492 results about "White noise" patented technology

The invention provides a high-voltage circuit breaker voiceprint denoising method based on data enhancement and a storage medium, and the method comprises the steps: processing a collected original voiceprint data sequence, extracting stable and effective Mel-frequency cepstrum coefficient features, and constructing a two-dimensional feature matrix; a parallel mixed data enhancement strategy is adopted to generate a positive sample pair, and an encoder is trained in combination with a contrast learning mechanism, so that the representation robustness of the model under different voiceprint change conditions is improved. The method comprises the following steps: decomposing an original signal containing noise fringes into a plurality of modal components by using variational modal decomposition, extracting low-frequency effective components, introducing Gaussian white noise, constructing a corrosion target signal as a decoder training target, learning through a denoising automatic encoder, and finally outputting a denoised voiceprint feature signal. The method has the advantages of high robustness, high noise suppression capability, excellent feature expression capability and the like, is suitable for the field of online monitoring and intelligent diagnosis of the state of high-voltage circuit breaker equipment, and has good application prospect and engineering value.

The invention discloses a park heterogeneous load prediction method based on EEMD and Kmeans adaptive optimization. The method comprises the following steps: carrying out missing value filling, abnormal value elimination and feature coding on park historical load data and external influence factors; dividing the park load into a plurality of typical scenes through a Kmeans clustering algorithm based on the similarity of load fluctuation characteristics; eEMD decomposition is carried out on the load sequence corresponding to each typical scene, Gaussian white noise is added to suppress modal aliasing, multi-frequency components are extracted and fused with external variables, and a prediction input data set is constructed; a prediction model is selected for different scenes, and adaptive adjustment and optimization are carried out through a Bayesian optimization algorithm; and carrying out dynamic weight fusion on the prediction result of each scene, dynamically adjusting the weight according to the historical load proportion of the scene, and outputting an overall load prediction value of the park.

The invention relates to the technical field of signal processing, and discloses a power equipment partial discharge signal denoising method and system, and the method comprises the steps: S1, generating an exponential decay type pulse signal simulating partial discharge, and superposing white noise and narrow-band interference into the exponential decay type pulse signal to generate a noisy signal; s2, decomposing the noisy signal into a plurality of intrinsic mode components through an ensemble empirical mode decomposition method, calculating a correlation coefficient and a kurtosis index of each intrinsic mode component, and identifying a noise component and a signal component based on the correlation coefficient and the kurtosis index; s3, wavelet threshold denoising is carried out on the noise components, and a wavelet threshold is dynamically adjusted for each noise component through a particle swarm optimization algorithm; and S4, reconstructing the de-noised noise component and the reserved signal component into a final de-noised signal, and evaluating the de-noising performance based on the signal-to-noise ratio, the root mean square error, the peak signal-to-noise ratio and the correlation coefficient.

The invention relates to the technical field of rolling bearing fault diagnosis, in particular to a rolling bearing fault diagnosis method based on data fusion and a neural network, and the method comprises the steps: S1, collecting vibration data of a plurality of sensors, superposing Gaussian white noise and pulse noise, and simulating a strong noise industrial environment; s2, extracting time domain and frequency domain features of vibration signals of each sensor, and constructing a preliminary feature matrix; s3, performing noise reduction processing and depth feature extraction on the feature matrix by using SDAE, fusing to generate a bearing data set, and dividing a training set and a test set; and S4, inputting the training set into an Adam optimization-based fractional order GRU model for training, and verifying the trained model by using the test set. According to the rolling bearing fault diagnosis method, the precision and robustness of rolling bearing fault diagnosis in a strong noise environment are remarkably improved by fusing the multi-source sensing data and the fractional order neural network model.

The invention relates to a building structure health monitoring method and system based on machine learning, and the method specifically comprises the following steps: firstly, building a target building three-dimensional numerical model through finite element simulation, generating a simulation signal, injecting Gaussian white noise, and adjusting model parameters to form a data set with health category labels; performing data enhancement by combining adaptive wavelet denoising with dynamic normalization, and extracting and enhancing high-resolution time-frequency features through adaptive window short-time Fourier transform and adaptive frequency band enhancement; then, a neural network model fusing structure physical prior guidance and multi-scale space-time interaction is constructed, a feature matrix is modulated, fused and coded to obtain a refined feature vector, and damage state probability distribution output is achieved; and training is carried out by using a feature consistency and prediction smoothness regularization term constraint model, and finally, the trained model is deployed, so that building structure health state evaluation and safety early warning are realized, the monitoring accuracy and reliability are improved, and effective technical support is provided for building safety guarantee.

The invention relates to the technical field of short-term power load prediction, in particular to a short-term power load prediction method combining improved empirical mode decomposition and a bidirectional long short-term memory network, which comprises the following steps: preprocessing historical load data; decomposing the load data into a plurality of intrinsic mode functions by adopting a complete adaptive noise integrated empirical mode decomposition algorithm, and adaptively adding Gaussian white noise to suppress a mode aliasing problem; constructing a prediction model based on a bidirectional long-short-term memory network and a self-attention mechanism for each component; extracting features by an input layer through a sliding window; the training layer adopts a bidirectional long short-term memory network to extract time sequence hidden features, and a key time point is focused through weighting of a self-attention mechanism; the output layer generates a component prediction value; and superposing all component prediction results, and reconstructing to obtain an accurate load prediction value of the to-be-predicted day. According to the method, the prediction precision and stability are effectively improved by improving signal decomposition and deep learning model fusion.

The invention relates to an audio control method and device, electronic equipment and a vehicle. The audio control method comprises the following steps: acquiring multi-modal data; performing feature analysis on the multi-modal data to obtain an analysis processing result, and constructing a user portrait based on the analysis processing result; generating audio white noise corresponding to the user portrait based on the user portrait; according to the embodiment of the invention, the multi-modal data can be acquired, the personalized characteristics of the user can be determined by analyzing the multi-modal data from multiple dimensions, the accuracy of user portrait construction is improved, on the basis, the demands of the user are deeply understood based on the user portrait, and the user experience is improved. According to the method, the audio white noise suitable for the user is generated, and audio control is performed through the audio white noise, so that the in-vehicle environment is improved to meet different requirements of the user, and the user experience is improved.

The invention discloses a structural damage detection method based on a bridge digital twinning group, and the method comprises the steps: building a general damage detection model through a multi-dimensional digital twinning technology: building a reference bridge model through Abaqus, and generating a digital twinning group containing a single / multiple damage scene; training the ethnic group by improving a WGAN-GP architecture, outputting a marked damage image by a generator, introducing a multi-scale attention module into a discriminator, and constructing an ethnic group characteristic spectrum (EFM) matrix through characteristic distillation; then collecting feature parameters to train a CNN model, and fusing features through SENet channel attention; through Gaussian white noise verification and n-fold cross optimization, the F1 score of the model is greater than or equal to 0.89 in a 40% noise environment, and the average accuracy rate reaches 95% or above. According to the technology, the limitation of single structure detection is broken through, the model generalization ability is improved by 300% and the detection efficiency is improved by 50% through a dynamic twin-driven convolutional network self-evolution mechanism, and complex scenes with + / -20% size change and 0-100% damage degree can be accurately identified; and an efficient anti-interference solution is provided for health monitoring of structures such as bridges, high-rise buildings and the like.

The invention relates to an intelligent hypnosis method and system based on a brain-computer interface and a storage medium, and belongs to the technical field of human-computer interaction and artificial intelligence. The method comprises the following steps: acquiring an EEG signal of a user in real time through EEG acquisition equipment, and extracting a brain wave segment power value as state input through preprocessing; a deep reinforcement learning model (such as DRQN) selects music type actions (such as classical music and white noise) based on an epsilon-greedy strategy; calculating a reward value (maximizing delta wave increment and inhibiting beta wave) according to the electroencephalogram state change after playing, and optimizing model parameters by adopting Q-Learning; and dynamically adjusting the strategy through iterative interaction until the user reaches a preset sleep target. The system comprises an electroencephalogram acquisition module, a preprocessing module, a reinforcement learning module and a music control module, and realizes closed-loop regulation and control. The method has the advantages of high personalization, high hypnosis efficiency (induction to sleep in 1-7 minutes), flexible adaptation to different users and self-evolution optimization capability, and effectively improves the sleep induction effect.

The invention provides a power load prediction method and system based on adaptive modal decomposition, and the method comprises the steps: obtaining an original load sequence and a multi-modal data set based on time sequence distribution, and the multi-modal data set comprises meteorological data, equipment state data, and economic and social activity data; copying the original load sequence into a plurality of copies, and adding adaptive white noise to each original load sequence to obtain a plurality of noise-added sequences; performing modal decomposition, component screening and modal reconstruction on each noise adding sequence, and performing superposition to obtain a de-noising load sequence; performing feature screening based on causal driving on the de-noising load sequence and the multi-modal data set to obtain a feature vector; and inputting the feature vector into a preset load prediction model, so that the load prediction model outputs a corresponding load prediction result based on an attention mechanism and a preset physical rule constraint, and the accuracy of power load prediction in a complex scene is improved.

The invention relates to the technical field of geodetic survey and structural health monitoring, and discloses a GNSS-RTK coordinate domain error correction method, which comprises the following steps: acquiring GNSS-RTK dynamic observation data to form a mixed signal time sequence; executing an improved adaptive noise complete empirical mode decomposition algorithm on the sequence to obtain an intrinsic mode function component; identifying and eliminating high-frequency components representing Gaussian white noise based on an energy coefficient, and reconstructing residual components into a signal sequence after primary noise reduction; inputting the noise reduction sequence into a rapid independent component analysis model for blind source separation; and finally, carrying out sorting and phase and amplitude uncertainty correction on the separated independent components, and outputting a multi-path error model and a structure dynamic deformation signal. According to the method, the problem of blind source separation failure or low precision caused by strong noise covering source signal statistical characteristics is solved through a strategy of first noise reduction and then separation, and an effective physical signal can be accurately extracted from a strong noise background.

The invention discloses a baby cry recognition method for white noise equipment, and relates to the field of audio processing and intelligent acoustic recognition, and the method comprises the steps: collecting a pure reference signal, a first-path signal and a second-path signal, carrying out the synchronization and preprocessing, and obtaining a mixed audio signal based on the first-path signal; a residual signal is calculated through an adaptive echo cancellation algorithm; calculating an acoustic masking parameter based on the residual signal, the pure reference signal, and the ambient noise estimate; constructing a three-level recognition processing path including lightweight feature detection, registration voiceprint comparison and multi-modal information fusion; based on the numerical range of the acoustic masking parameter, selecting an identification processing path, and determining a crying event of the target infant from the residual signal; and triggering a corresponding grading alarm based on a crying event confirmation result. A three-level identification path is scheduled through acoustic masking parameters, and accurate and low-power-consumption baby crying monitoring under strong interference is realized by fusing multi-modal information.

The invention relates to the technical field of deep sleep management, in particular to a deep sleep staging management method and system based on electroencephalogram feedback. Comprising the steps that electroencephalogram feedback of a target user is collected, an actual deep sleep waveform is extracted and judged with an expected spectral domain of an ideal state transfer stage, and an unreached stage is locked; in an unreached stage, constructing a signal mode prediction model, and determining an optimal guide window stage by combining with intervention entry point migration analysis; performing wavelet basis decomposition to obtain an excellent wave band and a clutter band of the optimal guide window period; when the signal-to-noise ratio is high, crosstalk coupling analysis is executed, and white noise is generated by using the rhythm coding library for guide management; when the signal-to-noise ratio is low, light environment and temperature response regulation is planned, and deep sleep is managed in combination with a color coding chain and a signal excitation result. According to the method and the system, guide management of different decisions can be executed in the stage of entering the deep sleep of the target user, and the deep sleep quality, stability and continuity are improved.

The transformer substation connection fitting damage identification method provided by the invention comprises the following steps: emitting white noise to a transformer substation connection fitting to obtain a vibration signal of the transformer substation connection fitting; an optimal influence coefficient of the preset harmonic sine structural element is determined by adopting a starfish optimization strategy, an optimal harmonic sine structural element is obtained, and the starfish optimization strategy is that after influence parameters of the preset harmonic sine structural element are initialized, an objective function corresponding to the vibration signal is adopted to perform optimal influence parameter solution search; and carrying out morphological gradient combination product operation on the vibration signal by adopting the optimal harmonic sine structural element, and then carrying out spectral analysis to obtain a damage identification result of the transformer substation connection fitting. Therefore, accurate identification of the damage of the connection fitting is realized.

The invention belongs to the technical field of partial discharge detection, particularly relates to a partial discharge signal denoising method based on SVDTQWT, and aims to effectively remove periodic narrow-band interference and white noise in partial discharge signals. Comprising the following steps: performing Fourier transform on a noisy partial discharge signal to obtain a frequency spectrum, determining the number of periodic narrowband interferences through singular value decomposition, constructing a Hankel matrix to eliminate the periodic narrowband interferences, and obtaining a preliminary de-noised signal; and decomposing the preliminarily denoised signal by adopting adjustable quality factor wavelet transform to obtain a plurality of sub-bands. And dividing the plurality of sub-bands into high-frequency sub-bands and low-frequency sub-bands through sample entropy. Wherein the sample entropy indicates measurement of the complexity of the time series. And de-noising the high-frequency sub-band by using a group sparse total variation de-noising algorithm, de-noising the low-frequency sub-band by using an improved wavelet threshold de-noising algorithm, and reconstructing by using adjustable quality factor wavelet transform to obtain a pure partial discharge signal.

The present disclosure relates to methods and systems based on a new type of radar waveform, taking the form of a continuous aperiodic signal, effectively indistinguishable from white noise, though actually it is a signal allowing both accurate and ambiguity free range and radial velocity, i.e. Doppler, measurements.

The invention discloses a GIS partial discharge fault positioning method based on multi-feature fusion and kernel density estimation optimization, and the method comprises the following steps: collecting a partial discharge original pulse signal, carrying out the signal processing, and outputting a first processing signal; performing adaptive robust baseline determination on the first processing signal, detecting a wave head position based on multi-feature fusion and a dynamic threshold voting mechanism, and outputting a wave head arrival time; carrying out time synchronization matching on the wave head arrival time, and calculating and outputting a preliminary positioning result based on a time difference of arrival method; and accumulating a plurality of preliminary positioning results, performing optimization by adopting a kernel density estimation method, and outputting a final partial discharge fault position. According to the method, impulse noise and white noise are effectively suppressed, stable signals are provided for subsequent processing, the detection precision of the wave head arrival time is remarkably improved, the problem that false detection and missing detection are caused by the fact that a traditional fixed threshold value or a single feature is prone to noise interference is solved, the random error of single-time positioning is greatly compressed to be within 0.07 m from 0.33 m on average, and the overall positioning precision is improved by 80%.

Methods are generally described for inaudible orthogonal signal communication. An example method includes determining, based on an input message, a series of symbols, where each symbol from the series of symbols represents a numerical value, wherein each symbol from the series of symbols corresponds to a band-limited white noise tone orthogonal to each other band-limited white noise tone corresponding to each other symbol. The example method also includes generating a sub-token comprising each symbol from the series of symbols, where each symbol overlaps in time with at least one other symbol. The example method also includes modulating a carrier wave with the sub-token to produce a signal audio waveform, where the modulation uses a spread spectrum technique, and providing the signal audio waveform to a television device, where the television emits the signal audio waveform as a low-amplitude audio signal.

The invention discloses an online modeling and control method for a narrowband active noise control system, which comprises the following steps of: (1) performing frequency estimation on a signal picked up by a reference microphone by using a triangular cascade adaptive notch filter algorithm, and calculating an online estimated value of a signal-to-noise ratio by using energy of an input signal and an output signal of a notch filter; (2) when the online estimated value of the signal-to-noise ratio is larger than a preset value, modeling and control are started, a reference signal and auxiliary noise are generated based on the estimated frequency, and a control signal is calculated; (3) estimating auxiliary secondary sound by using the auxiliary noise; (4) estimating the sum of the primary sound and the secondary sound of the error signal at the noise frequency by using the reference signal; (5) updating the estimation of the frequency response of the secondary path at the auxiliary noise frequency and the frequency response of the auxiliary filter at the narrow-band noise frequency; (6) obtaining the estimation of the frequency response of the secondary path at the narrowband noise frequency in an interpolation mode; (7) updating the filter coefficient of the controller; and (8) outputting a control signal to drive a secondary source loudspeaker to produce sound, propagating the sound through a secondary path, and coherently superposing the sound with the initial noise signal at the error microphone to realize a noise reduction function. According to the method, the secondary path can be accurately modeled online, compared with an existing algorithm, the auxiliary noise energy is low, the convergence speed is high, the noise reduction effect is good, meanwhile, the problem that the same parameter operation results are different in a traditional auxiliary white noise online modeling algorithm is solved, and the algorithm stability is better.

The invention discloses a quantitative calculation method for analyzing injection-production dynamic response by using time sequence interference. The quantitative calculation method comprises the following steps: S1, acquiring time sequence data of output and water injection rate; s2, taking the change of the water injection rate as an intervention event, and establishing a virtual interference variable; s3, taking the oil well output as a target variable and substituting the target variable and the virtual interference variable in the S2 into an SARMAX time sequence model for fitting; s4, judging the significance degree of the model by using the residual error of the model, and if the residual error is a random time sequence, ending; if the residual sequence of the model is non-white noise, going to S5; s5, determining autoregression and moving average orders by using an ARMA (p, q) model; s6, substituting order information into the model in the S3 for re-fitting; and S7, repeating the steps S3-S6 until the model is significant. The method can quantitatively evaluate the influence degree of the injection well on the output. Quantifiable data are provided for evaluation of production and injection dynamic influences in oil reservoir dynamic management, reasonable injection and production adjustment countermeasures are formulated accordingly, oil reservoir production dynamic management is guided, and the oilfield exploitation effect is guaranteed.

The invention discloses a secondary path online modeling and real-time ANC system observation method without additional white noise, and the method comprises the following steps: monitoring the stability of an ANC system through a real-time observer, and preventing divergence caused by the change of a secondary path or an overlarge step length; a secondary path online modeling method is designed, secondary path online precise modeling without adding white noise is realized by combining an offline modeling initial value and online self-adaptive updating, and residual noise rise and human ear discomfort caused by introduction of the white noise in a traditional online modeling method are avoided; secondary path coupling is eliminated by using a loudspeaker time-sharing sounding strategy, and the modeling calculation amount is reduced; and an intelligent ANC system is constructed, a secondary path model is dynamically updated, the step length is adaptively adjusted, and stable operation of the system is ensured. According to the invention, the stability and the noise reduction performance of the ANC system are improved, and the calculation complexity and the human ear interference risk are reduced at the same time.

The invention discloses a ring stationary signal extraction method based on a multi-scale gating attention mechanism and BiLSTM, and belongs to the technical field of mechanical fault diagnosis and signal processing. The objective of the invention is to solve the problem that weak fault features are difficult to accurately extract in a strong background noise and non-Gaussian abnormal interference environment in the prior art. Comprising the steps that a deep neural network model is constructed, local waveforms of different receptive fields are captured through parallel convolution branches, channel masks are generated through an attention mechanism, and hierarchical adaptive suppression of background noise is achieved; biLSTM is embedded in a bottleneck layer, a bidirectional memory unit of the BiLSTM is used for carrying out complete sequence time sequence modeling on compression features, and random abnormal pulses which do not conform to periodic logic are accurately eliminated through a long-term evolution rule of the sequence features; and finally, a decoder is adopted to carry out nonlinear shaping and detail repairing on the up-sampling features after transposed convolution. Gaussian white noise and non-Gaussian outliers can be efficiently filtered out, and pure ring stationary fault impact signals are reconstructed in a high-fidelity mode.

The invention discloses an EEMD (ensemble empirical mode decomposition)-based industrial sensor missing data filling method and system, and relates to the technical field of sensor data processing.The method comprises the steps that firstly, self-adaptive dynamic EEMD is carried out, the amplitude of white noise is dynamically adjusted through Bayesian noise modeling, modal aliasing is inhibited, and a physically interpretable IMF component is generated; thirdly, based on missing mode classification, a divide-and-conquer strategy is adopted, dynamic weight fusion IMF components are combined, and filling precision and real-time performance are optimized; and finally, evaluating the authenticity of the reconstructed signal by using a GAN framework, dynamically adjusting a smoothing factor, and verifying the multi-scale consistency. According to the method, the accuracy and the physical interpretability of data are improved through adaptive dynamic EEMD decomposition, the accuracy and the real-time performance of missing data filling are optimized through a divide-and-conquer strategy, the reliability and the multi-scale consistency of signals are enhanced through the generative adversarial network, and therefore the limitation of a traditional method in a high-noise and non-stationary industrial environment is effectively solved.

The invention provides a differential microphone array speech enhancement device and method based on reference point optimization, and relates to the technical field of speech signal processing. The device comprises a microphone array, a pre-processing module, a reference point selection module and a signal processing module. According to the invention, the reference point of the microphone array is improved from the first microphone in the prior art to the geometric center of the invention, the maximum value of the distance rm from the microphones in the microphone array to the reference point is reduced by half, the high frequency band is significantly reduced, and the approximation precision of Jacobi-Anger expansion is ensured. Through reference point optimization, the invention aims to realize a beam pattern with invariable frequency in a broadband range and combined optimization of DF and WNG. While the beam pattern approximation requirement and no distortion constraint are met, the white noise gain can be maximized, and the robustness is improved.

The invention discloses a method and a device for detecting structural anomaly of a switch cabinet, which are used for solving the technical problem of non-ideal detection effect caused by the existing method for detecting the structural anomaly of the switch cabinet. The method comprises the following steps: preprocessing an obtained switch cabinet original vibration signal based on a plurality of preset white noise sequences, and outputting a normalized switch cabinet digital signal and a plurality of normalized switch cabinet disturbance signals; performing first-order mode successive extraction according to the normalized switch cabinet digital signal and the plurality of normalized switch cabinet disturbance signals based on an empirical mode decomposition algorithm, and outputting a total residual error and a plurality of initial adaptive intrinsic mode function components; screening each initial adaptive intrinsic mode function component, and outputting a plurality of target adaptive intrinsic mode function components; performing superposition reconstruction according to the multiple target adaptive intrinsic mode function components and the total residual error, and outputting a switch cabinet noise reduction signal; and performing structural anomaly detection on the switch cabinet based on the switch cabinet noise reduction signal, and outputting a target detection result.

The invention provides a privacy protection noise generation method based on black box adversarial samples, and relates to the technical field of noise generation. Comprising the steps of constructing a voice data set; randomly generating initial white noise data; superposing the initial white noise data and target voice in the voice data set to obtain a superposed voice data set; performing data amplification on the superposed voice data set to obtain an amplified data set; inputting the amplified data set into a local white-box speech recognition model to obtain final white noise data; performing noise amplification on the final white noise data, and superposing the final white noise data with the voice data set to obtain a multi-intensity noise-added voice data set; and performing fine tuning training on the local white box model by using the multi-intensity noise-added voice data set to obtain a final noise generation model so as to generate final privacy protection noise. The technical problem that interference noise cannot be optimized based on an adversarial sample technology when the black box model parameters and architecture cannot be obtained and only the final recognition result can be obtained in the prior art is solved.

The invention provides a high-speed ADC clock jitter measurement method and device, and relates to the technical field of high-speed and high-precision analog-digital conversion. Comprising the following steps: setting coprime coherence of an input sinusoidal signal and a sampling clock signal, and determining a sampling sequence according to the frequency and clock jitter of the input sinusoidal signal; according to the gain mismatch, the phase offset and the detuning mismatch, separating an output sequence of the sub-channel from the sampling sequence, and determining an output sequence amplitude according to the amplitude and the gain mismatch; determining a sub-channel jitter error sequence according to the output sequence of the sub-channel, and calculating an autocorrelation function of the sub-channel jitter error sequence according to a mean value of the sub-channel jitter error sequence; according to the autocorrelation function of the sub-channel jitter error sequence, the standard deviation of additive white noise and the excitation of time delay, the total cyclic spectral density is determined through Fourier transform; and determining the standard deviation of clock jitter according to the total cyclic spectral density, the output sequence amplitude and the frequency of the input sinusoidal signal. Therefore, the test precision of the clock jitter is improved.

The invention discloses a beam structure fault intelligent diagnosis method based on a DIC and a CNN. The method comprises the steps that a beam structure model is established in finite element analysis software, a plurality of node positions corresponding to DIC measuring points are arranged on the surface of a structure, loads are applied, time domain vibration response signals of the multiple measuring points are extracted, and Gaussian white noise is superposed; performing wavelet packet analysis and multi-band adaptive noise reduction on the acquired multi-channel vibration signals, judging the energy distribution characteristics of the signals by using the change rate of wavelet energy entropy, and adaptively adjusting the length of a segmented window to decompose original non-stationary signals into a plurality of stationary time periods; the inherent frequency, the damping ratio and the vibration mode modal parameters of the structure are obtained through working modal analysis; and constructing a feature sample set by taking the extracted modal parameters and time domain features as inputs, inputting the feature sample set into a convolutional neural network model for training and prediction, and realizing classification and identification of various states such as structure health, early cracks, middle cracks and late cracks.

The invention relates to the field of railway data analysis and processing, in particular to a railway irregularity data processing method based on cEEMDAN-Hilbert. The method comprises the following steps: acquiring and preprocessing track irregularity original signals to obtain a target track irregularity signal sequence; performing CEEMDAN (adaptive noise complete ensemble empirical mode decomposition) on the target track irregularity signal sequence, and performing iterative decomposition on the signal added with Gaussian white noise to obtain a plurality of IMF (intrinsic mode function) components and a residual component; performing Hilbert transformation on each IMF component, calculating an instantaneous frequency and an instantaneous amplitude, constructing an analysis signal in a complex form, and extracting time and frequency joint distribution information; according to the instantaneous frequency and the energy result, false wavelength components are removed, signal reconstruction is carried out, and reconstructed track irregularity signals are generated. According to the invention, the identification capability of local structure disturbance can be enhanced.

The invention relates to an electroencephalogram signal preprocessing method based on particle swarm optimization (PSO), ensemble empirical mode decomposition (EEMD) and independent component analysis (ICA), and aims to improve the quality of electroencephalogram signals and enhance analyzability of the signals. The method comprises the following steps: firstly, carrying out EEMD (Ensemble Empirical Mode Decomposition) on an original electroencephalogram signal, generating a plurality of noise auxiliary signals by adding white noise with different intensities, carrying out EMD on each noise auxiliary signal, and then averaging intrinsic mode functions (IMF) of all the noise auxiliary signals to obtain a final IMF, thereby reducing the problem of mode aliasing and improving the reliability of the electroencephalogram signal. Useful components and noise components are preliminarily separated out; secondly, optimizing EEMD parameters by using a PSO algorithm; a particle swarm is initialized, each particle represents a possible parameter combination (such as Gaussian white noise standard deviation and noise adding times), a fitness function is defined, a quality index of a signal is taken as a target, positions and speeds of the particles are iteratively updated, the parameter combination is optimized, and finally optimal parameters are applied to perform EEMD decomposition, so that an optimized IMF is obtained. Therefore, the complexity of manual parameter adjustment is avoided, and the decomposition accuracy and stability are improved. Then, the sample entropy of each IMF is calculated, a sample entropy threshold value is set, the IMFs with the sample entropy values higher than the threshold value are screened out, IMF components with low information content are effectively removed, effective components with high information content are reserved, and the analyzability of the signals is further improved. And finally, combining the screened effective IMF component with the original electroencephalogram signal to generate a virtual multi-channel signal, performing Fast ICA (Independent Component Analysis), separating out independent electroencephalogram signal components, further removing noise, and improving the purity and the signal-to-noise ratio of the signal. Through the steps, the quality of the electroencephalogram signals can be remarkably improved, and a solid foundation is provided for subsequent signal analysis and application.