44results about How to "Improve real-time processing performance" patented technology

The invention relates to the technical field of anesthesia effect evaluation, in particular to an anesthesia effect evaluation system based on big data, which comprises a physiological data processing module, a patient grouping and analysis module, an anesthesia risk evaluation module and an anesthesia adjustment support module. And collecting physiological indexes and types and doses of anesthetic drugs. According to the method, the real-time processing capacity of physiological data is enhanced by integrating historical health records and operation information, accurate evaluation of the instant effect of anesthetic is provided by deeply analyzing time sequence changes of heart rate and blood pressure, the response of different patient groups to the anesthetic can be recognized through grouping analysis, and the accuracy of the anesthetic is improved. Therefore, a targeted anesthesia scheme is realized, the anesthetic injection time and dosage are analyzed by using the LSTM network, the dynamic prediction of drug response is optimized, a doctor is supported to adjust the anesthetic dosage according to real-time data, the precision and safety of anesthesia management are greatly improved, and the anesthesia side effect risk is reduced.

The invention relates to the technical field of deep learning, in particular to a new energy station illegal behavior intelligent recognition system and method based on machine vision. The method comprises the steps of obtaining new energy station image data, performing preprocessing and data enhancement, and creating a behavior image data set; a CNN-based behavior recognition model is constructed, and training is carried out through the data set; meanwhile, a clustering algorithm based on comparative learning is constructed for judging whether the unknown behavior belongs to a new violation type or not; if a new violation type is found, transmitting a sample of the new violation type to a behavior recognition model for incremental learning, and expanding the number of the sample; and through a knowledge distillation technology, knowledge of the trained CNN behavior recognition model is migrated to a miniaturized student model, and lightweight deployment is realized. According to the method, the illegal behaviors of the new energy station can be automatically identified, new illegal behavior types can be continuously found, the model is dynamically updated, and the adaptive capacity and the real-time performance of the system are improved.

The invention provides a power distribution network coordination evaluation method based on a graph convolutional neural network, and relates to the technical field of power system optimization control, and the method comprises the steps: carrying out the coordination evaluation of a power distribution network containing a high-proportion distributed power supply, so as to obtain a coordination evaluation index of the power distribution network; obtaining power flow data of the power distribution network, and processing the power flow data through the graph convolutional neural network; calculating the index weight of the coordination evaluation index by using an analytic hierarchy process, and adding the index weight as a hidden layer into the graph convolutional neural network to construct an improved graph convolutional neural network; inputting the relation characteristics between the nodes and the topological structure information of the network into an improved graph convolutional neural network model to obtain an evaluation index of the power distribution network; performing coordination evaluation according to the evaluation indexes and the index weights; according to the method, comprehensive and accurate coordination evaluation is carried out on the power distribution network connected with the high-proportion distributed power supply, the evaluation accuracy is improved, the real-time processing capacity is improved, and the coordination state of the power distribution network is comprehensively reflected.

The invention discloses a photoelectric reconnaissance character recognition method, system and device and a medium, and relates to the technical field of optical character recognition, and the method comprises the steps: collecting photoelectric reconnaissance image data to be recognized; photoelectric reconnaissance image data to be recognized are input into the character recognition model, and a fused feature map is obtained through the PP-LCNet backbone network; inputting the fused feature maps into two predictors to respectively obtain a prediction probability map and a prediction threshold map, optimizing the prediction probability map by adopting a knowledge distillation-based mutual learning strategy CML, generating a binary image according to the optimized prediction probability map and prediction threshold map, and further obtaining an image containing a required text region; inputting the image containing the required text area into a text recognition module, outputting a text label matrix, and obtaining a character recognition result of the photoelectric reconnaissance image data to be recognized according to the text label matrix; according to the method, the accuracy of character recognition in a complex environment is improved.

The invention relates to an operation and maintenance processing method and system applied to a solid beverage intelligent production control platform, and belongs to the technical field of equipment operation and maintenance processing. Comprising the steps of obtaining sensor monitoring data and control platform state data; the edge detection module checks the collected system data, diagnoses the operation states of the sensor and the control platform, and gives an alarm when the network performance data is abnormal; integrating the authenticated data into a data packet based on the device code, encrypting the data packet, and uploading the encrypted data packet to a cloud computing center; the cloud computing center predicts the equipment fault state through the received system data; and according to the prediction result of the equipment fault state, generating an operation state score of each piece of processing equipment. According to the invention, through cloud-side cooperation, system data is uploaded after the state of the control platform is detected, and fault prediction of the processing equipment is carried out, so that predictive operation and maintenance of the equipment are realized, and the system stability and the production efficiency are improved.

The invention discloses a vehicle detection method and system based on an unmanned aerial vehicle and an improved YOLO algorithm, and the method comprises the following steps: data acquisition: collecting an aerial image of the unmanned aerial vehicle; a data preprocessing step: carrying out image preprocessing operation on the acquired image data; and a target detection step of pre-building a target detection model based on a YOLO algorithm, performing target detection according to the preprocessed unmanned aerial vehicle aerial image, obtaining a target detection result, and completing vehicle detection and identification. According to the invention, by improving the target detection model, efficient and accurate detection of the vehicle can be realized under the view angle of the unmanned aerial vehicle.

The invention discloses an equipment maintenance management method and device. The method comprises the following steps: collecting vibration data of equipment in different motion states; performing feature extraction and labeling on the vibration data; a training sample is generated according to the extracted feature values and corresponding labeling information, the labeling information is used for indicating equipment states corresponding to the vibration data, and the equipment states comprise a normal state and a specific fault state; training a fault diagnosis model by using the training sample; and performing fault diagnosis on the equipment by using the real-time monitoring data of the equipment and the fault diagnosis model, and determining the current state of the equipment. By means of the scheme, the real-time performance and intelligence of equipment vibration monitoring can be improved, and normal operation of equipment can be better guaranteed.

The invention discloses an intelligent connection method and device for a measurement and control instrument system based on a multi-modal large model, and belongs to the technical field of intelligent control. By utilizing the powerful information processing capability of a multi-modal large model and a man-machine interaction interface which is easy to operate, through a target detection algorithm based on deep learning and the like, positions of various operations such as grabbing, plugging, moving and the like of the multi-modal large model are accurately positioned; the intelligent agent executing action arrangement and the action skill knowledge base help the mechanical arm to complete complex tasks. The multi-mode large model can receive various information such as pictures, texts and voices, process the information and then call proper actions for combination, and control over the mechanical arm is completed. According to the method, the dependence on the environment is effectively reduced, the operation method of the mechanical arm is simplified, and meanwhile the real-time processing capacity is improved.

The invention discloses a decision-making system and method based on a Dores database, and the system comprises a data access layer which is used for accessing multi-source heterogeneous financial data in real time, a Dores storage calculation layer, and achieves the data storage, column-type bucket separation and materialized view pre-calculation based on the Dores database; the visual application layer is used for providing bidirectional interactive financial data visualization and decision support; wherein the data access layer acquires data in real time through a streaming computing engine and transmits the data to the Dores storage computing layer, and the visual application layer performs real-time data interaction with the Dores storage computing layer through a two-way communication protocol. According to the method, the synchronous delay of financial data is shortened from 6 hours to 1 second, second-level real-time integration is realized, the aggregation query response time of billion-level data is reduced to be within 1 second based on a Dores column type storage and materialized view technology, the user interaction response delay is lower than 500 milliseconds through WebSocket two-way communication, and real-time adjustment of parameters and dynamic refreshing of a billboard are supported.

The present invention relates to the field of navigation technology, and proposes a point-line fusion visual SLAM navigation method and system based on image entropy perception, which includes the following steps: preprocessing a part of the acquired image; detecting the environmental anomaly of the image that has not been preprocessed; extracting features from the preprocessed image to obtain point features, determining the number of line features to be extracted according to the environmental anomaly, and using local one-dimensional image entropy for weighted fitting to obtain line features; performing optical flow tracking on the point features and line features; for the tracking of line features, constructing an optical flow tracking model based on the grayscale invariant assumption, and adjusting the line features by selecting endpoint adjustment or line segmentation method according to the environmental perception result; for the tracking result, minimizing the error between the observed value and the predicted value to obtain the optimal pose estimation result. Real-time and accurate navigation and positioning can be achieved in complex environments such as weak texture, exposure, and blurry environments.

The invention relates to a sketch general configuration method and system based on a data processing unit, and belongs to the technical field of data processing. The method comprises the following steps: (1) identifying resource parameters; (2) selecting a process model; (3) resource constraint input; (4) optimizing target configuration; and (5) parameter configuration recommendation. The invention aims to optimize the communication efficiency of the DPU and the host, reasonably allocate hardware resources, and realize flexible process model selection and efficient state operation configuration, thereby achieving the goals of high accuracy and high throughput of network measurement under limited resources.

The embodiment of the invention discloses a communication signal multi-parameter intelligent parallel extraction method. The method comprises the following steps: firstly, acquiring a to-be-measured signal and extracting an I / Q component; processing the I / Q components through a multi-scale feature extraction structure to obtain high-dimensional features; extracting shared features from the high-dimensional features through a plurality of residual block groups, inputting a plurality of parallel task branches, and enhancing the shared features in the branches by using a channel attention mechanism to obtain output features; the output features of all task branches are transformed through a learnable linear layer to form a key matrix and a value matrix, the current task features are transformed through a learnable linear layer to form a query matrix, and the input features are segmented into a plurality of heads by means of a multi-head attention mechanism; each head obtains output of a single head based on zoom dot product self-attention operation, and results of all the heads are integrated to obtain enhanced features of the current task; and determining a communication signal parameter prediction label according to the enhanced features, and realizing multi-parameter efficient parallel extraction.

The embodiment of the invention provides a video processing method and a video processing device. The method comprises the following steps: decomposing a video into multiple frames of video images; dividing each frame of video image in the plurality of frames of video images into a plurality of data units; determining the similarity between the central data unit in the first sliding window in the second frame of video image and each data unit in the second sliding window in the first frame of video image to obtain a first result; determining the similarity between the central data unit in the first sliding window and the surrounding data units in the first sliding window to obtain a second result; and obtaining a characteristic value of each data unit in the first sliding window based on the first result and the second result. According to the method and the device provided by the embodiment of the invention, the calculation amount in the video processing task can be reduced, and the real-time processing capability of the system can be improved.

This invention relates to an orbit recognition method based on EfficientNet, comprising the following steps: constructing a network structure based on the EfficientNetB0 structure; obtaining an orbit recognition model based on an orbit dataset through a network training method; acquiring real-time orbit images as input to the orbit recognition model to obtain the coordinates of the recognized orbit.

The invention relates to the field of non-contact physical sign perception, in particular to a human body vital sign perception method based on VMD + ACMD joint decomposition, which comprises the following steps of data acquisition, distance angle heat map generation, direct current bias removal, expansion DACM phase extraction, signal decomposition, signal reconstruction, feature extraction, time-frequency analysis and the like. According to the invention, the VMD and the ACMD are combined for use, and the frequency resolution capability of the VMD and the adaptive decomposition capability of the ACMD are fully utilized. Through the combined decomposition, the analysis capability of the micro-motion signal is enhanced, the sign extraction precision is remarkably improved, compared with a traditional single decomposition method, the scheme improves the separation precision of the sign signal, the modal aliasing problem is reduced, the higher signal quality is ensured, and by enhancing the main frequency component of the sign signal, the sign extraction precision is improved. Interference of background noise is effectively suppressed, especially in a complex or high-noise environment, high signal-to-noise ratio of signals can be ensured, and accurate extraction of sign signals such as heart rate and respiratory rate can be improved.

The invention relates to a target signal enhancement method based on a generative adversarial network, which belongs to the technical field of signal processing, and comprises the following steps: S1, carrying out adversarial training on a generator and a discriminator of the generative adversarial network by using a training set, and adding an attention mechanism and an audio restoration module in the training process, obtaining a trained target signal enhancement model; s2, using the trained target signal enhancement model to carry out identification, extraction, denoising and restoration processing on the complex signal sample to obtain a target signal; and S3, carrying out model pruning and retraining on the trained target signal enhancement model so as to optimize the target signal enhancement model. According to the method, the precision and effect of target signal processing in complex signals are improved, and more real-time and reliable signal data processing capability is provided for complex signal scene application.

The invention discloses an electric power communication system optimization method based on big data analysis, and the method comprises the steps: collecting operation data of an electric power communication system, carrying out the format conversion, and carrying out the fusion; identifying system operation characteristics based on the fused data, and evaluating the system to obtain an evaluation result; and carrying out optimization scheduling on the network resources according to the system operation characteristics and the evaluation result. According to the method, comprehensive optimization, including real-time data processing, operation feature recognition, intelligent resource scheduling and dynamic path optimization, of the electric power communication system is achieved by integrating the big data analysis technology, the operation efficiency, stability, safety and reliability of the system are remarkably improved, and efficient data support and decision basis are provided for intelligent power grid management.

The invention discloses a non-vision-field stable imaging method and a non-vision-field stable imaging system. The non-vision-field stable imaging method comprises the following steps: acquiring transformation from a previous frame to a current frame of all frames in a video; accumulating and transforming to obtain an image track; smoothing the trajectory by using an average window; generating a new transformation matrix from the previous frame to the current frame; and applying the new transformation matrix to each frame of the video, and synthesizing the processed frame sequence into a final video. The system comprises a camera module, a processor module and a helmet module. According to the method, target identification is carried out in the vision field, hidden targets can be identified under the non-vision field condition, the exposure risk is remarkably reduced, the investigation efficiency and safety are improved, and the reliability of performance under different scenes can be ensured.

The invention discloses a digital compression method for a power monitoring image, and the method comprises the following steps: image preprocessing: carrying out the preprocessing of an input power monitoring image, including denoising and contrast enhancement, and obtaining a preprocessed power monitoring image; feature extraction: extracting key features of the preprocessed power monitoring image by using a deep learning model; feature compression: compressing the key features extracted in the step S2 to obtain compressed power monitoring image features; feature reconstruction: decompressing the compressed power monitoring image, and reconstructing the image by using the decompressed power monitoring image features; and post-processing: carrying out post-processing on the reconstructed image, wherein the post-processing comprises removing artifacts and enhancing details. According to the method, efficient compression and high-quality reconstruction of the power monitoring image are realized through deep learning and image processing technologies.

The application relates to the technical field of map construction, and provides a SLAM map optimization method and device based on local characteristic values. The method comprises the following steps: acquiring sensor data, and constructing a state matrix of a SLAM system based on the sensor data; performing eigenvalue decomposition on the state matrix, acquiring each eigenvalue, and calculating the change rate of each eigenvalue; determining an eigenvalue with a change rate exceeding a preset threshold as a to-be-optimized eigenvalue, selecting the first M eigenvalues with the largest change rates from the to-be-optimized eigenvalues, and forming a target eigenvalue set; updating a local block corresponding to the first M eigenvalues in the state matrix according to the target eigenvalue set, obtaining an updated state matrix; and optimizing and outputting pose data and map data based on the updated state matrix. By performing local recalculation on a small number of eigenvalues with significant changes in the state matrix, the calculation complexity is significantly reduced while ensuring the positioning and mapping accuracy, and the real-time requirement of high-frequency positioning scenes such as intelligent driving is met.

The utility model provides a high-speed signal generation and acquisition module. The high-speed signal generation and acquisition module comprises a power supply module, an FPGA module, a DSP module, an AD / DA module, a clock module, a communication module, a debugging interface and an FMC interface, the FPGA module, the DSP module, the AD / DA module and the clock module are in communication connection; the communication module, the FMC interface and the debugging interface are in external communication connection; the AD / DA module comprises two paths of AD acquisition chips and two paths of DA acquisition chips; the FPGA module is composed of three domestic FPGA chips which are respectively an FPGA (field programmable gate array) A1, an FPGA A2 and an FPGA A3, and the FPGA A1 is respectively connected with the FPGA A2 and the FPGA A3. According to the utility model, in a signal processing part, a localized FPGA and DSP architecture is adopted, and the operational capability of the DSP is applied, so that the real-time processing performance of the module on data is improved.

The invention discloses a broadband radar signal reconnaissance and parameter estimation system and method and a storage medium. The system comprises a frequency spectrum reconnaissance module used for performing Fourier transform on reconnaissance channel input data and detecting a first carrier frequency and a bandwidth of a signal in a frequency domain; the signal processing module is used for carrying out self-adaptive frequency conversion and filtering on the input data according to the parameters; the pulse detection module is used for performing pulse detection on the filtering signal and finely estimating a second or third set of carrier frequency and bandwidth parameters of the frequency modulation or phase modulation type signal; the direction-finding data FFT module is used for performing FFT conversion on the lateral channel data to extract phase information; and the packaging module is used for packaging the pulse detection result and the phase information into pulse description words and outputting the pulse description words. According to the invention, the integrated processing of the broadband radar signal from rapid reconnaissance, precise processing, modulation identification to direction finding information fusion is realized, and the real-time processing capability of the system, the throughput of the broadband signal and the parameter estimation precision are obviously improved.

The invention discloses a vehicle identification and path tracking method in a video based on big data analysis. The method comprises the following steps: collecting video data through a camera; performing image enhancement, noise elimination and feature extraction on the video data; identifying the vehicle by using an image analysis technology; tracking a vehicle path in real time by adopting an algorithm; predicting a vehicle path using an intelligent algorithm; a large amount of video data is stored, processed and analyzed by using a big data technology; multiple vehicles are identified and tracked at the same time; real-time processing is achieved through an optimization algorithm and hardware acceleration, and delay is reduced through edge calculation. The method combines various technologies, efficiently and accurately identifies and tracks the vehicle, and is suitable for a complex traffic environment. Through distributed computing and parallel processing technologies, a large amount of video data can be processed, and real-time requirements can be quickly responded. The method can manage a plurality of vehicles at the same time, improves the image quality, enhances the recognition and tracking accuracy, and is suitable for intelligent traffic management and automatic driving application.

The present invention belongs to the technical field of separation and identification of crops and weeds, and specifically is a method and system for separation and identification of crops and weeds based on image processing and machine vision. The present invention makes full use of advanced image processing technology, deep learning, machine vision algorithms and intelligent post-processing mechanisms, and describes in detail the complete process from image acquisition, preprocessing, segmentation, feature extraction and identification to intelligent post-processing and comprehensive management decision support. This method has the advantages of high-precision identification technology, real-time processing capability, adaptability to complex environments, self-learning and adaptive mechanisms, and integrated management system integration, thereby achieving efficient and accurate identification of crops and weeds, and providing strong technical support for precision agriculture.

The invention discloses a processing method for industrial internet identification analysis data, and the method specifically comprises the following steps: 1, carrying out the data processing near a device or a sensor, reducing the delay and bandwidth consumption, analyzing the data in real time, generating feedback, achieving the real-time decision through edge calculation, reducing the dependence on a cloud end, and improving the response speed; 2, dynamically adjusting a data stream processing strategy according to a real-time analysis result, preferentially processing important data, reducing unnecessary data processing burden, processing the data by adopting an event-driven mode, and triggering corresponding processing logic based on a specific event; the data transmission delay is reduced, the response time is shortened, the real-time processing capability is improved, unnecessary calculation is reduced, and the resource utilization rate is improved; the data security is enhanced by adopting a block chain technology, and the non-tampering and traceability of the data are ensured; the decision quality is improved through self-adaptive data stream processing, and the strategy is dynamically adjusted through real-time data.

The present invention relates to the field of medical care information technology of artificial intelligence, and specifically discloses a non-contact vital sign detection method based on deep learning, including frame processing of visible light video of human face; obtaining color face image and frame difference image; extracting color face image and frame difference image by multi-channel input, and forming stacked feature image; extracting spatial features; weighting spatial feature image; performing channel transformation fusion processing on weighted spatial feature image; transforming the fusion feature image dimension into a feature image vector; storing the feature image vector in sequence; and capturing the heart rate timing characteristics, respiratory rate timing characteristics, and blood oxygen saturation timing characteristics in rPPG signals respectively. The present invention greatly reduces parameters, saves the amount of calculation parameters and memory consumption, and has the technical characteristics of relatively simple analysis process, easy implementation, and fast calculation response speed.

The invention discloses a memory allocation method and system, a medium and a product, which are applied to the technical field of storage, and the method comprises the following steps: monitoring a memory access condition of a running task to obtain memory space release information of the running task; determining a memory space physical address to be addressed based on the memory space release information; addressing the memory space physical address to be addressed to obtain a continuous addressed address; and allocating the memory space for the memory space application instruction based on the continuous addressing address and the unaddressed memory space physical address. In this way, the memory utilization rate can be increased, and therefore the real-time task processing performance is improved.

The present application aims at the problems of poor real-time performance, easy interference of complex environment and high misjudgment rate in the existing loading and unloading operation of truck lifting anomaly detection, and provides a truck lifting detection method and system based on detection positioning and KCF tracking combination, belonging to the technical field of port intelligent monitoring and computer vision detection. The method comprises the following steps: real-time acquisition of port operation area video stream and extraction of key frame image; using a target detection model to locate the container hole as a first tracking area; calculating the truck pallet position as a second tracking area by a space correlation derivation model; continuously tracking the motion trajectory of the two areas by using KCF algorithm; three-dimensional judgment based on distance, angle and distance change rate to determine the separation state of truck and container; triggering alarm in abnormal situation. Through multi-dimensional feature judgment, the stability and accuracy of container separation state recognition in complex environment are improved, which is suitable for safety monitoring of automatic container loading and unloading scene in port, wharf and the like.

The invention discloses a real-time object static detection method and system based on image fuzziness. The real-time object static detection method comprises the steps of converting a current frame color image in a video stream into a gray level image and updating a gray level image sequence with a fixed length in real time; calculating the ambiguity of the grayscale image and updating the ambiguity sequence in real time; updating the average image in real time by using an incremental updating method and calculating the ambiguity of the average image; and taking the ratio of the ambiguity of the average image to the maximum ambiguity in the ambiguity sequence as a relative static coefficient, comparing the relative static coefficient with a set threshold value, and judging the motion state of the scene. According to the method, the relative static coefficient is obtained by fusing continuous multi-frame information, so that the method is real-time, accurate and good in stability, and interference of noise and illumination variation can be effectively resisted.

The invention relates to the technical field of speech recognition, in particular to a speech recognition method based on multi-scale residual convolution and Longform, which comprises the following steps: S1, preprocessing a speech signal acquired in advance to obtain a corresponding acoustic signal; s2, inputting the acoustic signal into a trained speech recognition model to obtain a speech text corresponding to the speech signal; wherein the trained speech recognition model is obtained by training a pre-created speech recognition model by adopting a training data set; the training data set comprises a plurality of segments of first acoustic signals and pre-labeled voice texts in one-to-one correspondence with the segments of first acoustic signals; the multiple segments of first acoustic signals are obtained by preprocessing multiple segments of voice signals which are obtained in advance and used for training; the speech recognition model is constructed by a multi-scale residual error convolutional neural network and a longform model.