95 results about "Spectral entropy" patented technology

The invention belongs to the technical field of circuit-breaker failure diagnosis, and particularly relates to a method and device for diagnosing mechanical characteristic failures of a high-voltage circuit-breaker. The device comprises the circuit-breaker and further comprises a vibration sensor, a voltage conditioning element, an AD conversion element, a clock element, a power element, a central processing unit, a communication unit and a failure diagnosis upper computer. According to mechanical vibration signals in the motion process of the circuit-breaker, the vibration sensor, the voltage conditioning element, the AD conversion element, the clock element, the power element, the central processing unit, the communication unit and the failure diagnosis upper computer are utilized for achieving the mechanical characteristic failure diagnosis of the circuit-breaker. The method for diagnosing mechanical characteristic failures comprises the steps of conducting wavelet packet decomposition on vibration signals in the operation process of the high-voltage circuit-breaker, extracting characteristic vectors of the vibration signals in spectral entropy of each frequency band, and adopting a relevance vector machine algorithm to conduct failure diagnosis on the mechanical characteristics of the high-voltage circuit-breaker. The method and device can effectively diagnoses the mechanical characteristic failures of the circuit-breaker, and provide a basis for the state maintenance of the circuit-breaker.

The invention relates to the technical field of speech, in particular to a real-time speech endpoint detection method and device. The method comprises the following steps that signal framing and emphasis are carried out; pulse removal processing is carried out; direct current components are removed; the short-time energy and zero-crossing rate of each frame of signal are calculated; windowing processing is carried out; spectrum reduction processing is carried out; spectral entropy is calculated; transformation smooth spectral entropy is calculated; a speech frame and a noise frame are preliminarily judged; the transformation smooth spectral entropy and a threshold are processed; a start frame and end frame in a speech segment are judged. The real-time speech endpoint detection method and device have the advantages that according to the conditions under which a signal is judged and a judged result, thresholds of parameters, such as a spectrum reduction threshold, the transformation smooth spectral entropy, the corresponding short-time energy, corresponding short-time average energy and a spectrum reduction power spectrum are weighted and updated, so that the thresholds are more andmore accurate, and finally the judged speech start frame and end frame are also more and more accurate; the method can efficiently and accurately detect speech in real time.

A spectrum calculating unit calculates, for each of the frames, a spectrum by performing a frequency analysis on an acoustic signal. An estimating unit estimates a noise spectrum. An energy calculating unit calculates an energy characteristic amount. An entropy calculating unit calculates a normalized spectral entropy value. A generating unit generates a characteristic vector based on the energy characteristic amounts and the normalized spectral entropy values that have been calculated for a plurality of frames. A likelihood calculating unit calculates a speech likelihood value of a target frame that corresponds to the characteristic vector. In a case where the speech likelihood value is larger than a threshold value, a judging unit judges that the target frame is a speech frame.

The invention aims at providing a linear frequency modulation (FM) signal parameter estimation method based on small-wave-packet denoising and power spectral entropy, which includes the following steps: denoising signals in a multi-dimension small-wave-packet mode, and determining small-wave-packet function and small-wave-packet decomposition level; calculating power spectral entropy of signals denoised in a small-wave-packet mode, and setting an entropy feature data base of linear FM signals with different FM slope under the condition of different signal to noise ratios; carrying out interpolation operation for obtained discretized entropy features; fitting the curve after interpolation of a cubic spline function with a polynomial function, and obtaining FM slop of linear FM signals under the condition of different signal to noise ratios and relational expression with input entropy features; and estimating FM slop of linear frequency modulation (LFM) signals received by a receiver by utilizing the fit expression. The linear FM signal parameter estimation method based on small-wave-packet denoising and power spectral entropy is small in calculated amount and capable of estimating FM slop of LFM in real time under the premise of meeting the requirement for parameter estimation accuracy.

A speech segment determination device includes a frame division portion, a power spectrum calculation portion, a power spectrum operation portion, a spectral entropy calculation portion and a determination portion. The frame division portion divides an input signal in units of frames. The power spectrum calculation portion calculates, using an analysis length, a power spectrum of the input signal for each of the frames that have been divided. The power spectrum operation portion adds a value of the calculated power spectrum to a value of power spectrum in each of frequency bins. The spectral entropy calculation portion calculates spectral entropy using the power spectrum whose value has been increased. The determination portion determines, based on a value of the spectral entropy, whether the input signal is a signal in a speech segment.

The invention provides a multi-point leakage location method based on VMD-PSE (variation mode decomposition and power spectral entropy) for a pressure pipeline. The multi-point leakage location methodbased on the VMD-PSE comprises the following steps of: introducing power spectrum entropy by combining the variation mode decomposition and the power spectral entropy for distinguishing the low frequency region and high frequency region of a leakage signal, compensating for the shortcoming in the unsatisfactory signal separation of the variation mode decomposition in the high frequency region, and effectively processing the high and low frequency regions of the signal to obtain more and effective leakage information; extracting characteristic values through singular value decomposition for estimating a source signal, and separating out each source leakage signal by combining a blind source separation method; and finally extracting the delay of each source leakage signal by utilizing a time-frequency analysis method, and realizing the accurate location of the multi-point leakage of the pipeline.

The invention provides an underwater acoustic communication signal modulation mode identification method based on a recurrent neural network. The method comprises steps of acquiring underwater acoustic communication signal simulation data or actual measurement data, and dividing the data into a training set and a test set; sampling each signal data, and carrying out standardization and normalization processing; extracting instantaneous frequency features and spectral entropy features from the data samples of the training set and the test set respectively; labeling data samples of the trainingset and the test set with labels; establishing a Bi-LSTM recurrent neural network model and setting parameters; inputting the training set into a network model, and training to obtain optimal trainingnetwork parameters; and switching the input of the deep learning recurrent neural network into a test set, and verifying automatic identification of the network. According to the method, the characteristic that the underwater acoustic communication signals have time sequence is adopted, the Bi-LSTM recurrent neural network capable of processing the time sequence input sequence is adopted, the network suitable for the underwater communication signals is obtained through training, and the network has a high recognition rate for the non-cooperative underwater acoustic communication signals.

The invention discloses a method for diagnosing a main circuit fault of a power converter of a switched reluctance motor. The method comprises the following steps of: detecting a transient value of a direct current bus filter capacitor voltage of a main circuit of the power converter of the switched reluctance motor; calculating the serial variance and the power spectral entropy of the direct current bus filter capacitor voltage, which serve as fault characteristic quantities; and diagnosing an upper tube short-circuit fault, a lower tube short-circuit fault and a single-phase open-circuit fault of the main circuit of the power converter of the switched reluctance motor by adopting a serial variance curve of the direct current bus filter capacitor voltage and a power spectral entropy curve of the direct current bus filter capacitor voltage of the main circuit of the power converter of the switched reluctance motor in an overall rotating speed range. The method has the characteristics of reliable fault diagnosis, obvious effect, reduction of requirement on a sensor, high cost performance, high practicality, good effect and broad application prospect.

The invention discloses an automatic splitting method and an automatic splitting system for audio punctuation. The method comprises the following steps: acquiring multiple framed segments according to an audio; acquiring an energy threshold value according to energy values of each framed segment, acquiring the framed segment of which the energy value exceeds the energy threshold value Et from each framed segment according to the energy threshold value, scanning a previous frame or a subsequent frame of the frame by taking the framed segment as a sentence middle frame, and if the energy value of the previous frame or the subsequent frame is lower than the set energy threshold value Et, combining the frame and the sentence middle frame into an independent sentence according to a frame starting sequence, and performing spectral entropy analysis on each independent sentence to obtain final analyzed sentences. Therefore, the problem of incapability in automatic punctuation in an existing caption corresponding process is solved. Therefore, not only can recorded audios and videos be processed, but also live audios and videos can be processed. For a webcast stream, webcast voices can be automatically cut, so that convenience is brought to parallel processing of a subsequent link such as a dictation link, and the processing time is shortened.

The invention discloses a spectrum sensing method and device based on time-domain energy and frequency-domain spectral entropy. The method comprises the following steps: detecting by a mode in combination of energy detection and frequency-domain spectral entropy detection; detecting the time-domain energy in case of large signal-to-noise ratio or poor channel condition; otherwise, detecting the frequency-domain spectral entropy; generating an energy pool and a spectral entropy pool for an idle frequency band according to the detection result; sequencing in the pools according to the statistical value. According to the method, the channel state of the idle frequency brand can be initially estimated, which is beneficial for a cognitive user to flexibly select the idle frequency band to communicate, and therefore, the influence of noise uncertainty on the spectrum detection can be effectively lowered down, and the detection accuracy can be improved; in addition, the self-adaptive double-threshold method is carried out, and high and low thresholds are self-adaptively adjusted according to the detection conditions, thus the mis-judgment probability can be reduced, and the detection accuracy can be raised; moreover, times for performing frequency domain detection again are decreased, the time cost for detection is reduced, and as a result, the data transmission time of the cognitive user is prolonged.

The invention provides a fault diagnosis method and device of a rolling bearing of low-speed heavy-duty equipment and a medium. The method comprises a step of acquiring a vibration signal of the rolling bearing of the low-speed heavy-duty equipment and analyzing the vibration signal to obtain a plurality of state signals, a step of filtering and denoising the multiple state signals to obtain denoised signals, and a step of constructing a three-dimensional feature on the denoised signals, obtaining a feature vector of the three-dimensional feature, taking a part of the feature vector in the feature vector of the three-dimensional feature as an input, establishing a rolling bearing fault diagnosis model based on a nuclear pole nuclear limit learning machine algorithm, inputting the remainingpart of the feature vector is into the rolling bearing fault diagnosis model to obtain a fault diagnosis result of the rolling bearing of the low-speed heavy-duty equipment. The three-dimensional feature vector based on 'EEMD energy entropy-morphological fractal dimension-morphological spectral entropy' is used an input of the classification model, the working state of the bearing can be accurately characterized and identified, and a good fault state identification effect can be obtained.

The invention relates to an audio sentence boundary detection method. The method comprises the following steps: in accordance with a phrasing problem on song audio frequency, implementing time-frequency transformation on a sung song via CQT in accordance with human ear auditory perception characteristics, and implementing fine grit end detection on the sung song via a sub-band spectral entropy method; and on the basis, implementing clustering analysis based on a K-means algorithm so as to obtain end positions of sub-sentences based on the characteristics that sentences of the sung song keep relatively long pause and pronunciation intervals therebetween, so that boundary points between unaccompanied audio sentences can be obtained relatively well, and self-adapting to musics of different rhythms can be achieved. The audio sentence boundary detection method provided by the invention is simple, flexible to implement and relatively strong in practicability.

The invention discloses a feature extraction method for performance degradation evaluation of a rolling bearing. The method comprises the following steps of S1, acquiring vibration signal informationof the rolling bearing; S2, conducting self-adaptive EEMD decomposition on a vibration signal of the rolling bearing; S3, adopting a Bayesian information criterion and a correlation kurtosis method for screening sensitive IMF components, wherein firstly, the Bayesian information criterion is adopted for calculating the number of the sensitive IMF components, secondly, the sensitive components arescreened out according to the values of the correlation kurt (CK), finally, composite spectral analysis is conducted on the sensitive IMF components, and a calculated composite spectral entropy servesas a feature parameter of the performance degradation of the rolling bearing. According to the method, a composite spectral analysis method is adopted for fusing the selected IMF components, the composite spectral entropy is extracted as the degradation feature of the rolling bearing, the sensitivity to the degradation process is high, and the capability of characterizing the degradation processof the rolling bearing by the feature is improved.

The invention discloses an emotion cognition method based on electroencephalogram signal feature analysis. The method comprises the following steps of: S1, obtaining a corresponding type of subject, and collecting electroencephalogram signals which are induced by the subject under different emotions and are used for reference and analysis; S2, performing denoising and separating processing on the electroencephalogram signals, and performing feature extraction and analysis based on a method of combining Hilbert transform and information entropy; and S3, calculating the Hilbert spectral entropy of the electroencephalogram signals in different emotional states, and performing statistical analysis. According to the method, the electroencephalogram signals of the subject under different emotions are obtained, then Hilbert transformation and information entropy are combined, the Hilbert spectral entropy of electroencephalogram rhythms of different brain regions and different genders under different emotional states is analyzed, better statistical performance is achieved, changes of time-frequency domain complexity of the electroencephalogram signals are represented, the change rule of the amplitude of the signals along with time and frequency in the whole frequency band is accurately described, the signal analysis efficiency is improved, the Hilbert spectral entropy is more reliable than approximate entropy, and the method is more comprehensive than single time domain analysis and frequency domain analysis.

The invention relates to an attention rehabilitation training and evaluation method based on spectral entropy. The method comprises the step S1 of collecting an EEG related potential of a rehabilitation training object to a simple speech attention task and preprocessing the EEG data; the step S2 of calculating the power spectral density of the preprocessed EEG data; the step S3 of calculating thespectral entropy value; the step S4 of utilizing a support vector machine to train attention EEG signals in the rehabilitation training with the spectral entropy value as the characteristic to obtaina classification model; the step S5 of collecting the resting state data before training of the rehabilitation training object; the step S6 of conducting rehabilitation training on the rehabilitationtraining object to classify the EEG data during the rehabilitation training process in real time; the step S7 of feeding back based on a real-time classification result and giving corresponding prompts to help pay attention; the step S8 of collecting the resting state data after the training of the rehabilitation training object; the step S9 of judging the rehabilitation training effect accordingto the change of the spectral entropy value collected by the resting state data before and after the training.

The invention discloses a vibration signal-based mechanical link looseness detection and early warning method. According to the method, the looseness condition of a mechanical link piece is judged bycollecting the vibration signals of the vibration device of the mechanical link piece, and then the early-warning information is given. The method comprises the following steps: step 1, constructing amechanical link looseness detection and early warning system based on vibration signals; step 2, acquiring the vibration signal of the mechanical link piece; step 3, processing the vibration signal to obtain the morphological gradient spectral entropy of the vibration signal, and then judging the operation state of the mechanical link piece; step 4, carrying out the performance degradation analysis of the mechanical link piece according to the information in the preceding steps, comprehensively evaluating the health condition of the mechanical link piece, and giving the early-warning information. The method provided by the invention is used for carrying out real-time and on-line monitoring and early warning analysis on the running state of the mechanical link. Therefore, the reliability and the safety of the operation of the mechanical link piece are improved. The better implementation performance is achieved.

The invention discloses a milling chatter online detection method based on the power spectral entropy difference. Through an acceleration sensor, vibration information in the milling process can be obtained; variation modal decomposition is used for decomposing a signal into a set of basic mode components, a basic mode component reconstruction signal of the high-frequency part is taken to obtain asignal of a frequency segment where a chatter component is located, the reconstruction signal is subjected to self-adaptation filtering, the power spectral entropy before and after signal filtering can be calculated, the obtained power spectral entropy is subjected to differencing to obtain the power spectral entropy difference, and the influence of filtering on chatter frequency band signal frequency spectrum distribution can be reflected. Compared with a traditional chatter detection method, the method can effectively separate chatter frequency segment signal, the threshold value reflects that influence of filtering on different state signals has the clear physical meaning, randomness of threshold selection can be avoided, accuracy and reliability of milling chatter detection can be improved, and the misjudgment rate and the misdetection rate can be reduced.