45 results about "Incidence matrix" patented technology

The invention relates to the technical field of production process intelligent regulation and control, in particular to an intelligent regulation and control method and system for an instant gelatin production process, and the method specifically comprises the following steps: firstly, collecting process parameters such as reaction kettle temperature, acid-base concentration and the like in a production line sensor and control system to form production batch data; then, an optimization model and an intelligent regulation and control strategy are constructed, a dynamic incidence matrix is constructed through time-varying mutual information entropy, multiple types of features are fused to generate a process state descriptor, a constraint dominating relation and a population initialization strategy are optimized, and a self-adaptive genetic manipulation adjustment mechanism and a state-guided search and constraint processing method are designed; iteratively executing an improved NSGA-II algorithm, obtaining an optimized Pareto solution set, and evaluating diversity; and finally, selecting an optimal solution through a multi-criterion decision and a dynamic regulation and control strategy, and converting the optimal solution into an executable process parameter set value. According to the method, the accuracy and the high efficiency of the production process can be improved by realizing intelligent optimization regulation and control of the process parameters.

The invention provides a multi-domain power grid data collaborative modeling method and system based on a tensor game diagram, and the method comprises the following steps: firstly constructing a six-dimensional enhanced tensor model, constructing a six-order tensor based on the number of nodes, timestamps and other six dimensions, and obtaining a kernel tensor and a factor matrix through CPD-Tucker mixed decomposition; a joint optimization objective function containing reconstruction errors, game equilibrium and privacy risks is constructed, and an optimization kernel tensor and factor matrix is solved; and finally, inputting an optimization result into the MGGCN, processing double targets through a leader branch (a physical topology adjacency matrix guarantees reliability) and a follower branch (a market transaction incidence matrix optimizes economic cost), and generating a collaborative decision result through multi-head attention fusion. According to the method, the problems of low multi-source heterogeneous data fusion efficiency and difficulty in considering cross-domain collaborative privacy security and dynamic optimization are solved, the new energy output prediction error can be reduced, and the data utility loss caused by global encryption is reduced.

The invention relates to the technical field of model interpretability, in particular to an interpretability analysis method and device for an anomaly detection model, equipment and a medium, and the method comprises the steps: training a constructed initial anomaly detection model through employing an obtained model training sample set, and obtaining a target anomaly detection model; inputting the obtained to-be-detected data into a target anomaly detection model for anomaly detection to obtain an anomaly detection result, extracting model parameters after anomaly detection is completed to obtain a time sequence concept feature set, a sequence incidence matrix and a gradient set, performing time sequence decomposition processing on the time sequence concept feature set and a model training sample set to obtain a time sequence concept feature set; performing path analysis processing on the sequence incidence matrix and the gradient set to obtain a global attention contribution matrix, performing interpretability analysis based on the exception driving characteristics and the global attention contribution matrix, and generating an exception interpretation report based on obtained exception interpretation data and an exception detection result. And the transparency of the abnormal decision logic is improved.

The invention discloses a water supply network safety monitoring system based on a hydraulic model, and belongs to the technical field of water supply network monitoring. The Caputo fractional order operator introduces historical memory characteristics through an integral term, explicit modeling is carried out on disturbance such as user water consumption fluctuation and sensor noise, and a clear disturbance compensation object can be provided for subsequent multi-source data fusion verification and self-adaptive correction; the pipeline-node coupling relation is embedded into the model through the topological incidence matrix, and structured input can be provided for a subsequent topological consistency verification algorithm; in combination with historical memory characteristics of a Caputo operator, the observer has robust estimation capability on slowly changing leakage disturbance and sudden water impact; the time-varying scaling item decays along with the time index, the gain of the observer is dynamically reduced, and quantization noise and electromagnetic interference of the SCADA sensor can be effectively filtered out in the steady-state stage; and through the positive error convergence index and the negative error convergence index, respectively optimizing the convergence speeds of the positive and negative pressure fluctuations.

The invention provides an industrial network security situation prediction method and system based on a generative large model, and relates to the technical field of industrial network security, and the method comprises the steps: obtaining time sequence network behavior data of a plurality of monitoring nodes in an industrial network, and extracting multi-dimensional features to obtain a security feature vector; decomposing the feature vector into a periodic baseline component and a transient disturbance component through frequency domain transformation, and obtaining a decomposition situation feature through sparsity constraint screening; performing semantic space mapping by utilizing a generative large model to obtain semantic enhanced situation representation, calculating security anomaly correlation between nodes based on decomposed situation characteristics, and constructing a dynamic incidence matrix; and constructing a propagation operator based on the dynamic incidence matrix, carrying out multi-step iterative propagation on the semantic enhancement situation representation, introducing an attenuation factor to simulate abnormal influence diffusion, and obtaining a security situation prediction result in a future time window.

The invention relates to the technical field of data processing, and provides a feature selection-based multi-modal data classification method, which comprises the following steps of: obtaining original data of at least two heterogeneous modals and corresponding initial feature sets; performing intra-modal selection on each modal, constructing a feature association graph based on a graph theory, and screening a core feature subset meeting a threshold requirement through mutual information; constructing a cross-modal feature incidence matrix, realizing inter-modal fusion based on a weighted graph model and condition mutual information, and screening a cross-modal key feature set; and inputting the key features into a classification model to train a multi-modal classifier, and repeating a feature selection process on to-be-classified data to complete classification prediction. According to the method, the optimal threshold value is adaptively determined through innovative fusion of double-stage feature selection, the graph theory and the information theory, the classification accuracy and the data processing efficiency are remarkably improved, and the method can be widely applied to the fields of medical image and pathological report combined diagnosis, automatic driving data fusion, internet multimedia understanding and the like.

The invention relates to the technical field of artificial intelligence content generation and meta-universe scene generation, and discloses a meta-universe scene dynamic generation method based on AIGC cross-media, and the method comprises the steps: decomposing a source memory scene feature into a style dimension feature vector and a content dimension feature vector through a memory semantic decoupling analyzer; the method comprises the following steps: establishing a cross-scene context style mapping relation by using a memory semantic projection matrix, identifying style sensitive anchor point nodes and generating an anchor point-medium incidence matrix, ensuring cross-medium style coordination through a style consistency constraint propagation algorithm, and generating scene content comprising a 3D model, a texture image and audio by using a cross-medium AIGC generator. The technical problems that style features are coupled with scene contexts, cross-context styles are not coordinated, and cross-medium conversion degrees are not consistent are solved, and cross-context and cross-medium migration of personalized memory styles is achieved.

The invention relates to a material processing method and system based on sensitive content detection and a medium, and belongs to the technical field of multimedia content security. The material processing method comprises the following steps: receiving an input original material, and analyzing the input original material into an image stream, a video stream and a text stream; executing HSV color space conversion, and extracting texture features; performing semantic analysis on the text stream to generate a word segmentation sequence and a named entity recognition result; random frequency noise is injected based on the saturation channel, and local binary pattern features are extracted from the saturation channel after noise injection; generating a semantic vector according to the word segmentation sequence and a corresponding named entity recognition result, and calculating a risk entropy value; and establishing a space-time mapping relationship between the image stream / video stream and the text stream, constructing a cross-modal incidence matrix, executing a grading decision according to an output result, executing a material processing operation, and feeding back a processing result to the cross-modal incidence matrix for weight updating. The sensitive content identification accuracy and timeliness can be improved.

The invention belongs to the technical field of breeding planning, and particularly relates to a large model construction method and system for multi-source data collaboration pear precision breeding planning, and the method comprises the steps: obtaining a multi-source collaboration data set, constructing a five-dimensional incidence matrix, and building a unified correlation basis of multi-dimensional data; constructing a constrained knowledge expression structure and training a constraint consistency reasoning model on the basis, analyzing a user breeding instruction in a constraint driving mode, and generating a combined constraint condition set containing a gene feasible region and a multi-phenotype collaborative target; calling germplasm resources under a cross-domain feature alignment and feature-level desensitization framework, executing phenotype-genotype collaborative screening, parent dynamic matching and multi-generation genetic evolution deduction, and generating breeding planning simulation data; and finally, outputting an optimal breeding scheme through a multi-dimensional evaluation system, and continuously updating the model by utilizing verification data. According to the invention, unification of multi-phenotype balanced synergistic improvement and stable gene transfer is realized, and the breeding efficiency and variety adaptability are remarkably improved.

The invention provides a transformer health state assessment method and system, and the method comprises the steps: obtaining the multi-source operation data of a target transformer, generating a multi-criterion discrimination value sequence, carrying out the deep feature obtaining of the operation features of the transformer obtained by the sequence, and obtaining the operation deep features of the transformer, obtaining a criterion weight sequence under the weight transformation matrix and the bias term according to the transformer operation deep characteristics; generating a health state for maintenance according to the multi-criterion discriminant value sequence and the criterion weight sequence; the deep feature acquisition step comprises the following steps: performing multi-head self-attention processing according to transformer operation features to obtain an operation feature incidence matrix, and obtaining preliminary transformer deep features according to the incidence matrix and the transformer operation features; and repeatedly executing the deep feature acquisition step according to the preliminary transformer deep features until the execution times are not lower than the coding level, and taking the preliminary transformer deep features as transformer operation deep features. According to the technical scheme, the stability of transformer health state evaluation can be improved.

The invention relates to the technical field of numerical control, in particular to a real-time data driven cutting bed automatic control method and system and a storage medium. The method comprises the following steps: collecting dynamic stress data of a cutting bed cutter and microscopic deformation data of a fabric in real time through a multi-source sensing network, and constructing a physical coupling state space of the cutter and the fabric; according to the invention, through a collaborative optimization mechanism driven by multi-source sensing data, high-precision collaborative control of cutter dynamics and fabric deformation is realized in the field of cutting bed control, and through hardware synchronous triggering of the piezoelectric ceramic array sensor and the distributed optical fiber sensor, space-time alignment monitoring of cutter stress fluctuation and fabric interlayer shear deformation is realized. According to the design, the problem of time sequence mismatch of a traditional system in cross-scale data fusion is solved, the thermodynamic state of the cutter and fabric strain energy distribution form a computable incidence matrix, and a solvable space foundation of an optimization algorithm is laid.

The invention provides a science popularization content intelligent recommendation method and system based on deep learning. The method comprises the following steps: acquiring science popularization image / video data, performing space-time decoupling processing to generate a multi-scale space-time feature tensor, and constructing multi-level feature representation of a local scene, a dynamic process and an overall context; utilizing a multi-head cross-modal attention network to calculate a semantic association degree matrix of the features and knowledge graph entities; a time sequence causal chain extraction module is introduced to generate a causal path constraint vector, a dynamic semantic alignment constraint is formed through matching with a knowledge graph relation path, and a semantic incidence matrix is optimized; and generating a uniform space-time semantic fusion vector by adopting an adaptive mapping decoder, and outputting a structured semantic tag set through a deep semantic classifier. According to the method, the limitation of traditional shallow recommendation is broken through, understanding of deep semantics and causal logic of scientific contents is realized, and the accuracy, semantic depth and interpretability of science popularization content recommendation are remarkably improved.

The invention relates to a three-dimensional reconstruction method based on Gaussian point information redistribution and geometric structure constraint. The method comprises the following steps: acquiring target scene image data, and extracting a visual Gaussian point attribute parameter set based on a 3D Gaussian sputtering framework; randomly rejecting part of Gaussian points, determining a neighborhood point set of the Gaussian points, and constructing an approximate model to obtain a gradient item set; constructing an incidence matrix and calculating an information distribution coefficient, and compensating the opacity and color information of the rejected points to neighborhood points to update a parameter set; constructing a total loss function containing basic reconstruction, local direction consistency and main direction alignment loss, and iteratively optimizing parameters; and dynamically adjusting the Gaussian point shielding rate and modifying opacity associated parameters, maintaining physical constraints, and repeating information redistribution until training is completed, thereby realizing high-quality three-dimensional reconstruction. By adopting the method, the definition of the geometric structure can be improved.

The invention discloses an early language ability evaluation and improvement method and system based on deep learning, relates to the technical field of child language development evaluation, and proposes the following scheme: obtaining discontinuous voice stream data of a child, carrying out endpoint detection through a pre-trained deep neural network model, and segmenting to generate voice segments, and recognizing each voice segment and incorporating the voice segments into different types of grammar slots, and combining the voice segments in the same grammar slots into a competition node set. According to the method, the incidence matrix is constructed by taking the co-occurrence probability value between the competition node sets of the adjacent syntax slots as the element, the competition node combination with the co-occurrence probability value greater than the preset co-occurrence threshold value is screened, and the initial logic path is generated through the Cartesian product operation, so that the problem of path stiffness caused by a fixed rule is solved; the probability-driven dynamic path generation is realized, so that the language logic path is closer to the actual language use probability and habit of children, and the authenticity and individuation level of evaluation are improved.

The application discloses a hydraulic pipe network simulation system, and relates to the technical field of pipe network simulation, which comprises the following steps: establishing a connection relation graph of nodes and pipe sections; determining a peak array, and deriving an augmented incidence matrix, a basic incidence matrix, a basic loop matrix, a tree branch matrix, a residual branch matrix, a tree branch vector and a residual branch pipe section vector; inputting corresponding parameters; adopting a basic loop method optimized by a magnetotactic bacteria method to perform first layer iteration, and calculating a residual branch pipe section flow increment iteration step; adopting an LM algorithm, and combining with the magnetotactic bacteria method to perform second layer iteration and calculate an end point flow iteration step; adopting an LM algorithm modified by a trust region, and combining with the magnetotactic bacteria method to perform third layer iteration and calculate a controlled variable iteration step; and calculating and outputting hydraulic properties of each node and pipe section. The application solves the problem that the existing iteration algorithm causes singular values due to too many control points or numerical solutions cannot converge due to improper initial value setting.

The invention relates to the technical field of boiler water treatment, and particularly discloses a boiler feed water pH value water treatment method and system. Dynamic fluctuation data and static ion component data of water quality are synchronously collected; performing time sequence decomposition and normalization on the dynamic data to form a dynamic feature sequence, and performing normalization on the static data to form a static feature vector; mining an implicit coupling relationship between the two through an iterative optimization algorithm, and constructing a dynamic incidence matrix for quantifying the influence of each component on the water quality stability; on the basis of the matrix and a target pH interval, simulating a plurality of regulation and control paths in the state transition network and reversely generating an operation variable correction instruction set; finally, an execution mechanism is driven through asynchronous analysis and signal mapping, closed-loop optimization is achieved based on feedback, accurate recognition and advanced regulation and control of buffer components are achieved, and complete resource utilization of complex water sources is supported while it is guaranteed that pH stably reaches the standard.

The invention discloses a few-sample graph classification method based on orbital enhancement and residual prompt, relates to the technical field of artificial intelligence and graph deep learning, and is suitable for graph learning tasks such as molecular property prediction and drug screening. According to the method, node structure roles are obtained through primitive orbit statistics, an orbit incidence matrix is constructed, random walk is executed in an orbit topological space, and an orbit enhancement view is generated to extract high-order topological semantics; a learnable global prototype is introduced in a pre-training stage, and the prototype allocation probability of an original view and a track enhanced view is subjected to consistency constraint, so that a graph encoder which is more robust to a structural mode is learned; in the downstream stage, a small number of category samples are aggregated to form orbital anchor representation, probability prompt representation is generated through parameterization of the anchor representation, a graph prompt vector is obtained in a differentiable mode, and weighted aggregation and classification prediction of node-level representation are guided. According to the invention, the method can adapt to the distribution difference in the category, and improves the generalization performance and robustness under the condition of few samples.

The invention provides an abnormal event analysis monitoring method and system, and belongs to the technical field of informatization supervision, and the method comprises the steps: integrating multi-source data, and constructing a cross-domain incidence matrix; service rules are extracted from the cross-domain incidence matrix, an analysis model is constructed and trained according to the extracted service rules, and the analysis model comprises a logistic regression model and a neural network model; according to the linear separability of the to-be-analyzed data, calling the corresponding logistic regression model or neural network model, analyzing, studying and judging the business operation behaviors in the to-be-analyzed data, and monitoring abnormal behavior events and contradictory behavior events. The method has the beneficial effects that a mode of combining logistic regression and a neural network is adopted, all research and judgment tasks are prevented from being processed by the neural network, a logistic regression model is adopted for simple tasks, and the analysis efficiency is improved; according to the method, contradictory behavior recognition is combined with the rule engine, not only can behaviors which do not conform to rules be recognized, but also contradictory behaviors can be effectively recognized, and the recognition rate of abnormal behaviors is increased.

The invention belongs to the field of graph data processing, and discloses a multi-modal graph representation learning method and system based on a dynamic hyperbolic hypergraph Transform model.The method comprises the steps that multi-modal data are obtained and preprocessed, and multi-modal graph data are obtained; generating hyperedges in a self-adaptive manner on the basis of the multi-modal graph data, and constructing a sparse incidence matrix between nodes of the graph data and the hyperedges; nodes and hyperedges of graph data are mapped to a hyperbolic geometric space, and bidirectional feature propagation and aggregation between the nodes and the hyperedges are executed according to a sparse incidence matrix to update node representation; performing global dependency modeling on the node representation based on the updated node representation to obtain a fused node representation; and based on a category weighted loss function and a multi-label threshold search strategy, performing classification prediction on the fusion node representation, and completing a graph representation learning task. According to the method, the problems of insufficient high-order relation modeling, hierarchical structure representation distortion, high calculation complexity, poor category imbalance adaptability and the like of a graph representation learning method in the prior art are solved.

The invention relates to the technical field of network security and artificial intelligence data processing, and discloses an AI call-out detection method and system based on multi-modal fusion and dynamic evolvement, and the method comprises the steps: constructing multi-source data containing call behaviors, audios and semantics; extracting depth behavior features representing the mechanical features; calculating fusion features by using a cross-modal attention mechanism of an embedded incidence matrix; inputting the decision model to output scores and intercepting early warning; and updating the model by using an elastic weight solidification algorithm in combination with an adversarial sample based on the feature distribution difference. The system comprises a multi-source data construction module, a microcosmic interaction modeling module, a cross-modal fusion module, a decision and execution module and a double-closed-loop dynamic evolution module. According to the method, the problems of feature splitting and model aging are solved by injecting modal association and fusing deep microscopic behavior features and a dynamic evolution mechanism, long-acting active defense is realized, and the recognition accuracy of variant AI outbound is improved.

The invention discloses an expressway multi-interval flow control method based on reinforcement learning, and relates to the technical field of expressway flow control, and the method comprises the steps: constructing a local state vector of each interval according to traffic state data; constructing an interval incidence matrix according to the highway road network topological structure, wherein the interval incidence matrix is used for quantifying the traffic association strength between the intervals; fusing the local state vector of each interval based on matrix multiplication according to the interval incidence matrix to obtain a global state vector; the global state vector is input into the trained reinforcement learning model, an optimal control action combination is output, the optimal control action combination is used for forming a control instruction, and the control instruction is used for controlling the highway network. The method can adapt to dynamic fluctuation of traffic flow, has a global view angle, and can perform adaptive optimization according to real-time road conditions.

The invention discloses a production and environment data fusion and cooperative control method and system, and relates to the field of industrial informatization and intelligent manufacturing. According to the method, production execution data and environment monitoring data are collected, and a unified data structure is constructed; calculating a spatial correlation degree according to a workshop topological structure, determining a monitoring point set and a weight coefficient, and calculating a representative environment parameter value based on a time window; fusing the process parameters and the environment parameters to form a data set; environment influence factors and process response factors are extracted, an incidence matrix is constructed, a constraint function is established, a process parameter value range is determined, and a constraint rule is generated; environmental parameters are monitored in real time, constraint rules are triggered when threshold values are exceeded, adjustment values are calculated, a scheme is generated, and automatic parameter adjustment is achieved. According to the method, the environmental factors are used as constraint conditions for production scheduling and process parameter setting, and the production efficiency is optimized on the premise of environment compliance.

The invention provides a thermal power generating unit safety interlocking method and system based on protection logic, and relates to the technical field of thermal power generation control, and the method comprises the steps: collecting operation parameters, analyzing a time sequence change rule, and constructing a protection logic incidence matrix; the association strength is converted into a distance relation, and a multi-level interlocking protection structure is constructed through adaptive clustering; tracking a parameter change transmission path, and selecting a control strategy based on risk assessment; and adjusting a control period through the adaptive factor, and determining an execution time sequence. According to the method, the accuracy, the real-time performance and the adaptability of safety interlocking control of the thermal power generating unit can be improved.

The invention provides a foggy day target detection method based on dynamic dilated convolution and cross-dimensional attention, relates to the crossing field of computer vision and artificial intelligence, and designs three parallel multi-scale dilated convolution to capture multi-scale features of near, middle and far fog clusters, design scale perception attention weights for different scales, and improve the detection accuracy of the foggy day target. And then the multi-scale features are spliced by adopting jump connection to generate a defogged clear image, so that the defogging effect is effectively improved, and the detail features of the target are reserved. Secondly, performing scale division of large targets and small targets on the input features, dividing channels into information branches and redundant branches, generating a space-channel incidence matrix, and completing bidirectional weighted optimization of space enhancement and channel compression at the same time; and finally, lightweight operation is carried out on information and redundant branches of the channels by adopting a depth separable convolution machine and sparse 11 convolution, and dual-channel features are fused to generate enhanced features, so that the detection precision of the unmanned aerial vehicle on small targets in foggy days is improved.

The invention discloses a sample set construction method based on a negative sample selection strategy and an lncRNA-miRNA association prediction method, and belongs to the technical field of biological information. In order to solve the problem that high-quality and high-credibility negative sample data cannot be properly used in an existing method, the method comprises the following steps: respectively obtaining sequence similarity matrixes by utilizing respective sequence similarities on the basis of sequence information of lncRNA and miRNA; based on the lncRNA-miRNA incidence matrix, respectively constructing meta-paths for lncRNA and miRNA, and respectively utilizing meta-path similarity to obtain a standardized meta-path similarity matrix; accumulative similarity scores are respectively determined based on the sequence similarity matrix and the meta-path similarity matrix of the positive samples and the candidate negative samples, then the total similarity score is obtained, the negative samples are determined, and a sample set is constructed. According to the method, features of sample data are extracted based on a sample set, a graph convolutional network and a hypergraph convolutional network are combined, cooperative capture of local and global association features is realized, and finally the accuracy and reliability of lncRNA-miRNA association relationship prediction are improved.

The invention discloses a marine traffic network mining method, which belongs to the field of marine traffic, and comprises the following steps: S1, obtaining a standardized feature matrix, a ship trajectory set and a meteorological interference coefficient; s2, based on the standardized feature matrix, AIS data, ship static parameters and wind power plant data, fusing ship 3D security domain parameters and adaptive graph learning, and constructing a maritime traffic dynamic graph of 3D space constraint; s3, based on the maritime traffic dynamic graph, fusing Patch-Transform time sequence coding and information geometry multi-target coupling to obtain a global time-space incidence matrix and a key incidence path; and S4, based on the feedback of the key node and the path, correcting the global space-time incidence matrix through iterative optimization. By adopting the marine traffic network mining method, global space-time association can be accurately captured, key association paths can be efficiently identified, and the dynamic adaptability and decision support value of network mining are improved.

The invention discloses a CT bulb tube fault early warning method and system capable of monitoring multi-dimensional physical parameters in real time, and relates to the field of CT bulb tube fault early warning, and the method comprises the steps: determining the multi-dimensional physical parameters of a CT bulb tube, collecting the parameter data, and carrying out the preprocessing of the data; based on fault tree analysis, constructing a CT bulb tube fault and characteristic parameter incidence matrix; based on the bidirectional LSTM network, constructing a CT bulb tube fault early warning model, and outputting the occurrence probability of each fault mode in a future period of time; setting a model updating mechanism, and dynamically adjusting and optimizing model parameters; and setting a fault probability threshold value, and judging whether to carry out fault early warning or not according to a comparison result of the probability and the threshold value. The CT bulb tube fault early warning method has the advantages that through combination of multi-dimensional physical parameter fusion, bidirectional LSTM dynamic prediction and fault tree analysis, accurate and adaptive early warning of the CT bulb tube fault is realized, and the equipment reliability and the maintenance efficiency are remarkably improved.

The invention discloses a bridge jacking force-displacement dynamic prediction and counterweight optimization method based on CNN-LSTM and edge calculation, and belongs to the field of intelligent bridge engineering monitoring and structure control, and the method comprises the steps: S1, integrating multi-dimensional sensing data, constructing a space-time incidence matrix, extracting force-displacement characteristics such as curvature, gradient change rate, hysteresis effect parameters, and the like, and carrying out the optimization of the force-displacement characteristics; multi-scale convolution sampling is carried out to capture abrupt change and stationary features; s2, constructing a CNN-LSTM hybrid model, extracting local features by CNN, capturing time sequence dependence by LSTM, introducing a moment balance constraint Fc * R = Mz + Mun into a loss function, outputting a prediction confidence interval through Dropout, and fusing multi-branch features to form a final feature vector; s3, deploying an edge-cloud cooperative system, realizing real-time reasoning and counterweight optimization by an edge end, and executing incremental training by a cloud end to update a model weight; and S4, calculating the balance weight adjustment amount according to the predicted critical jacking force, and realizing closed-loop control of hoisting operation by linkage with BIM visualization. The method has the beneficial effects that the construction safety and the real-time decision-making efficiency are remarkably improved.

The invention discloses a property project operation cost measuring and calculating method, which comprises the following steps of dividing a property project to be received to obtain a plurality of functional areas; obtaining historical operation data of the functional areas, and evaluating a cost value and an income value of each functional area based on the historical operation data; constructing a sub-block matrix of each functional area based on the cost value and the profit value, and constructing an initial cost matrix according to the sub-block matrixes; the influence degree between every two sub-block matrixes is evaluated, an incidence matrix is constructed, the incidence matrix is supplemented into the initial cost matrix, and a final cost matrix is obtained; accounting the operation cost of the property project to be received according to the final cost matrix, and judging whether to receive the property project based on the operation cost; after the incidence matrix is determined by introducing mutual influence among different functional areas, the incidence matrix is combined with the sub-block matrix to form the final cost matrix, and the operation cost of the property project to be received can be accurately calculated through the final cost matrix so as to guide a property company to judge whether the project is received or not.

The invention relates to an image feature transmission method, device and system, and relates to the technical field of communication. The transmission method includes the following steps: extracting a feature matrix of a to-be-processed image for each channel by using a machine learning model; determining one or more incidence matrix pairs according to a comparison result of the correlation degree between the feature matrixes and a first threshold value; according to the information amount, determining a representation matrix and a represented matrix in the two feature matrixes of each incidence matrix pair; determining a corresponding relation between each representation matrix and each represented matrix; and carrying out quantization processing and coding processing on each representation matrix, the corresponding relation and the maximum characteristic value and the minimum characteristic value in each represented matrix, and then transmitting to a decoding end.