17 results about "Preprocessing algorithm" patented technology

This invention provides an intelligent analysis system for photoelectric detection spectral data based on deep learning, belonging to the fields of photoelectric detection, spectral data analysis, and deep learning application technology. The system includes core units such as a spectral data acquisition unit, a preprocessing unit, a feature extraction unit, and a deep learning analysis unit, as well as auxiliary units for anomaly detection and data storage, forming a closed-loop architecture of acquisition-processing-analysis-output-optimization. It improves data quality through step-by-step preprocessing algorithms, reduces computational load by adaptively selecting key features, deeply mines spectral patterns using a model that integrates convolutional neural networks and attention mechanisms, achieves rapid scenario adaptation through transfer learning, and dynamically optimizes model parameters based on a feedback mechanism. The system solves the problems of low accuracy, poor adaptability, and long training cycles in traditional spectral analysis, achieving accurate, efficient, and intelligent spectral data analysis, and is suitable for various photoelectric detection scenarios such as qualitative and quantitative analysis of substances and environmental monitoring.

This invention discloses an energy-saving motor speed data acquisition method and system based on cloud computing, belonging to the field of data acquisition technology. The method includes: acquiring factory production plans, personnel shift schedules, and historical motor operation data; generating production load level information based on the factory production plans, personnel shift schedules, and historical motor operation data; encoding the production load level information to obtain rhythm commands; and sending the rhythm commands to a local data acquisition terminal, which adjusts data acquisition parameters according to the rhythm commands and performs motor speed data acquisition processing based on the data acquisition parameters to obtain motor speed data. The data acquisition parameters include operating frequency, data sampling rate, and preprocessing algorithm complexity. This invention can combine production load level information with rhythm commands to adjust data acquisition parameters to achieve motor speed data acquisition, reducing the power consumption and cost of acquisition equipment.

The present application relates to a kind of PSO-based forklift low voltage management optimization method, comprising the following steps: data acquisition and preprocessing, algorithm model construction and initialization, parallel optimization calculation and automatic optimization, if judging optimization result is effective, then directly output, then parameter deployment is carried out, if it is invalid, then the weight of multi-objective fitness function is dynamically adjusted, and return parallel optimization stage.The PSO-based forklift low voltage management optimization method, by the ternary mapping model constructed accurately quantifies the coupling loss relationship of working condition, the quantitative relationship model of exponential temperature correction factor is integrated to realize the accurate characterization of temperature nonlinear characteristics, multi-objective function realizes demand balance, WOA-PSO efficiently optimizes core parameter, dynamic temperature compensation improves adaptability, engineering determination and targeted weight adjustment form full working condition closed loop optimization, battery safety and operation efficiency are considered, solve the forklift battery working condition coupling and temperature nonlinear characterization deficiency, the problem of single-objective optimization losing one and gaining another.

This invention discloses an airborne intelligent information processing platform based on MPSOC, comprising a program loading module, a clock fan-out module, a memory cache module, an intelligent information processing module, and a VPX interface module. Upon power-up, the program loading module loads the signal processing program. The clock fan-out module uses an intermediate frequency input clock and two onboard differential crystal oscillators as references to output four 50MHz clock signals, three 125MHz clock signals, and twelve 156.25MHz clock signals to the intelligent information processing module. The memory cache module stores the operating system and data cache. After receiving data from the preceding stage, the intelligent information processing module performs pulse compression processing and transmits the processed data to DSP1 via the SRIO protocol to perform functions such as CFAR detection. Then, it transmits the data to DSP2 via Hyperlink to complete image preprocessing algorithms. Finally, it transmits the data back to the intelligent information processing module via the SRIO protocol for intelligent algorithm processing. The calculation results are then transmitted to the servo unit and the telemetry and control system via the VPX interface module.

This invention relates to the field of heart rate variability (HRV) characteristic analysis technology, and discloses a physical education classroom teaching load monitoring system based on HRV. The system includes: guiding users to complete combined movements; extracting four temporal feature points to fit the device time deviation curve; analyzing four-dimensional data of photoplethysmography (PPG) pulse waves to distinguish wearing status; and combining dynamic stable baselines and disease-specific parameters to complete multimodal data acquisition. It employs a temporal feature-based exercise state classification method to divide exercise states, adaptively switching preprocessing algorithms and parameters, identifying preset individual anaerobic thresholds, and introducing HRV feature interpretation logic for smooth transition state switching. During rest periods, it extracts dynamic features of the RMSSD recovery curve, assesses the degree of recovery based on the user's personal historical recovery model, and groups classes based on current recovery, historical physical fitness, and cumulative load. It also employs a three-level directional voice warning mechanism to generate a multi-dimensional teaching analysis report after class, including individual performance and session effectiveness.

The application relates to the technical field of task compilation, and discloses a task compilation optimization method, device, equipment and medium: all preset tasks in a task demand library are traversed, and feature parameters are extracted; according to a preprocessing algorithm, task collection data corresponding to the same preset task is simplified to obtain a simplified data set corresponding to each preset task; according to an instruction module corresponding to the feature parameters, all the simplified data in the simplified data set is subjected to data compilation to obtain data compilation information corresponding to each preset task; and according to a preset mapping model and real-time state data, the data compilation information is subjected to optimization processing to obtain task compilation optimization information corresponding to each preset task. The application is applied to a financial scene and a medical scene. Different tasks are respectively optimized in a targeted manner, the flexibility and efficiency of optimization are improved, and the efficiency and precision of compilation are improved.

Embodiments of the present application provide a material constitutive model construction and finite element integration method, device, computer equipment and medium, relating to the technical field of computing materials, wherein the method comprises the following steps: obtaining experimental data of a target material in a wide strain rate and a wide temperature range, processing the experimental data using an automatic preprocessing algorithm, and generating a standard data set; selecting and outputting the AI algorithm with the highest fitting degree, and outputting the preprocessing parameters and training parameters corresponding to the AI algorithm; extracting strain, strain rate and temperature from the standard data set and reconstructing based on material mechanics constraints, outputting the trained material constitutive model through the reconstructed feature matrix and the constraint term reflecting the nonlinear mechanical behavior of the material; and converting the data format output by the trained material constitutive model into the input data format of the finite element software through a standardized interface. Through the constructed material constitutive model, the prediction accuracy of the constitutive model under wide working conditions is improved.

The present application belongs to the technical field of computer vision and deep learning, and discloses a common fundus disease image multi-label classification method, which comprises the following steps: first, an adaptive preprocessing algorithm based on edge detection and contour recognition is used to position, crop and translate the original image to compensate for the extraction of the standardized fundus target area; then, the preprocessed image is input into the constructed ConvNeXt-Tiny hybrid architecture model, and the deep multi-dimensional pathological features are extracted through global average pooling and high proportion Dropout layer; finally, combined with the class-specific dynamic threshold selected by the optimal F1-score in the training stage, multiple disease labels are determined in parallel and the structured classification results are output. In the training stage, the Mixup data enhancement strategy is used to improve the generalization ability of the model in the complex multi-disease concurrent scene, and the adaptive threshold allocation mechanism is used to reduce the judgment deviation caused by class imbalance, which significantly improves the accuracy of fundus disease multi-label recognition.

This invention relates to the fields of computer vision and edge computing technology, and discloses a multi-computing-power collaborative video structured analysis method. The method includes: acquiring multiple raw video streams and synchronously decoding them; using gradient analysis to assess the quality of the decoded raw video streams and selecting a combination of preprocessing algorithms based on quality feature vectors; establishing a data transmission channel using an asynchronous communication protocol and a multi-level caching system; performing content complexity analysis and obtaining scheduling decisions through model matching based on the analysis; distributing the model and performing pipelined inference on the deployed model; fusing the pipelined inference results and verifying the consistency of the fused results; and optimizing system configuration parameters based on the video structured analysis results to achieve video structured analysis. This invention improves the real-time performance, accuracy, and system stability of video analysis.

This invention discloses a method for constructing an intangible cultural heritage (ICH) causal graph based on a large language model. The method involves acquiring ICH information data through a data acquisition device; preprocessing the ICH dataset using a preset data preprocessing algorithm; structuring the standardized ICH dataset using a preset event structuring algorithm; identifying entities in the structured ICH corpus using a preset entity recognition algorithm; extracting events from the structured ICH corpus using a preset event extraction algorithm; and mapping the obtained event entity data and the extracted events to a preset causal graph model to obtain an updatable ICH causal graph. Compared with existing technologies, this invention significantly improves the accuracy and completeness of ICH knowledge extraction; achieves standardized modeling of causal relationships, improving the semantic consistency and reliability of the graph; and possesses dynamic adaptation capabilities to meet the continuous growth and updating needs of ICH data.

This technology provides a preprocessing method that differs from the conventional approach of employing a single preprocessing technique. [Solution] The disclosed method is a computer implementation method performed by a computer. The computer implementation method includes numerically analyzing eddy currents by solving the discretized governing equations using a preprocessing algorithm with the finite element method. The preprocessing in the preprocessing algorithm is a composite preprocessing that performs different types of preprocessing on multiple parts of the governing equations.

The application discloses a sperm activity detection method, device, equipment and storage medium. The method comprises the following steps: acquiring sperm video data collected by a video collection device; wherein the sperm video data comprises at least two video frame images; for each video frame image, at least two levels of preprocessing algorithms are adopted to sequentially perform preprocessing operations on the video frame image, and reference preprocessing image sets and target preprocessing images are obtained respectively; based on the reference preprocessing image sets and the target preprocessing images, target trajectory data corresponding to the sperm video data is determined; and based on the target trajectory data, sperm activity corresponding to the sperm video data is determined. The embodiment of the application takes into account the algorithm processing speed and the algorithm accuracy.

This invention discloses a data analysis and prediction system based on deep learning, belonging to the fields of data analysis and artificial intelligence technology. The system includes a data acquisition module, a data preprocessing module, a feature extraction module, a deep learning prediction module, a result output module, a model optimization module, a data storage module, an anomaly monitoring module, a human-computer interaction module, and a permission management module. Through a unique multi-source data fusion preprocessing algorithm, an improved deep learning prediction model, and a dynamic optimization mechanism, it achieves accurate analysis and efficient prediction of multi-domain and multi-type data. Its prediction accuracy is improved by more than 15% compared to existing technologies, and the response latency is controlled within 500ms. It also possesses strong environmental adaptability and data security, and can be widely applied to data analysis and prediction scenarios in finance, industry, healthcare, and other fields.

The present application relates to the field of ship machinery bearing life prediction, and particularly relates to a ship machinery bearing anti-noise type life prediction method fusing a cross-modal attention mechanism. By constructing a mechanical bearing life prediction experiment platform, real-time simulation and collection of ship machinery bearing operation data are carried out, and multi-modal data are obtained covering normal operation, slight wear and severe failure conditions; a data preprocessing algorithm based on wavelet transform and dynamic threshold denoising is studied to suppress high-frequency noise in the marine environment; the collected multi-modal data are optimized and processed; and the processed multi-modal data are used for anti-noise life prediction of the ship machinery bearing.

This invention relates to the field of edge computing, and discloses a multimodal edge computing method, device, and storage medium. The method includes: acquiring multimodal data at the edge and reading the current resource occupancy rate; setting an attention head for an attention mechanism based on the resource occupancy rate to obtain an adjusted attention mechanism; performing classification preprocessing on the multimodal data based on a preprocessing algorithm to generate multimodal preprocessed data; performing feature extraction processing on the multimodal preprocessed data according to a preset lightweight neural network model to generate multimodal extraction vectors; calculating association weights on the multimodal extraction vectors according to the adjusted attention mechanism to generate multimodal association vectors; performing weighted mapping processing on the multimodal extraction vectors based on the multimodal association vectors to generate fusion vectors; and performing inference analysis processing on the fusion vectors to generate edge inference results. In this embodiment of the invention, a multimodal processing mechanism is introduced into IoT computing and adapted to the computing resources of mobile terminals.

This application relates to the field of computer technology. It provides a method and system for secure mining of large model training data based on a TEE (Train the Execution Environment). The method includes: securely partitioning the original training dataset into logical data block units; securely encapsulating the logical data block units, a pre-built preprocessing algorithm library, and a user-configured task descriptor to obtain a secure input packet; parsing the secure input packet within the TEE using a pre-loaded access pattern planner to generate a secure execution plan; performing obfuscation calculations on the secure input packet according to the secure execution plan using an integrated obfuscation execution engine to obtain intermediate processing data; auditing the access patterns of the intermediate processing data to generate verification results; and when the verification result is successful, performing enclave encryption on the intermediate processing data to obtain encrypted training data. This achieves the technical effects of improving the access pattern obfuscation strength in the preprocessing stage, enhancing the security adaptability of data mining scenarios, and reducing the risk of side-channel leakage.