219 results about "Feature extraction algorithm" patented technology

The invention discloses a VLM model intelligent decision-making-based driving method and device and a storage medium, and relates to the technical field of data processing, and the method comprises the steps: extracting key visual features from continuous multi-frame driving scene images based on a preset visual feature extraction algorithm; inputting a user instruction and a time sequence visual Token corresponding to the key visual features into a preset VLM model for multi-modal alignment, and generating a target planning Token; inputting the target planning Token and the time sequence vision Token into a preset trajectory generation model to obtain a predicted trajectory; and converting the predicted trajectory into a control signal, and controlling the mobile device to complete a moving action based on the control signal. The problem of modal difference between a semantic space and an action space is solved.

The invention discloses a motion recognition method and system based on a redox photoelectric memristor, a terminal and a storage medium, and the method comprises the steps: selecting classical motions based on a human body motion data set, extracting time sequence data, and coding the time sequence data into an optical pulse sequence; constructing a reservoir array composed of a plurality of photoelectric memristors, and expanding an optical pulse sequence signal into a high-dimensional state vector; constructing a supervised training model, solving a weight matrix, and constructing a memristive cross array; and outputting an action classification result through simulation domain operation based on the output current multi-path light current signals in combination with the memristor cross array. According to the method, the motion features can be directly fed into a rear-end classification network for action recognition without depending on a complex digital feature extraction algorithm, so that the transmission and processing overhead of redundant data is fundamentally eliminated; a high-efficiency, low-delay and high-robustness hardware solution is provided for real-time and anti-noise motion recognition in scenes such as intelligent monitoring and man-machine interaction.

The invention discloses a bridge safety early warning method and system, and the method comprises the steps: collecting bridge structure response and environment parameter data, and carrying out the preprocessing of the data, and obtaining a distributed multi-source sensing information matrix; constructing a distributed neuron cell automaton network, interacting state information through unit local communication, operating a lightweight feature extraction algorithm, and generating a multi-dimensional feature vector; carrying out distributed modeling by utilizing a pre-trained distributed LSTM-Transform hybrid model, and carrying out parallel calculation on a local prediction result by each unit to obtain a global health state evaluation result; a local gradient change anomaly detection mechanism is established, data anomaly is identified by comparing prediction results of adjacent units, and a potential risk area is positioned by combining a related algorithm; and constructing a hierarchical early warning threshold system, activating a corresponding early warning response by means of local decision logic, executing an alarm operation, generating recovery guidance information and monitoring uploaded data. According to the invention, two types of network models are combined, so that the system reliability and early warning accuracy can be remarkably improved.

An apparatus an method for generating three-dimensional representations of serial sections, wherein the method includes creating a plurality of slide images using a slide scanner, the slide scanner comprising an image sensor and a stage, receiving, using at least a processor, the plurality of slide images from the slide scanner, extracting, using the at least a processor, a plurality of features from the plurality of serial slide section images using a feature extraction algorithm, registering, using the at least a processor, the plurality of serial slide section images as a function of the plurality of feature, generating, using the at least a processor, a three-dimensional (3D) stack view, wherein the 3D stack view comprises the plurality of serial slide section images, and displaying the 3D stack view through a display device.

The invention relates to the technical field of nutritious food configuration, and discloses a system and a method for configuring personalized nutritious food, and the method for configuring personalized nutritious food comprises the following steps: collecting spectral characteristic data and chemical component data of a food material sample by adopting a hyperspectral imaging and mass spectrum combined technology, and obtaining holographic chemical fingerprint information; processing the collected spectral data by using a multi-scale feature extraction algorithm, extracting multi-level feature information of the food materials, and carrying out standardization and dimension reduction processing on the feature data; according to the method, high-precision prediction of the nutrient content of the food material is realized through combination of hyperspectral imaging and mass spectrometry technologies and a deep learning model, so that the nutrient content prediction accuracy and the configuration scheme accuracy are improved, the nutrient balance achievement rate is also improved, and meanwhile, the limitation of a static database is broken through; real-time nutritional ingredient analysis of any food material is achieved, the method can adapt to individual differences and freshness changes of the food materials, and accurate configuration based on actual food material characteristics is provided.

The invention relates to an industrial supply chain big data inventory dynamic optimization method, and belongs to the field of industrial supply chain management. Comprising the steps of supply chain multi-dimensional data acquisition and preprocessing, dynamic inventory demand prediction based on a time sequence interactive feature extraction (SIFM) algorithm, inventory safety line intelligent regulation and control in combination with credit and response factors, and real-time dynamic inventory adjustment and optimization based on an inventory circulation active early warning and adjustment and optimization algorithm (PIAW). And continuous adaptive optimization based on incremental distributed feedback learning is carried out. According to the method, the inventory transfer and replenishment scheme can be automatically decided under the emergencies such as order sudden increase or logistics abnormity, and the inventory distribution model is dynamically optimized. Through multi-source data real-time integration and intelligent algorithm cooperation, optimal distribution and efficient and stable operation of supply chain inventory are realized, and the supply chain response speed and the inventory turnover efficiency are greatly improved.

The invention belongs to the technical field of equipment early warning, and particularly relates to a multi-source information fusion real-time monitoring and data analysis method and system, and the method comprises the steps: achieving the automatic recognition and data collection, remote assistance and support, and intelligent path planning and navigation functions in the inspection process of a chemical device, so as to standardize the inspection process; the inspection efficiency and the result accuracy are improved; establishing a chemical device image recognition model, collecting and preprocessing chemical device image data, performing model training by using a convolutional neural network deep learning model, and deploying the model to actual equipment; an abnormal feature extraction algorithm is developed, the operation state of the equipment is monitored in real time in combination with an intelligent early warning and processing system, early warning is quickly given out, and corresponding processing measures are taken; by adopting the anomaly detection and classification technology, an anomaly detection algorithm is realized, an efficient alarm information transmission mechanism and an intelligent alarm information processing system are developed, and an alarm information feedback and closed-loop management mechanism is established, so that the accuracy and timeliness of anomaly detection are improved.

The invention belongs to the technical field of metal plates, and particularly relates to a metal plate surface quality rapid positioning and recognition system based on image processing, which comprises a high-speed synchronous data acquisition module, an image acquisition unit and a transmission and positioning synchronization unit, and is used for acquiring a high-definition image of the surface of a metal plate and transmitting and recognizing the surface quality of the metal plate based on the transmission and positioning synchronization unit. Determining that the acquired image corresponds to the plate detection area; after interference such as light reflection and noise is eliminated through cooperation of the structure and image preprocessing, time consumed by image preprocessing is greatly reduced, non-defect interference factors such as uneven illumination, plate texture noise, equipment vibration artifacts and dust and stains can be filtered out through interference elimination in combination with a lightweight multi-dimensional feature extraction algorithm, the situation that irrelevant noise points are misjudged as defects is avoided, and the accuracy of image processing is improved. The omission ratio and the false detection rate are reduced, the multi-dimensional feature extraction can describe the essence of the defects from multiple dimensions, the similar defect types are distinguished, the limitation of single feature recognition is solved, and the accuracy of defect classification is improved.

The invention discloses a BIM (Building Information Modeling)-based pile support foundation pit whole-process safety management system, relates to the technical field of foundation pit safety management, and aims to solve the technical problem of frequent false alarm caused by insufficient construction interference identification capability of a fixed threshold alarm mechanism. And a decision intervention module. According to the method, the multi-source information atlas is constructed, and construction activities, equipment states, environmental factors and monitoring data are dynamically associated. When the data is abnormal, the root cause correlation analysis unit is started, and the intensity and duration characteristics of the abnormal signal are calculated by using a time sequence characteristic extraction algorithm and are compared with the historical pattern library. The mechanism can intelligently judge whether data abnormity is derived from construction interference such as substantive soil body / structure instability or heavy equipment operation, the alarm accuracy is remarkably improved, and the problem of frequent false alarm caused by insufficient construction interference identification capability of a fixed threshold value alarm mechanism is solved.

The invention discloses a spinal cord electrical stimulation control method and system based on tremor detection. The method comprises the following steps: acquiring motion sensing data of a user; extracting tremor signal features of the user according to the motion sensing data and a preset feature extraction algorithm; determining an electrical stimulation parameter corresponding to the tremor signal feature according to a preset electrical stimulation parameter determination rule; according to the electrical stimulation parameters, generating an electrical stimulation control instruction corresponding to electrical stimulation equipment arranged in a spinal cord area of the user; the electrical stimulation control instruction is used for indicating current output of the electrical stimulation equipment. Therefore, accurate nerve intervention control in combination with tremor monitoring and spinal cord electrical stimulation can be realized, the physical therapy inhibition effect of electrical stimulation physical therapy measures on the tremor of the user is improved, and more efficient and accurate nerve regulation and control service is provided for the user.

The invention relates to the field of mines, in particular to a mine cemented filling body strength intelligent monitoring method. The method comprises the following steps: firstly, preprocessing parameters in obtained data from different data sources to obtain preprocessed parameters; in each observation age, splicing the preprocessing parameters to obtain a fusion feature vector, and obtaining a standardized input sequence; based on the standardized input sequence, a final output feature vector is obtained through an adaptive physical gating multi-scale causal expansion convolution feature extraction algorithm; then, based on the final output feature vector, introducing a double-branch mixed attention mechanism to obtain an attention enhancement feature vector; obtaining a final compressive strength prediction value based on the attention enhancement feature vector; and finally, based on the final compressive strength prediction value, setting a safety early warning mechanism to obtain a monitoring result. The technical problems that the multi-scale receptive field cannot be adaptively adjusted, key stage features are insufficiently captured or redundant calculation is performed, and generalization is poor are solved.

The invention discloses a bridge tower footing differential settlement monitoring method based on visual inspection, and relates to the technical field of bridge structure health monitoring, and the method comprises the following steps: selecting natural feature points and stable reference points on the surface of a bridge tower footing, arranging a camera system, and collecting a reference image; a time sequence image is collected through a camera system and preprocessed, natural feature points are matched through an improved SIFT feature extraction algorithm, a homography matrix is estimated through an RANSAC algorithm in combination with stable reference points, coordinate system unification is achieved, and accurate feature point coordinates are obtained; the method comprises the following steps: calculating a two-dimensional pixel displacement amount by comparing a reference with a feature point coordinate in a current image, converting the two-dimensional pixel displacement amount into a settlement amount in the vertical direction of a tower footing based on a camera parameter, carrying out time sequence analysis on the settlement amount to obtain a differential settlement value, and generating a differential settlement early warning according to a preset multi-stage safety threshold value. According to the invention, non-contact, high-precision and full-automatic monitoring of differential settlement of the bridge tower footing is realized.

The invention relates to a rock mechanics testing method and system based on artificial intelligence, and the method comprises the steps: carrying out the three-dimensional scanning of a rock sample, and obtaining three-dimensional model data; sending into a convolutional neural network to analyze the surface characteristics of the rock sample, and obtaining the initial state parameters of the rock sample; according to the initial state parameters, a rock mechanics test loading scheme is formulated through an adaptive optimization algorithm, and a test execution instruction is obtained; applying a preset load to the rock sample according to the test execution instruction, and synchronously acquiring mechanical response data and a deformation image sequence; according to the deformation image sequence, constructing a full displacement field and a strain field of the rock sample, and obtaining global deformation characteristics; and based on the global deformation characteristics and the mechanical response data, identifying damage behavior parameters of the rock sample through a time sequence characteristic extraction algorithm, and obtaining a damage type. Intellectualization, data closed loop and self-evolution of the whole rock mechanics test process are achieved, and the test precision and the reliability of rock damage prediction are improved.

The invention discloses a power plant wastewater treatment intelligent control method based on digital twinning, and particularly relates to the field of prediction.The power plant wastewater treatment intelligent control method comprises the steps that S1, a whole-process digital twinning body mapping a power plant wastewater treatment physical system is constructed, and multi-dimensional core parameters are collected in real time; s3, training a BP neural network model optimized by a genetic algorithm to realize prediction of water quality parameters; s4, solving a multi-objective optimization problem through an NSGA-III algorithm, and generating an optimal process parameter formula to support a decision; s6, dynamically iterating the digital twin and the prediction model based on the actual operation data, and continuously improving the prediction and optimization reliability; according to the method, through construction of the whole-process digital twinborn body and the improved LSTM time sequence feature extraction algorithm, accurate prediction of the water quality change of the power plant wastewater is realized, the problem of prediction lag in the prior art is solved, sufficient data support is provided for process adaptation, the risk that effluent exceeds the standard is effectively reduced, and stable operation of a treatment system is guaranteed.

The application relates to the technical field of data processing, in particular to an accelerometer vibration rectification error analysis method, which comprises the following steps: receiving multiple original current signals of accelerometers, processing collected data to output a characteristic point coordinate sequence with confidence; dynamically scheduling an optimization algorithm combination in a knowledge base according to the confidence and frequency band distribution characteristics to calculate a local statistical data set; inputting the data set into a mechanical dynamics model to perform trajectory simulation, simultaneously inputting a control model to reconstruct a signal to generate an error correction coefficient matrix by comparing the differences of double-path positioning drifts; adopting the matrix to reconstruct a vibration rectification error quantitative value to output a knowledge base update instruction and a parameter reset instruction; importing the quantitative value to calculate a theoretical angle deviation and verify convergence, and triggering neural network retraining when the deviation exceeds a threshold value. Through a non-smooth feature extraction algorithm and a closed-loop feedback mechanism, the signal distortion problem caused by the smoothing operation in the prior art is solved.

The application discloses a brain-computer interface method and system based on real-time closed-loop vibration enhancement, relates to the technical field of brain-computer interface, and comprises the following steps: processing EEG signals through a filtering and feature extraction algorithm, generating a regression vector and identifying an event-related potential as an expected signal; constructing a distributed EEG signal decoding network, optimizing node cooperation through task clustering; adopting an adaptive algorithm update node estimation of a diffusion strategy, optimizing signal decoding through dynamic adjustment of combination coefficients and inter-node cooperation; generating a virtual feedback signal according to a decoding result, fusing the virtual feedback signal with the EEG signal and feeding back to the system, adopting machine learning to optimize classification output, and adjusting model parameters to enhance brain rhythm. The method disclosed by the application can accurately capture the change characteristics of the electroencephalogram signal, dynamically adjust the cooperation mode between nodes, and improve the signal decoding accuracy. The decoder parameters can be adjusted in real time, the electroencephalogram rhythm synchronization is enhanced, and the recognition effect of the motor imagination task is improved.

This invention discloses an AI-interactive AR (Augmented Reality) cultural wall system for cultural tourism, belonging to the field of AR system technology. The surface of the cultural wall is equipped with a pre-embedded feature set, including visual markers, differentiated texture features, 3D depth variation areas, and specifically designed interactive hotspots. These features provide the system with stable visual benchmarks, semantic distinction criteria, depth cues, and precise positioning anchors. The system includes: a feature point acquisition module, used to collect the feature set on the cultural wall through a multimodal feature extraction algorithm to construct a 3D spatial map with semantic annotations; a user positioning and pose estimation module; an AI content generation engine; and a spatial adaptation and rendering module. Through the cultural wall feature set and the multi-module collaborative system, precise spatial integration of virtual content and the physical wall surface is achieved, along with intelligent dynamic content generation based on user behavior, effectively enhancing the immersion, interactivity, and personalization of the AR cultural experience.

The application discloses a kind of based on graph coding's diagnosis and surgical matching check method and system, the method includes: obtaining the diagnosis data and surgical plan data of target object;Based on feature extraction algorithm, extract the diagnosis change feature and surgical step feature in the diagnosis data and the surgical plan data;Based on graph network coding algorithm, the diagnosis change feature and surgical step feature are encoded into corresponding node relationship graph;The node relationship graph is input into trained graph neural network to obtain the matching degree parameter of output diagnosis and surgical plan.It can be seen that the present application can realize accurate medical diagnosis and surgery matching evaluation based on graph structure feature association, improve the scientificity and adaptability of surgical plan selection, reduce the surgical risk and treatment deviation caused by mismatch between scheme and diagnosis.

The invention relates to an infrared ship feature extraction system and method based on depth expansion. The system comprises a background estimation module which adopts a time recursion coding and decoding architecture, the input end and the output end of the time recursion coding and decoding architecture are in residual connection, and the background estimation module is used for processing an input first foreground feature map and a first intermediate variable under approximate low-rank constraint and outputting a second background feature map; the foreground estimation module sets a CNN layer and an activation function based on Lipschitz continuity constraint and is used for processing the input first foreground feature map, the first intermediate variable and the second background feature map and outputting a second foreground feature map; and the image enhancement module maps the second background feature map and the second foreground feature map into the infrared image through convolution, batch normalization and activation operations to obtain a second intermediate variable. According to the method, the robustness and generalization ability of a ship wake flow infrared feature extraction algorithm can be improved.

The invention discloses an IPC image intelligent optimization method based on a Tiong operating system. The method comprises the following steps: acquiring an image data stream and corresponding equipment hardware parameters; according to each parameter item in the equipment hardware parameters, matching the image data stream with various image optimization branch conditions; when it is determined that the real-time image data stream is matched with the target type of image optimization branch condition, adjusting at least one optimization parameter of the image video stream by using a target optimization processing strategy matched with the target type of image optimization branch in combination with equipment hardware parameters; according to the method, the image video stream after matching or parameter optimization is coded to obtain the target data stream, so that the captured image can be directly optimized, third-party software is not needed, on the premise of ensuring the basic attributes of the image, image defects are restored to the maximum extent, the image quality is improved, hardware parameters are taken as optimization basis, and the optimization efficiency is improved. The dependence on an image feature extraction algorithm is reduced, and the processing efficiency is improved.

The invention provides a bubble line defect detection method and related equipment, and relates to the technical field of transparent tubular defect detection. Multi-focal-plane scanning is performed by using the zoom image acquisition system to obtain sequence image sets with different focal lengths, so that the problem of blurring of bubble lines under a single focal plane is effectively solved. By introducing a rotating target detection model, linear defects at any inclination angle can be accurately identified, a rotating detection frame closely surrounding the defects is output, and the problem that a traditional horizontal detection frame introduces irrelevant backgrounds is avoided. Furthermore, in combination with a linear feature extraction algorithm, the position information, the weighted average slope and the total length of the linear defect can be accurately obtained from the region of interest, and high-speed, high-precision and full-automatic quantitative detection of the fine bubble line defect in the transparent tubular object to be detected is realized.

This invention discloses an artificial intelligence-based method, system, and storage medium for generating panoramic sound audio. The method includes the following steps: S1. Inputting the original audio signal and scene parameters, obtaining the time-domain-frequency domain features and spatial scene features of the original audio signal through a multimodal feature extraction algorithm, and weighted fusing the time-domain-frequency domain features and spatial scene features to obtain fused features; S2. Constructing the mathematical expression basis of the 3D spatial sound field model based on the spherical harmonic function, mapping the fused features to three-dimensional spatial coordinates through a sound field modeling algorithm to obtain the 3D spatial sound field model; S3. Using an adaptive rendering algorithm, converting the 3D spatial sound field model into multi-channel panoramic sound audio output according to the parameters of the target playback device. This invention achieves automated panoramic sound audio generation by inputting the original audio signal and scene parameters for feature extraction, fusion, and sound field modeling, and converting the output based on adaptive rendering. It has the advantages of automation, high efficiency, and high precision.

The embodiment of the invention provides a railway limit intrusion detection method and related equipment, and belongs to the technical field of railway detection. The method comprises the following steps: acquiring a track detection result of track detection and a railway line image; carrying out time difference matching according to the rail inspection result and the railway line image to obtain an image extrinsic parameter; performing image correction processing on the railway line image to obtain a distorted image; performing target detection according to the distorted image to obtain target information of the to-be-detected railway equipment; based on two different feature extraction algorithms, performing feature matching on the distorted image according to the target information to obtain a target feature set; performing triangulation on the target feature sets of the images along the different railways according to the image extrinsic parameters to obtain boundary position coordinates of the railway equipment to be measured in a track coordinate system; and according to the boundary position coordinates and the track detection result, carrying out intrusion detection to obtain an intrusion result. According to the invention, automatic detection can be realized, the accuracy and operation efficiency of limit invasion detection are improved, and data support is provided for track adjustment.

A method for extracting features of transmission line conductors based on image segmentation is disclosed. The method includes: coarsely segmenting the conductor portion of an image using an image segmentation model; further processing using a local thresholding method to obtain key points of the transmission line, resulting in a key point map; obtaining an expanded region for the key points and determining the main direction of the key points; obtaining the feature extraction region for the key points; and using a feature extraction algorithm to obtain the conductor features of the image. This invention uses an image segmentation model and a local thresholding method to obtain and extract key points of the transmission line conductors from images captured by a visual monitoring device. It can extract not only points with drastic changes, such as corners, but also points with relatively gentle changes. The subsequent determination of the main direction of the key points primarily serves to reduce errors caused by rotation of the shooting perspective.

The invention discloses a panoramic sound audio generation method and system based on artificial intelligence, and a storage medium, and the method comprises the following steps: S1, inputting an original audio signal and scene parameters, obtaining the time domain-frequency domain features and space scene features of the original audio signal through a multi-modal feature extraction algorithm, carrying out weighted fusion on the time domain-frequency domain features and the space scene features to obtain fusion features; s2, constructing a mathematical expression basis of a 3D space sound field model based on a spherical harmonic function, and mapping fusion features to three-dimensional space coordinates and the 3D space sound field model through a sound field modeling algorithm; and S3, converting the 3D space sound field model into multi-channel panoramic sound audio output according to the parameters of the target playing equipment by using an adaptive rendering algorithm. According to the method, feature extraction, fusion and sound field modeling are carried out by inputting the original audio signals and scene parameters, automatic panoramic sound audio generation is realized based on adaptive rendering conversion output, and the method has the advantages of automation, high efficiency, high precision and the like.

The application discloses a kind of underground pipeline impact event identification method, device, equipment and medium, it is related to underground pipeline safety monitoring technical field, comprising: initial piezoelectric signal of target underground pipeline is collected using piezoelectric sensing array;The initial piezoelectric signal is preprocessed by preset composite algorithm, to generate piezoelectric signal after processing;Using multidimensional feature extraction algorithm, multidimensional feature extraction and fusion operation are carried out to the piezoelectric signal after processing, corresponding fusion feature is obtained;Impact event of the target underground pipeline is identified and classified based on the fusion feature by preset identification model;If the identification result represents that the impact event exists, corresponding target impact event report is generated, and the impact point of the impact event is positioned using preset positioning algorithm, to complete the identification of target underground pipeline impact event.

The invention relates to the technical field of medical image processing, and discloses a cardiovascular plaque component unmarked recognition system and method, and the system comprises a data collection unit, a time domain and frequency domain combined feature extraction algorithm module, a data processing unit, a data transmission unit, a three-dimensional imaging unit and a background processing unit. The time domain and frequency domain combined feature extraction algorithm module maps original measurement pulse sound pressure data to a Hilbert function space to form multi-dimensional feature representation, a tissue specificity functional constraint model is constructed, and multi-scale feature integration is achieved through geodesic distance calculation; the data processing unit generates three kinds of tissue simulation data based on a dual-wavelength optical absorption spectrum principle; the three-dimensional imaging unit obtains different phase data in a two-dimensional rotation acquisition mode, a three-dimensional image is generated through a multi-wavelength optical component decomposition imaging algorithm, a contrast agent or a marker does not need to be used, and accurate distinguishing of a lipid core, a calcification area and a basic tissue is achieved.

The application relates to the field of vehicle detection, and provides a vehicle operation state real-time monitoring method and system based on Internet of Things, which comprises the following steps: performing multistage pretreatment on original signals collected by a plurality of source sensors of a vehicle to obtain standardized multimodal data flow; inputting an edge computing unit, performing feature extraction on the standardized multimodal data flow through a multi-domain feature extraction algorithm to obtain a comprehensive feature vector, and performing abnormality detection through a manifold alignment algorithm to obtain a feature data packet with a priority label; performing channel allocation on a multi-channel communication interface through a game theory model to obtain a target transmission scheme, uploading the feature data packet to the cloud according to the target transmission scheme to obtain a time sequence feature database; and inputting the time sequence feature database into a cloud digital twin model, performing fault diagnosis on the time sequence feature database through a residual fusion algorithm to obtain a structured fault diagnosis report. The application improves the driving safety of the vehicle.

This invention discloses a stable combustion control system based on a multi-dimensional furnace combustion measurement method, specifically relating to the field of furnace combustion measurement. This includes multi-dimensional sensor data acquisition, feature parameter quantification and extraction, combustion stability modeling, intelligent graded stable combustion control, iterative optimization of control effect, and dynamic evaluation of combustion state. Multi-dimensional sensor data acquisition involves synchronously acquiring data through an array of sensors. Feature parameter quantification and extraction involves preprocessing the acquired data to extract key feature parameters. Combustion stability modeling involves constructing a quantitative evaluation model of the combustion stability index based on the key feature parameters. Intelligent graded stable combustion control involves triggering a three-stage direct injection response mechanism for the fuel gun based on the combustion stability index, combined with an adaptive energy matching algorithm to dynamically adjust the stable combustion energy injection. This invention, through a multi-dimensional sensor array and feature extraction algorithm, achieves advanced identification of early combustion deterioration, solving the problem of lag in single flame detector monitoring.

The application provides a blind person discrimination device, system and method based on a distributed optical fiber sensing device.The device comprises an optical time domain reflectometer module, a signal preprocessing module, a signal cutting module, a signal feature extraction module and a signal classification module.The method provides a high recognition rate without video monitoring equipment and wearable sensing equipment, more comprehensively and effectively provides application support for the field of security and blind person travel, enhances the adaptability for different human body recognition, and identifies the source of the human body walking signal accepted by the current device, whether it is a visually impaired user or a normal user, so as to provide related services for the visually impaired user in the future.The long-time and short-time feature joint recognition mode is adopted in the feature extraction algorithm, and the accuracy is further improved.