index (NDVI) at three-time scales in Loess Plateau (LP). The results indicate that: (1) The mode mixing phenomenon of EMD can be better solved by using ICEEMDAN, and the end effect can be mitigated by using SVM to expand both ends of the data before decomposition. (2) The residual NDVI stripped out using ICEEMDAN-SVM can represent the long-term trend of the original data. The results show that the NDVI and climate factors performed noticeable spatial and temporal differences under different climate zones. (3) The relationship between vegetation and climatic factors revealed obvious spatial heterogeneity at different time scales, and the climate change in the overall trend explained the vegetation change to the greatest extent, about 95%. The study has shown that temperature is a limiting factor affecting the growth of vegetation on the Loess Plateau. We propose ICEEMDAN-SVM to study the relationship between climate factors and vegetation indices at multiple timescales., revealing some hidden information in long time series and providing a new method to quantify the impact of climate change on vegetation dynamics.