Battery impedance spectrum prediction from partial charging voltage curve by machine learning

Electrochemical impedance spectroscopy(EIS) is an effective technique for Lithium-ion battery state of health diagnosis, and the impedance spectrum prediction by battery charging curve is expected to enable battery impedance testing during vehicle operation. However, the mechanistic relationship between charging curves and impedance spectrum remains unclear, which hinders the development as well as optimization of EIS-based prediction techniques. In this paper, we predicted the impedance spectrum by the battery charging voltage curve and optimized the input based on electrochemical mechanistic analysis and machine learning. The internal electrochemical relationships between the charging curve,incremental capacity curve, and the impedance spectrum are explored, which improves the physical interpretability for this prediction and helps define the proper partial voltage range for the input for machine learning models. Different machine learning algorithms have been adopted for the verification of the proposed framework based on the sequence-to-sequence predictions. In addition, the predictions with different partial voltage ranges, at different state of charge, and with different training data ratio are evaluated to prove the proposed method have high generalization and robustness. The experimental results show that the proper partial voltage range has high accuracy and converges to the findings of the electrochemical analysis. The predicted errors for impedance spectrum are less than 1.9 mΩ with the proper partial voltage range selected by the corelative analysis of the electrochemical reactions inside the batteries. Even with the voltage range reduced to 3.65–3.75 V, the predictions are still reliable with most RMSEs less than 4 mO.

Deep Reinforcement Learning Based Spectrum Prediction for Bursty Bands

Spectrum prediction plays an important role for the secondary user(SU)to utilize the shared spectrum resources.However,currently utilized prediction methods are not well applied to spectrum with high burstiness,as parameters of prediction models cannot be adjusted properly.This paper studies the prediction problem of bursty bands.Specifically,we first collect real Wi Fi transmission data in 2.4GHz Industrial,Scientific,Medical(ISM)band which is considered to have bursty characteristics.Feature analysis of the data indicates that the spectrum occupancy law of the data is time-variant,which suggests that the performance of commonly used single prediction model could be restricted.Considering that the match between diverse spectrum states and multiple prediction models may essentially improve the prediction performance,we then propose a deep-reinforcement learning based multilayer perceptron(DRL-MLP)method to address this matching problem.The state space of the method is composed of feature vectors,and each of the vectors contains multi-dimensional feature values.Meanwhile,the action space consists of several multilayer perceptrons(MLPs)that are trained on the basis of multiple classified data sets.We finally conduct experiments with the collected real data and simulations with generated data to verify the performance of the proposed method.The results demonstrate that the proposed method significantly outperforms the stateof-the-art methods in terms of the prediction accuracy.

Cross-Band Spectrum Prediction Algorithm Based on Data Conversion Using Generative Adversarial Networks

Spectrum prediction is one of the new techniques in cognitive radio that predicts changes in the spectrum state and plays a crucial role in improving spectrum sensing performance.Prediction models previously trained in the source band tend to perform poorly in the new target band because of changes in the channel.In addition,cognitive radio devices require dynamic spectrum access,which means that the time to retrain the model in the new band is minimal.To increase the amount of data in the target band,we use the GAN to convert the data of source band into target band.First,we analyze the data differences between bands and calculate FID scores to identify the available bands with the slightest difference from the target predicted band.The original GAN structure is unsuitable for converting spectrum data,and we propose the spectrum data conversion GAN(SDC-GAN).The generator module consists of a convolutional network and an LSTM module that can integrate multiple features of the data and can convert data from the source band to the target band.Finally,we use the generated target band data to train the prediction model.The experimental results validate the effectiveness of the proposed algorithm.

Deep Spectrum Prediction in High Frequency Communication Based on Temporal-Spectral Residual Network

High frequency(HF) communication is widely spread due to some merits like easy deployment and wide communication coverage. Spectrum prediction is a promising technique to facilitate the working frequency selection and enhance the function of automatic link establishment. Most of the existing spectrum prediction algorithms focus on predicting spectrum values in a slot-by-slot manner and therefore are lack of timeliness. Deep learning based spectrum prediction is developed in this paper by simultaneously predicting multi-slot ahead states of multiple spectrum points within a period of time. Specifically, we first employ supervised learning and construct samples depending on longterm and short-term HF spectrum data. Then, advanced residual units are introduced to build multiple residual network modules to respectively capture characteristics in these data with diverse time scales. Further, convolution neural network fuses the outputs of residual network modules above for temporal-spectral prediction, which is combined with residual network modules to construct the deep temporal-spectral residual network. Experiments have demonstrated that the approach proposed in this paper has a significant advantage over the benchmark schemes.

Spectrum Prediction Based on GAN and Deep Transfer Learning:A Cross-Band Data Augmentation Framework

This paper investigates the problem of data scarcity in spectrum prediction.A cognitive radio equipment may frequently switch the target frequency as the electromagnetic environment changes.The previously trained model for prediction often cannot maintain a good performance when facing small amount of historical data of the new target frequency.Moreover,the cognitive radio equipment usually implements the dynamic spectrum access in real time which means the time to recollect the data of the new task frequency band and retrain the model is very limited.To address the above issues,we develop a crossband data augmentation framework for spectrum prediction by leveraging the recent advances of generative adversarial network(GAN)and deep transfer learning.Firstly,through the similarity measurement,we pre-train a GAN model using the historical data of the frequency band that is the most similar to the target frequency band.Then,through the data augmentation by feeding the small amount of the target data into the pre-trained GAN,temporal-spectral residual network is further trained using deep transfer learning and the generated data with high similarity from GAN.Finally,experiment results demonstrate the effectiveness of the proposed framework.

Cross-Band Spectrum Prediction Based on Deep Transfer Learning

Spectrum prediction is a promising technology to infer future spectrum state by exploiting inherent patterns of historical spectrum data.In practice,for a given spectrum band of interest,when facing relatively scarce historical data,spectrum prediction based on traditional learning methods does not work well.Thus,this paper proposes a cross-band spectrum prediction model based on transfer learning.Firstly,by analysing service activities and computing the distances between various frequency points based on Dynamic Time Warping,the similarity between spectrum bands has been verified.Next,the features,which mainly affect the performance of transfer learning in the crossband spectrum prediction,are explored by leveraging transfer component analysis.Then,the effectiveness of transfer learning for the cross-band spectrum prediction has been demonstrated.Further,experimental results with real-world spectrum data demonstrate that the performance of the proposed model is better than the state-of-theart models when the historical spectrum data is limited.

A Spectrum Prediction-Based Frequency Band Pre-Selection over Deteriorating HF Electromagnetic Environment

As the earliest invented and utilized communication approach, shortwave, known as high frequency(HF) communication now experience the deterioration of HF electromagnetic environment. Finding quality frequency in efficient manner becomes one of the key challenges in HF communication. Spectrum prediction infers the future spectrum status from history spectrum data by exploring the inherent correlations and regularities. The investigation of HF electromagnetic environment data reveals the correlations and predictability of HF frequency band in both time and frequency domain. To solve this problem, we develop a Spectrum Prediction-based Frequency Band Pre-selection(SP-FBP) for HF communications. The pre-selection of HF frequency band mainly incorporated in prediction of HF spectrum occupancy and prediction of HF usable frequency, which provide the frequency band ranking of spectrum occupancy and alternative frequency for spectrum sensing, respectively. Performance evaluation via real-world HF spectrum data shows that SP-FBP significantly improves the efficiency of finding quality frequency in HF communications.

A novel spectrum prediction scheme based on SVM in cognitive radio networks

In cognitive radio networks, Secondary Users (SUs) have opportunities to access the spectrum channel when primary user would not use it, which will enhance the resource utilization. In order to avoid interference to primary users, it is very important and essential for SUs to sense the idle spectrum channels, but also it is very hard to detect all the channels in a short time due to the hardware restriction. This paper proposes a novel spectrum prediction scheme based on Support Vector Machines (SVM), to save the time and energy consumed by spectrum sensing via predicting the channels' state before detecting. Besides, spectrum utilization is further improved by using the cooperative mechanism, in which SUs could share spectrum channels' history state information and prediction results with neighbor nodes. The simulation results show that the algorithm has high prediction accuracy under the condition of small training sample case, and can obviously reduce the detecting energy, which also leads to the improvement of spectrum utilization.

Spectrum sensing sequence prediction in cognitive radio networks

Spectrum sensing is one of the key issues in cognitive radio networks. Most of previous work concenates on sensing the spectrum in a single spectrum band. In this paper, we propose a spectrum sensing sequence prediction scheme for cognitive radio networks with multiple spectrum bands to decrease the spectrum sensing time and increase the throughput of secondary users. The scheme is based on recent advances in computational learning theory, which has shown that prediction is synonymous with data compression. A Ziv-Lempel data compression algorithm is used to design our spectrum sensing sequence prediction scheme. The spectrum band usage history is used for the prediction in our proposed scheme. Simulation results show that the proposed scheme can reduce the average sensing time and improve the system throughput significantly.

Predictive Spectrum Sensing Strategy Based on Reinforcement Learning

In this paper,we consider a cognitive radio(CR) system with a single secondary user(SU) and multiple licensed channels.The SU requests a fixed number of licensed channels and must sense the licensed channels one by one before transmission.By leveraging prediction based on correlation between the licensed channels,we propose a novel spectrum sensing strategy,to decide which channel is the best choice to sense in order to reduce the sensing time overhead and further improve the SU's achievable throughput.Since the correlation coefficients between the licensed channels cannot be exactly known in advance,the spectrum sensing strategy is designed based on the model-free reinforcement learning(RL).The experimental results show that the proposed spectrum sensing strategy based on reinforcement learning converges and outperforms random sensing strategy in terms of long-term statistics.

The dependence of response spectrum on the tectonic ambient shear stress field

It has been analyzed the influence of the tectonic ambient shear stress value on response spectrum based on the previous theory. Based on the prediction equation BJF94 presented by the famous American researchers, CLB20, a new prediction formula is proposed by us, where it is introduced the influence of tectonic ambient shear stress value on response spectrum. BJF94 is the prediction equation, which mainly depends on strong ground motion data from western USA, while the prediction equation SEA99 is based on the strong ground motion data from exten-sional region all over the world. Comparing these two prediction equations in detail, it is found that after BJF94′s prediction value lg(Y) minus 0.16 logarithmic units, the value is very close to SEA99′s one. This case demonstrates that lg(Y) in extensional region is smaller; the differences of prediction equation are mainly owe to the differences of tectonic ambient shear stress value. If the factor of tectonic ambient shear stress value is included into the pre-diction equation, and the magnitude is used seismic moment magnitude to express, which is universal used around the world, and the distance is used the distance of fault project, which commonly used by many people, then re-gional differences of prediction equation will become much less, even vanish, and it can be constructed the uni-versal prediction equation proper to all over the world. The error in the earthquake-resistant design in China will be small if we directly use the results of response spectrum of USA (e.g. BJF94 or SEA99).

Novel mode-coupling vibrations of AlN thin film bulk acoustic resonator operating with thickness-extensional mode

The dispersion curves of bulk waves propagating in both AlN and ZnO film bulk acoustic resonators(FBARs)are presented to illustrate the mode flip of the thickness-extensional(TE)and 2nd thickness-shear(TSh2)modes.The frequency spectrum quantitative prediction(FSQP)method is used to solve the frequency spectra for predicting the coupling strength among the eigen-modes in AlN and ZnO FBARs.The results elaborate that the flip of the TE and TSh2 branches results in novel self-coupling vibration between the small-wavenumber TE and large-wavenumber TE modes,which has never been observed in the ZnO FBAR.Besides,the mode flip leads to the change in the relative positions of the frequency spectral curves about the TE cut-off frequency.The obtained frequency spectra can be used to predict the mode-coupling behaviors of the vibration modes in the AlN FBAR.The conclusions drawn from the results can help to distinguish the desirable operation modes of the AlN FBAR with very weak coupling strength from all vibration modes.

Spectrum Occupancy Prediction for Realistic Traffic Scenarios: Time Series versus Learning-Based Models

Spectrum occupancy information is neces-sary in a cognitive radio network(CRN)as it helps in modeling and predicting the spectrum availability for efficient dynamic spectrum access(DSA).However,in a CRN,it is difficult to ascertain a priori the pattern of the spectrum usage of the primary user due to its stochastic behavior.In this context,the spectrum occupancy predic-tion proves to be very useful in enhancing the quality of experience of the secondary user.This paper investigates the practical prowess of various time-series modeling approaches and the machine learning(ML)techniques for predicting spectrum occupancy,based on a spectrum measurement campaign conducted in Jaipur,Rajasthan,India.Moreover,the comparison analysis conducted between the above two approaches highlights the trade-off in terms of the respective performance depending upon the nature of the spectrum occupancy data.Nevertheless,prediction through ML-based recurrent neural network proves to perform reasonably well,thereby providing an accurate future spectrum occupancy information for DSA.

Mono-isotope Prediction for Mass Spectra Using Bayes Network

Mass spectrometry is one of the widely utilized important methods to study protein functions and components. The challenge of mono-isotope pattern recognition from large scale protein mass spectral data needs computational algorithms and tools to speed up the analysis and improve the analytic results. We utilized na¨？ve Bayes network as the classifier with the assumption that the selected features are independent to predict monoisotope pattern from mass spectrometry. Mono-isotopes detected from validated theoretical spectra were used as prior information in the Bayes method. Three main features extracted from the dataset were employed as independent variables in our model. The application of the proposed algorithm to public Mo dataset demonstrates that our na¨？ve Bayes classifier is advantageous over existing methods in both accuracy and sensitivity.