85 results about "Threshold model" patented technology

The invention discloses a tunnel deformation real-time early warning method and system based on an LSTM-CNN model, and belongs to the technical field of tunnel engineering safety monitoring. The method comprises four steps of data acquisition, spatio-temporal feature fusion processing, deformation prediction and risk assessment, and intelligent early warning and decision support: collecting multi-dimensional monitoring data through a distributed sensor network and recording spatio-temporal labels; after standardized noise reduction, extracting time trend and spatial distribution characteristics by using an LSTM-CNN fusion model, and constructing a time-space sequence data set; outputting a deformation prediction value based on the fusion features, and calculating a deviation degree in combination with a dynamic threshold model; and triggering multi-level early warning according to the deviation degree and performing visual display. The system comprises a data acquisition module, an edge calculation module, a cloud analysis module and an intelligent terminal module, and double-channel redundancy transmission is adopted to guarantee data continuity. Through spatio-temporal feature collaborative mining, dynamic threshold value adaptation and graded early warning, the deformation prediction precision and the risk assessment accuracy are improved, the transmission stability in an extreme environment is guaranteed, and the early warning decision efficiency is optimized.

The invention relates to a glass flaw recognition method, system and equipment based on computer vision and a medium. The method comprises the following steps: firstly, acquiring original image data of a glass product, and performing feature extraction by adopting a deep learning model to obtain a defect candidate region set; extracting pixel-level boundary information from the defect candidate region set by using an instance segmentation algorithm, dividing a feature cluster after fusing multi-scale morphological features, and matching the feature cluster with a preset defect feature library to obtain a defect labeling result containing a defect category and a defect region feature vector; constructing a reference threshold model based on the defect category, and dynamically calibrating the model in combination with the defect region feature vector and the production condition parameter to generate a real-time detection threshold range; and performing secondary scanning on the original image data by using the real-time detection threshold range to generate a defect distribution map. According to the method, the accuracy of defect identification is improved, the adaptability to environment change is enhanced, and the glass defect identification task can be efficiently and stably completed.

The invention provides a data exception early warning method and device based on a dynamic threshold model. The method comprises the following steps: firstly, constructing an index time sequence library and an event library for storing index and event information; fusing the historical same-period base line, the trend term and the event influence value and generating a base line through a dynamic weight formula; calculating upper and lower limits of a dynamic threshold by combining the fluctuation acceleration and a basic fluctuation coefficient; and finally, comparing the real-time index value with a threshold range to judge whether to give an alarm, and optimizing the model or updating the event library according to a manual labeling result. According to the method, dynamic and accurate early warning of data exception can be realized, service scene changes can be adapted, the early warning accuracy is continuously improved through feedback optimization, and powerful support is provided for stable service operation and efficient exception disposal.

The invention discloses a medical image analysis method and system based on a visual language model, and belongs to the technical field of medical image intelligent diagnosis, and the system comprises an image preprocessing unit which carries out the down-sampling of an original retina OCT image to 256 * 256 and carries out the normalization of the original retina OCT image; the feature encoding unit comprises an image encoder based on RET Found in combination with LoRA optimization and a text encoder based on BioClinicalBERT; the class balance comparison learning unit is used for adjusting loss through class balance coefficients so as to relieve the class imbalance problem; the uncertainty estimation unit is used for calculating confidence quality and uncertainty scores based on Dirichlet distribution, and determining a threshold value in combination with an improved Youden index; and the model training unit adopts a total loss function of class balance loss and uncertainty loss, outputs a diagnosis result and an uncertainty score through transfer learning, and further comprises an image input module, a result display module and a data storage module. Rare disease classification performance and reliability are improved, training efficiency is improved through LoRA optimization, and an accurate and reliable scheme is provided for detection of the rare retina diseases.

The invention relates to the technical field of natural disaster risk assessment, in particular to a multi-disaster disaster-inducing factor internal correlation analysis method, which comprises four steps of data preprocessing, Bayesian network correlation model construction, space-time dynamic evolution analysis and risk early warning threshold model construction. And deep coupling analysis of disaster-inducing factors of multiple disasters such as mountain torrent-debris flow and the like is realized. The method comprises the following steps: firstly, collecting 17 disaster-inducing factor data such as terrain and rainfall, performing dimensionality reduction through kernel principal component analysis, and calculating a dynamic weight by adopting an entropy evaluation method; then constructing a Bayesian network model to quantify conditional probability association; introducing a space-time attention mechanism to optimize a multi-scale coupling weight; and finally, establishing a coupling risk early warning threshold value and outputting a relevance intensity matrix. According to the method, the problem that a traditional single-disaster analysis method cannot capture the space-time linkage effect of disaster-inducing factors is solved, the risk early warning accuracy of multiple disasters is improved, and scientific decision support is provided for regional disaster prevention and reduction.

The invention discloses a global material intelligent management and control analysis platform based on multi-professional link data, relates to the technical field of material management, and solves the technical problems of insufficient consumption abnormality judgment accuracy and delayed inventory early warning and supply decision. According to the method, a weighted initial threshold model of historical data and business quota is innovatively constructed, a differential statistical method is adopted for periodic and non-periodic materials, meanwhile, a real-time influence factor is introduced, and a threshold is dynamically corrected through a weighting algorithm, so that the threshold can accurately adapt to business scene change, and the service scene change is accurately matched with the business scene change. The method effectively reduces the abnormal misjudgment rate and the missed judgment rate, achieves the precise monitoring of the consumption state, achieves the precise calculation of the basic remaining time, the scene-based remaining time and the safety buffer time through the scene-based consumption speed model and stock effective value measurement and calculation, achieves the scientific evaluation of the stock sufficiency through the combination with a preset threshold value, generates a purchase signal in advance, and improves the purchase efficiency. And the risks of stock overstock and material shortage are avoided.

The invention discloses a differential early warning method for a water turbine speed regulating system based on data driving, and aims to solve the problems of poor real-time performance, low accuracy, weak adaptability and lack of differential early warning of a traditional method. The method comprises the following steps: collecting data such as guide vane opening and unit frequency in real time, and preprocessing the data through median filtering and Kalman filtering; identifying working conditions by using a rule engine and fuzzy logic and processing according to the working conditions; extracting time domain and frequency domain characteristic parameters, and constructing statistical indexes; integrating statistics, nonlinearity and twinborn simulation to build a threshold model, and combining a production rule, a fault tree and a Bayesian network to build a rule model; and three types of early warning models are constructed, early warning levels are divided according to an expert scoring method, information is pushed, and a differentiation strategy is formulated. The method improves the early warning real-time performance and accuracy, adapts to complex working conditions, and guarantees the safe operation of the hydropower station.

A power transformation equipment oil temperature prediction and abnormity early warning method based on a fuzzy algorithm comprises the following steps that real-time operation parameters and oil temperature historical data of power transformation equipment are collected, the data are preprocessed, and a high-quality input data set is constructed; constructing a fuzzy inference system considering multiple input factors, and setting a fuzzy membership function and an inference rule base; according to the fuzzy prediction result and the real-time oil temperature change, a self-adaptive dynamic safety threshold model is compared, a potential abnormal trend is identified, and automatic grading early warning is carried out according to the overtemperature grade; and corresponding exception type identifiers and disposal suggestions are generated and are pushed to the operation and maintenance platform through the communication module, so that remote operation and maintenance scheduling and response control are supported, and intelligent cooperative processing is realized. According to the invention, intelligent prediction of the oil temperature change trend under the influence of multiple factors is realized, the capability of sensing temperature rise abnormity in advance is improved, the accuracy of sensing the operation state of the equipment and the timeliness of early warning response are also remarkably improved, and the risk of equipment failure caused by the abnormal oil temperature is reduced.

The invention discloses a multi-device state sensing driven intelligent storage abnormity early warning method, and relates to the field of storage cargo management, and the method comprises the steps: 1, collecting multi-source data through a multi-source sensor node, including odor intensity, volatile organic compound concentration, carbon dioxide or carbon monoxide content, and environment temperature and humidity, meanwhile, physical coordinates of the multi-source sensor nodes are obtained; 2, acquiring historical multi-source data in advance, constructing a dynamic threshold model based on the historical multi-source data, calculating odor fluctuation, a gas combination abnormal index and a temperature and humidity coupling risk value according to the multi-source data, and introducing time evolution trend analysis to obtain a gas combination abnormal index and a temperature and humidity coupling risk value; and a dynamic threshold model is combined to identify risk signals in the odor fluctuation degree, the gas combination anomaly index and the temperature and humidity coupling risk value, a space attenuation model is adopted to generate a three-dimensional anomaly thermodynamic diagram based on the physical coordinates of the multi-source sensor node and the risk signals, and a high-risk area is positioned.

The invention discloses a state monitoring method for a ring main unit, and relates to the technical field of dynamic monitoring. The method comprises the following steps: collecting and preprocessing historical data of ring main unit equipment; constructing an environment load dynamic threshold model by using a meta-learning algorithm, and after historical data training, inputting real-time environment data to obtain a threshold; comparing the real-time electrical data with a threshold value, judging abnormity and diagnosing a fault; acquiring an internal image of the ring main unit, and processing by using a multi-scale denoising method; constructing a model based on an open set recognition algorithm, and inputting a de-noised image to recognize physical defects; constructing a three-dimensional model according to the design drawing, and performing spatial registration with the physical defect to determine a physical fault; and grading early warning is carried out by integrating electrical and physical fault results. According to the invention, state monitoring of the ring main unit is realized through electrical and physical fault results.

The invention relates to the field of foundation construction monitoring, and discloses a multi-sensor fusion tubular pile sinking process real-time analysis system. The data acquisition and preprocessing module is connected with the sensor group; the real-time analysis module is connected with the data acquisition and preprocessing module and comprises a feature analysis sub-module used for generating a dynamic time sequence feature vector; the health diagnosis submodule is used for diagnosing the health state of the pile body; the dynamic threshold determination sub-module is used for determining a dynamic judgment threshold; a comprehensive judgment sub-module; the early warning module is connected with the real-time analysis module; and the data storage and visualization module is connected with the real-time analysis module. A dynamic early warning threshold value model is constructed through real-time geological changes and construction stages, the problems of false alarm and missing alarm caused by the fact that traditional fixed threshold value early warning cannot adapt to variable geology and working conditions are solved, the accuracy and reliability of early warning judgment are greatly improved, and the system can accurately recognize real risks.

The invention provides a dynamic threshold adjustment method based on a self-adaptive alarm rule engine, and belongs to the field of intelligent power grids, and the method comprises the steps: collecting equipment operation data in real time by an edge node, and building a personalized threshold model; the central server initiates a model aggregation request according to a preset period, each edge node uploads a personalized threshold model parameter increment, calculates an aggregated global model parameter, adds noise to the aggregated global model parameter to obtain the global model parameter, and establishes a global model; inputting the standardized feature vector into a global model to obtain a prediction probability, and calculating a basic threshold according to the prediction probability; if the prediction probability exceeds a basic threshold value, generating a weight vector for the feature vector after standardization processing to obtain a weighted feature; calculating an anomaly degree, calculating a smoothness anomaly degree, calculating an adjustment factor, and adjusting the basic threshold value to obtain a dynamic threshold value; the invention further provides a dynamic threshold adjusting system. The problem that a traditional fixed threshold value cannot cope with equipment parameter volatility enhancement is solved.

The invention relates to the technical field of big data mining, and particularly discloses a discrete particle swarm optimization-based influence maximization method on a hypergraph, which comprises the following steps of: S1, constructing a hypergraph model and a propagation rule: defining a hypergraph which is a node set and a hyperedge set, and satisfying hyperedges; a threshold model is adopted to describe the propagation process, a seed node set is activated at the beginning, other nodes and all hyperedges are not activated, the non-activated hyperedges are traversed, if the proportion of the activated nodes in the hyperedges is larger than or equal to a threshold value, all the nodes in the hyperedges are activated, and the process is repeated until no new hyperedges are activated; s2, initializing a particle swarm; s3, performing two-layer local influence evaluation; s4, particle speed and position updating; s5, local search optimization; and S6, iteration is terminated. By adopting the technical scheme of the invention, the global search capability, the convergence speed and the evaluation precision can be balanced, and efficient and accurate identification of high-influence nodes in the hypergraph is realized, so that the influence propagation effect and the algorithm expandability are improved.

The invention provides an underground coal mine unmanned aerial vehicle dynamic path planning system. Comprising a map construction unit, a risk prediction unit, a cost calculation unit, a path planning unit and an execution feedback unit. The system collects underground environment information through a multi-wire-harness laser radar, a depth camera, a gas sensor and an inertial measurement unit, constructs a real-time three-dimensional voxel map through point cloud splicing and inertial navigation fusion, and records gas concentration, temperature, dust density and SLAM uncertainty; according to the system, gas concentration, temperature, dust density and SLAM uncertainty are continuously recorded by using a three-dimensional voxel map, and a future delta t-oriented space-time risk field is constructed through a gas convection diffusion model, a temperature time sequence model and a micro-seismic threshold model, so that the unmanned aerial vehicle can know a risk evolution trend in advance in a path planning stage; and the problem that in a traditional scheme, a person enters a future high-risk area by mistake while dodging local obstacles is avoided.

The invention provides a ground GNSS differential data correction method and device, and the method comprises the steps: obtaining a pseudo range and a carrier phase, calculating a whole cycle jump value of the carrier phase through employing an M-W combination function and an ionosphere delay change rate, and correcting a carrier phase value through employing the calculated whole cycle jump value of the carrier phase. The influence of station-satellite geometric distance is eliminated, the influence of ionosphere change is considered, and the method is suitable for cycle slip detection in a dynamic environment; according to the method, a self-adaptive threshold model is constructed on the basis of Turbo Edit cycle slip detection in a flight dynamic observation environment, cycle slip misjudgment at a low satellite elevation angle is reduced, and the influence of too low sampling rate on GF combined cycle slip detection performance is weakened to a certain extent; and a technical guarantee is provided for applying the GNSS differential data to the flight test.

The application belongs to the technical field of information technology, and specifically discloses an enterprise financial management auxiliary method and system based on artificial intelligence, wherein the method comprises the following steps: obtaining multi-layer data from transaction records, processing characteristic differences according to hierarchical classification, and obtaining hierarchical financial data sets; analyzing distribution rules by using the hierarchical financial data sets, adjusting a unified threshold range if the distribution shows loose and missed report signs, and determining an optimized threshold group; monitoring strict false report conditions according to the optimized threshold group, fusing time evolution factors by using a self-adaptive threshold algorithm, and obtaining a dynamic threshold model; and obtaining risk degree indexes in the dynamic threshold model; the application aims to solve the problem in the prior art that how to establish different abnormal judgment standards that can be continuously self-adjusted with time and detection effect according to the unique distribution characteristics of different hierarchical financial data and specific business scenarios, so as to become a truly intelligent and efficient enterprise financial risk early warning system.

The invention relates to a deformation double-control early warning index dynamic drawing-up method for informatization integration. The method comprises the following steps: acquiring an early warning grade of an early warning index drawing-up measuring point; constructing measurement value sequence data of the measurement points; carrying out dynamic error estimation by adopting a generalized autoregressive conditional heterovariance model GARCH; judging whether the data sequence of the measuring point is converged or not by adopting a stationarity test ADF; based on data sequence convergence, a static threshold submodel and a dynamic threshold submodel are constructed, and the size of a data window and the moving average change rate of an intra-oral measured value are defined; and calculating different levels of measured value thresholds and measured value change rate thresholds, and drawing up double-control indexes for different levels of early warning of the measuring points. According to the dual-path threshold calculation architecture based on the data convergence state, the convergence state of the measuring point data is judged in real time, and the static threshold model or the dynamic threshold model is respectively started for differential processing, so that the calculation efficiency of the convergence data is ensured, the early warning sensitivity of the non-convergence data is remarkably improved, and the full-period accurate early warning is realized.

The application provides a measurement method for macroscopic and microscopic damage characteristics of early-age concrete under high-temperature tunnel blasting load, relates to the technical field of concrete macroscopic and microscopic damage measurement, and solves the technical problem of early-age concrete damage detection in a high-temperature tunnel blasting environment through multi-scale data feature fusion. Firstly, a cross-scale data synchronization mechanism of a high-temperature resistant sensor network and a microfocus CT is constructed to capture the cooperative damage characteristics of temperature gradient and blasting impact. Secondly, a temperature correction factor and a dynamic damage threshold model are introduced to quantize the coupling effect of high temperature and hydration reaction and improve the criterion adaptability under complex working conditions. Finally, based on the multi-modal feature fusion technology of deep learning, the correlation analysis of macroscopic and microscopic data and the adaptive classification of damage grades are realized, and the detection precision and anti-interference ability are improved. The applicability in a high-temperature scene is improved, and reliable technical support is provided for the whole life cycle health monitoring of concrete structures.

The invention discloses a low-voltage comprehensive distribution box monitoring method and device based on big data, and relates to the technical field of power distribution equipment, and the method comprises the following steps: constructing an operation fingerprint deviation model, and obtaining an operation fingerprint deviation; constructing an abnormal driving intensity model to obtain abnormal driving intensity; constructing a risk cumulative evolution model to obtain a risk cumulant; constructing a self-adaptive risk threshold model to obtain a risk transcendental degree; constructing a disposal demand intensity model, and generating disposal demand intensity; constructing a monitoring strategy adaptive adjustment model, and generating a monitoring adjustment factor; and the device is assembled based on the steps. According to the invention, the multi-source operation information is uniformly mapped into comparable and traceable risk representations by constructing models such as operation fingerprint deviation and the like, and a self-adaptive risk discrimination mechanism is formed in combination with group statistical characteristics, so that the dynamic evolution process of the operation risk of the low-voltage comprehensive distribution box can be accurately reflected, and the reliability of the low-voltage comprehensive distribution box is improved. And the problem of misjudgment caused by a fixed threshold value can be effectively avoided.

The application discloses a noise elimination and transformation model robust estimation method in multi-modal image registration, comprising: obtaining initial matching point pairs, establishing a homography transformation model to be estimated and residual error measurement; performing vector field consistency calculation on the initial matching point pairs, and constructing an edge expansion consensus set; performing initial fitting on the homography transformation model on the edge expansion consensus set to obtain an initial homography transformation model; estimating noise scale and initializing residual error threshold based on residual error statistics induced by the initial homography transformation model; regarding the residual error threshold and the homography transformation model parameters as same-order decision quantities, constructing a joint optimization framework of threshold-model cooperation; adaptively constructing a joint search domain according to regularity statistics of residual error distribution; constructing a threshold-coupled cost function under a mixed statistical modeling framework of inliers / outliers; performing collaborative iterative updating on the homography transformation model and the residual error threshold by adopting a group optimization strategy, and outputting the homography transformation model and a refined inlier set.

The invention belongs to the technical field of GNSS precise point positioning cycle slip detection, and discloses a DI-based adaptive GF cycle slip detection threshold model method, and the method comprises the steps: obtaining IGS tracking station data during an extra-large magnetic storm; preprocessing the data, and calculating an inter-epoch difference GF value and a DI value; obtaining 99.7 percentile in each interval section according to a 3 sigma criterion by using distribution of a time domain difference GF value and a DI value; performing quadratic polynomial fitting on the percentile to obtain a final threshold model; the GF adaptive threshold model based on the DI avoids the limitation that the ROTI is influenced by cycle slip, and effectively solves the problem of false detection.

The application discloses a continuous rigid frame aqueduct cantilever pouring hanging basket stress monitoring and early warning method and system, relates to the technical field of hydraulic engineering construction, and comprises the following steps: arranging stress monitoring points and displacement monitoring points, and constructing a stress-displacement double-system collaborative perception model; carrying out double-channel vibration noise separation on real-time collected stress signals and displacement signals; based on a stress-wet weight nonlinear mapping relationship and a displacement-wet weight geometric inversion model, respectively calculating wet weight values, and verifying the consistency of the wet weight value measurement results through a residual error checking mechanism; constructing a multi-source driven safety threshold model, and generating a dynamic safety threshold in real time; when the monitored wet weight value exceeds the dynamic safety threshold, triggering a hierarchical early warning and executing risk control measures. The stress-displacement double-system independent measurement and cross-checking method, through a vibration frequency domain compensation algorithm, suppresses vibration interference, constructs a multi-source driven safety threshold model, and realizes high-precision monitoring of the stress of the continuous rigid frame aqueduct cantilever pouring hanging basket.

Methods and devices for controlling a storage device including a non-volatile memory including a plurality of sectors including a first sector and a second sector; and a storage controller configured to: read first data from the first sector based on a plurality of threshold models; based on determining that the first data is invalid, read second data corresponding to the second sector based on the plurality of threshold models; based on determining that the second data is valid, update at least one parameter of the plurality of threshold models based on the second data; and generate updated first data corresponding to the first sector based on the plurality of threshold models.

The invention provides a full-process simulation method and system based on a Hall thruster and electronic equipment, and relates to the field of Hall thruster simulation. According to the method, a two-dimensional full-particle numerical simulation model, a sputtering deposition dynamic competition mechanism model, a channel wall surface deposition layer cracking, warping and stripping judgment threshold model and a discharge disturbance induction and evolution simulation model are constructed, so that sputtering, deposition, stripping and discharge disturbance full-process co-simulation is realized; a cross-scale particle microdynamics simulation method, an exfoliation macroscopic motion simulation method and a discharge induction simulation method are incorporated into the same simulation model, so that a complex evolution mechanism in the thruster can be comprehensively reflected. According to the invention, plasma dynamics, wall evolution and discharge disturbance coupling behaviors can be accurately and efficiently simulated in a self-consistent manner, and performance optimization, service life prediction and stability regulation and control of the Hall thruster are supported.

The invention discloses an auditing risk early warning method, device and equipment based on a dynamic AI model and a storage medium, and relates to the technical field of digital auditing, and the method comprises the steps: obtaining historical auditing data and real-time business data, carrying out the dynamic risk recognition of the historical auditing data and the real-time business data based on an industry vertical large model, and obtaining an initial risk score; inputting the initial risk score into an adaptive threshold model to obtain a dynamic risk threshold; performing risk judgment on the real-time service data based on a dynamic risk threshold value, and generating a risk early warning signal; inputting the risk early warning signal into a knowledge graph module for correlation analysis to obtain a risk propagation network; and generating a risk early warning report based on the risk propagation network, uploading the risk early warning report to the group auditing digital comprehensive platform, obtaining auditing risk early warning result data, realizing dynamic risk identification on historical and real-time data at the same time, and automatically updating a risk score along with the change of a service.

The invention discloses an intelligent aquaculture management system and method integrating water quality monitoring, and relates to the field of aquaculture, and the method comprises the steps: building a three-dimensional monitoring network through the deployment of a multi-parameter sensor array, and completing the data cleaning and fusion through edge calculation; in combination with the breeding variety growth feature library and the dynamic threshold model, water quality state evaluation and 24-hour risk prediction are realized by adopting an LSTM-random forest fusion algorithm; an optimization regulation and control instruction is generated based on the three-level strategy library, and the execution equipment cluster is driven to cooperatively work and form closed-loop control; and finally, establishing a full-life-cycle data file, continuously optimizing the model through transfer learning and outputting a management report. The method has the advantages that through three-dimensional accurate monitoring and intelligent model analysis and pre-judgment, the precision and intelligence of breeding management are realized by combining personalized dynamic regulation and control with equipment collaborative linkage, and the effect can be continuously optimized through data iteration.

A method for determining a threshold for entropy coding that skips potential features in image and video coding using a neural network is described. A threshold for estimating a mean of standard deviations based on all latent variables is proposed. For an autoregressive neural network, in order to avoid drifting between context model parameter estimations calculated during training and reasoning, two separate entropy parameter estimation networks, namely an initial entropy estimation network and a refined entropy estimation network, are adopted, the initial entropy estimation network is used for entropy skipping coding and decoding of quantized latent variables, and the refined entropy estimation network is used for entropy skipping coding and decoding of quantized latent variables. A refined entropy estimation network is used for arithmetic encoding and decoding of quantized latent variables. Methods and systems for multi-stage entropy skipping are also presented.

The application relates to the technical field of medical health information, and particularly discloses a heart rehabilitation training control method based on real-time monitoring of muscle strength and electrocardiosignal. The method first calculates real-time electromyographic intensity to quantify muscle load, synchronously analyzes electrocardiosignal and respiratory frequency to obtain cardiopulmonary state, and corrects physiological indexes in combination with trunk movement state. Then, a weighted fusion algorithm is used to generate a multi-parameter fusion result, which comprehensively reflects the correlation between muscle load and heart response. Next, the system judges training intensity through a preset threshold model, generates a voice feedback instruction in real time to adjust the action of a patient, and adaptively updates training parameters to match individual rehabilitation progress. Meanwhile, the system continuously monitors multi-parameter changes, automatically triggers an early warning mechanism when detecting abnormal risks, and ensures training safety. The method overcomes the limitation of traditional schemes relying on a single index, and realizes personalized and dynamic closed-loop control of heart rehabilitation training.

The invention relates to the technical field of carbon monitoring, in particular to a brewing carbon emission abnormity identification system and method, and the system comprises an equipment identification management module, a multi-source sensor network module, a dynamic weighting accounting module, an abnormity detection engine module, a knowledge graph builder module, a GIS visual platform module and the like. According to the method, self-adaptive weight distribution can be carried out based on brewing equipment and a brewing process, accurate mapping of the equipment, the process and carbon emission is realized, so that a basis is provided for multi-source carbon emission anomaly identification, an anomaly detection engine module in which a threshold model and an isolated forest model are matched is adopted, and the threshold model carries out static threshold anomaly identification based on a baseline, so that the probability of abnormal identification is reduced. And the two models are combined to realize a dual-model collaborative anomaly detection architecture, and closed-loop feedback optimization is adopted in the method level to adjust the weight of the brewing equipment according to the early warning condition in actual use, so that the subsequent carbon emission anomaly identification and monitoring sensitivity is improved.

The invention belongs to the technical field of sensors and machine learning, and provides a hydropower station flooding factory building early warning method based on a multi-source monitoring signal and a dynamic threshold, and the main scheme is as follows: constructing a Prophet-XGBoost joint prediction model by using a historical data set; priori knowledge in the water conservancy project field is introduced, multi-source sensing monitoring data and environmental parameters are coupled, and an XGBoost dynamic early warning threshold model is constructed; on the basis of a Prophet-XGBoost combined prediction model, parameters related to the flooded factory building in a future period of time are predicted, and prediction results are obtained; inputting the prediction result into an XGBoost dynamic early warning threshold model, and calculating through the XGBoost dynamic early warning threshold model to obtain an early warning index value under the combined action of multiple parameters; and comparing the prediction result with the early warning index value in real time, and sequentially carrying out flooding risk comprehensive evaluation and risk source positioning according to the comparison result.