1471 results about "Principal component analysis" patented technology

The invention relates to the technical field of oil well fault monitoring, and discloses an oil well indicator diagram real-time fault prediction method and system. The method comprises the following steps: acquiring an oil well sensor data stream, buffering and checking data integrity through a sliding window, and aligning multi-channel sensor data by applying a dynamic time warping algorithm to generate a standardized data stream; extracting time domain features based on the data stream, and comparing the time domain features with a historical feature library after principal component analysis dimension reduction to generate a feature difference index; triggering a multi-level threshold strategy according to the difference index, collecting an incremental training data set, finely tuning the model by adopting an elastic weight preserving algorithm, and generating a hot switching ready model; after the model is loaded, a fault probability value is generated through GPU accelerated reasoning, and an early warning event with a timestamp is generated; and finally analyzing the message into an early warning protocol message edge for transmission, and dynamically optimizing system resources based on logs. According to the method, the delay problem of high-frequency data flow is effectively solved, and the fault prediction accuracy and the system response speed are remarkably improved.

According to the cooling tower early fault early warning method based on vibration state monitoring, vibration signals and working condition labels of key parts of the cooling tower are synchronously collected through multiple channels, and data quality is improved through preprocessing operation such as band-pass filtering and normalization; time-frequency features are extracted in a multi-scale mode through self-adaptive variational mode decomposition and wavelet packet transformation, signal complexity is quantized through energy entropy, and weak fault detection capacity is enhanced; the obtained features are input into a deep belief network after being subjected to principal component analysis dimensionality reduction, and automatic classification and recognition of the equipment operation state are achieved; dynamic early warning grade adaptation is carried out according to an identification result in combination with a working condition label, the environmental adaptability and stability of early warning are effectively improved, the method further has the functions of early warning sample recording and periodic model iterative optimization, and the fault identification precision and robustness in a complex noise environment are remarkably improved.

The invention discloses a construction emergency early warning method and system, and the method comprises the following steps: collecting the three-dimensional coordinates of a constructor, combining a sliding time window with a wavelet packet energy entropy and other indexes, and generating a multi-scale movement disorder index through principal component analysis and fusion; continuously unstable persons are recognized according to the disorder index, after the trajectory of the persons is segmented, a weighted graph is constructed in combination with hidden Markov and building information model environment parameters, and the cognitive mismatch degree is calculated; mapping the cognitive mismatch degree to a space grid, calculating a local Moran index, fusing a density gradient and a mechanical operation sequence resonance result, and constructing a propagation weight matrix; constructing a heterogeneous graph based on weight matrix guidance, calculating risk influence propagation potential energy, combining historical disorder sequence analysis and relative entropy, and coupling to obtain a group-level instability pre-judgment value; and outputting a comprehensive critical level for the pre-judgment value over-limit individuals through motion trend prediction, spatial intersection and shortest path algorithms. The safety during building construction operation is improved.

The invention belongs to the technical field of data processing, and particularly relates to a hidden ore body evaluation and positioning method based on multi-source data processing. The method mainly aims at the problems of incompleteness and isomerism of multi-source geological data in acquisition, fusion and modeling. Comprising the following steps: acquiring hyperspectral, geochemical and magnetic anomaly multi-source data of an evaluation area; intelligently complementing missing modal data by using a generative adversarial network based on geological constraints and modal outburst to form a complete multi-source data set; an unsupervised clustering algorithm combining geological correlation and entropy weight analysis is adopted to construct high-confidence-coefficient pseudo-label data, and knowledge mining of unlabeled samples is achieved; feature purification and dimension reduction are carried out through multi-modal feature fusion and hierarchical principal component analysis, and key feature vectors representing the existence of the ore body are extracted; and finally realizing space prediction of the concealed ore body by utilizing the classification model. According to the method, a high-quality data basis and a unified processing framework are provided for intelligent recognition of the hidden ore body, and efficient and accurate positioning of the hidden ore body is achieved.

The invention provides an intelligent warehouse goods posture recognition and automatic sorting method and system, and the method comprises the steps: S2, extracting the edge contour and surface features of goods from point cloud data for a preliminary value set, calculating the direction vector and shape distribution characteristics of the goods through a principal component analysis method, and obtaining a quantitative result of shape feature extraction; s6, historical sorting data and real-time sensor data are extracted from the warehousing system database according to the classification basis adapted to the complex scene, the dynamic posture change trend of the goods is predicted through a time sequence analysis method, and optimization parameters of real-time processing are obtained; and S7, the movement track and the grabbing angle of the sorting mechanical arm are adjusted through the optimization parameters subjected to real-time processing, a deep reinforcement learning algorithm is adopted to conduct iterative optimization on the sorting action sequence, and an execution scheme of sorting accuracy is determined. According to the method, the accuracy of goods posture recognition and the automatic sorting efficiency in the complex storage environment are remarkably improved.

The invention relates to the technical field of intelligent monitoring, in particular to a municipal building engineering construction progress monitoring method based on an unmanned aerial vehicle and laser scanning, and the method comprises the following steps: obtaining point cloud data through carrying laser scanning by the unmanned aerial vehicle, calibrating a space coordinate, binding a design coordinate system, carrying out Gaussian filtering denoising processing, and analyzing point cloud through coordinate transformation. The method comprises the following steps: extracting a main axis vector by adopting a principal component analysis method, dividing a sectioning region, calculating a cosine value of a normal included angle, classifying and identifying a difference region by Euclidean distance, carrying out secondary scanning clustering segmentation, and calculating an offset judgment state label set. According to the method, a three-dimensional reference is established by adopting laser scanning and coordinate calibration, Gaussian filtering is combined to eliminate noise, PCA is used to extract a geometric main axis, a normal included angle matching degree is calculated, Euclidean distance classification detection deviation is carried out, geometric difference is quantitatively identified, graph-model matching precision is improved, millimeter-level capture is realized, an automatic monitoring system is constructed, and manual errors are reduced. And controlling the power-assisted progress.

The invention discloses a soil heavy metal distribution prediction method and system based on machine learning, and the method comprises the steps: obtaining topographic factor land use industrial activity data, carrying out the fusion remote sensing information processing, and determining a multi-source data set; performing standardization processing according to the multi-source data set, and performing space-time registration if the scale difference after standardization processing exceeds a preset threshold value to obtain data in a unified format; key features are extracted according to the unified format data, and a dimension reduction feature set is obtained through principal component analysis; a random forest model is constructed according to the dimension reduction feature set, parameters are optimized, and a heavy metal content prediction model is determined; inputting the sampling finite point location data into a heavy metal content prediction model, judging an industrial activity influence area, and obtaining a preliminary distribution estimation result; based on the preliminary distribution estimation result, high-resolution grid data are obtained by fusing topographic factors for the space complex region; and generating a pollution distribution diagram according to the high-resolution grid data, judging a low-prediction-precision region, and obtaining a final optimized layer.

The invention discloses an automatic identification and state evaluation method for rail transit unmanned aerial vehicle inspection, particularly relates to the field of rail transit unmanned aerial vehicle inspection, and is used for solving the problem that existing unmanned aerial vehicle inspection is influenced by rail attitude change and environment interference, so that rail defect identification and positioning are not comprehensive. An attitude time sequence matrix is constructed by controlling an unmanned aerial vehicle cluster to fly in the track direction, and an analysis unit is obtained by adopting space-time alignment and sliding window segmentation; based on principal component analysis, attitude cooperative oscillation characteristics are extracted, an attitude reference curved surface is constructed, and individual attitude dissimilation degree indexes are calculated; identifying an abnormal section by using a dynamic threshold value to realize track defect positioning, and controlling a cluster to execute matrix scanning and cross validation flight according to the defect distribution density; and finally, outputting a continuous track smoothness evaluation map. The method can keep stable inspection performance in a complex environment, improves the accuracy and coverage range of track anomaly recognition, and has high engineering application value.

According to the springback prediction compensation method in the precise stamping process of the high-pitch panel, the stamping force, the die temperature and the material thickness tolerance are collected through the multi-channel sensor, and the data quality is improved by combining standardization and abnormal value elimination; realizing parameter dimension reduction by adopting principal component analysis, obtaining working condition feature representation, quantifying working condition similarity by utilizing a weighted Euclidean-Mahalanobis distance, judging unseen working conditions in real time, and automatically triggering meta-learning model compensation; through model-independent element learning, rapid adaptive error prediction based on newest stamping data, and combination of confidence evaluation and dynamic adjustment compensation strategies, efficient prediction and robust compensation of springback errors are realized, and model parameters are continuously optimized in follow-up combination with actual compensation feedback. According to the method, the self-adaptive capacity and the forming precision to unknown stamping working conditions are improved.

The invention relates to the technical field of power equipment operation and maintenance and state monitoring, in particular to an optical fiber transformer health state assessment method and system based on feature fusion. According to the invention, sampling signals of a plurality of optical fiber current transformers are collected and preprocessed; constructing a historical data matrix for principal component analysis, establishing a principal component space and calculating a square prediction error control limit value; constructing a simulated fault data set and projecting the simulated fault data set to a principal component space to calculate a square prediction error time sequence; performing wavelet packet decomposition on the square prediction error time sequence to extract a normalized energy feature vector, and training a one-dimensional convolutional neural network model; projecting the real-time sampling signal to a principal component space to judge whether the real-time sampling signal exceeds the limit or not; if yes, the health state level is output through the trained model. According to the invention, accurate grading and real-time online evaluation of the health state of the optical fiber transformer are realized.

The invention discloses a building safety intelligent monitoring, early warning, prevention and control method, and the method comprises the steps: collecting a physical state parameter and an environment disturbance parameter of a building structure body in real time through a distributed monitoring node group disposed at a key part of the building structure body, and forming an original monitoring data flow; performing space-time alignment and noise reduction processing on the original monitoring data stream by using an adaptive weighted fusion algorithm to generate a standardized structure response data set; extracting a multi-dimensional time-frequency domain feature vector representing the health state of the structure from the standardized structure response data set based on a wavelet packet transformation and principal component analysis combination method; and inputting the multi-dimensional time-frequency domain feature vector into a pre-trained twin neural network, and outputting abnormal region positioning information and an abnormal degree quantitative index. According to the method, the building mechanics mechanism and the artificial intelligence technology are deeply fused, a full-closed-loop intelligent prevention and control system from accurate risk identification to active regulation and control is constructed, and the reliability and timeliness of building safety monitoring in a complex environment are remarkably improved.

The invention discloses an ecological shoreline diagnosis method and system based on hydrological-biological communication, and is applied to the technical field of water environment ecological restoration and shoreline health assessment. Comprising the following steps: dividing intertidal zones of a shoreline research area, and generating uniformly distributed sampling points in each tidal zone; extracting a plurality of indexes related to hydrological and biological connectivity, and constructing a comprehensive hydrological-biological connectivity state index through principal component analysis; dam feature data of a shoreline area are obtained, correlation analysis and collinearity screening are carried out, and key shoreline features are screened through a random forest algorithm; constructing a mechanism connectivity index based on a circuit theory; constructing and training a Bayesian network prediction model; and performing connectivity state diagnosis, attribution analysis and restoration scheme effect simulation on the target shore section by using a Bayesian network prediction model. By coupling hydrological and biological connectivity, static element evaluation is changed into dynamic ecological process diagnosis, and dominant factors are accurately positioned.

The invention discloses an acoustic emission signal noise reduction and feature extraction method and system suitable for a deep roadway, and relates to the technical field of safety monitoring of deep mineral resource mining, and the method comprises the specific steps: arranging an acoustic emission sensor array along the deep roadway, collecting multi-source data, converting the multi-source data into digital signals, and storing the digital signals; identifying an interference type through a wavelet packet decomposition and environment correction algorithm; self-adaptive noise reduction is carried out by using a complexity sensitive penalty algorithm; time domain and frequency domain features are extracted, coupling parameters are calculated, and time domain and frequency domain features are obtained through Hilbert-Huang transform; and finally, through principal component analysis dimensionality reduction and mutual information entropy screening, constructing a feature vector and transmitting the feature vector to a safety early warning system. According to the invention, the processing precision and reliability of the acoustic emission signal are improved through the multi-source signal acquisition module, the interference identification module and the adaptive noise reduction module; signal characteristics are comprehensively described through algorithm optimization, key characteristics are output through characteristic optimization, real-time monitoring and early warning are achieved through a dynamic damage vector algorithm, and deep roadway construction safety is guaranteed.

The invention belongs to the technical field of petroleum and natural gas engineering, and particularly relates to a priori knowledge embedded unsupervised learning geology-engineering dessert comprehensive evaluation method, which comprises the following steps: S1, obtaining an evaluation data set of a research area, and preprocessing and standardizing the evaluation data set; s2, dividing the evaluation data set into a positive index and a negative index according to the prior knowledge, constructing positive and negative ideal points through the prior knowledge, and fusing to generate an unsupervised evaluation data set; s3, clustering is carried out on the unsupervised evaluation data set by adopting principal component analysis and an incremental online k-means algorithm; and S4, according to category labels obtained by clustering, determining geological-engineering dessert grades by adopting a standardized variable distance grading evaluation method, and carrying out segmented clustering optimization and perforation position optimization on the geological-engineering dessert grades. According to the method, a supervised evaluation effect is achieved through an unsupervised algorithm, comprehensive evaluation of the dessert of the fractured horizontal well section is achieved, and the problem of unlabeled data evaluation is effectively solved.

The invention relates to the technical field of image processing, and discloses a pitaya fruit pose estimation method based on point cloud three-dimensional reconstruction. The pitaya fruit pose estimation method comprises the steps that RGB color images and depth images of pitaya fruits are collected, expanded and marked; the DeepLabV3 + network model is improved and trained, and then an RGB color image is segmented; segmenting the depth image by using the semantic segmentation mask image, then matching the depth image with the RGB color image to obtain a pitaya tree point cloud, and preprocessing the pitaya tree point cloud; registering the pitaya fruit tree point cloud by using an improved point cloud registration algorithm to obtain a fruit tree three-dimensional point cloud model, and segmenting fruit local point clouds from the fruit tree three-dimensional point cloud model; establishing a pitaya fruit point cloud coordinate system by using a PCA principal component analysis method; and performing spherical fitting and ellipsoid fitting on the pitaya fruit point cloud coordinate system to obtain an ellipsoid model with unique pose parameters. According to the invention, a fruit tree three-dimensional point cloud model and accurate fruit three-dimensional space pose information can be provided for a pitaya fruit picking system.

The invention provides a polarization modulation coupling time control phase locking high-temperature noise suppression in-situ Raman method and device, and relates to the technical field of spectrum detection. The method comprises the following steps: exciting pulse linear polarization laser to irradiate a sample to obtain Raman scattering light; a parallel vibration component (P) and a vertical vibration component (S) are obtained, and original spectral data containing polarization information and residual high-temperature noise are collected by adopting an ISCCD detector synchronously triggered by gating. Background noise is removed and a parallel vibration component (P) is extracted through polarization differential processing in combination with adaptive filtering (dynamically adjusting parameters according to signal distribution) and principal component analysis (PCA); and finally, performing signal-to-noise ratio optimization on the pure component by applying a phase locking algorithm to finally obtain a Raman spectrum result with a high signal-to-noise ratio. According to the invention, the in-situ Raman spectrum of the material within the temperature range of room temperature to 3000 DEG C can be measured, and the detection of weak signals is more sensitive.

The invention discloses a data processing method for geological disasters, and particularly relates to the technical field of geological disaster monitoring and data processing. Collecting and normalizing multi-source monitoring data, and constructing a unified time sequence fusion vector; identifying a weak abnormal signal based on the disturbance change rate and a sensitivity model; high-disturbance feature points are extracted through sliding window clustering, and a dynamic risk factor matrix is constructed in combination with a remote sensing image; extracting dominant risk factor feature vectors by using principal component analysis, inputting the dominant risk factor feature vectors into a disaster evolution simulation model, and predicting a future high-risk area in combination with a historical path library; and finally, potential micro-disaster trigger points and risk levels are output, and graded early warning processing is realized. The system has the advantages of high disturbance identification precision, strong simulation prediction capability, automatic early warning response and the like, and is suitable for an intelligent early warning system for sudden geological disasters such as landslide and debris flow.

The invention discloses an efficient lithium battery life prediction method and system based on machine learning, and the method comprises the steps: obtaining the voltage, current and temperature data of a lithium battery, and extracting a multi-dimensional feature set through principal component analysis; inputting the feature set and the number of charge and discharge cycles into a long short-term memory network model, and outputting an initial life prediction value; according to whether the deviation between the initial predicted value and the reference value exceeds a first threshold value or not and whether the predicted deviation after model updating continuously exceeds a second threshold value or not, multi-stage correction operation including first-stage correction and second-stage correction is executed on the initial predicted value, and a final life predicted value is obtained; according to the lithium battery life prediction method, through multi-stage correction and model adaptation, the accuracy, the adaptability and the robustness of lithium battery life prediction are remarkably improved.

The invention discloses a fracturing equipment state monitoring and fault diagnosis system and method, and belongs to the technical field of oil and gas field fracturing equipment intelligence. The invention aims to solve the problems that in the prior art, monitoring depends on a single signal, fault early warning lags behind, and the misjudgment rate is high. The method comprises the following steps: collecting multi-source operation data of the fracturing pump in real time; establishing a theoretical pressure indicator diagram, and comparing the theoretical pressure indicator diagram with an actual indicator diagram generated by real-time data to realize first-stage fault judgment; an improved wavelet threshold noise reduction method is adopted to process the signals, and time domain and frequency domain features are extracted; and inputting the processed feature data into the combined diagnosis model for second-stage fault judgment. The joint diagnosis model combines a principal component analysis (PCA) model used for uncalibrated data anomaly detection and a BP neural network model used for calibrated data fault classification. Through deep fusion of the mechanism model and the data driving model, environmental interference is effectively resisted, and the diagnosis accuracy and the operation and maintenance efficiency are remarkably improved.

The invention relates to the field of retrieval enhancement, in particular to a multi-view knowledge-intensive retrieval enhancement generation system and method, which are characterized in that structural vectors and semantic topics are extracted from professional corpora through principal component analysis and non-negative matrix factorization technologies, and a multi-dimensional professional view set is constructed. After a user query is received, a potential intention is identified, a view angle weight vector is generated, the query is rewritten according to the view angle weight vector, multiple groups of view angle retrieval requests are constructed, targeted document retrieval is executed, and a structured prompt input language generation model is constructed based on a view angle weight reordering result and fusion of an original query and a multi-view angle retrieval result. The method is suitable for scenes of law assistance, intelligent diagnosis, academic questions and answers and the like, so that the accuracy, the interpretation and the reliability of retrieval and generation in the complex field are remarkably improved.

The invention belongs to the technical field of computer vision and three-dimensional point cloud processing. The invention provides an anisotropic point cloud convolution method based on local geometric self-adaption. According to the embodiment of the invention, the principal component analysis is carried out on the neighborhood point cloud to estimate the principal direction of the local geometric structure, and the affine transformation matrix is constructed; and performing dynamic deformation on the basic kernel point of the convolution operator by using the matrix, so that the shape and the direction of the convolution kernel are dynamically aligned with the local structure of the point cloud. The convolution operator is integrated into a point cloud registration network, and can be used for feature extraction of three-dimensional point clouds, especially underwater multi-beam depth finder point clouds. The convolution operator constructed by the method can effectively capture the structure priori of the strong anisotropic point cloud, and solves the problems of structure information loss and limited feature extraction capability caused by the isotropic kernel processing the data.

The invention discloses a wheat yield remote sensing prediction method combining phenological parameters, which comprises the following steps: S1, performing field observation in a key growth period of winter wheat, synchronously collecting canopy hyperspectral reflectivity data, leaf area index (LAI) and SPAD value of each observation sample point, and recording wheat grain yield of the corresponding sample point; s2, performing noise reduction preprocessing on the acquired canopy hyperspectral data, and extracting sensitive spectral parameters by combining principal component analysis (PCA) and correlation analysis methods; s3, taking the sensitive spectrum parameters, LAI and SPAD values as independent variables, taking the wheat grain yield as a dependent variable, and adopting partial least squares regression (PLSR) to construct a multivariable yield prediction model; and S4, performing wheat yield prediction on an independent test sample or regional scale remote sensing data by using the trained model. The method overcomes the defect that only yield sensitive spectrum parameters are used for predicting the effect, and accurate estimation of the model is achieved.

The invention discloses an ecological environment intelligent pre-auditing method based on partition management and control and capacity control, and relates to the technical field of ecological environment informatization management. According to the method, parallel superposition is carried out on multiple types of ecological environment sensitive layers, dimension reduction processing and weighted integration are carried out on the multi-dimensional space conflict factors through principal component analysis, a quantifiable ecological sensitivity risk score is generated, the single logic that a traditional method can only judge whether overlapping exists or not is effectively overcome, fine expression of multi-layer composite conflicts is achieved, and the risk assessment efficiency is improved. And the resolution and judgment basis of spatial analysis are enhanced. A pollution factor-control unit-emission period three-dimensional capacity comparison model is established, and for capacity accounting requirements of multiple pollutant types, multiple space areas and multiple time dimensions, limitation of static threshold comparison of a traditional method is broken through, and dynamic adaptability and accuracy of capacity evaluation under a complex emission structure are realized.

A power transformation equipment temperature anomaly identification and diagnosis method comprises the following steps: collecting state variables, performing cleaning, interpolation complementation and abnormal point elimination on multi-source data through a time synchronization mechanism, and constructing a unified data matrix; extracting statistical features and time sequence dynamic features in the time sequence based on the data matrix, and performing dimensionality reduction on redundant information in combination with a principal component analysis method to form a multi-dimensional fusion feature vector; an unsupervised learning model based on LSTM-AE is constructed, a normal working condition data learning feature reconstruction mode is utilized, and a reconstruction error is taken as a criterion to identify potential temperature anomaly; and calling a preset expert rule base and a knowledge graph, automatically analyzing dominant factors causing anomalies, and identifying typical anomaly types. According to the invention, automatic identification and classification diagnosis of the temperature abnormity of the power transformation equipment under an unsupervised condition are realized, the accuracy and response speed of fault identification are obviously improved, and the intelligence and practicability of equipment operation state monitoring are enhanced.

The invention provides a method and a system for processing content measurement data of multi-index components in folium cortex eucommiae, and relates to the technical field of traditional Chinese medicine quality control and data analysis crossing. The method comprises the following steps: aligning standardized content data of a current batch with data of a corresponding batch in a predicted content data set; obtaining an actual detection value and predicted value pairing sequence of each index; the deviation ratio of each data pair in the actual detection value and predicted value pairing sequence of each index is calculated, and a deviation ratio analysis result is formed; and based on the deviation ratio analysis result, in combination with a pre-stored legal standard threshold and historical batch data, performing comprehensive evaluation through principal component analysis and analytic hierarchy process to obtain a multi-dimensional quality comprehensive evaluation value including specification conformity, component proportional relation and process stability. According to the method, an integrated data processing system is constructed, so that the intelligence and precision level of multi-index quality evaluation of the folium cortex eucommiae is improved.

The invention relates to the technical field of industrial automation control, and discloses a chemical production-oriented multi-stage cooperative intelligent control and equipment operation and maintenance data management platform, which comprises a multi-source data acquisition module for acquiring multi-source heterogeneous data and performing time alignment processing and quality verification; the working condition identification and feature extraction module is used for obtaining equipment operation state data by adopting weighted fusion and principal component analysis; the equipment health assessment module is used for calculating an equipment health degree score and predicting a degradation trend through exponential smoothing; the self-adaptive control module is used for designing a multivariable coordination controller and dynamically generating operation constraints of the multivariable coordination controller according to the equipment health degree score; the layered optimization control module is used for establishing an interlayer feedback mechanism; the collaborative decision-making module generates a collaborative decision-making trigger signal according to the equipment health degree information and generates a collaborative decision-making scheme; according to the invention, by establishing a collaborative decision-making mechanism, a decision-making scheme considering the production task completion degree, the equipment health degree and the maintenance cost is generated.

The invention discloses a wind field downscaling method based on physical constraint and artificial intelligence, and relates to the technical field of meteorological numerical forecasting, and the method comprises the steps: obtaining an input data set which comprises topographic data, low-resolution forecasting data and observation data; feature extraction and data processing are carried out on the input data set to obtain a training data set, the training data set is utilized to train a neural network model fused with the attention mechanism, and a loss function of the neural network model comprises physical constraint terms. According to the method, feature information of the terrain and the air pressure field is brought into model input, feature extraction is carried out on the air pressure field through principal component analysis, and the adaptability of the model to different terrain and weather backgrounds is improved.

The invention belongs to the technical field of fault diagnosis, and discloses a thermal defect identification method and system for a high-voltage switchgear, and computer equipment, and the method comprises the steps: firstly segmenting an infrared image through employing a transfer learning optimized Mask R-CNN model, reducing the dependence of annotated data through sharing pre-training parameters, and achieving the region extraction of pixel-level equipment; secondly, multi-dimensional temperature information is extracted in combination with a gray histogram and a gray co-occurrence matrix, and key features are screened through PCA to enhance noise immunity; and finally, the LSSVM is adopted for classification, so that the training efficiency is remarkably improved. According to the method, the equipment area is automatically segmented through deep learning, temperature distribution is quantified in combination with multi-dimensional features, man-made misjudgment is reduced, pre-training model parameter sharing is utilized, new tasks are rapidly adapted in a small sample scene, the generalization ability and efficiency are improved, feature dimensions are compressed through principal component analysis, the real-time monitoring requirement is met, and the monitoring efficiency is improved.

The invention discloses an edge reasoning optimization method and system based on segmented knowledge distillation. Firstly, a segmented knowledge distillation framework is constructed, a teacher-student model is divided into corresponding sub-modules with balanced parameters, parallel distillation training is adopted, middle feature dimension reduction and space alignment are achieved in combination with a principal component analysis method, the knowledge transmission efficiency is improved, and convergence is accelerated. Secondly, proposing an equipment perception self-adaptive pruning strategy, dynamically distributing a differential pruning proportion according to the real-time calculation capability and resource state of heterogeneous edge equipment, and balancing the load of low-performance equipment and the precision of high-performance equipment; and finally, establishing a deep reinforcement learning dynamic scheduling mechanism, generating a module delay-energy consumption file through offline analysis, adaptively selecting a device combination by an intelligent agent in an online stage, determining an optimal partition and deployment strategy through a threshold value distribution algorithm, and realizing joint optimization of energy consumption and reasoning time while meeting delay constraint.

The invention discloses a multi-dimensional attribution analysis method and system for a cross-media advertisement putting effect, relates to the technical field of data analysis and advertisement, and is used for solving the problems that the advertisement conversion path recognition accuracy is reduced and the putting strategy optimization strategy of an attribution result is low. The method comprises the following steps: collecting contacts, time sequences and advertisement creative information of a user in a multimedia channel through a point burying technology, classifying according to a channel mapping rule, calculating the conversion contribution degree and propagation efficiency of each channel in combination with historical conversion data, obtaining an influence weight, constructing a user cross-channel access path diagram, and counting node switching frequency. And a multi-layer perceptron model is adopted to calculate a key node coupling coefficient, extract a nonlinear interaction relationship, fuse the coupling coefficient, a time sequence and advertisement originality, extract an attribution variable based on principal component analysis, generate a structured attribution result, and realize refined optimization of an advertisement strategy.