106results about How to "Improve Fusion Accuracy" patented technology

The invention discloses a method for constructing high temporal-spatial remote sensing data. A CDSTARFM (combining downscale mixed pixel with spatial and temporal adaptive reflectance fusion model (STARFM)) method performs downscaling treatment on low-resolution data by using a downscale mixed pixel algorithm, replaces direct resample low-resolution data in the STARFM algorithm with the downscaled data, and performs data fusion by using a CDSTARFM algorithm. The method may effectively solve a phenomenon of image spots in downscale mixed pixel data fusion and a problem that it is difficult for the STARFM model to find a pure MODIS pixel. The method constructs the high temporal-spatial-resolution remote sensing data by using Landsat8 and MODIS data. A result manifests that the method has higher fusion precision than the STARFM and downscale mixed pixel algorithms.

The present invention discloses a measurement method on the basis of multi-sensor date fusion. The method provided by the invention comprises the following steps: when a sensor is used for measuring a parameter, the rejection of careless errors of acquired original data according to the Grubbs criterion is performed to minimize the influence on the fusion precision caused by the careless errors; local fusion of measurement data of the same observation object through adoption of the same sensor at different times through a batch estimation method is carried out to obtain local estimated values of relevant parameters of the observation object, and the measurement error of the sensor at this time is obtained through calculation; and weighting coefficients of different sensors are determined through measurement errors of the sensors obtained through the batch estimation, and weighted fusion of the obtained local estimated value of each sensor is performed to obtain final global estimated values of the parameters. The measurement method on the basis of multi-sensor date fusion is applicable to the practical engineering application, and weighting coefficients of data fusion are determined according to prior knowledge.

The invention discloses a ground three-dimensional laser point cloud and ground penetrating radar image fusion method. On the basis of ground penetrating radar image and GPS data synchronization acquisition, an improved electromagnetic wave path propagation model is used for realizing time-depth conversion of the ground penetrating radar image, with a coordinate system of the ground three-dimensional laser data as reference, a geo reference absolute positioning equation is combined to convert two-different source isomerism data to the same coordinate system, and data fusion is realized. According to data features of a ground three-dimensional laser scanner and the ground penetrating radar, a three-dimensional laser point cloud and ground penetrating radar image fusion method is provided. The experiment result and error analysis prove that the process is simple, the fusion precision is high, and the adaptability is wide. Differences between resolutions and data amounts of the two-different source isomerism data are overcome, advantages of each of the three-dimensional laser scanner and the ground penetrating radar are executed thoroughly, and effective fusion of two different typesof data is realized.

The invention relates to a CAN bus-based pig house environmental temperature intelligent monitoring system. The system is characterized in that the intelligent monitoring system is composed of a CAN bus-based pig house environment parameter acquisition and intelligent prediction platform, a pig house environment multi-point temperature fusion model and a pig house environment intelligent prediction model. With the CAN bus-based pig house environmental temperature intelligent monitoring system of the present invention adopted, an existing pig house monitoring system cannot intelligently monitor and predict the temperature of the environment of a pig house according to the non-linearity and large lag of the environmental temperature change of the pig house, the large area of the pig house, complicated temperature change and other characteristics, as a result, control and adjustment of the environmental temperature of the pig house are seriously affected, while, with the CAN bus-based pig house environmental temperature intelligent monitoring system adopted, the above problem can be solved.

The invention provides a multi-sensor fusion farmland environment perception method used for a unmanned agricultural machine in an agricultural machine path control technology field. The method comprises the following steps of step1, before an agricultural machine works, calibrating a camera and fusing a radar and visual sense information in space; step2, during agricultural machine working, adjusting a conversion relation of the radar and a camera coordinate in real time so that the radar and a camera realize space synchronization; step3, resolving received millimeter wave radar data by an industrial control computer, determining an effective target, selecting an area which the radar is interested in at a front portion of an agricultural machine working area, determining a most dangerous target and synchronously collecting camera images; and step4, according to radar information, determining a most dangerous target state, and according to most dangerous target image data collected by the radar and the camera, determining a type of the most dangerous target, and making the agricultural machine carry out corresponding motion by a navigation box. Accuracy of data fusion is high and identification accuracy of an obstacle in front of the agricultural machine is increased.

The invention discloses a networking navigation radar target tracking system based on data fusion. The system comprises at least one radar front end, at least one radar terminal and an Ethernet. The radar front end and the radar terminal are in bidirectional data transmission through the Ethernet. The radar front end comprises an antenna, a transceiver, a data acquisition module, a signal preprocessor module and a control command processing module. The radar terminal comprises a display unit, a signal processing unit and a control unit. The transceiver generates radio-frequency pulses which are radiated to the space through an antenna. The system belongs to a trace point serial processing centralized data fusion structure, has the features of high fusion precision and small system time delay, and meanwhile, has the advantages of eliminating detection blind areas, improving detection probability and improving tracking performance. After the terminal carries out data fusion and comprehensive optimization processing on the data of the plurality of front ends, the problem of blind area of the single radar due to antenna erection location can be effectively solved, marine target detection and tracking accuracy performance is improved, and navigation safety is guaranteed.

The invention provides a farm machine unmanned navigation method based on farmland environment perception. Particularly, the method comprises the following steps of 1, before operation of the farm machine, performing calibration on a video camera, and performing spatial fusion on a radar and visual information; 2, in farm machine operation, detecting height change between the radar and ground by a first distance detecting device, detecting height change between the image camera and the ground by a second distance detecting device, and performing real-time adjustment on coordinate conversion between the radar and the image camera so that spatial synchronization between the radar and the image camera is realized; 3, resolving received millimeter wave radar data by an industrial control computer, determining an effective target and a most dangerous target, and synchronously acquiring video camera images; and 4, determining the state of the most dangerous target and planning a walking path by the industrial control computer, determining a target type according to the image data of the most dangerous target which is acquired by the radar and the video camera, and making the farm machine act by a navigation box. The arm machine unmanned navigation method has advantages of high data fusion precision and effective accuracy in obstacle identification.

The present invention provides a method for production of high spatial and temporal resolution NDVI (Normalized Difference Vegetation Index) based on spatial and temporal weight. The method comprises: NDVI data with a low spatial and temporal resolution at t1 time, NDVI data with a high spatial and temporal resolution and NDVI data with a low spatial and temporal resolution at tp time are subjected to mixed pixel decomposition and spatial interpolation to obtain two NDVI increments with the high spatial and temporal resolution data in time domain; employing Bayesian model average method to perform weight combination calculation of the two DNVI increments, and obtaining the NDVI synthesis increment of each pixel; and finally adding the NDVI synthesis increment to the NDVI with the high spatial and temporal resolution to obtain required NDVI with the high spatial and temporal resolution. The two NDVI increments are subjected to Bayesian integration, and the spatial and temporal relationship between geographical data is fully employed to provide an effective method for production of vegetation indexes with the high spatial and temporal resolution.

The present invention discloses a method for fusion unwrapping of spaceborne InSAR long and short baselines. The method comprises: employing a spaceborne InSAR system interference processing flow to respectively perform inversion of digital elevation models (DEM) of a long baseline and a short baseline; employing three-dimensional coordinates of ground points corresponding to each pixel point in the images of the long baseline and the short baseline, combining data such as satellite parameters, offset and the like, calculating a vertical baseline and an effective baseline respectively corresponding to the images of the long baseline and the short baseline, selecting an appropriate reference ground, and calculating the long baseline interferometric phase and the short baseline interferometric phase corresponding to the reference plane; and finally, taking correlation coefficients, reference plane phases, baselines and the like obtained before as auxiliary data, employing a maximum likelihood method, and performing fusion processing of the long baseline phase data and the short baseline phase data so as to obtain a long baselined unwrapping phase with higher precision. The method for fusion and unwrapping of the spaceborne InSAR long and short baselines is used in the multi-baseline interference processing, the maximum likelihood method is employed to fuse the long and short baseline unwrapping phases to obtain high-quality DEM; and the method for fusion and unwrapping of the spaceborne InSAR long and short baselines can be also used for fusion processing of data such as interferometric phase and elevation, etc.

The invention discloses a fusion method of a high-temporal-spatial-resolution vegetation index NDVI. The fusion method comprises the following steps of A, performing downscaling processing on a low-resolution vegetation index NDVI image by utilizing a pixel downscaling algorithm; B, calculating increment NDVI by utilizing the vegetation index NDVI image subjected to downscaling processing and thehigh-resolution vegetation index NDVI image, counting an increment NDVI cumulative distribution histogram, and clustering to construct an increment NDVI classification map; C, constructing a linear hybrid model of the high-resolution increment and the low-resolution increment delta NDVI by utilizing the increment delta NDVI classification diagram, and respectively calculating the high-resolution increment by utilizing a ridge regression method; and D, calculating a high-resolution vegetation index NDVI value at the moment tp. According to the method, the classification graph is generated through incremental NDVI clustering. The NDVI change characteristics are fully utilized. Compared with other methods, the fusion effect of the method is more satisfactory. An effective method is provided for constructing the vegetation index NDVI with the high temporal-spatial resolution.

The invention discloses a multi-source information unequal interval federated filtering method based on filter variance matrix correction. The method comprises the particular steps that a fused federated filter without a reset mode of an aviation onboard inertial navigation system (INS) and other navigation systems is established, and then the filtering accuracy of a sub-filter is improved by the multi-source information unequal interval federated filtering method based on sub-filter variance matrix correction. An INS / CNS (Communications, Navigation and Surveillance) / SAR (Synthetic Aperture Radar) / TER integrated unequal interval federated filtering realization scheme suitable for a high-altitude long-endurance unmanned aerial vehicle is designed by taking the high-altitude long-endurance unmanned aerial vehicle as an application object and combining the actual sensor use characteristic. The method can improve the fusion accuracy of the filter by correcting a sub-filter filtering variance matrix under the condition that additional calculation burden is increased hardly; the adaptability of the reset-free federated filter applied to multi-source information unequal interval federated filtering is achieved while the characteristics of no reset structure and high fault tolerance are kept; the navigation accuracy of a system is improved further; and the robustness of the integrated navigation system is improved.

The invention relates to the field of image processing, and specifically relates to a multi-source image fusion method based on enhanced learning. The method comprises the steps: decomposing a multi-source image to obtain a high-frequency coefficient signal and a low-frequency coefficient signal, and performing weighted reconstruction of the high-frequency coefficient signal and the low-frequencycoefficient signal to obtain a fusion image. The method adopts a Q learning algorithm in enhanced learning for the training of the weights of the high-frequency coefficient signal and the low-frequency coefficient signal, selects a weight which enables the fusion evaluation criteria to be optimal for the weighted fusion, and obtains the fusion coefficient signal. The fusion image is reconstructedto obtain the fusion image. Based on the Q learning algorithm, the method achieves the training of the weights of the high-frequency coefficient signal and the low-frequency coefficient signal obtained through the decomposition of the multi-source image, and improves the reliability and accuracy of weight estimation.

A measurement data fusion method for a two-dimensional radar and infrared sensors at different sites. The method includes the following steps that: step 1, successful measurement target pair information is received; second step, a target observation equation is constructed through transformation; third step, state vectors that satisfy Markov estimation conditions are constructed according to measurement azimuth angle characteristics; fourth step, Markov estimation is performed on a target state vector X; step five, error covariance calculation is performed on the target state vector X; and step six, state vectors on a Z axis and the variance of the state vectors are calculated. The measurement data fusion method is advantageous in simple operation. According to the measurement data fusion method of the invention, the target observation equation is constructed through combining the measurement data of the two-dimensional radar and the infrared sensors at different time points; the state vectors and the covariance thereof of the target observation equation are estimated according to the Markov estimation principle; and therefore, complex time-space alignment calculation of vacant-dimension data can be avoided. The method is easy to implement and apply to engineering practice, and has a target state estimation function as well as fusion and filtering functions, and has excellent real-time performance and high fusion accuracy.

The invention discloses a safety data aggregation method based on dynamic segmentation technology, and relates to a network safety data aggregation method. A wireless sensor network is abstracted as a connected graph G(V, E), and adopts the tree type fusion network topology; all nodes in the network are divided into three types: the BS (Base Station) node, the aggregator node and the leaf node; the aggregator node upward transmits the data received from the sub node to the father node after performing the aggregation operation; the leaf sub-node is only responsible for acquiring data and transmitting to the father node; therefore, the data aggregation function is defined as shown in description; the method comprises the aggregation tree establishment stage, the data mixing stage, the data aggregation stage, and the data aggregation stage. The data aggregation precision is improved, the energy consumption is reduced, the network life cycle is prolonged, the privacy protectiveness of the data is improved, and the insufficiency of the SMART algorithm is effectively solved.

The invention discloses an unmanned aerial vehicle height measuring method based on multilayer information fusion. The method comprises the following steps: installing a plurality of height sensors at different spatial positions of an unmanned aerial vehicle, and acquiring unmanned aerial vehicle height observation data of each height sensor at different times and spaces; performing two-step delay time fusion for an observed value of each height sensor at different moments and estimated values at two moments before the moment, and obtaining a time fusion estimated value of the height sensor at different moments; performing spatial fusion for the time fusion estimated values of each height sensor at the same moment and obtaining a time-spatial fusion estimated value; and finally, performing self-adaptive complementary filter fusion based on parameter identification for the time-spatial fusion estimated value obtained in step 3 and measured data of an accelerometer to obtain a final fusion estimated value. The unmanned aerial vehicle height measuring method is small in calculation quantity, simple in programming and apparent in application value in the field of unmanned aerial vehicle height measurement.

The present invention provides a farmland environment sensing method used for the agricultural machinery navigation in the agricultural machinery path control technology field. The farmland environment sensing method concretely comprises the following steps of 1, before an agricultural machinery works, calibrating a camera, and fusing the radar and visual information in space; 2 when the agricultural machinery works, using a distance detection device 1 to detect the height change between the radar and the ground, using a distance detection device 2 to detect the height change between the camera and the ground, and real-timely adjusting to convert the coordinates of the radar and the camera to enable the radar and the camera to synchronize in space; 3 using an industrial control computer tocalculate the received millimeter wave radar data, determining an effective target and a most dangerous target, and synchronously acquiring the images of the camera; 4 according to the radar information, determining the state of the most dangerous target, and according to the image data of the most dangerous target acquired by the radar and the camera, determining the type of the most dangerous target, and using a navigation box to control the agricultural machinery to act correspondingly. The method of the present invention is high in data fusion precision, and enables the obstacle identification accuracy to be improved.

The invention discloses a pig house environment temperature detection system based on a CAN (controller area network) bus. The pig house environment temperature detection system is characterized by comprising a CAN bus based pig house environment parameter collection platform and a pig house environment multipoint temperature fusion model, wherein pig house environment factor parameters are detected, adjusted and monitored by the CAN bus based pig house environment parameter collection platform, and the pig house environment multipoint temperature fusion model realizes pig house environment multipoint temperature fusion on the basis of support degree fusion weight and grey relational degree fusion weight which are obtained by a support degree matrix and a grey relational degree matrix of temperature fuzzy values of multiple detection point temperature sensors of the pig house environment as well as linear combination weight. According to the system, the problem that regulation and control for the temperature of the pig house environment is greatly affected by the fact that a conventional pig house monitoring system does not detect the temperature of the pig house environment according to the characteristics of nonlinearity and large lag of the pig house environment temperature change, large area of a pig house, complicated temperature change and the like is effectively solved.

The invention discloses a WSN data layered fusion method for plant growth monitoring. According to the method, a wireless sensor network is divided into a plurality of fixed layers, data transmission volume can be decreased, and errors can be reduced layer by layer. The method includes the following steps that: gross errors in a single node datum are eliminated by using Grubbs criterion, and the fusion value of the node datum per unit time is obtained through batched estimation; data adaptively weighted fusion at a sub layer level is carried out; and finally, data weighted adaptively fusion at a base station level is carried out. As indicated by a simulation experiment result, the fusion result of the method of the invention is closer to a true value compared with an arithmetic average method and a single adaptive weighting algorithm. With the method adopted, data transmission volume can be reduced. The method has high expansibility.

The invention discloses a dinner plate edging method based on vision control. The method comprises the following steps: clamping and locating a dinner plate to be edged, and acquiring images of the dinner plate by virtue of two cameras which are symmetrically arranged at the two sides of the dinner plate; integrating the images through pre-treatment, image registration and fusion processing, and then extracting contour features to generate a complete dinner plate contour drawing; comparing the dinner plate contour drawing with a theoretical CAD model of the dinner plate, and obtaining a central position and a deviation angle of the current clamping and locating; and carrying out coordinate offset in a numerical control system according to the central position and the deviation angle, and controlling an edging unit to realize edging for the dinner plate by the numerical control system after the coordinate offset according to G code corresponding to the theoretical CAD model. According to the method disclosed by the invention, the two cameras which are symmetrically distributed at the two sides of the dinner plate are used for carrying out contour acquisition on the edge of the dinner plate, at a machining station, comprehensive acquisition and control for the contour of the dinner plate can be realized, and the G code is finally used for machining; and the method has the advantages of high edging quality, high machining efficiency and the like.

The invention provides a sensor-based multi-target trajectory fusion method in target tracking, and relates to the technical field of target tracking. The method comprises: in a target tracking process, firstly, judging whether detection data of different sensors come from the same target; adopting an adaptive trajectory association algorithm; selecting a corresponding association algorithm from aspeed-based double-threshold track association algorithm and a nearest domain track association algorithm by judging the track state; and then track fusion is carried out on tracks from the same target, and a weighted track fusion algorithm based on local state estimation is adopted: the local state estimation of each sensor is utilized to complete the weighted combined filtering of multiple sensors, so that the error problem caused by the precision change of the sensors is solved, and the track fusion precision is improved. According to the method, the target tracks can be effectively fusedin a dense or sparse environment of the target tracks, and the target track detection precision of the sensor is improved.

The invention provides an agricultural machinery unmanned navigation method based on farmland environment perception, which is in the technical field of in agricultural machinery control. The agricultural machinery unmanned navigation method specifically comprises the steps of: step 1, calibrating a camera before agricultural machinery operates, and fusing radar and visual information spatially; step 2, during the operation of the agricultural machinery, detecting height variations between a radar and the ground by means of a distance detection device 1, detecting height variations between thecamera and the ground by means of a distance detection device 2, and adjusting conversion of radar and camera coordinates in real time, so that the radar and the camera are spatially synchronized; step 3, calculating millimeter-wave radar data by means of an industrial control computer, determining an effective target, determining the most dangerous target, and acquiring camera images synchronously; step 4, and judging a state of the most dangerous target based on the radar information, planning a walking path according to a motion state by means of the industrial control computer, judging atype of the most dangerous target according to image data of the most dangerous target acquired by means of the radar and the camera, and controlling the agricultural machinery to do corresponding actions by means of a navigation box. The agricultural machinery unmanned navigation method is high in data fusion precision, and improves the accuracy degree of identifying an obstacle.

The present invention discloses an information processing method. The method comprises the steps of: identifying a base feature point of the source object in the source image and a base feature pointof the material object in the material image; identifying, according to the distribution of basic feature points in each of the images, auxiliary feature points that meet a filling condition accordingto the distribution from corresponding images; determining, feature points of the same type in the source object and the material object, the feature points of the same type comprising basic featurepoints of the same type and auxiliary feature points of the same type in the source object and the material object; determining an average feature point according to the positions of feature points ofthe same type; deforming the source object and the material object according to the average feature point; and performing fusion on the deformed source object and material object. Also disclosed arean image processing apparatus, terminal and storage medium.

The invention provides a multi-sensor data synchronization method and device thereof, and the method is applied to a data synchronization controller and comprises the steps: obtaining the state information reported by all sensors; if it is determined that each sensor is in a normal state according to the state information, sending a time synchronization command to each sensor while obtaining the reference time, so that each sensor is synchronized with the reference time and a beat timer is started; and sending a data acquisition command to each sensor, so that each sensor acquires data after receiving the data acquisition command, determines a current timestamp according to the reference time and the beat number of the beat timer, and reports the acquired data associated with the current timestamp to a data fusion processor. Therefore, the sensors work under the same time reference to acquire data, and the subsequent data fusion accuracy is improved.

The present invention provides a multilayer image fusion method and system based on the auto-focus technology. The method comprises: according to a fusion time and a hierarchical division on a processing layer of a fused input image, transforming the image from a spatial domain and a transform domain respectively; introducing the wavelet transform operation method into a convolution process, obtaining high-frequency information and low-frequency information after the transformed image is subjected to the lifting wavelet decomposition, and optimizing multiple thresholds in the fusion process; using the average structural similarity as a weight-optimized evaluation function in the fusion process; and performing high-frequency fusion on the high-frequency sub-band by using the image contrastcombined with the PCNN, and performing low-frequency fusion on the low-frequency sub-band by using the regional energy combined with the regional gradient. According to the technical scheme provided by the present invention, adaptive registration of high-resolution biological imaging can be implemented through an automatic and intelligent means, and the technical scheme has a simple operation, high detection efficiency, high reliability and strong versatility.

The invention provides a point cloud correction method and device and a readable storage medium,relating to technical field of unmanned driving. The method comprises the steps of obtaining first pointcloud data and second point cloud data at the same moment, mapping the first point cloud data and the second point cloud data to the same coordinate system, wherein the overlapped area of the first point cloud data and the second point cloud data is third point cloud data; constructing a multi-stage error equation according to the third point cloud data; calculating a correction result, obtainingthe corrected first point cloud data and / or the corrected second point cloud data according to the correction result so that the fusion precision of the laser point cloud and the visual point cloud is improved, then the overall precision of the output fusion point cloud is improved, and a stable and reliable data source is provided for a follow-up technical scheme of unmanned driving.

The invention discloses a networking radar tracking method and system based on conversion measurement reconstruction, and belongs to the technical field of radar target tracking. The method comprisesthe following steps: S1, initializing a filter; S2, predicting a target state; S3, reconstructing a conversion measurement model; S4, applying a CMKF filter; S5: estimating a target state. According to the method, an angle prediction value is estimated based on state prediction, then angle measurement and prediction are introduced into conversion measurement, and the proportion of the two is adjusted according to the proportion of the angle measurement error to the prediction error, so that a new conversion measurement model is constructed; in the reconstruction conversion measurement model, the cosine term of the angle error is effectively reduced, and the radial estimation performance is improved accordingly; and in the networking radar, information of a plurality of radars is fused in sequence based on the reconstructed conversion measurement model, so that for target tracking estimation, fusion precision is greatly improved compared with an original method, a satisfactory result isobtained; and the fusion method can be expanded to application of 2D radar networking tracking.

The invention discloses a single out-of-order measurement centralized fusion method of an arbitrary delay step of a nonlinear system. The generalized Gaussian smoothing method is used to solve a smoothing probability density function of the arbitrary nonlinear system, and the particle filter algorithm is used to achieve update. Firstly, it is determined that the measurement is sequential or not, asequential posterior state is obtained by using the particle filter algorithm if so, otherwise, out-of-order measurement fusion is performed, i.e., the generalized Gaussian smoothing method is used to solve the smoothing probability density function at the generation time of the out-of-order measurement, then the smoothing probability density function and a local likelihood function are combinedto solve an asynchronous likelihood function, and finally, the asynchronous likelihood function is used to update the sequential posterior state at the previous time to obtain the fusion posterior state, and the fusion posterior state represented by particle samples is approximately stored as Gaussian parameters, which effectively solves the problem that the out-of-order measurement data in asynchronous sensor networks are difficult to be fused due to data processing and communication delay.

The invention discloses a multi-sensor data weighted fusion method, which is used for a distributed multi-sensor detection system and belongs to the technical field of multi-sensor information fusion.The method comprising the steps of 1) taking i (i=1, 2, 3,.., N) and all the data before the k-th time as a group, filtering and correcting the k-th measurement data according to the group data, replacing the k-th measurement data with the corrected data, and participating in the correction of the next measurement data; 2) carrying out weighted fusion among the N sensors on the corrected measurement data; 3) combining all the fused data into a fused data group, and performing secondary weighting fusion on the data of the fused data group. The method corrects the measured data of sensors and accurately calculates the data fusion weights of different sensors, so as to improve the data fusion accuracy of sensors, and make the results of distributed multi-sensor fusion closer to the actual situation and more reliable.

The invention provides an improved evidence dissimilarity index and an improved evidence fusion algorithm. The improved evidence dissimilarity index and the improved evidence fusion algorithm comprisethe following steps 1-6. The method comprises the following steps: step 1, calculating dissimilarity indexes among different evidence bodies based on proposed improved evidence dissimilarity; step 2,calculating the support degree between different evidence bodies according to the evidence dissimilarity degree, and constructing a support degree matrix; step 3, calculating a characteristic value and a characteristic vector of the support degree matrix; step 4, taking the feature vector corresponding to the maximum feature value as a weight coefficient of each evidence body; step 5, except forthe evidence with the maximum weight, correcting the rest evidence bodies by taking the weight coefficient as a discount factor; and step 6, performing evidence fusion on the corrected evidence body through a D-S combination formula, and completing a final decision. According to the method, the conflict degree between different evidence bodies can be evaluated more effectively, and the accuracy and precision of evidence fusion are improved, so that the decision risk is reduced.