64 results about "Basic probability" patented technology

The invention discloses an uncertainty mapping method and device based on selective learnable depreciation, and the method comprises the steps: S1, collecting color images and depth maps at different time points, and projecting the color images and the depth maps to a three-dimensional coordinate system; s2, predicting an evidence vector for each pixel by using an evidence generation network; s3, calculating total evidence intensity based on the evidence vector, and defining credibility and uncertainty based on the total evidence intensity; s4, constructing a noise mask to represent the observation quality, and predicting a damage coefficient by using a selective damage network; s5, the evidence intensity is adjusted in a zooming and depreciation mode; and S6, outputting a final probability, carrying out basic probability distribution of multi-frame fusion observation on the three-dimensional voxel grid, and constructing a semantic map containing uncertainty estimation. According to the method, the conflict rate and uncertainty in the semantic map fusion process can be effectively reduced, and the accuracy and credibility of the map are remarkably improved.

The invention discloses a fire detection method and device based on multi-modal perception and D-S evidence theory fusion. The method comprises the following steps: collecting multi-modal perception big data at least comprising video image data and temperature sensing data in a monitoring area; performing fire visual feature analysis on the video image data to generate first basic probability distribution; performing fire temperature characteristic analysis on the temperature sensing data to generate second basic probability distribution; taking the first basic probability distribution and the second basic probability distribution as two independent evidence sources, and fusing by adopting a D-S evidence theory to obtain a fused third basic probability distribution; converting the third basic probability distribution into a fire occurrence probability for decision making; and when the fire occurrence probability exceeds a preset alarm threshold, determining that a fire occurs and triggering an alarm. According to the fire detection method, multi-modal sensing big data are synchronously collected and fused, multi-source uncertain information is processed and decided under the framework of the D-S evidence theory, and the early stage of fire detection is remarkably improved.

The invention relates to a multi-modal data fusion method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: respectively carrying out feature extraction on image input data from at least two modes to obtain an image feature vector of each mode; on the basis of the image feature vector of each mode, Dirichlet distribution parameters corresponding to each mode are generated, the distribution difference between the modes is evaluated, and the Dirichlet distribution parameters are used for representing the evidence intensity of each category in the classification or clustering task; according to the distribution difference, fusing the Dirichlet distribution parameters of each mode to obtain a basic probability distribution result after fusion; and dynamically adjusting the basic probability distribution result by using a Kalman filter, and outputting a final classification or clustering result. By adopting the method, the fusion weight of each modal data can be dynamically adjusted, and the accuracy and reliability of a multi-modal fusion system are improved.

The invention provides a TLF-GV signal correlation earthquake three-element joint probability prediction method and system, belongs to the technical field of earthquake prediction, and aims to solve the problems that traditional three-element prediction lacks joint fusion, evidence conflicts are difficult to process, and weights are fixed. Comprising the following steps: extracting physical features of TLF-GV signals, carrying out preorder three-element prediction results and model credibility parameters, splicing the physical features into feature vectors, inputting the feature vectors into a dynamic weight prediction model constructed by a shallow MLP, and outputting time, magnitude and position weights; dividing the three elements into 3 * 3 * 3 = 27'time-magnitude-position 'propositions, generating basic probability assignment (BPA) of each proposition based on weights, and quantifying evidence contradictions through a conflict coefficient K; and adopting a Dempster combination rule to obtain a joint BPA, calculating a mean value of a reliability function and a likelihood function of each proposition as a comprehensive probability, and extracting a maximum probability proposition as a coupling result. According to the method, the scientificity and practicability of strong earthquake short-term and temporary prediction are remarkably improved.

The invention relates to the technical field of generalized quantum evidence theory and image processing, in particular to a ship image data target identification method based on the generalized quantum evidence theory, which comprises the following steps: converting synthetic aperture radar ship image data into complex field synthetic aperture radar data, and constructing training data; generating a composite quantum evidence label according to the potential structure; training the quantum evidence neural network to obtain a standard quantum evidence neural network; inputting to-be-recognized complex field synthetic aperture radar data into the standard quantum evidence neural network, and extracting a complex field feature vector of a to-be-recognized sample; normalizing the complex field feature vector into a generalized quantum basic probability amplitude function, and defining a proposition space under a quantum identification framework; and calculating generalized quantum basic probability distribution of the proposition in the proposition space, and executing double judgment to obtain a recognition result of the original synthetic aperture radar ship image data to be recognized. According to the method, the accuracy of ship image data target identification based on the generalized quantum evidence theory can be improved.

The invention discloses an interval generation and probability correction method and system for new energy power prediction, and belongs to the technical field of power system operation and control, and the method comprises the steps: generating a multi-dimensional environment feature code through numerical weather forecast data, and calling a basic point prediction model to obtain a point prediction result and basic probability distribution; inputting the multi-dimensional environment feature code and the point prediction result into a dual-channel dynamic interval generator to form a preliminary prediction interval, calculating a short-term error sequence based on the actual power of new energy power generation and historical prediction data, analyzing the trend characteristics of the short-term error sequence, and correcting the basic probability distribution according to the trend characteristics; and extracting a probability verification signal from the corrected probability distribution, feeding back the probability verification signal to an interval generator, carrying out optimization adjustment on the preliminary prediction interval, and outputting an optimized prediction interval. According to the technical scheme, the preliminary interval is generated by adopting a dual-channel mechanism, and feedback correction and closed-loop optimization are performed in combination with the short-term error trend, so that the adaptive capacity, accuracy and reliability of the prediction interval can be improved.

The invention relates to the technical field of product reliability evaluation, in particular to a product reliability evaluation method and system based on multi-information measurement fusion. The method comprises the following steps: acquiring an unknown parameter of life distribution adapted to an evidence source, generating a parameter interval for the unknown parameter, and generating basic probability distribution associated with the parameter interval based on the evidence source; calculating correlation coefficients between the evidence sources based on the basic probability distribution, and generating importance weights according to a plurality of correlation coefficients of the evidence sources and other evidence sources; calculating the information entropy of the evidence source, and obtaining a credibility weight through normalization based on the information entropy; carrying out normalization on an operation result of the importance weight and the credibility weight to obtain an evidence weight; taking the evidence weight as a weight coefficient of basic probability distribution, and obtaining comprehensive probability distribution through evidence fusion; and carrying out parameter estimation on unknown parameters based on comprehensive probability distribution, and substituting an estimation result into a reliability function to carry out reliability evaluation, so that reliability evaluation under fault-free data is realized.

The invention relates to a financial equipment security authentication method, system and device and a medium, and the method comprises the steps: carrying out the feature extraction and matching of user data collected in multiple modes, and generating a matching score and a quality score of each mode; uncertainty quantification and evidence conversion are carried out based on scores and scores, and a basic probability distribution function is constructed; carrying out dynamic weighting on probability distribution in combination with a sensor state and historical performance, and generating a preprocessing evidence set; through conflict adaptive evidence fusion processing, calculating conflict quality between evidences and redistributing conflict values according to a weight ratio, and generating a global fusion evidence body; and finally, the authentication confidence is calculated according to the fusion result, and a binary authentication instruction is generated through comparison with a dynamic risk threshold. The method can effectively solve the problem of evidence conflict in multi-modal authentication, dynamically adapts to system state change, and remarkably improves the accuracy and reliability of financial equipment security authentication.

The invention discloses a URL (Uniform Resource Locator) security analysis method and device for a browser-side large language model and a medium, and relates to a network information security technology, the method comprises the following steps: determining a session-level security contract based on terminal performance and user preference; page static scripts, dynamic behaviors, network requests and text information are collected and packaged into a unified evidence unit, and an attack causal graph is constructed; based on the risk ontology and the rule atom library, mapping the evidence into basic probability distribution for the hypothesis set, and generating evidence constraints; under the constraint of a session contract, performing multiple rounds of attack and defense games on a causal graph through a defense agent and an attack simulation agent which are realized by an end-side large language model; calculating a session risk value through a risk quantification model on the basis of a game result in combination with scores fused by a D-S evidence theory and subjected to multi-prompt LLM conjunctive calibration, and making a decision according to a self-adaptive threshold value; and high-confidence, explainable and sustainable-evolution end-side URL security analysis in an off-network environment is realized.

The application provides a kind of management method and system for scenic spot multi-source heterogeneous data, belong to data management technical field;The application obtains comprehensive credibility by real-time calculation of the accuracy, real-time performance, coverage and stability index of each data source, weighted fusion;The measured value is converted into the basic probability assignment of evidence theory, and the comprehensive credibility is used for discount correction;The Jousselme distance between evidences is calculated as the conflict degree, and the fusion is carried out by using Dempster combination rule, credibility weighted average and deep conflict resolution according to low, medium and high levels;The fusion result is verified by multi-source cross, time sequence consistency and physical constraint, and the quality evaluation parameter, conflict threshold and fusion strategy are dynamically adjusted according to the deviation to form a closed loop feedback;The application significantly improves the accuracy, robustness and adaptive ability of multi-source heterogeneous data fusion, and is suitable for scenic spot passenger flow statistics, congestion warning and other scenes.

The invention discloses a method for measuring the difference degree between a plurality of evidence theory models, the method comprises a plurality of evidence models, a complex value distribution model, a probability distribution model and a calculation KL divergence, the plurality of evidence models comprise a plurality of basic probability distributions, the complex value distribution model comprises a complex value probability distribution, and the probability distribution model comprises a probability distribution model. The probability distribution model comprises probability distribution, and the calculation of the KL divergence mainly comprises the calculation of the KL divergence of RBetM1 and RBet2 and the calculation of the KL divergence of RBetM12 and RBet1. Through the provided mapping method, the complex evidence theory model can be mapped to a single distribution model in a real number field, so that understanding and analysis of the complex evidence theory model in the real number field are facilitated; by calculating the KL divergence in a real number field, the model can measure the difference degree between two complex evidence theory models, and the difference degree can be further practically applied.

Systems and methods are provided for implementing an adaptive threshold for user absence prediction for power state transitions. A controller of a device monitors a user presence and / or absence of a user, in some cases, correlated with time, and determines whether the user presence and / or absence at the correlated time is consistent with a power management model that is based on a time-based probability of user absence relative to a dynamic probability threshold. The dynamic probability threshold is a function (e.g., a sum) of a base probability threshold corresponding to device-specific time-usage characteristics and a heuristic probability threshold corresponding to user-specific time-usage characteristics. When inconsistent, the controller adjusts the dynamic probability threshold to account for the user presence and / or absence at the correlated time. The controller updates the power management model based on the dynamic probability threshold, and sets a sleep or wake state of the device accordingly.

The application discloses a kind of Internet of Things multi-source data fusion method, device, equipment, medium and product, it is related to Internet of Things technical field.The method is first constructed dynamic object model, and for each common equipment attribute, respectively for each brand equipment configuration corresponding data acquisition and pre-processing scheme, then for target common equipment attribute, using corresponding processing scheme is received Internet of Things multi-source data is pre-processed, obtain with uniform format and consistent on time dimension and corresponding multiple parameter values, then for each brand equipment, according to corresponding multiple data credibility influence factors, determine the credibility of the corresponding equipment about target attribute upload data, and calculate the corresponding equipment according to this correction basic probability distribution under D-S evidence theory, finally according to the calculation result, application D-S evidence theory is fused and handled, so it can realize the efficient integration of Internet of Things multi-source data and the accurate presentation of physical equipment.

The invention relates to the technical field of automobile electronic safety, in particular to a vehicle-mounted abnormity comprehensive detection method and system based on evidence deep learning. The evidence deep learning model in the invention comprises four layers of core structures. The feature input layer receives information entropy of a vehicle-mounted CAN message sequence; model parameters are built in the evidence reasoning layer, and input features are mapped into basic probability distribution corresponding to a plurality of categories; the uncertainty calculation layer quantifies uncertainty measurement in the classification process; and the decision output layer fuses the basic probability distribution and the uncertainty measurement result and outputs a classification result. In the training stage, a loss function formed by a multilayer perceptron evidence function and a product of a preset coefficient and uncertainty measurement is adopted, and in the verification stage, on the basis of the loss function in the training stage, an optimal annealing factor is added for weight adjustment. The optimal annealing factor is determined through correct classification on a maximized verification set and the proportion of samples with the uncertainty lower than an annealing threshold value, and the classification precision is effectively improved.

The invention discloses a rotating machinery fault feature extraction and fault diagnosis method based on feature fusion. A bearing vibration signal collected by a sensor is subjected to wavelet packet transform (WPT) to obtain time-frequency domain features. And for time-frequency domain features obtained by different sensors, training corresponding probabilistic SVM multi-classifiers as basic probability distribution functions. And finally, fusing the basic probability assignment functions BPA of the multiple sensors through a D-S evidence theory algorithm to obtain a final diagnosis result. According to the method, the defect that uncertain information processing is insufficient in a traditional method is overcome.

The invention discloses a multi-source information fusion-oriented equipment fault diagnosis method and a related device, and relates to the field of industrial equipment fault diagnos.The method comprises the steps that validity verification is carried out on obtained interval probability data of multiple fault modes of each expert on the basis of constructed validity verification conditions, and a verification result is obtained; when the verification result is that the verification is passed, solving the constructed optimal conversion model based on interval probability data of a plurality of fault modes of each expert to obtain a basic probability assignment distribution result of each expert; fusing the basic probability assignment distribution results of all experts to obtain a fused basic probability assignment distribution result; according to the fused basic probability assignment distribution result, the decision probability of each fault mode is obtained, and a fault diagnosis result is determined; and when the verification result is that the verification is not passed, obtaining the corresponding interval probability data again and verifying. According to the invention, the accuracy, reliability and timeliness of fault diagnosis are improved.

The invention discloses a wide-area backup protection method and system based on confidence weighting and conflict self-adaption. The method comprises the following steps: selecting a suspected fault line according to positive sequence, negative sequence and zero sequence voltages; constructing a plurality of probability functions, and selecting a plurality of probability functions from the results of all the probability functions as basic probabilities according to whether the zero-sequence current magnitudes at the two ends of the suspected fault line are 0 or not; calculating a confidence coefficient and a multi-source criterion consistency index of each basic probability, and fusing the confidence coefficient and the multi-source criterion consistency index to generate a fusion weight; performing weighted summation on the basic probabilities according to the fusion weight to serve as a weighted average fault probability, and calculating a fault probability of high-confidence correction according to the basic probabilities of which the confidence is greater than a set confidence threshold; and carrying out weighted summation on the two fault probabilities according to the adaptive conflict weight to obtain a line fault probability, and if the line fault probability is greater than a set fault threshold, determining that the line is a fault line. According to the method, the fault identification accuracy is ensured, and support is provided for rapid fault isolation of a power grid.

The invention discloses a frame-level network packet loss simulation method, device and equipment based on receiving end buffer time length and bandwidth correction and a storage medium, and relates to the technical field of video coding, decoding and transmission, and the method comprises the steps: obtaining the receiving end buffer time length, the network available bandwidth and a video frame split data packet; resolving the basic probability of a video frame corresponding to a data packet which cannot arrive in the split data packets of the video frame based on the time length of the buffer area of the receiving end, and determining the basic probability of a late packet loss frame; correcting the basic probability of the late packet loss frame based on the time length of the buffer area of the receiving end, the available bandwidth of the network and the total data volume of the video frames corresponding to the split data packets of the video frames, and determining the probability of the late packet loss frame after correction; and performing simulation truncation on a data packet in a corresponding target frame based on the late packet loss frame probability to complete frame-level network packet loss simulation based on the duration and bandwidth correction of the buffer area of the receiving end.

The application discloses a multi-source information fusion method for solving model prediction conflicts, characterized in that a plurality of single-source prediction models are constructed according to multi-source historical data in an industrial scene, basic probability distribution of the single-source prediction models is adaptively corrected in combination with operating conditions, and on this basis, evidence fusion rules are adopted to complete fusion of prediction results.

This invention discloses a decision-level contextual information inconsistency elimination system and method, relating to the field of artificial intelligence technology. Its features include: a human behavior data acquisition module, a human behavior data preprocessing module, a bidirectional long short-term memory human behavior recognition network module, an inconsistency detection module, an initial basic probability allocation mapping module, an evidence set conflict degree division module, an adaptive credibility assessment module, and a human behavior information application module, connected sequentially. The human behavior data acquisition module includes an inertial motion sensing unit and a physiological signal monitoring unit. The technical problem this invention aims to solve is to provide a decision-level contextual information inconsistency elimination system and method to achieve effective fusion of conflicting evidence, fully utilize the complementarity of multi-source information, and improve the reliability and accuracy of decision-making in complex scenarios.

The application provides a ship collision arc gate failure risk method, system, device and medium, and relates to the field of water conservancy engineering safety evaluation. The method comprises the following steps: constructing an index system for evaluating the failure risk of the arc gate; simulating the dynamic response of the arc gate after being hit by ships under different ship tonnages and ship speeds according to a finite element model, so as to extract actual values of various evaluation indexes; sending the actual values of the various evaluation indexes into a forward cloud generator model to generate corresponding cloud droplets, so as to calculate the membership degrees of the various evaluation indexes under different risk levels; calculating the combined weight values of the various evaluation indexes by using an entropy weight method and a variation coefficient method, and combining the membership degrees to obtain corresponding basic probability assignment functions; and fusing the basic probability assignment functions of the different evaluation indexes according to a Dempster combination rule, so as to determine the final failure risk evaluation grade of the arc gate under the ship collision. The technical scheme can accurately measure the failure risk of the gate under the dynamic load.

This invention relates to the field of computational toxicology, and in particular to a precise method for comparing the consistency of animal and non-animal toxicity evaluation results. This method acquires in vivo baseline data from animals and in vitro test data from non-animals. It employs a dynamic time warping algorithm and radial basis function kernel function to construct correction coefficients and perform equivalent mapping to generate a converted sequence. Based on a binary classification mapping of toxicity thresholds, a confusion matrix is ​​constructed to calculate sensitivity, specificity, and predicted values. Each indicator is treated as an independent source of evidence, and the Mahalanobis distance is calculated using the covariance matrix to generate a basic probability allocation function. The D-S evidence theory combination rule is applied for fusion, and a secondary factor allocation is introduced when there is conflict. Finally, the result is compared with a preset threshold to determine the feasibility of substitution. This invention adaptively eliminates the nonlinear misalignment difference between in vivo and in vitro dose responses, providing an objective quantitative judgment standard for toxicological substitution verification.

The application relates to the technical field of emergency management, and discloses a post-disaster road recovery graph network updating method fusing heterogeneous disaster situation sensing data, which comprises the following steps: converting the obtained heterogeneous data into basic probability distribution of observation evidence; predicting the physical field state based on a physical process simulator to generate basic probability distribution of prediction evidence; fusing the two to obtain posterior basic probability distribution; deciding and updating the road graph network state according to the posterior basic probability distribution; and feeding back the updated state to the simulator as a dynamic boundary condition to form bidirectional coupling of physical simulation and road network state. Through the construction of the bidirectional coupling simulation assimilation framework, the technical problem of the separation of physical simulation and infrastructure state in the traditional method is solved, the uncertainty of data is effectively processed by using the evidence theory, and the authenticity of disaster situation evolution prediction and the reliability of road state updating are improved.

The application provides a kind of array group target data association method and system based on adaptive clustering, comprising: accumulating same load data;Each load data is clustered, and initial array group division is obtained;Single cluster target set is screened;According to array group determination condition, remove non-array group target;The characteristics of array group target of two time points before and after are described respectively, and the characteristics of formation and formation are modeled;According to the array group target association matching model of the characteristics of formation and formation, the characteristics of formation and formation, the corresponding basic probability distribution function is calculated respectively;Evidence synthesis is carried out, and the comprehensive basic probability distribution function is established;Based on two-dimensional distribution, the comprehensive basic probability distribution function is associated with decision, and the array group target matching result is obtained.The application can effectively reduce the influence of array group target motion state change, sensor's missed detection and observation error, and significantly improve the correct association rate of target.

The invention discloses an anti-falling method, device and equipment for a working well platform, and relates to the technical field of engineering protection, and the method comprises the following steps: an anti-falling decision-making layer is selected according to decision-making time, and the anti-falling decision-making layer comprises a state decision-making layer, an image decision-making layer and a mixed decision-making layer; the state decision-making layer obtains a first anti-falling decision according to the state data; the image decision-making layer obtains a second anti-falling decision according to the falling credibility and the falling trend criticality, the falling credibility is obtained by processing state data to obtain a basic probability assignment and synthesizing the basic probability assignment based on an evidence synthesis algorithm, and the falling trend criticality is obtained by obtaining state data statistical characteristics and performing inspection processing; and the mixed decision-making layer obtains a third anti-falling decision by combining the predicted falling risk value, falling credibility and falling trend criticality. The method is used for solving the problem that an existing anti-falling technology of a working well platform is insufficient in adaptability.

This invention provides a similarity-based method for anomaly detection and diagnosis of multivariate time series, comprising: S1, acquiring the multivariate time series to be tested; S2, obtaining a similarity matrix by combining a basic probability allocation method and the Tanimoto coefficient; S3, reconstructing the similarity matrix using an attention-based reconstruction model; S4, comparing the similarity matrix and the reconstruction matrix to obtain a similarity error matrix, and using the number of abnormal elements in the similarity error matrix as the anomaly score for the corresponding time moment; S5, if the anomaly score reaches a first threshold, the time moment is considered abnormal, and the anomaly attribute is determined based on the distribution of abnormal elements. The similarity matrix calculation method based on basic probability allocation and the Tanimoto coefficient realizes interpretable relationships between attributes of multivariate time series. The attention-based reconstruction model enhances the ability to process long-term, high-dimensional data, better describing long-term temporal dependencies, improving anomaly diagnosis capabilities, and detecting abnormal times and attributes.

The invention relates to the technical field of smart power grids, and discloses a flexible straight valve capacitor fault diagnosis method and system, and the method comprises the steps: obtaining an original signal data set; wavelet packet decomposition is combined with the Leimeh-Tiangers complexity, empirical mode decomposition is integrated with permutation entropy, feature extraction, fusion and dimensionality quantitative optimization are performed on the original signal data set, and the optimal fault feature dimensionality is obtained; a deep belief network model with the visible layer node number equal to the optimal fault feature dimension is constructed, and a target fault diagnosis model is obtained after optimization; in the diagnosis process, the target fault diagnosis model outputs each fault probability, the D-S evidence theory is adopted to fuse the basic probability distribution values obtained by weighting and adjusting the fault probabilities according to the global credibility and the local credibility, and the fused basic probability distribution value is obtained and used for fault diagnosis. The method provided by the invention improves the precision of fault diagnosis, does not need complex diagnosis equipment, and has higher adaptability.

The invention relates to the technical field of model construction, and discloses a dynamic inference method and system based on a Bayesian inference evolution probability model, and the method comprises the steps: constructing an initial network topology of an initial data segment, initializing the probability distribution parameters of nodes in the initial network topology, and generating a basic probability graph model; variational inference is carried out on the basic probability graph model and the new data fragments to obtain an updated probability model; performing fitting degree evaluation on the updated probability model and the new data fragments to obtain a model fitting abnormal signal; performing mismatch degree evaluation on the updated probability model, and adjusting a network structure of the updated probability model to obtain an evolved probability model; calculating similarity measurement, and performing component removal and combination on the evolved probability model to obtain a simplified probability model; optimizing parameters of the simplified probability model to obtain a robust probability model; according to the method, the accuracy of dynamic inference based on the Bayesian inference evolution probability model can be improved.

The present application belongs to the field of gas turbine fault diagnosis, and discloses a basic probability assignment method for gas turbine fault diagnosis. The present application mainly comprises the following steps: step one: determining a fault identification framework according to the fault category of the gas turbine; step two: obtaining data after normalization processing of the fault sample features; step three: calculating the mean value of the historical fault samples of each fault category; step four: calculating the distance proximity coefficient according to the current fault sample; step five: calculating the direction proximity coefficient according to the current fault sample; step six: calculating the membership of the current fault sample to each fault probability according to the distance proximity coefficient and the direction proximity coefficient; and step seven: normalizing the membership fault probability to obtain the basic probability assignment of each fault category. The present application solves the problems of the existing basic probability assignment method, such as limitation, poor generalization ability, and insufficient sample information mining, when the information fusion technology based on D-S evidence theory is used in the field of gas turbine fault diagnosis.

The application discloses a social network information credibility evaluation method based on D-S evidence theory, which comprises the following steps: 1, feature extraction, using BERT pre-training model, ResNet50 neural network, TF-IDF algorithm and cosine similarity method to extract text features, image-text consistency features and title-text consistency features; 2, evidence generation, using the basic probability assignment method and related formula to convert the original evidence into the corresponding BPA; 3, evidence fusion, calculating the evidence clarity, evidence credibility and evidence weight, and combining the evidence synthesis rule to realize the evidence fusion. The method firstly respectively utilizes the text features, image-text consistency features and title-text consistency features to evaluate the information credibility; then converts the credibility evaluation results based on the three single features into evidence, and uses the improved D-S evidence theory fusion rule to make fusion and decision, and finally outputs the final result of the comprehensive credibility evaluation.