496 results about "Model testing" patented technology

The invention provides an urban rail transit panoramic monitoring video fault detection method based on depth learning. The method comprises a data set construction process, a model training generation process and an image classification recognition process. The data set construction process processes a definition abnormity video, a colour cast abnormity video and a normal video in an urban rail transit panoramic monitoring video. A training set and a test set are classified. The model training generation process comprises model training and model test. The model training is to train a fault video image recognition model based on a convolution neural network. The convolutional neural network comprises a plurality of convolution layers and a plurality of full connection layers. The model test is to calculate the test accuracy. If expectation is not fulfilled, the fault video image recognition model is optimized. The image classification recognition process comprises the steps that a single-frame image to be recognized is input into the model, and the fault video image recognition model outputs an image classification result to complete the fault image detection of the urban rail transit panoramic monitoring video.

A testing system and various methods involving testing of a device under test (DUT) use a device model to model a stimulus-response behavior of a the DUT. The testing system includes a device model of the DUT that is fitted to the stimulus-response behavior of the DUT and a measurement projector connected to an output of the device model. The device model includes a block diagram model and a difference model. Test metrics for the DUT are produced by the measurement projector from an output of the fitted device model.

The invention discloses a multi-angle face recognition model training and testing system which comprises a face sample library angle division module, a multi-angle sample combination training module, a face angle loss function calculation module and a multi-angle face recognition model testing module. According to the invention, the system is applicable to multi-angle face recognition; the problem of poor non-positive-face recognition effect in a face recognition process can be overcome; and the accuracy of non-positive-face recognition in the face recognition process is improved.

The invention relates to a computed tomography (CT) scanning heterogeneous model testing system. The system comprises a CT scanning system, a displacement system, an overburden pressure system, a pressure measurement system, a heterogeneous multilayer core clamp holder and a metering system, wherein the heterogeneous multilayer core clamp holder consists of a shell, a rubber cylinder, a core left plug, a core right plug, a left fixing sleeve, a right fixing sleeve, a left fastening sleeve, a right fastening sleeve and a fixed bracket; the core right plug is provided with a plurality of liquid outlets; each liquid outlet is respectively aligned with one layer of core model; and a strip liquid outlet sealing gasket is formed on the core right plug, which correspond to a seam between two adjacent layers of core models, so that liquid flow passing through each layer of core model flows out from the corresponding liquid outlet of each layer of core model, and thus, intraformational heterogeneous water displacing oil layered measurement is realized. By the CT scanning system, the real-time on-line monitoring of the on-way distribution of saturation of fluid on each layer section in a heterosphere is realized so as to observe the interlayer communication phenomenon.

The invention discloses a visual model testing device and method for simulating water gushing in a tunnel. The device comprises a model test box, a water tank, a high-pressure water pump, a flow meter, a runoff flow and filling collecting device, a digital photographing non-contact measuring system, an osmotic pressure monitoring system, a fiber grating displacement monitoring system and the like. The testing device is characterized in that the catastrophe evolution process of water gushing in the tunnel is simulated, and anti-gushing rock mass crack growth and water gushing channel formation during water gushing can be directly observed. The device can quantitatively research water gushing behavior of the tunnel under different water pressure, different lithological characters and different thicknesses of anti-gushing rock masses, acquire coupling information of multiple fields such as stress fields, displacement fields, seepage fields and the like during water gushing of the tunnel, determine formation processes and forms of water gushing channels under the effect of different factors, disclose mechanical characteristics and evolution laws in anti-gushing rock mass structure progressive failure processes and provide an effective basis for the tunnel water gushing disaster incoming criterion and the minimum safe thickness analysis method.

The invention discloses a visual-model testing apparatus and testing method for seepage-stress-coupled internal piping penetration. The testing apparatus is composed of four parts, i.e., a variable-head control and liquid circulation system, a vertical loading and penetration system, a laser control system and an image acquisition system and can carry out experimental study on internal piping penetration of non-steady seepage under the conditions of different stresses, different contents of fines, different grading and the like and research on the characteristics of particle loss and deformation and fracture in the process of seepage. The invention also provides a visual testing method for seepage-stress-coupled simulated erosion-type internal piping. According to the method, head difference is applied to a piping soil sample in a stress field in the vertical loading and penetration system in virtue of the variable-head control and liquid circulation system, so a seepage-erosion-stress three-field coupled internal piping test is realized. The visual-model testing apparatus and testing method provided by the invention are simple in procedure, easy in fabrication, capable of realizing direct and in-depth observation, economic and good in operability, so the meso-scopic mechanism of seepage-stress-coupled internal piping is revealed preliminarily, and a testing foundation is provided for subsequent research.

The invention discloses a testing device and a testing device method for simulating collapse of tunnel surrounding rock under conditions of rainfall and underground water seepage. The testing device comprises a model testing box, a rainfall adjustable simulating system, an underground water seepage simulating system, an underground water supplying system, a water content testing system, a stress monitoring system, a digital radiography noncontact measuring system and an optical fiber bragg grating displacement monitoring system. By simulating a progressive failure process of tunnel weak broken surrounding rock under the action of seepage of surface water and underground water, mechanical behaviors of crushing the surrounding rock under different rainfall conditions and different underground water seepage path conditions are quantitatively researched. By obtaining multiple-physical field information in a tunnel surrounding rock collapse process, multi-information features in different evolutionary phases and internal links of information of all fields of the collapse are analyzed, a progressive failure process and fracture morphology and ranges of the tunnel weak broken surrounding rock under different factors are determined, and the effective basis is provided for disclosing a progressive failure principle and a dynamic pressure arch effect of the weak broken surrounding rock under conditions of rainfall and underground water seepage.

The invention provides a pedestrian re-identification method based on transfer learning and deep feature fusion. The pedestrian re-identification method comprises the following steps of pre-training,human body posture correction and segmentation, feature vector, deep feature fusion, model training, model testing and result identification. According to the invention, global and local features of pedestrians are extracted by using a deep convolutional neural network; deep fusion is carried out on the two features to obtain final pedestrian feature representation; then, in the training process of the deep convolutional neural network, a pedestrian re-identification network model with a better effect is obtained by adopting a transfer learning mode, and finally, features extracted by the pedestrian re-identification network model have stronger resolution capability, so that the purpose of improving the pedestrian re-identification accuracy is achieved.

The invention discloses an end-to-end difference network learning method for image semantic segmentation. The method comprises the steps that a main network structure and a complete network structureare constructed respectively by using a Caffe deep learning framework, wherein the main network structure is used to generate a rough segmentation model and a small target area of each image in a training set and the complete network is used for final image semantic segmentation; the rough model of the main network is trained by using a part of the data of the training set, and a segmentation result acquired by the rough model is compared with a real segmentation image to acquire a mistaken segmentation area of the rough model; the acquired rough model is used as an initialization parameter totrain a complete network model to acquire a final segmentation result, and an image semantic segmentation model is established; the segmentation model is tested; and all test images are segmented according to the image semantic segmentation model acquired in the step 3. The method can be sensitive to small target areas, and can solve the problems of edge blur and misjudgment of similar parts to some extent.

The invention discloses a multiplexing-oriented embedded software testing method. According to the method, a model library is arranged in a testing system, a tested software model, a crosslinked environmental model, a testing exemplar model, a testing execution model and a testing result model are stored in the testing system, and testing is finished in the converting process of each model. The software testing method is an effective testing method specific to embedded system software; in a testing process, multiplex testing products can be produced, and multiplex products in the model library can be directly used simultaneously, so that the testing efficiency is increased, the generation of a plurality of man-made errors and undefined factors is avoided, and the controllability and the specificity of a simulated testing process are ensured. During testing of difference software, the testing model becomes a multiplex effective assembly in a testing process, so that repeated work in the testing process can be reduced greatly in particular for production line software, and the testing efficiency is increased to the maximum extent.

The invention discloses a single-picture three-dimensional face reconstruction method based on a cascade regression network. The whole single-picture three-dimensional face reconstruction method includes the steps of 1, data preparation, wherein three-dimensional face data is collected, and a three-dimensional face-shape point cloud of a face picture is obtained; 2, model designing, wherein a network model comprising two-stage multitask convolutional neural networks is established, face features are regressed step by step, and then modified shape features are extracted; 3, model training, wherein the two-stage multitask convolutional neural networks are trained respectively, and a multitask cascade regression convolutional neural network M is obtained; 4, model testing, wherein an image Iis input into the multitask cascade regression convolutional neural network M, and model testing is carried out. According to the single-picture three-dimensional face reconstruction method based on the cascade regression network, the multitask cascade regression convolutional neural network M is used, a three-dimensional face shape, face feature points and a face posture are regressed step by step, and meanwhile a traditional feature extracting method is improved; by computing visibility of the feature points, the confusion performance of the feature points on a two-dimensional face picture is eliminated, and the single-picture three-dimensional face reconstruction method is more robust for large gestures.

The invention discloses a pedestrian searching method and device based on structural perception self-attention and online instance aggregation matching, and belongs to the technical field of computervision technology processing. The method comprises the following steps: firstly, in a training phase, combining a convolutional neural network with a non-local layer; carrying out feature extraction on an input whole scene image to obtain feature representation of the scene image, designing structure-perceived anchor points for a special object of a pedestrian, improving the performance of a detection framework, framing the detected pedestrian into the same size, then sending the pedestrian into a pedestrian re-identification network, and carrying out training, storage, optimization and updating of pedestrian features with tags. In the model testing stage, the trained non-local convolutional neural network is used for carrying out pedestrian detection on an input scene image, and after a pedestrian frame is detected, a target pedestrian image is used for carrying out special similarity matching sorting and retrieval. Pedestrian detection and re-identification can be carried out on large-scale real scene images at the same time, and the method plays an important role in the security and protection fields of urban monitoring and the like.

The invention relates to an intraformationally heterogeneous model testing system. The rubber cylinder of a multilayered core holder is arranged in a shell; a left core pressing head and a right core pressing head are pressed to the two ends of a multilayered core model; a core accommodating chamber for accommodating the multilayered core model is formed among the inner wall of the rubber cylinder, the left core pressing head and the right core pressing head; a closed annular confining pressure room is formed among the outer wall of the rubber cylinder, a left fixing sleeve, a right fixing sleeve and the inner wall of the shell; the shell is provided with a confining pressure interface which is communicated with the annular confining pressure room and an overburden pressure system; a liquid inlet is formed in the left core pressing head and is communicated with a displacement system and the core accommodating chamber by a pressure measurement system; a liquid outlet is formed in the right core pressing head and is communicated with the core accommodating chamber and a metering system; and a strip liquid outlet gasket corresponds to a joint between two adjacent layers of core models on the right core pressing head, so that flowing liquid which passes through every layer of core model flows out of the liquid outlet of the corresponding layer. The system can accurately reflect the changing law of intraformationally heterogeneous oil-water displacement in mixed water injection layer mining.

The invention discloses an intelligent detection and quantitative recognition method for the defect of concrete. According to the method, a concrete test piece is subjected to impact echo signal sample acquisition, signal noise reduction treatment and characteristic value extraction so as to construct a recognition model for analysis components including feature extraction, defect inspection, defect diagnosis and defect quantification and positioning; and the model is used for detecting and recognizing to-be-detected concrete. The intelligent detection and quantitative recognition method provided by the invention is directed at disadvantages of conventional detection technology for concrete defects and based on theoretical analysis, value simulation and model testing, employs advanced signal processing and artificial intelligence technology and fully digs out characteristic information of a testing signal, thereby establishing the model for intelligent rapid detection and classified recognition based on wavelet analysis and an extreme learning machine; and the model has good classified recognition performance, realizes intelligent rapid quantitative recognition and evaluation of the variety, properties and scope of the defect of concrete and further improves the innovation and application level of non-destructive testing technology for the defect of concrete.

The present invention relates to a method for the automatic identification of at least one informative data filter from a data set that can be used to identify at least one relevant data subset against a target feature for subsequent hypothesis generation, model building and model testing. The present invention describes methods, and an initial implementation, for efficiently linking relevant data both within and across multiple domains and identifying informative statistical relationships across this data that can be integrated into agent-based models. The relationships, encoded by the agents, can then drive emergent behavior across the global system that is described in the integrated data environment.

The invention discloses a listed-company announcement classification and abstract generation method based on deep learning. The method comprises the following steps: step 1, acquiring announcement original-text data, extracting text, picture and form information, and establishing structured documents. step 2, establishing a classification rule word library of different announcements on the basis of industry knowledge of announcement fields according to various company operation change event keyword differences, and carrying out statistical judgment on announcement classes; and step 3, for the announcements of the different classes, extracting announcement document contents, combining the rule word library of corresponding class keywords to train an announcement content classification model, and automatically generating document abstract contents, wherein content extraction, training set selection, keyword model optimization, model training, model testing, result analysis and content generation are included. The method can solve technical problems of automatically classifying the announcements for a large amount of announcement information generated each day, automatically extracting key and important information according to classification situations, generating the abstract contents and the like.

The invention discloses a testing device for observing shear deformation failure characteristics between a tubular pile and a grouting soil body. The testing device comprises a fixing platform, a tubular pile model testing box, a loading mechanism and a monitoring system, wherein the fixing platform comprises a base plate, a top plate and a plurality of connecting screw rods; the tubular pile model testing box comprises a bottom wall, two side walls, a rear wall and a front wall, wherein the front wall is movably arranged between the side walls, and is made of colorless transparent organic glass; the monitoring system comprises a displacement sensor, a pressure sensor and a strain rosette group arranged on the outer surface of the tubular pile; the strain rosette group comprises a plurality of strain rosettes which are uniformly distributed along the central axis direction of the tubular pile, and are connected with an automatic recorder. According to the invention, a semi-circular pipe pile is adopted, so that the shear deformation failure characteristics between the tubular pile and the grouting soil body can be observed conveniently, and the shear resistant intensity between the tubular pile and the grouting soil body can be observed; a camera device is adopted for recording the video of the testing process; the visual testing enables the analysis to be relatively intuitive.

The invention relates to a device for testing a model with a three-dimensional gradient nonuniform loading structure, which comprises a table-shaped gradient nonuniform loading device and an opposite force loading device, wherein the opposite force loading device is connected with the table-shaped gradient nonuniform loading device which is arranged in the opposite force loading device; and a table-shaped gradient loading module of the table-shaped gradient nonuniform loading device is tightly attached to the surface of a testing model. Because the deep underground chamber engineering in the fields of hydroelectric power, traffics, power sources, resources and defense construction usually lies in three-dimensional loading states which are nonuniformly distributed in a gradient, the table-shaped gradient loading module is arranged on the surface of the model, thus the model testing device achieves the three-dimensional gradient nonuniform loading of an underground structure model and more truly discloses the nonlinear deformation characteristics and failure mechanisms of wall rocks of the deep underground chamber through the testing of the three-dimensional gradient nonuniform loading structure model so as to more reasonably evaluate the stability of the wall rocks of the underground chamber and take effective reinforcement measures, thereby having a quite important significance to guiding the design and the construction of the deep underground engineering and guaranteeing the safety of the deep underground engineering.

The invention discloses a large-sized landslide intelligent model testing system under a combined effect of rainfall and reservoir water. The system comprises a flexible slope testing groove and a distributed spray rainfall simulation device, wherein a slope hydraulic lifting device comprises a plurality of groups of electric hydraulic jacks. The flexible slop structure in the flexible slop testing groove mainly comprises two layers of high-strength steel wire grids and a high-strength geotechnical cloth sandwiched by the two layers of high-strength steel wire grids; the geotechnical cloth is prepared from polymer expressed by a formula I; Mw of the polymer in the formula I is 5.8*10<4>, and the molecular weight distribution index is 1.298; the distributed spray rainfall simulation device comprises a distributed spray rainfall device and a windshield structure; the rainfall device is formed by combining upper and lower layers of spray pipes of different diameters and sprinklers corresponding to the diameters of nozzles; the windshield structure is arranged below the rainfall device, can be fixedly arranged on the inner side of a test rack and is prepared from the polymer shown by a formula II; Mw of the high molecular polymer is 5.115*10<4>, and the molecular weight distribution index is 1.761.

The invention is the project realizing method of fiber top inertia measuring device fast starting and precision guarantee, which adopts the temperature demarcating system to detect the temperature mode; found the module of the affirming parts of the fiber top and the quartz accelerator and effectively reduce the zero-bias brought by the temperature; implement small-wave transforming procession to the fiber top noises and make the stability of the zero-bias be better than 0.5 Deg. / h and the zero-bias error be smaller than 10 to power 4 g. The invention effectively reduces the zero temperature floating and noises and improves the performance of the system.

The invention relates to the technical field of wireless communication, particularly relates to a model update method and device, and is used for solving the problems that a manner for carrying out model update artificially consumes a lot of labor costs and the model update efficiency is low. The model update method provided by the embodiment of the invention comprises the steps of obtaining a model update file from a remote server by a background management server, determining a model which is corresponding to the model update file and is after the update according to the model update file as well as the model which is corresponding to the model update file and is before the update in a shared database, and storing the model after the update to the shared database; indicating a test environment server to obtain the model after the update from the shared database, testing the model after the update, and after the test environment server successfully tests the model after the update, indicating a production environment server to replace the model being used before the update with the model after the update.

The invention belongs to the technical field of image recognition and detection and specifically discloses a tire X-ray defect / flaw detection method based on the deep convolutional neural network. The detection method comprises the steps of image preprocessing, data cleaning, convolutional neural network designing and training, parameter adjusting, input image preprocessing, model training and model testing. According to the invention, tire x-ray images are classified based on the convolutional neural network. Compared with the traditional method, image features can be automatically learned during the training process of the convolutional neural network, so that the defect of manually selecting features can be avoided. The occurrence rate of accidental errors is reduced. The detection result of the method is high in reliability and tire defect / flaw can be accurately detected. The outstanding feature of applying the convolutional neural network to the image classification is that, a lot of time needs to be consumed during the training process of the model, while only a short period of time is required for the usage of the convolutional neural network model. Therefore, the cost is saved. The method is convenient for wide popularization.

The invention discloses a method for simulating roadway rock burst based on coaction of a static load and a blast load. The method comprises the following steps of: A, selecting cement mortar and ramming a model in a rock model testing machine, and embedding a pressure sensor, an acceleration sensor and a strain gage in the model body; B, maintaining the model body within 28 days, applying initial crustal stress to the model body by utilizing the rock model testing machine, and testing force and deformation field distribution characteristics of the model body under the action of the static load; C, simulating actual cavern excavation characteristics, and measuring force and deformation field distribution characteristics in the model body after excavation each time; and D, drilling in the model body, mounting explosive cartridges and detonators in the drill bores, filling with loess, and detonating the explosive. By utilizing the method, the occurrence scale and time of deep roadway rock burst can be effectively predicted, damage of rock burst disasters on mine production settings can be reduced, the mine safety production risk is greatly reduced, and the production efficiency of the mine is greatly improved.

The invention provides a multifunctional three-dimensional model testing platform for a roadbed slope and underground engineering. The multifunctional three-dimensional model testing platform is composed of an underground groove body, a ground reaction wall, a ground groove body, a servo actuator, a loading plate, a middle separation plate, an underground water system and an artificial rainfall system, wherein the ground reaction wall is located at one end of the underground groove body; the ground groove body is mounted on the side wall of the underground groove body; the servo actuator is arranged between the ground reaction wall and the ground groove body; the underground water system is arranged in the ground groove; the artificial rainfall system and a shower nozzle are arranged above the ground groove body. The construction of the multifunctional three-dimensional model testing platform for the roadbed slope and the underground engineering can further improve an existing large-size foundation and slope model testing system, so as to provide a good testing platform for a large-scale model test of engineering projects including roadbed slope supporting structures, foundation treatment, tunnels, foundation pits and the like; the construction cost is reduced and the model testing efficiency is improved.

The invention relates to a test bench of a true three-dimensional self-balancing loading model with a slide wall, the test bench has a loading system that is arranged outside a structural wall and comprises a door-typed reaction frame, the structural wall and the slide wall, the door-typed reaction frame can realize the loading of the upper, lower, left and right surfaces of a model body, only a hydraulic loading device is arranged in the outer side of the back structural wall in the forward and backward direction of the model body, and a plurality of reaction cross beams and a lateral pull rod can realize the symmetrically self-balancing loading in the forward and backward direction. The slide wall is arranged between the model body and the structural wall, and a plurality of bar observing windows are arranged surrounding the position of a perforated chamber on the structural wall. The test bench has the advantages that the test bench can realize three-dimensional graded loading, has convenient operations of excavation and shoring, and can remarkably reduce the surface friction force of the model body when loading, etc.

This symbol model detection method is a method to make the model testing practical; the main principle is to use Boolean formula quantitative determination to instead of the CTL model testing. Their method: 1.) extract state machine M of the tested system and the attributes of the tested tree temporal logic description; 2.) taking border k as the form parameters, describe k sector semantics of the attributes f; 3.) taking border k as the form parameters, convert the k sector semantic of attribute f to quantify Boolean formula [M, f] k: 4.) progressive determine whether the system M meets the reliable attribute f: the approach integrates the advantages of symbol model testing based on binary decision diagram and bounded model checking based proposition equitation meeting judgment, also overcomes their shortcomings, which are not dependent on the sort of Boolean variables, or dependent on the property to verify the full name attribute.

The invention discloses a decision-making tree based prediction method and device and relates to the field of data processing, and the method and the device can improve the visualization effect of a decision-making tree model and the decision-making tree model testing process. The decision-making tree based prediction method includes steps of generating a target training set by feature selection of data in a defective data set of prestored products according to feature attributes; training a training set by a decision-making tree algorithm to generate a decision-making tree, wherein the training set belongs to the target training set; compressing the decision-making tree to obtain a first decision-making tree; displaying the first decision-making tree by means of the visualization technology; selecting at least one test case from a test set to be sequentially input into the first decision-making tree for testing, and generating classification paths for the test cases; displaying the classification paths of the test cases in the first decision-making tree by means of the visualization technology. The decision-making tree based prediction method and device is applied to defect prediction of the products.

The invention discloses an underwater vacuum preloading centrifugal model testing device and an underwater vacuum preloading centrifugal model testing method. The device comprises a centrifugal machine and also comprises an underwater soil body self-weight consolidation system and a vacuum driving system, wherein the underwater soil body self-weight consolidation system comprises a test model box loaded with soil body; and the vacuum driving system comprises a vacuum gas-water separation box, a vacuum aspirator pump, a rotating adapter, a high-pressure air supply device and the like. The invention can simulate soil body self-weight consolidation, namely can simulate an initial stress state of the soil body. Meanwhile, the invention adopts high-pressure air to acquire vacuum and is applied to centrifugal model testing technology, underwater vacuum preloading can be simulated, land vacuum preloading also can be simulated, vacuum preloading consolidation process and the final stress state of the consolidated soil body can be simulated, and the aim of testing underwater testing soil body displacement is fulfilled. The testing simulation can perform two-dimensional (namely a plane strain problem) or three-dimensional vacuum preloading-related simulation, and the internal stress and displacement field in a soil body consolidating area can be effectively measured.

The invention relates to the field of data mining of internet broadcasting, in particular to a mining method and device of a potential anchor user, and a server. The method comprises the following steps: acquiring feature data of new anchor users and direct broadcasting rooms of the new anchor users within preset time after first-time broadcasting starting; testing the feature data according to amining model established based on a decision tree algorithm and obtaining corresponding decision-making data; calculating the decision-making data by using an information entropy computing rule to obtain potential values of the new anchor users; and selecting the new anchor user with the potential value meeting a preset condition as a potential anchor user. According the invention, a new potentialanchor user is discovered by a direct broadcasting platform automatically and fast with a high coverage rate; and thus problems of slow time effectiveness, low coverage rate, and high subjectivity during the existing manual potential anchor user discovery are solved.