Search

Due to its rapid response time and a high degree of robustness, the selective fixed-filter active noise control (SFANC) method appears to be a viable candidate for widespread use in a variety of practical active noise control (ANC) systems. In comparison to conventional fixed-filter ANC methods, SFANC can select from a variety of pre-trained control filters for different types of noise. Deep learning technologies, thus, can be used in SFANC methods to enable a more flexible selection of the most appropriate control filters in the presence of varying incoming noise. Furthermore, deep neural network parameters can be learned automatically from noise data rather than through trial and error, significantly simplifying and improving the practicability of ANC design. Therefore, this paper compared the performance of SFANC using several different neural networks. Two-dimensional and one-dimensional neural networks are used to classify noise types in the frequency and time domains, respectively. Additionally, we compare training from scratch versus fine-tuning training to determine which network training strategy is superior in SFANC methods.

Working from home went from a temporary result of the pandemic to a way of life. However, peace and quiet are hard to come by. To get rid of distraction, Active Noise Control (ANC) particularly the ANC headphones are widely used to attenuate the undesired noise in daily life. While it works, the prolonged usage of headphone can cause some discomfort. This work describes a wireless ANC system implementation designed for the headrest to reduce the noise distraction while doing away with discomfort caused by prolonged usage of headphone. Wireless microphones are utilized to pick up the noise with a high reference-to-interference ratio, which helps in improving the noise reduction performance of the ANC headrest. This paper investigates the feedforward, feedback, and hybrid ANC structures' noise reduction performances in the ANC headrest. Comparative experimental results shown in this paper indicate the effectiveness of the proposed approach in reducing noise around the headrest.

For active noise control systems in an open aperture application, virtual sensing system is often needed to overcome design constraint in terms of microphone placement. The virtual sensing system, however, make assumptions to the acoustical field and thus is highly sensitive to any changes in the primary field. Taking the window aperture ANC application for example, incoming disturbance signals could impinge from multiple locations, altering the spatial correlation between the physical and virtual microphones. For instance, in the context of a high-rise apartment window, aircraft noise would propagate downwards from the sky, whereas traffic noise propagate upward from the ground. This paper considers the estimation performance of the remote microphone technique, an example of a virtual sensing method, in an open aperture application when faced with changes in the primary noise source. Additionally, this paper introduces a new method for estimating the spectral density due to multiple sources through the incoherent decomposition method. It has been shown experimentally that this algorithm is able to improve the nearfield estimation of the system.

The minimum output variance filtered reference least mean square (MOV-FxLMS algorithm is a effective algorithm that utilizes the penalty mechanism to help the active noise control (ANC system achieve noise cancellation with constrained output variance or power. As it can constrain output power, the MOV-FxLMS algorithm can freely determine the ANC system's control effort, avoid output saturation, and improve system stability. However, its performance is determined by a penalty factor, which is normally chosen by trial and error. Hence, this work proposes an optimal penalty factor and its feasible estimation that does not require any assumptions of Gaussian reference signal or input independence. This factor assists MOV-FxLMS in achieving the optimal solution under the target output-variance constraint. Numerical simulations on measure paths demonstrate its effectiveness for various types of noise.

The selective fixed-filter active noise control (SFANC) method selecting the best pre-trained control filters for various types of noise can achieve a fast response time. However, it may lead to large steady-state errors due to inaccurate filter selection and the lack of adaptability. In comparison, the filtered-X normalized least-mean-square (FxNLMS) algorithm can obtain lower steady-state errors through adaptive optimization. Nonetheless, its slow convergence has a detrimental effect on dynamic noise attenuation. Therefore, this paper proposes a hybrid SFANC-FxNLMS approach to overcome the adaptive algorithm's slow convergence and provide a better noise reduction level than the SFANC method. A lightweight one-dimensional convolutional neural network (1D CNN) is designed to automatically select the most suitable pre-trained control filter for each frame of the primary noise. Meanwhile, the FxNLMS algorithm continues to update the coefficients of the chosen pre-trained control filter at the sampling rate. Owing to the effective combination of the two algorithms, experimental results show that the hybrid SFANC-FxNLMS algorithm can achieve a rapid response time, a low noise reduction error, and a high degree of robustness.

Active noise control (ANC) headphone is widely used to attenuate the noise around human' s ear. The microphone mounted on the conventional ANC headphones collected the mixed reference signals when more than one noise sources are often present in the surrounding. In this case, the uncorrelated noise sources involved in the mixed reference usually deteriorate the noise reduction performance of the ANC headphones. To solve this problem, wireless microphones are proposed to install close to each potential noise source in the environment. The microphones pick up the clean reference signals and transmit them to the ANC controller embedded in the headphones with time-advance wirelessly. Every reference signal selected by a coherence-based-selection algorithm is provided individual control filter in each ear. Each control filter updated by using a single clean reference offers better noise reduction performance for ANC headphones. Furthermore, numerical simulations and real-time experiment results in this paper demonstrate the improvement of the proposed method compared with conventional ANC headphones.

Hierarchical state estimation (HSE) is often deployed to evaluate the states of an interconnected power system from telemetered measurements. By HSE, each low-level control center (LCC) takes charge of the estimation of its internal states, whereas a trusted high-level control center (HCC) assumes the coordination of boundary states. However, a trusted HCC may not always exist in practice; a cloud server can take the role of an HCC in case no such facility is available. Since it is prohibited to release sensitive power grid data to untrustworthy cloud environments, considerations need to be given to avoid breaches of LCCs’ privacy when outsourcing the coordination tasks to the cloud server. To this end, this article proposes a privacy-preserving HSE framework, which rearranges the regular HSE procedure to integrate a degree-2 variant of the Thresholded Paillier Cryptosystem (D2TPC). Attributed to D2TPC, computations by the cloud-based HCC can be conducted entirely in the ciphertext space. Even if the HCC and some LCCs conspire together to share the information they have, the privacy of non-conspiring LCCs is still assured. Experiments on various scales of test systems demonstrate a high level of accuracy, efficiency, and scalability of the proposed framework.

This paper presents a hybrid metaheuristic optimizer called CHDJ to identify accurate and reliable values for the model parameters of solid oxide fuel cells (SOFC). CHDJ is a hybridizer of differential evolution (DE) and Jaya algorithm. The core of CHDJ lies in a developed competitive hybrid operation. In this operation, three vectors including the DE mutant vector, the Jaya mutant vector, and the crossover vector that constituted by the foregoing two mutant vectors form a tri-competitive mechanism together. This mechanism can contribute to exploring the global space extensively and exploiting the local region meticulously to generate promising candidate solutions. To verify the performance of CHDJ, two SOFC models including a cylindrical cell and a tubular stack operating at different pressures and temperatures are employed. Experimental results demonstrate that these three vectors interact and cooperate organically. They make CHDJ take full advantages of the exploration of DE and the exploitation of Jaya effectively. The optimal values of the mean square error (MSE) between the measured voltage and the calculated voltage obtained by CHDJ on the cylindrical cell at 1073K, 1173K, 1213K, and 1273K are 2.9215E-06, 1.4610E-06, 1.9093E-06, and 2.4682E-06, respectively. On the tubular stack, for the operating conditions of 3atm constant pressure at 873K, 923K, 973K, 1023K, and 1073K, the optimal MSE values are 2.7489E-05, 1.6169E-04, 6.9207E-04, 1.4833E-03, and 1.5876E-03, respectively. For the operating conditions of 1073K constant temperature under 1atm, 2atm, 3atm, 4atm, and 5atm, the results are 1.5293E-03, 1.6289E-03, 1.5876E-03, 1.5868E-03, and 1.5870E-03, respectively. Comparisons with other algorithms show that CHDJ achieves better overall performance. It can skip local optima and find optimal solutions, and thus can serve as a hopeful candidate method for parameter identification of SOFC models.

Output saturation, mainly caused by the power amplifier, is a critical issue influencing the performance and stability of an adaptive system, such as in active noise control. In this paper, a quadratically constrained quadratic program (QCQP) is defined to achieve optimal control under the averaging-output-power constraint, which ensures the output of the system operates linearly and hence, avoids the output saturation. To solve this QCQP problem recursively in practice, this paper utilizes one of the leaky-based filtered-x least mean square algorithm with an optimal leak factor. However, this method only can be applied when the statistical feature of the control signal with maximum output-power is known, which is difficult to obtain in practice. Hence, by incorporating the adaptive inverse modeling technique, we can derive a practical estimation of the optimal leaky factor, which is applicable to different noise types. Furthermore, as the optimal output-constraint control forces the output to operate linearly, the nonlinear amplifier model is not required for the leak factor estimate. The simulation of the proposed algorithm is carried out on measured nonlinear paths to validate its efficacy.

Leveraging multiple heterogeneous measurements to predict wind power has long been a challenging task in the electrical community. In this paper, a deep architecture incorporated with multitask learning and multimodal learning for wind power prediction, termed predictive stacked autoencoder (PSAE), is presented. PSAE is a unified framework integrating multiple stacked autoencoders (SAEs), one feature fusion layer, and one prediction terminal layer, which expands the architecture from two spatial dimensions, including the depth and width, compared to conventional prediction models. Initially, the SAEs at the bottom of PSAE extracted features from multiple kinds of measurements respectively. Following, the feature fusion layer encodes the high-order features extracted by different SAEs into a unified feature that is more informative and representative for wind power prediction. Finally, the prediction terminal layer functions as a regression machine which generates the predicted targets based on the fusion features. Trained in an end-to-end (E2E) manner, PSAE is capable of learning heterogeneous features jointly and achieving the prediction task of sequence-to-sequence (S2S). Experiments for multi-step short-term predictions are conducted on real-world data, and the results demonstrate the superiority of PSAE to prior methods.

Shutting the window is usually the last resort in mitigating environmental noise, at the expense of natural ventilation. We describe an active sound control system fitted onto the opening of the domestic window that attenuates the incident sound, achieving a global reduction in the room interior while maintaining natural ventilation. The incident sound is actively attenuated by an array of control modules (a small loudspeaker) distributed optimally across the aperture. A single reference microphone provides advance information for the controller to compute the anti-noise signal input to the loudspeakers in real-time. A numerical analysis revealed that the maximum active attenuation potential outperforms the perfect acoustic insulation provided by a fully shut single-glazed window in ideal conditions. To determine the real-world performance of such an active control system, an experimental system is realized in the aperture of a full-sized window installed on a mockup room. Up to 10-dB reduction in energy-averaged sound pressure level was achieved by the active control system in the presence of a recorded real-world broadband noise. However, attenuation in the low-frequency range and its maximum power output is limited by the size of the loudspeakers. Ministry of National Development (MND) National Research Foundation (NRF) Published version This research is supported by the Singapore Ministry of National Development and the National Research Foundation, Prime Minister’s Office under the Cities of Tomorrow (CoT) Research Programme (CoT Award No. COT-V4-2019-1). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not reflect the views of the Singapore Ministry of National Development and National Research Foundation, Prime Minister’s Office, Singapore.

Multichannel active noise control (MCANC) is widely recognized as an effective and efficient solution for acoustic noise and vibration cancellation, such as in high-dimensional ventilation ducts, open windows, and mechanical structures. The feedforward multichannel filtered-x least mean square (FFMCFxLMS) algorithm is commonly used to dynamically adjust the transfer function of the multichannel controllers for different noise environments. The computational load incurred by the FFMCFxLMS algorithm, however, increases exponentially with increasing channel count, thus requiring high-end field-programmable gate array (FPGA) processors. Nevertheless, such processors still need specific configurations to cope with soaring computing loads as the channel count increases. To achieve a high-efficiency implementation of the FFMCFxLMS algorithm with floating-point arithmetic, a novel architecture based on multiple-parallel-branch with folding (MPBF) technique is proposed. This architecture parallelizes the branches and reuses the multiplier and adder in each folded branch so that the tradeoff between throughput and the usage of the hardware resources is balanced. The proposed architecture is validated in an experimental setup that implements the FFMCFxLMS algorithm for the MCANC system with 24 reference sensors, 24 secondary sources, and 24 error sensors, at a sampling and throughput rates of 25 kHz and 260 Mb/s, respectively. NRF (Natl Research Foundation, S’pore) Accepted version

The parameter extraction problem of solar photovoltaic (PV) models is a highly nonlinear multimodal optimization problem. In this paper, quadratic interpolation learning differential evolution (QILDE) is proposed to solve it. Differential evolution (DE) is a preeminent metaheuristic algorithm with good exploration. However, its exploitation is poor, resulting in low searching precision when applied to the problem. To overcome this deficiency, in QILDE, quadratic interpolation (QI) is embedded in the crossover operation of DE to construct a QI learning-backup crossover operation to enhance the performance of DE. The mutation scheme of DE is primarily responsible for exploring the new search space while QI is mainly in charge of exploiting the local solution space around the best individual, which, therefore, can achieve a good trade-off between exploitation and exploration. QILDE is applied to six different PV cases. The experimental results demonstrate that QI coupled with the mutation scheme DE/best/2 can obtain superior results in solving the parameter extraction problem of PV models. Besides, compared with other advanced algorithms, QILDE shows highly competitive performance in terms of solution quality, extraction accuracy, robust stability, convergence property, computational time, and statistical significance. In addition, the current–voltage characteristics provided by QILDE agree well with the measured data for different PV models under different operating conditions.

Leveraging both temporal and spatial correlations to predict wind speed remains one of the most challenging and less studied areas of wind speed prediction. In this paper, the problem of predicting wind speeds for multiple sites is investigated by using the spatio-temporal correlation. We proposed a deep architecture termed predictive spatio-temporal network (PSTN), which is a unified framework integrating a convolutional neural network (CNN) and a long short-term memory (LSTM). Initially, the spatial features are extracted from the spatial wind speed matrices by the CNN at the bottom of the model. Then, the LSTM captures the temporal dependencies among the spatial features extracted from contiguous time points. Finally, the predicted wind speeds are given by the last state of the top layer of the LSTM, which are generated by using the spatial features and temporal dependencies. Though composed of two kinds of architectures, PSTN is trained with one loss function in an end-to-end manner, which can learn temporal and spatial correlations jointly. Experiments for short-term predictions are conducted on real-world data, whose results demonstrate that PSTN outperforms prior methods.

This paper presents an evaluation method of strength measure for the static voltage stability analysis of the hybrid multi-infeed dc (HMIdc) system, which comprises both the line-commutated converter-based high voltage dc (LCC-HVdc) and voltage source converter-based HVdc (VSC-HVdc) links. For this purpose, the coupled single-port model proposed originally to assess the static voltage stability of ac power systems is first extended in the HMIdc system. In particular, how the LCC-HVdc and VSC-HVdc links should be modeled in the extended coupled single-port model is further investigated. Then, a generalized effective short-circuit ratio (GESCR) index is proposed as a strength measure for the static voltage stability analysis of the HMIdc system using the extended coupled single-port model. The GESCR can resolve the ambiguity encountered by earlier indices arising from empirical reasoning; thus, it is able to indicate the static voltage stability of the HMIdc system more accurately. Finally, the simulated results based on a classical simplified HMIdc system and a modified IEEE 118-bus system, both including two LCC-HVdc links and a VSC-HVdc link, validate the proposed GESCR.

The leaky filtered-input least mean square (LFxLMS) algorithm is widely used in active noise control applications to minimize the degradation of attenuation performance due to output saturation distortion. However, the leak factor, which is critical in determining the steady-state error and robustness of the algorithm, is usually selected through trial and error. This letter proposes a leak factor selection approach, which ensures the LFxLMS algorithm converges to its optimal solution under the average-output-power constraint and can be readily derived in practice. Both broadband and narrowband cases are considered in the derivation without the independence assumption, and the simulations are conducted based on real primary and secondary paths to verify its effectiveness. Accepted version

This study presents a novel faulted line identification method that is applicable in power systems with only sparse synchronised measurements, and at the same time, with significantly reduced computational burden so that it can be a competitive choice for wide-area backup protection schemes. By revealing the behaviour pattern of bus voltages from the perspective of high-dimensional unitary space, the method is enabled to avoid unnecessary computation and thus to remarkably improve its efficiency. Analysis in high-dimensional unitary space also gives a direct and feasible guideline for phasor measurement unit (PMU) placement, from which a PMU optimal placement strategy is developed. Performed on the IEEE benchmark systems of 39 and 118 buses, the proposed method gives encouraging results.

Feedforward active noise control (ANC) is widely utilized to attenuate the broadband noise picked up by the reference microphone. However, in some situations, it is impractical to obtain a clean reference signal when the noise source is far away from the controller. Hence, we adopt a wireless reference microphone to pick up the reference signals around the noise sources. Furthermore, a coherence-based-selection algorithm is proposed to select the reference signals with high coherence. The proposed method improves the quality of the reference signals and the noise reduction performance of the ANC system. Numerical simulations and real-time experiments are conducted to validate the effectiveness of the proposed algorithm.

The growing integration of information and communication technology exposes more exploitable cyber-attack vectors to electric power systems. Measurement manipulation is an impactive type of cyber-attacks that can be launched to cause massive damages by interfering decision-making of control centers. Existing data-driven measurement manipulation detection models often bear criticisms due to the deficient mining of spatial-temporal correlations. In this paper, based on the Gramian Angular Summation Field (GASF) algorithm, various types of phasor time-series measured at different locations are transformed into a series of generalized multichannel images. From these generalized images, Convolutional Neural Networks (CNNs) are trained to discover suspicious patterns. GASF can visualize the temporal features of a single time-series and facilitate collaborative CNNs to figure out spatial-temporal correlations between different measurements. The improvements of the proposed framework in the detection performance are demonstrated through case studies.

Fast and accurate transient stability assessment (TSA) is significant for power system. Topological information is often ignored by some machine learning methods, which may have a profound influence for TSA. For the purpose of improving the assessment accuracy and the adaptability to topology changes for online TSA, graph convolutional network (GCN) is introduced. GCN can aggregate node features and topological information. A jumping knowledge layer is added to prevent overfitting by concatenating the output of every GCN layer. Steady-state features of power system that are got through state estimation from the SCADA system are employed as the input. Tests on IEEE 39-Bus system indicate that the GCN model has a higher performance than some other methods. Several system topology changes are considered, and good adaptability to different topologies is indicated.

The goal of optimal dynamic topology evolution is to establish a distribution network which accommodates a high penetration of renewable energy with least cost. In this paper, the evolution process is formulated as a bi-level optimization model. The upper-level model optimizes the location and size of renewable distributed generation (RDG) as well as flexible resources and aims to minimize the total cost of evolution. Besides, out of simplification, the upper-level model discretizes the continues evolution process into several stages and becomes an integer programming solved by an improved genetic algorithm (GA). The lower-level model is a linear programming after the linearization of nonconvex constraints and minimizes the operation cost of the distribution system. In order to handle the uncertainty in distribution system, an hourly simulation is performed in the lower-level model which is solved by SeDuMi. Finally, the model is applied to a 13-bus distribution network to show its effectiveness.

Urban noise pollution is an omnipresent but often neglected threat to public health that must be addressed urgently. Passive noise control measures, which are less effective at reducing low-frequency noise and are often bulky and may impede airflow. As evidenced in automobiles, active control of cabin noise has resulted in lighter cars due to reduced passive insulation. Despite its long history and recent popularisation by consumer head- phones, the implementation of active noise control in the built environment is still rare. To date, active noise control (ANC) has been demonstrated, at source, in construction machines and, in the transmission path, in noise barriers. Recent demand for naturally-ventilated buildings has also spurred the development of active control solutions at the receiving end, such as on windows. The ten questions aim to demystify the principles of ANC and highlight areas in which environmental noise can be actively mitigated. Since the implementation of active control in the built environment usually involves multiple stakeholders, operational concerns are addressed. To conclude, research gaps are identified that would enable increased adoption of ANC in the built environment. There is also renewed interest in applying intelligent ANC to tackle environmentally complex applications, such as varying noise levels in the earcup of ANC headphones, particularly with the advent of the low-cost, low-power, highly-efficient embedded electronics; advancing speaker technology; and new impetus from digital signal processing and artificial intelligence Algorithms. Ministry of National Development (MND) National Research Foundation (NRF) Published version This research is supported by the Singapore Ministry of National Development and the National Research Foundation, Prime Minister’s Office under the Cities of Tomorrow (CoT) Research Programme (CoT Award No. COT-V4-2019-1). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not reflect the views of the Singapore Ministry of National Development and National Research Foundation, Prime Minister’s Office, Singapore.

With the continuous growing of wind power grid connected capacity, the requirements for the accuracy of wind farm equivalent models in power system analysis are increasing. However, traditional equivalence methods have disadvantages such as irrational number of clusters and poor adaptability of clustering results, which makes the methods unable to meet the needs in certain fields. Therefore, this paper proposes an equivalence method for wind farm based on clustering of output power time series data. This method uses the DTW algorithm to measure the similarity between the time series output power from each wind turbine generator. Then the DBSCAN algorithm is used to achieve a reasonable number of clusters and to identify the outliers. Finally, simulation verification is carried out under the conditions of wind speed disturbance and three-phase short-circuit fault, and the results are compared with those of the single-machine equivalent model and the k-means multi-machine equivalent model. Simulation results show the effectiveness of the proposed equivalence method, and the established model can reflect the operating characteristics of wind farm more accurately.

The gradual adaptation and possibility of divergence hinder the active noise control system from being applied to a wider range of applications. Selective active noise control has been proposed to rapidly reduce noise by selecting a pre-trained control filter for different primary noise detected without an error microphone. For stationary noise, considerable noise reduction performance with a short selection period is obtained. For non-stationary noise, more restrictive requirements are imposed on instant convergence, as it leads to faster tracking and better noise reduction performance. To speed up a selective filtered active noise control system, empirical wavelet transform is introduced here to accurately and instantaneously extract the frequency information of primary noise. The boundary of the first intrinsic mode function of random noises is extracted as the instant signal feature. Primary noise is attenuated immediately by picking the optimal pre-trained control filter labeled by the nearest boundary. The storage requirement for a pre-trained control filter library is reduced. Instant control is obtained, and the instability caused by output saturation is overcome. With more concentrated energy distribution, better noise reduction performance is achieved by the proposed algorithm compared to conventional and selective active noise control algorithms. Simulation results validate these advantages of the proposed algorithm. NRF (Natl Research Foundation, S’pore) Published version

The gradual adaptation and possibility of divergence have been the two main obstacles in the efficient implementation of conventional adaptive active noise control (ANC) to a wider range of applications. Selective ANC (SANC) has been proposed to rapidly reduce noise by selecting a pre-trained control filter for different primary noise detected, and improve the robustness of the system. For stationary noise, considerable noise reduction performance and system stability are obtained by SANC. However, for non-stationary noise, in order to track the variability of the signal, frequency-band-match and selection have to be conducted constantly, which results in high computational burden. To confront this problem, empirical wavelet transform (EWT) is introduced to simplify the matching and selection of SANC in this paper. This EWT based SANC (SANC_EWT) algorithm extracts the first mode of random noises, and attenuates the noise immediately by picking the optimal pre-trained control filter labeled by the first boundary. Therefore, computational complexity is reduced drastically. Simulation results show that convergence could be reached rapidly. Better noise reduction performance is achieved by SANC_EWT compared to conventional FxLMS and SANC algorithms.

Active noise control is an efficient approach in dealing with unwanted acoustic disturbances. However, most of the active noise control algorithms aim to control the signal of the error sensor leading to local noise attenuation only around the error microphones. One of the approaches to enlarge the quiet zone is by restraining the derivative of the sound field around the error microphone to zero. It achieves noise cancellation not only at the error microphone but also in its vicinity. This paper proposes a time-domain adaptive algorithm to implement the spatial derivative constraint, which avoids the complex analytic acoustic calculations. Furthermore, the proposed algorithm does not require extra microphones to acquire the sound field information during control. Numerical simulations are carried out to validate the effectiveness of the proposed method.

The multichannel implementation of the auxiliary-filter-based virtual-sensing (AF-VS) technique for active noise control applications is revisited and realized in the paper. Frequency-domain analysis based on random primary noise proves that the multichannel virtual-sensing active noise control (MVANC) technique can achieve optimal control at the desired virtual locations even if the signals at the physical and virtual microphones are not causally related. Further analysis on a number of sensor-actuator configurations shows that the MVANC technique achieves optimal control at the desired locations as long as the number of secondary sources does not exceed that of the physical error microphones. Furthermore, the simulations with measured transfer functions and real-time experiments conducted on a four-channel system validate the frequency domain analyses NRF (Natl Research Foundation, S’pore) Published version

Online dynamic security assessment is one of the important applications of online situation awareness for power systems, providing essential information for secure operation and preventive control. Different from the existing methods which are fault-dependent or requiring for transient-state data, this paper proposes a novel deep learning model for online power system multi-fault dynamic security assessment driven by hybrid information of anticipated faults and static operating points. The proposed model is universal and suitable for different fault situations. In addition, this model avoids the acquisition of transient information, thus reducing the time-consuming process of online evaluation. For the discrete nominal features in the hybrid information, one-hot encoding is introduced to preprocess the discrete nominal features. Both the performance of the dynamic security assessment model and the principal component analysis are promoted. Furthermore, an improved adaptive synthetic sampling algorithm is proposed and applied to alleviate the problem of power system data imbalance. Finally, the feasibility and effectiveness of the proposed method is verified by testing on the anticipated faults in the New England 39-bus system and the IEEE 54-machine 118-bus system.