32 results about "Waterfall plot" patented technology

The invention discloses a blade vibration blade tip timing online monitoring method and device during inconstant rotation speed, wherein the method comprises the following steps: 1) obtaining a bladetip timing signal sequence and a rotation speed signal sequence during blade vibration; 2) constructing a blade tip timing angular domain uniformly sampled blade vibration displacement model; 3) establishing a reconstruction formula for the blade vibration blade tip timing angular domain undersampled reconstruction signal; 4) calculating the order spectrum of the blade vibration blade tip timing angular domain undersampled reconstruction signal and extracting synchronous vibration order; 5) calculating the spectrum waterfall plot of the blade vibration blade tip timing angular domain undersampled reconstruction signal to extract the asynchronous vibration frequency. The method can reconstruct the blade tip timing vibration signal with inconstant rotation speed, and the reconstructed signalhas no order aliasing. The method can simultaneously recognize the synchronous and asynchronous vibration characteristics of the blades, and the calculation process is simple and easy to implement. The high-order vibrations can be identified with very few blade tip timing sensors so as to save installation costs and space.

The invention discloses a railway health monitoring method based on a distributed optical fiber vibration sensing system. The railway health monitoring method comprises steps of using an existing communication optical fiber which is paved along a railway to perform sensing, capturing vibration information along the railway after a train passes, using data to constitute a waterfall plot which uses a gray scale to express intensity along a timeline, adopting a rim extraction algorithm based on dynamic planning to extract a track generated by a moving train after wavelet decomposition filtering is performed, marking off two areas, wherein one area is before the track and the other area is after the track, performing time frequency analysis on the track area and areas before and after the track, extracting a wheeltrack vibration characteristic frequency spectrum and a steel rail resonance characteristic frequency spectrum, obtaining wheeltrack vibration characteristic frequency and steel rail resonance characteristic frequency through a characteristic spectrum extraction algorithm, constantly repeating a monitoring process to obtain a change of a steel rail vibration characteristic along the time, and establishing a corresponding database. Newly collected data is compared with data inside the database and the database is updated, and health monitoring on the train and the steel rail is realized.

The present invention relates to a radio signal identification method based on an end-to-end convolutional neural network. The method is characterized in that: an original I/Q sampling data of an observation window is subjected to execution of preprocessing and identification through a convolutional neural network in order. The preprocessing step is that: the original I/Q sampling data of the observation window is taken as input, and a frequency spectrum waterfall plot is output after discrete Fourier transform and data format alignment processing; the step of identification through the convolutional neural network is that: the frequency spectrum waterfall plot obtained by preprocessing is taken as input, and a one-dimensional boolean vector configured to show whether all the signals to beidentified are existed or not is output after the input passes through a CNN feature extraction layer, an MLP feature mapping layer and a BR multi-tag classification layer. Compared with the mode offeature extraction and classification identification, the radio signal identification method employs the end-to-end technical solution thinking to avoid complex and low-efficient feature engineering,improve the signal identification accuracy, robustness and intelligence level, and has important meaning of radio monitoring of important areas and important activity scenes.

The invention discloses a rocket bomb muzzle velocity measurement method based on improved short-time Fourier transform, comprising the following steps: (10) echo signal windowing: discretizing echo data of a rocket bomb measured by radar, and using a Blackman window to carry out time-domain windowing on the discretized data; (20) power spectrum waterfall plot acquiring: constantly sliding time-domain windows, carrying out time-domain signal reconstruction, low-frequency coefficient extraction and instantaneous velocity calculation on the echo data in each time-domain window, and drawing a power spectrum waterfall plot according to the power spectrums in all the time-domain windows; and (30) muzzle velocity determining: finding out the position of the signal according to the power spectrum waterfall plot, fitting out the muzzle velocity of the rocket bomb according to the Doppler frequency-velocity relationship corresponding to the peak, and drawing a velocity fitted curve of the rocket bomb. The invention aims to provide a rocket bomb muzzle velocity measurement method based on short-time Fourier transform, which has the advantages of small amount of calculation and high measurement accuracy.

The invention discloses a real-time waterfall plot implementing method, and relates to a method used for implementing real-time waterfall plot in the frequency spectrum monitoring field. The method comprises the steps: a frequency spectrum data generation unit reads frequency spectrum data through a DMA mode, and circularly stores the frequency spectrum data in a data region by means of a ping-pong operation; the frequency spectrum data processing unit uses a weighting factor method to perform weight calculation on data at the same location of the frequency spectrum data in each frame; according to a time order, the amplitude of the frequency spectrum data after being subjected to weighting factor weighing is converted into color data and display on a display screen according to a self-defined color-amplitude ratio. The method disclosed by the invention has characteristics of real-time collection, real-time analysis, and real-time display; important information features of the frequency spectrum cannot be lost; the problem of frame loss caused by differences between processing speeds among hardware can be effectively avoided; and the method is particularly adapted to manufacturing a real-time frequency spectrum machine that can rapidly detect a momentary signal such as a hop signal in the frequency spectrum monitoring field.

The invention belongs to the technical field of aerospace experiments, and particularly relates to a method for reconstructing, displaying and analyzing an engine test data three-dimensional waterfall plot. The method comprises the main steps of: 1, exporting a text-format file from an original three-dimensional waterfall plot of Desktop-Advanced software of LMS Test.Lab Rev; 2, reading amplitude column data in spectrum data from the text-format file; 3, storing the spectrum data as a two-dimensional array; and 4, reconstructing a three-dimensional waterfall plot and carrying out display and analysis, wherein an analysis method comprises analysis steps of extracting data by spectral slices, carrying out dynamic perspective viewing, carrying out data comparison and the like. By the method disclosed by the invention, the defects of display aliasing, fuzzy display effect and poor discrimination of spectrum data of a single-color three-dimensional waterfall plot automatically generated by Desktop-Advanced software of a current LMS Test.Lab Rev are overcome, and the problem of inconvenience for an operator to carry out analysis processing is solved.

The invention discloses a vehicle calibration method and system. The method comprises the steps: inputting a shifting strategy mapping table and an accelerator pedal curve mapping table into a vehiclesimulation model; selecting a plurality of accelerator fixed values from the accelerator pedal curve mapping table, and carrying out the fixed acceleration simulation according to the plurality of accelerator fixed values through a simulation model; selecting a vehicle speed and a corresponding acceleration from a simulation result, and drawing a fixed acceleration waterfall plot according to theselected vehicle speed and the corresponding acceleration; and optimizing a torque in an accelerator pedal curve under the corresponding accelerator value and a shifting point in the shifting strategy mapping table according to the acceleration waterfall plot. According to the invention, the method can achieve the fixed acceleration simulation through the model, can achieve the quick and efficient searching of the most ideal shifting point, improves the efficiency, reduces the independence on a real-vehicle test, saves the development cost of a vehicle, and reduces the test risks.

The invention relates to a method for rapidly evaluating electromagnetic disturbance under dynamic working conditions of an electric vehicle. The method includes the steps of A, enabling the vehicle to be driven on a dynamometer and undergo a series of working conditions, and at a sampling frequency of not lower than 5 Hz, synchronously collecting a drum linear speed of the dynamometer and time-domain data of drum force as well as time-domain signals in a vertical direction of an electric field and three orthogonal directions of a magnetic field and conducted emission time-domain signals, wherein the drum speed of the dynamometer is equal to a vehicle speed under the condition that wheels don't slip; B, transforming the time-domain signals in the vertical direction of the electric field and the three orthogonal directions of the magnetic field and the conducted emission time-domain signals into frequency-domain signals of corresponding time points in real time through a fast Fourier transform algorithm and displaying the frequency-domain signals as a waterfall plot; C, displaying the drum linear speed of the dynamometer and the time-domain data of the drum force correspondingly to the frequency-domain signals of the time points corresponding to the vertical direction of the electric field, the three orthogonal directions of the magnetic field and conducted emission through a data fusion algorithm. The method is capable of evaluating the real conditions of the electromagnetic disturbance of the electric vehicle more reliably.

The invention relates to a rotating machinery frequency-doubled fault diagnosis method based on a three-dimensional holographic difference spectrum. The method comprises the following steps of: firstly, determining a measuring point with the most sensitive fault response on a unit shaft system, analyzing the main excitation frequency and amplitude variation trend of the measuring point by utilizing a frequency spectrum waterfall chart or a characteristic trend chart, and judging whether the fault type is a frequency-doubled fault or not by combining a rule; then calculating a double-frequencythree-dimensional holographic spectrum matrix of the shaft system in a normal state and a plurality of fault states, and constructing a double-frequency three-dimensional holographic differential spectrum matrix corresponding to the fault states on the basis; and finally, drawing a frequency-doubled three-dimensional holographic differential spectrogram of the shaft system in a plurality of faultstates to obtain a pure fault frequency-doubled ellipse of the whole shaft system in a fault development process and change characteristics of characteristics of the pure fault frequency-doubled ellipse, determining a fault source position by combining the characteristics, and carrying out qualitative diagnosis on the frequency-doubled fault. According to the method, the difference of the fault response signs is represented from time and space dimensions, and key diagnosis information is provided for accurately distinguishing rotor cracks and misalignment faults.

The invention discloses a method for calculating radiation noise of a silencer of an exhaust system. The method comprises the following steps: S1, extracting a static pressure curve of each cavity inthe silencer to be analyzed from simulation software; S2, performing pretreatment through a conversion tool completed during analysis process development, and converting static pressure into frequencyspectrum excitation; S3, conducting finite element modeling on the silencer, and making frequency response calculation; S4, extracting a calculation result, and synthesizing the calculation result into a waterfall plot and a total sound pressure curve; S5, evaluating the total sound pressure curve and the waterfall plot, if the total sound pressure curve and the waterfall plot do not pass the evaluation, returning to the adjustment structure, and starting evaluation again from the step S1; and if so, completing the step of analyzing the radiation noise of the silencer, and then entering a sample piece manufacturing stage. Compared with the prior art, the invention has the advantages that the recognition capability of the earlier-stage risk of silencer development is improved, the originaldevelopment process depending on tests and repeated iteration is simplified, and the development cost and time are saved.

The invention provides an enhanced phase waterfall plot-based rotor crack weak fraction harmonic feature identification method. Identification of weak frequency components is realized by using signal frequency domain phase spectrum features based on the characteristic that the phase value of the frequency components is independent of energy. The method comprises the following steps: 1, carrying out short time mobile sampling on signals to be analyzed to obtain serial short time signals, and adopting fast Fourier transform (FFT) to obtain frequency domain phase spectra respectively corresponding to the serial short time signals; 2, processing the obtained frequency domain phase spectra through adopting a line unit form filtering technology to obtain filtered feature enhanced frequency domain phase spectra in order to complete feature enhancement of the phase spectra; and 3, sequentially arranging the enhanced frequency domain phase spectra in a sampling time sequence to form an enhanced phase waterfall plot to realize separation of the regular phase band in the real frequency components and the random phase value caused by FFT calculating error and noises in order to complete identification of the rotor crack weak fraction frequency components.

The invention provides an intrusion alarm method and system based on vibration optical fiber detection and complete action extraction, and the method comprises the steps: S1, solving and storing a signal, and setting a threshold value to judge an intrusion action signal and give an alarm; s2, obtaining an original vibration signal curve and carrying out normalized leveling processing on the original vibration signal curve; s3, performing high-frequency high-pass filtering processing on the leveled signal, performing binarization processing on the image, and performing external contour extraction to obtain a minimum enclosing rectangle; s4, low-frequency high-pass filtering processing is carried out on the signals after leveling processing, a time-frequency image data set of a classification network is constructed, and an EfficientNet classification network model is trained; and S5, according to the information of the minimum enclosing rectangle, in combination with a matrix obtained through an OfficientNet classification network, realizing classified alarm. According to the method, the energy waterfall plot with a relatively good effect can be quickly obtained, so that the position point of the suspicious intrusion signal can be quickly obtained in the spatial dimension, and the coverage time of the suspicious intrusion signal can be accurately obtained in the time dimension.

The invention discloses a sonar image recognition method and device, and the method comprises the steps: intercepting a target sonar image from a sonar waterfall plot according to a preset frame number interval, and recording a frame identifier of a specified echo data frame; recognizing the target sonar image based on a preset target recognition model, acquiring the preset information of a target object in the target sonar image, and acquiring the preset target recognition model by training a one-stage type target detection algorithm based on a first training set formed by synthesizing the sonar image, wherein the preset information comprises position information and category information; based on the frame identifier, determining position information of a target sonar image in a sonar waterfall plot; and marking the target object in the sonar waterfall plot based on the position information of the target sonar image in the sonar waterfall plot and the preset information of the target object in the target sonar image. The method and the device are high in recognition speed, can improve the recognition efficiency of the sonar image, and can be applied to actual production.

The invention discloses a qualitative diagnosis method for a power frequency fault of rotary machinery. The method comprises the following steps: firstly, synchronously acquiring multiple groups of shaft vibration displacement signals and corresponding key phase signals at two measuring points which are vertically arranged at a rotor fault response sensitive section; then extracting a pure fault power frequency component in the vibration signal by using a two-dimensional holographic difference spectrum, and finally constructing a two-dimensional holographic difference spectrum waterfall plot corresponding to the power frequency and a characteristic trend plot thereof to completely describe the change characteristics of pure fault procession trajectory characteristics. According to the method, the change rule of pure fault power frequency track characteristics can be visually reflected, differentiated information is effectively mined from the fault development process, and accurate qualitative diagnosis of the rotor power frequency fault is realized.

The invention belongs to the technical field of modal tests and particularly relates to a time-varying modal parameter identification method which comprises the following steps: constructing a spectrum analysis module; performing time domain acquisition on acceleration sensor data and force sensor signals through the spectrum analysis module; constructing a time-varying analysis module; analyzingthe data acquired in the step 2 through the time-varying analysis module to obtain a waterfall plot of a frequency response function; selecting different frequency response functions from the frequency response function waterfall plot for analysis, and obtaining modal parameters. According to the time-varying modal parameter identification method provided by the invention, the time-varying analysis module is constructed to analyze the time-domain acquisition data, and finally, the modal parameters are obtained, so that the modal parameter identification problem of the time-varying system can be solved.