19results about How to "Increase sampling rate" patented technology

The invention provides a physical information long-short-term memory network site seismic response prediction method based on kinematics derivative constraint and data folding. The method comprises the following steps: collecting and preprocessing bedrock and ground surface seismic oscillation data; displacement, speed and acceleration supervision labels are constructed through surface acceleration integration; constructing a long-short-term memory network containing a data folding and unfolding module; physical constraint loss is constructed by utilizing kinematics derivative relations between displacement and speed and between speed and acceleration, and a network is trained by combining an adaptive weight strategy; and outputting earth surface displacement, speed and acceleration time history by adopting the trained network. According to the method, under the condition that constitutive parameters of a soil layer are not explicitly given, physical consistency constraint can be applied to the field seismic response prediction process, the long-time program train training burden is reduced through sequence folding processing, and the method is suitable for field seismic response prediction.

The invention belongs to the technical field of path planning and robot motion planning, and particularly relates to a multi-strategy fusion improved RRT * sampling planning method and device and a storage medium, and the method comprises the steps: constructing a dual-search tree through a starting point and a target point in a three-dimensional working space, and selecting a to-be-expanded search tree through a random mode; halton-Bridge mixed sampling is adopted to generate sampling points, so that the global coverage uniformity is improved, and the narrow channel sampling capability is enhanced; calculating the dynamic step length of the exponential decay based on the local obstacle proportion; adopting a hierarchical guidance expansion strategy, preferentially expanding towards a target, introducing an improved artificial potential field to guide expansion when the expansion fails, and taking the bottom with random expansion; performing RRT * neighborhood father node reselection and reconnection optimization on the new node; greedy pruning, self-adaptive interpolation and cubic B-spline smoothing are carried out on a path after the double trees are connected, and an executable smoothing track is output; according to the scheme, planning efficiency, success rate and path quality are improved.

This invention provides a method, apparatus, device, and readable storage medium for improving audio generation quality, relating to the field of speech conversion and recognition technology. The method includes acquiring low-sampling-rate audio, a target audio sampling rate, and an audio processing model; calculating an initial high-sampling-rate audio based on the low-sampling-rate audio, the target audio sampling rate, and the audio preprocessing mathematical model; calculating the target audio time-domain signal and target audio wavelet coefficients based on the initial high-sampling-rate audio and an audio signal reconstruction mathematical model; and solving the mathematical model based on the target audio time-domain signal, the target audio wavelet coefficients, and a preset fused audio signal mathematical model to obtain the target high-sampling-rate audio. This invention uses discrete wavelet transform to capture multi-scale details of the audio signal and reconstructs high-sampling-rate audio by combining the time-domain and frequency-domain information of the audio signal, further improving the overall generation quality of the audio.

This invention relates to the field of audio circuit hardware technology, and in particular to a low-latency ARC circuit configuration method and system based on I2S audio switching. External video data is received through an ARC circuit compatibility interface and input to a video conversion chip for data separation, achieving synchronous separation of I2S audio signals and MIPI video signals. The separated I2S audio signals are then simulated and sent to the host SOC and external I / O terminals respectively by a first-stage audio switching chip. Through link delay analysis and instantaneous delay judgment, the audio path is dynamically configured to achieve switching control between the internal processing path and the direct path. A second-stage audio switching chip realizes multi-channel audio multiplexing and direct transmission simulation, achieving a low-latency, highly compatible ARC audio output scheme while ensuring audio and video timing alignment. This invention can significantly reduce the direct transmission latency of ARC audio while ensuring audio and video synchronization and system compatibility, and improves the flexibility and reliability of audio path configuration.

The invention relates to a data set decomposition-based large language model training method, which comprises the following steps of: performing token processing on an original document in a training data set to obtain a one-dimensional token sequence and calculating an original length; performing binary decomposition on the original length, splitting the token sequence into sub-token sequences, storing the sub-token sequences in buckets according to the length, and adding an original document traceability identifier; calculating a single sampling number of each effective bucket, and configuring a sampling probability; executing random sampling according to the sampling probability, and performing homologous constraint splicing to generate a training batch; and inputting the training batch into the to-be-trained large language model, and performing fine adjustment on the sampling probability adjustment rule and the training hyper-parameter according to the results of the conventional reference performance evaluation, the long context capability evaluation and the training efficiency evaluation. By means of the large language model training method based on data set decomposition, cross-document attention interference can be reduced, calculation waste can be reduced, and training efficiency can be improved.

The invention discloses an enhanced imaging method for a microwave ESM sparse scanning system, and the method comprises the steps: obtaining the amplitude and phase information of a part of radiation sources on a far-field sampling plane, and obtaining sparse sampling data; on the basis, an image reconstruction problem is converted into a sparse optimization problem on a gradient domain by using total variation regularization, the gradient of the image in a smooth region approaches to zero by applying gradient constraint, and a relatively large gradient value is only reserved at a significant feature position. And an alternating direction multiplier method is further adopted to carry out iterative solution on the total variation minimization problem, and complete amplitude and phase information is gradually reconstructed by decomposing and optimizing a target and alternately updating variables. And dynamically adjusting the sampling rate according to the SSIM difference of the reconstructed image between adjacent iterations, and stopping sampling in advance when the SSIM value is higher than the quality threshold and the change tends to be stable. And finally, transforming the reconstructed amplitude and phase information through a far and near field inversion algorithm to obtain a near field diffraction pattern of the radiation source, and completing high-precision positioning and imaging of the radiation source.

The application discloses a fault detection device based on fast edge pulse frequency scanning, comprising: a digital code type generation module configured to generate a controllable square wave signal based on a preset fast edge pulse period; a delay line module configured to apply two different delay times to the controllable square wave signal to obtain two delay signals; a combination logic module configured to perform logic operation on the two delay signals to obtain a fast edge pulse signal; a signal shaping module configured to shape the fast edge pulse signal; a signal synthesis and distribution module for synthesizing the shaped signal; a collection and spectrum analysis module configured to obtain an input reference spectrum and an output spectrum; and an ADC data processing module configured to calculate a frequency response function based on the input reference spectrum and the output spectrum, and a host computer configured to obtain a fault condition of a measured system according to the spectrum result. The application effectively reduces the system hardware cost and implementation complexity while improving the frequency scanning capability.

The application relates to the technical field of motor control, and particularly discloses a resolver software decoding method based on an MCU and a device thereof, the decoding device comprising an MCU, an amplification circuit, two conditioning circuits, two multipliers and two low-pass filters; the MCU comprises a timer, a DAC module, a first ADC module and a second ADC module; the DAC module is connected with a primary side winding of a resolver, a first input end of a first multiplier and a first input end of a second multiplier through the amplification circuit; two secondary side windings of the resolver are connected with a second input end of the first multiplier and a second input end of the second multiplier; and an output end of the first multiplier is connected with the first ADC module through a first filter and a first conditioning circuit, and an output end of the second multiplier is connected with the second ADC module through a second filter and a second conditioning circuit. The application can reduce the cost of a resolver shaft angle calculation system and realize high-precision shaft angle calculation.

The present application relates to a kind of three-stage pipeline type successive approximation analog-digital converter, SAR ADC circuit includes first stage sub-SAR ADC module, second stage sub-SAR ADC module and third stage sub-SAR ADC module connected in turn, shared residual operational amplifier includes two independent first differential input stage and second differential input stage;First stage sub-SAR ADC module, for sampling quantization input signal output first residual signal;First differential input stage, for amplifying first residual signal;Second stage sub-SAR ADC module, for sampling quantization amplified first residual signal after first differential input stage, output second residual signal;Second differential input stage, for amplifying second residual signal;Third stage sub-SAR ADC module, for sampling quantization amplified second residual signal after second differential input stage and output.Two-stage sub-SAR ADC module share a shared residual operational amplifier and amplify, effectively save the area of analog-digital converter overall circuit and reduce power consumption.

This invention provides a three-dimensional quantitative analysis method, medium, and system for millimeter-level cracks on infrastructure surfaces, belonging to the field of crack detection technology. The invention identifies crack contours using a minimum cut-maximum flow-based crack-background segmentation algorithm, extracts the crack skeleton centerline using an ant colony optimization algorithm, and performs network connectivity and branch identification of the crack network through a crack topology connectivity optimization model. Cracks are classified according to their direction angle and length characteristics. Inclined laser scanning lines are set in the crack cross-section area to form cross-scan coverage. Weak echo signals are amplified by adaptive gain and supplemented with neighborhood depth interpolation to complete missing depth data. This accurately calculates the cross-sectional crack width and depth, solving the technical problem of inaccurate crack depth measurement due to missing bottom point cloud data caused by laser occlusion in deep and narrow cracks.

The application discloses an industrial sensor data anomaly detection and repair method based on an attention mechanism, belongs to the technical field of industrial data processing, and carries out preprocessing on original data to obtain standardized data segments, a training set and a test set; end-to-end collaborative training is carried out on the training set to obtain a dynamic space-time attention detection model and a multi-scale self-attention repair model; the standardized data segments are input into the dynamic space-time attention detection model for anomaly detection to obtain abnormal related information; the standardized data segments and the abnormal related information are input into the multi-scale self-attention repair model for data repair to obtain repaired data; secondary verification and parameter fine-tuning operations are performed on the repaired data to obtain compliant repair data; and the POT algorithm is used to dynamically adjust an abnormality judgment threshold value, and normal industrial sensor data or compliant repair data is output. By using the above method, accurate detection and compliant repair of industrial sensor data anomalies are realized.

The invention discloses a near-infrared signal quality optimization method, system and device and a storage medium, and the method comprises the steps: collecting a near-infrared light intensity signal, carrying out the filtering processing of the near-infrared light intensity signal, extracting a heart rate frequency band signal, calculating the power spectrum energy of the heart rate frequency band signal, judging whether the power spectrum energy is lower than a preset threshold value or not, and if the power spectrum energy is lower than the preset threshold value, carrying out the filtering processing of the near-infrared light intensity signal; if yes, increasing the sampling rate of the near-infrared light intensity signal, and if not, calculating a dynamic amplitude ratio and a correlation coefficient between an oxygen-containing hemoglobin signal and a deoxidized hemoglobin signal in the near-infrared light intensity signal based on a preset first algorithm; the near-infrared light intensity signal with the dynamic amplitude ratio and the correlation coefficient not meeting the preset range value is corrected to obtain a corrected signal, quality evaluation is performed on the corrected signal, and the corrected signal with the quality evaluation score exceeding the preset score value is output, so that automatic processing and quality analysis are performed on the collected signal, manual intervention is reduced, and the detection efficiency is improved. Real-time feedback is improved, and invalid experiments are avoided.

An image acquisition device based on terahertz signals, an image generation system and method, the image acquisition device comprising: a swing reflection plate adapted to reflect terahertz wave signals emitted in a sampling area including a target to be measured; a lens assembly adapted to focus the terahertz wave signals reflected by the swing reflection plate; a radiometer array including n rows of radiometers installed on the focal plane of the lens assembly, the n rows of radiometers being configured to sample the terahertz wave signals emitted by the target to be measured and reflected by the swing reflection plate at the same height, wherein n>=2, and the terahertz wave image has both high resolution and contrast.

The present application relates to a kind of motor train unit net pressure zero-crossing rapid detection system and method, system includes the net pressure acquisition circuit of acquisition analog net pressure signal, the analog-digital converter of the analog net pressure signal into digital net pressure signal, FPGA;FPGA is sampled to the digital net pressure signal sent by analog-digital converter, and the sampled digital net pressure signal is carried out sliding window filtering and FIR filtering and obtains the sine wave net pressure signal of frequency 50Hz;Zero-crossing detection module is set in FPGA, zero-crossing detection module is set to: when reaching the preset threshold T of continuous net pressure data greater than zero, mark Flag=1;When reaching the preset threshold T of continuous net pressure data less than zero, mark Flag=0;When the continuous preset threshold N packet net pressure data less than zero, then output zero-crossing interrupt signal.The present application can be real-time fast and accurately detect motor train unit net pressure zero-crossing, so that motor train unit still maintains normal operation in the case where net pressure distortion is serious.

The utility model discloses a high-speed multi-phase clock generation circuit and electronic equipment, and belongs to the technical field of wired optical communication. The high-speed multi-phase clock generation circuit comprises a ring oscillator group 1, a phase insertion resistor group 2 and a bridging resistor group 3, the ring oscillator group 1 comprises four ring oscillators 10, and each ring oscillator 10 comprises four phase inverters 100 which are connected end to end and form a closed loop; in a ring oscillator 10, the output end of each phase inverter 100 is connected with a phase node, and the phase difference of clocks output by two adjacent phase nodes is 90 degrees; four phase insertion resistors 20 which are connected in series are inserted between the input end and the output end of each phase inverter 100, a phase node between every two adjacent phase insertion resistors 20 generates a clock phase, and the difference between the clock phases generated by every two adjacent phase nodes is 22.5 degrees; and phase nodes for outputting corresponding clock phases in the four ring oscillators 10 are connected and coupled through a bridging resistor 30.

The application discloses a circuit automatic design method based on large language model distillation and self-evolution, comprising the following steps: inputting a circuit type and a design target specification; constructing a reference netlist set, and performing initial supervision fine-tuning on a generation strategy model; generating a candidate netlist and performing simulation, calculating a target accuracy and a standard mark of each candidate netlist; for the candidate netlist that does not meet the standard, calculating a constraint violation degree and attempting automatic repair; for the candidate netlist that meets the standard, calculating a comprehensive performance vector and performing a Pareto sorting and diversity screening; performing preference optimization training by using good and bad sample pairs; performing security strategy optimization; and outputting a finally generated netlist. The application can reduce invalid candidate netlists and invalid simulation, improve the output rate of candidate netlists meeting the design target specification, and enhance the training stability and the circuit automatic design effect under multi-index constraints.

The utility model discloses a high-speed data acquisition and storage device, and relates to the technical field of data acquisition and storage, and the device comprises a CML interface which is used for obtaining video data; the serial-to-parallel conversion circuit is connected with the CML interface, and the serial-to-parallel conversion circuit is used for converting a serial signal input by the CML interface into a parallel signal; the main control chip is connected with the serial-parallel conversion circuit, and the main control chip adopts an XC7Z045 processor with a ZYNQ framework; and the storage unit is connected with the main control chip and is used for storing the video data. According to the invention, the CML interface is adopted to collect video data, and compared with an existing LVDS interface, the CML interface is greatly improved in sampling rate and conversion precision.

A photonic analog-to-digital conversion system and chip based on wavelength multiplexing and optical capture, which utilizes a wavelength division multiplexer to combine N continuous lights of different wavelengths into one light, loads sampled analog signals on the N light paths, makes the analog signals on the light paths of different wavelengths obtain different time delay amounts through a time delay control module, the time delay amount difference of adjacent two light paths is equal, and the time domain discretization of N equal time interval analog signals is realized by using an optical capture module. The present application utilizes passive devices such as a wavelength division multiplexer, greatly reduces the number of active devices such as electro-optic modulators, can effectively reduce the chip complexity and link loss, and thus improves the performance of the photonic analog-to-digital conversion chip. The use of passive devices instead of active devices can effectively reduce the chip power consumption and improve the stability. The chip can increase the number of continuous laser light sources and the channel number of the wavelength division multiplexer / demultiplexer through simple configuration, and can multiply the sampling rate of the photonic analog-to-digital conversion chip, and has the ability of large-scale expansion.

The embodiment of the invention provides a communication method and a related device, which are used for improving time-frequency synchronization performance and / or reducing power consumption of a first communication device. The first communication device comprises a first module for data transmission and a second module for waking up the first module; the method comprises the following steps: receiving a first synchronization signal from a second communication device on a first time domain resource through a second module, wherein the first synchronization signal is used for performing time-frequency synchronization on the first communication device and the second communication device; monitoring a first wake-up signal from a second communication device at a first wake-up signal opportunity through a second module, wherein a time interval between the first time domain resource and a second time domain resource occupied by the first wake-up signal opportunity is not less than a first duration; the time interval between the first time domain resource and a second time domain resource occupied by the first wake-up signal is used for the second module to adjust the sampling rate; and determining whether to wake up the first module according to the first wake-up signal.