14 results about "Ship noise" patented technology

The invention discloses a ship identification method and system based on dual physical model fusion verification, and relates to the technical field of optical fiber sensing and ocean monitoring. The method aims at solving the problems that ship noise in submarine optical cable DAS signals is prone to interference of environment noise such as low-speed guided waves, the false alarm rate is high, and accurate positioning cannot be achieved. The method comprises the steps that DAS phase data is acquired and preprocessed to generate a channel-time energy diagram; suppressing low-speed guided wave noise through frequency-wavenumber domain slowness scanning, detecting an energy ridge line on the purified energy diagram, and calculating an observation apparent velocity; acquiring AIS data to calculate and predict the apparent velocity; the relative error of the observed apparent velocity and the predicted apparent velocity is firstly matched, and then linear regression is carried out on the arrival time of the candidate ridge line and the AIS distance to verify and fit the root mean square error; the validated ridge line is confirmed as an effective ship event and a directory is generated. According to the technical scheme, through physical mechanism denoising and dual model verification, the accuracy and reliability of ship identification are effectively improved.

The application provides a marine environment noise prediction method for ship abnormal behavior monitoring, and comprises the following steps: establishing a marine environment noise prediction model; determining the type of the marine environment noise numerical prediction model; and performing marine environment noise prediction according to the determined type of the prediction model. The application constructs a fast and refined marine environment noise prediction model, the fast model considers three noise sources of wind, rain and ships, and realizes the purpose of real-time fast prediction of marine environment noise according to the source level empirical formula of different noise sources summarized according to the measured data; the refined model considers two noise sources of wind and ships, and realizes the purpose of refined prediction of marine environment noise by using the source level empirical formula of different noise sources, combining real-time gridded wind speed and ship AIS data to complete the construction of wind and ship noise source models, and comprehensively considering the factors of gridded hydrological information and ocean depth database and calculation frequency.

The invention provides an unmanned ship navigation noise measurement system and a data processing method, which are used for solving the problems that the adaptability of an existing standard to an unmanned ship point sound source is poor, the laying and the recovery of a measurement system are tedious, and the deviation of a data processing result is large. Specifically, the measurement system is composed of a computer, a data acquisition module and the like, and the attitude and stability of the hydrophone are ensured through a single-point suspension design; according to data processing, 1-2 s time domain data about the maximum value of the broadband sound pressure level is selected, a line spectrum and a 1 / 3Oct spectrum are obtained through FIR band-pass filtering, framing windowing, FFT and the like, and the total sound pressure level and the sound source level are calculated. The noise of the unmanned ship is conveniently and accurately measured and processed, the test cost is saved, and the real level of the noise is accurately reflected.

The application provides a ship noise sequence generation method based on frequency domain mask guided diffusion, multi-granularity frequency information is constructed by introducing a frequency domain random mask to guide the diffusion process, thereby improving the quality and diversity of the generated samples. Traditional diffusion methods usually rely on complete initial information for generation, and less explore the use of incomplete intermediate information as guidance. Existing methods mostly use time domain fixed filters to construct multi-granularity data as intermediate information, but ignore the randomness of the data, limiting the flexibility and diversity of the generation process. Therefore, the application proposes to apply a random mask in the frequency domain to construct multi-granularity guide information as intermediate information to guide diffusion, and combine the correlation coefficient matrix embedding as a condition to supplement, to supplement the channel correlation information lost due to the mask operation.

The invention relates to the technical field of noise recognition, in particular to a ship noise recognition method and system based on deep learning, and the method comprises the following steps: collecting a ship noise signal, and generating a two-dimensional time-frequency spectrogram of the ship noise signal through adaptive time-frequency transformation; according to the two-dimensional time-frequency spectrogram, generating a time sequence feature vector sequence through a global time sequence perception convolutional neural network; inputting the time sequence feature vector sequence into a multi-scale time sequence mode extraction network to obtain a multi-scale time sequence fusion feature of the ship noise signal; and inputting the multi-scale time sequence fusion features into a noise classification network to realize noise identification. According to the method, a more adaptive time-frequency spectrogram can be generated through adaptive time-frequency transformation, the capability of capturing a noise periodic mode and the sensitivity of anomaly detection are improved through the global time sequence perception convolutional neural network and the multi-scale time sequence mode extraction network, and the accuracy and the intelligent degree of noise recognition in a complex marine environment are improved.

The invention relates to a multi-modal deep learning marine ship noise classification method based on bilinear fusion, and belongs to the technical field of acoustic signal processing. The multi-modal deep learning marine ship noise classification method based on bilinear fusion comprises the steps of data preprocessing, construction of a double-flow feature extraction network, bilinear feature fusion and classification. And the bilinear feature fusion step comprises the steps of performing dimension reduction processing on the waveform feature vector and the spectrum feature vector respectively, calculating an outer product weighted sum of the waveform feature vector after dimension reduction and the spectrum feature vector after dimension reduction through bilinear interaction, obtaining a fusion feature vector, and performing nonlinear processing on the fusion feature vector to obtain a final fusion feature vector. According to the method, fine-grained interaction of the waveform features and the spectrum features is realized through bilinear fusion, complementary information of two modes can be more fully mined, and the expression ability and discrimination of the fused features are improved.

The application discloses a dolphin sound driving method, system and equipment based on underwater acoustic signal feedback adjustment, and relates to the field of marine ecological protection.According to the hearing threshold curve characteristics of dolphins, the application adopts a superimposed signal of a downward frequency sweeping signal and a high-speed yacht noise signal as an artificial synthetic sound wave for sound driving, and in order to avoid the situation that the sound driving effect of dolphins is significantly reduced or disappears with time under continuous sound wave stimulation (because dolphins are adaptive to sound), the application method designs an amplitude frequency shifter based on Hilbert transform, and adds a feedback adjustment mechanism for the signal to simulate the frequency shift and amplitude change of the real ship noise propagation, so that the variability of the emitted sound wave frequency and amplitude is maintained.Therefore, the application method can effectively drive dolphins for a long time, so that the purpose of protecting dolphins from being hurt in underwater engineering activities is achieved.

The invention relates to a dynamic stretching towing sonar system and method capable of reducing flow noise, and the system comprises a towing winch which is disposed on a ship and is used for collecting and releasing a towing cable; the towing array sonar is connected with the towing winch through the towing cable; wherein the total length of the towing cable is 5-10 km, and the towing winch dynamically controls the retracting and releasing length of the towing cable in the sailing process of the ship. According to the invention, the winch provided with a longer trailing cable is used for dynamically and safely controlling and releasing the trailing cable, and the trailing sonar moves at a speed lower than the driving speed of the ship and is far away from the ship, so that the interference of flow noise and the noise of the ship is greatly reduced.

The application discloses an online prediction method for underwater radiation noise based on body vibration and self-noise, and belongs to the technical field of ship noise monitoring and evaluation. The method is characterized in that: a plurality of test working conditions under different working conditions are designed according to a corresponding navigation body, and parameters such as body vibration, self-noise and underwater radiation noise of the navigation body under the test working conditions are accurately obtained; based on the above-mentioned collected parameters, a transfer function and a working condition clustering center under each working condition are accurately calculated; parameter information of the navigation body in an actual operation process is collected in real time, and underwater radiation noise is accurately predicted according to the transfer function and the working condition clustering center. The application provides an online prediction method for underwater radiation noise based on body vibration and self-noise, can reduce nonlinearity and ill-conditioned property of solving of underwater radiation noise transfer function under influence of external disturbance, accurately calculate the transfer function, and enable underwater radiation noise to be accurately obtained in real time under different running environments and running states of the navigation body.

On the basis of a finite element and statistical energy method, a ship underwater radiation noise time domain passing characteristic space distribution simulation calculation method is provided, and ship sound field passing characteristics at different positions in a flow field are obtained on the premise that a sound source is not moved by adopting frequency domain analysis and spatial discretization; the method can be used for calculating the full-band passing characteristics of the complex sound field under the coupling action of a whole ship noise source, the time domain-frequency domain-space characteristics of the sound field are reflected, and the problem that a high-precision simulation calculation method for the passing characteristics of the radiation sound field of the complex sound source is lacked currently is solved.

The application discloses a high-fidelity power pulse sound source for ship noise simulation and a control method thereof, and contains an electromagnetic transducer and a driving circuit thereof, the electromagnetic transducer driving circuit includes elements for actively controlling the charging current of an energy storage capacitor, and a bipolar Marx circuit for generating a bipolar power pulse, the working mode includes a bipolar pulse output mode and a power amplifier output mode, and a circuit control method required for simulating ship noise. The application adopts the above-mentioned high-fidelity power pulse sound source for ship noise simulation and the control method thereof, realizes accurate injection and extraction of the current in the electromagnetic transducer coil, and further accurately controls the transducer output sound wave, the two working modes of bipolar pulse output and power amplifier output simulate the continuous spectrum and line spectrum in the ship noise respectively, and finally, a neural network for generating a pulse power trigger signal is trained based on a cycle gating unit, and high-fidelity ship noise simulation is realized.

The invention relates to a ship noise signal simulation method based on a deep neural network. The method comprises the following steps: simulating a continuous spectrum, a modulation spectrum and a line spectrum of ship noise; simulating the continuous spectrum by adopting an FIR filter design based on a Welch method; an improved model based on fusion of bidirectional LSTM and a self-attention mechanism is provided, so that the time-frequency characteristic of a modulation envelope is effectively captured, and high-precision simulation of a modulation spectrum is realized; a physical information neural network (PINN) is innovatively introduced, physical constraints are embedded into a loss function, and high-fidelity simulation of line spectrum features is achieved.

The invention discloses an underwater acoustic communication codebook generation method, device and equipment and a readable storage medium, and the method comprises the steps: normalizing the characteristics of candidate signal segments through preprocessing, and weakening the inherent correlation influence caused by a stable line spectrum and a similar envelope in real ship noise; a similarity evaluation index between candidate signal segments is defined, and similarity misjudgment caused by similar segment characteristics is avoided; based on the similarity, a simulated annealing optimization algorithm is adopted to screen a target signal segment from a candidate signal set, the algorithm does not adopt a simple random or greedy selection mode, and low-cross-correlation segments are screened in a targeted mode through heuristic search logic, so that the defect that random or greedy selection is easy to select high-cross-correlation segments is effectively avoided, and the selection accuracy of the target signal segment is improved. Further, the codebook separability is improved, and the problem that the codebook separability is poor due to the fact that real ship noise contains stable line spectrums and similar envelopes is solved; meanwhile, in the screening process of the method, a low-correlation starting point set is obtained through farthest point sampling, and a traditional exhaustion or global optimization mode does not need to be adopted.

The invention provides an SEA subsystem automatic identification method based on a grid model, and belongs to the technical field of ship noise control. For a ship finite element model, a non-triangular grid is converted into a triangular grid by adopting an ear cutting method; dividing unit sets on the basis of the attribute information of the grid units, and searching for each unit set through the topological relation between the grids to form a complete plate structure corresponding to the deck, the bulkhead and the outer plate of the ship; the method comprises the following steps: after board subsystem identification is completed, constructing a topological relation for an imported finite element model based on a half-side and half-surface data structure, and filtering out internal unit components by using a single side; selecting a surface, continuously searching adjacent surfaces by adopting a region growing method until all surfaces of the vocal cavity to which the surface belongs are found, completing vocal cavity recognition, completing recognition of all vocal cavities through circulating search, and finally removing redundant vocal cavities to complete establishment of a vocal cavity subsystem; the method is suitable for rapid construction of the SEA model in ship cabin noise prediction.