30results about How to "Accurate inference" patented technology

The invention relates to a method for determining a crankshaft position of an internal combustion engine; a speed curve of the crankshaft is detected via a time; wherein the crankshaft position (OT *) is determined by comparing the speed characteristic with a known speed course of a working cycle of the internal combustion engine; wherein the known speed curve has a characteristic of the crankshaft position portion; wherein a signal of a crank shaft coupled to the electric machine is detected; wherein a least once per working cycle occurring value (OT 'OT' 'OT' '') of the signal is taken into account in determining the crankshaft position; and wherein determining the crankshaft position by superimposing and / or combining crankshaft position as determined by comparison of the speed curve is determined by the crankshaft position as from the value (OT 'OT' 'OT' '').

The invention relates to a key with a function of displaying locking and unlocking state of a door. The key comprises a key rod and a key handle fixedly arranged at the tail end of the key rod. The key is characterized in that a micro touch switch, a controller and a time storage module are arranged in the key handle; a display screen is arranged on the surface of the key handle; a telescopic rod is inserted into the end face of one side, which is used for fixing the key rod, of the key handle in a telescopic manner; the micro touch switch, the time storage module and the display screen are respectively connected with the controller; the telescopic rod is parallel to the key rod; the telescopic rod is provided with a first position for touching the micro touch switch in the key handle; and in addition, a spring for separating the telescopic rod from the first position is arranged in the key handle. The key has the advantages that when the key is inserted into the key hole, the current time is automatically recorded and displayed on the screen. If a user needs to check whether the door is locked, the user takes out the key and checks the time recorded in the key so as to accurately judge whether the door is locked.

The invention discloses a differential detection control plane fault diagnosis system and an implementation method thereof. The system comprises a data acquisition module, an online monitoring module,an offline fault diagnosis module and a network event replay engine. The implementation method comprises steps of acquiring code calling and state change information in the controller through the online monitoring module and the data acquisition module, and constructing a context-aware system behavior model; when the current system is not abnormal, extracting a system behavior model and storing the system behavior model in a reference library as a reference model; when the system finds out abnormity, finding out a system behavior model with a fault and a corresponding reference model according to the symptom of the fault; Difference comparison is carried out on the nodes from top to bottom, and abnormal nodes are found out; And then, carrying out static analysis at the abnormal nodes, finding out the source of the fault at the code level, confirming the source through a network event replay engine in a simulation environment, and submitting the source to an administrator as a fault diagnosis report. The fault source can be positioned, and the applicability is high.

The invention relates to a method for obtaining abstracts of multiple texts in a same-topic text set. The method comprises the steps of firstly preprocessing the texts, wherein the preprocessing comprises word segmentation, stop word processing, feature selection, dimension reduction and the like; secondly, building a spatial vector model by utilizing processed feature words, and generating a distance matrix; thirdly, adding a sample density sorting method in a clustering method, constructing a circle by taking a central vector as the center of the circle and using a mean value of eigenvalue distances in a vector space as the radius, and automatically determining an initial clustering center according to sample density generated by the content similarity of the texts sorted in the circle,thereby automatically discovering the quantity of potential corresponding sub-topic sets in a document set; after the corresponding sub-topic sets are generated, performing supervised training on clustered sub-topic texts, performing scoring and marking on sentences, and extracting central sentences from different sub-topics to serve as the abstracts of the texts; and finally, outputting contentsof the abstracts. The quality of the abstracts of the texts is improved.

The invention discloses an inter-domain aggregation path analysis method based on edge node request information characteristics, and the method can achieve the detection of whether an aggregation path exists in a to-be-analyzed network or not through the edge node request information, achieves the decomposition of the aggregation path, and improves the topology reconstruction accuracy of the inter-domain network. The method comprises the steps of firstly obtaining a network topology structure with prefix aggregation; secondly, extracting an inter-domain path identifier sequence from the edge node set, and obtaining an inter-domain path identifier from the ordered path prefix sequence for prefix extraction to obtain an ordered path prefix sequence; and finally, obtaining an ordered prefix sequence from the to-be-analyzed network, comparing the ordered prefix sequence with an inter-domain path identification sequence extracted from the edge node set to obtain an aggregated prefix, identifying the aggregated inter-domain path identification prefix, and splitting the aggregated inter-domain path identification prefix; and repeating the process until there is no inter-domain path identification prefix which can be continuously split in the network. The method can effectively improve the accuracy of topology reconstruction between network domains and reduce the number of information required by network topology inference.

The present invention relates to a method and a device for determining the actual capacity of a battery using a set of characteristic curves, wherein at the beginning of a discharge process the initial voltage (UBeginn) of the battery, the discharge current strength and the temperature of the battery are determined. From this, by comparison with the data provided in the set of characteristic curves, a characteristic curve is selected and an initial state of the battery is determined. At the end of the discharge process, the final voltage (UMess) of the battery is measured. The final voltage (UMess) is compared, at the end of the discharge, with the voltage (UErwartet) expected by reason of the selected characteristic curve and, in the event that the final voltage (UMess) is smaller than the voltage (UErwartet) expected by reason of the discharge process, it is determined that the actual capacity of the battery is smaller than the available capacity assumed at the start of the discharge process.

The invention relates to the technical field of multimedia, and particularly relates to a video release prompting method and device and a storage medium. The method is used for simplifying the operation complexity of video publishing. The method comprises the following steps: an intelligent terminal running a short video application program and presenting an interface of the application program; detecting a video stored in the intelligent terminal according to a preset time period; determining the video meeting a preset publishing condition as a to-be-published video; and presenting a prompt window in an interface of the application program, the prompt window comprising information of the to-be-published video. Thus, the video publishing intention of the user can be accurately inferred, the user can be prompted to publish the video in time. Meanwhile, the operation steps of video publishing are simplified, and the operation cost of video publishing is reduced.

The invention discloses a method for quickly reconstructing a propagation network structure, which comprises the following steps of: firstly, recording the association strength between each node Vi ina network and other nodes, namely each node in Lvi, and recording the association strength in a set MIvi, the Lvi being the remaining nodes except the Vi in the network; calculating the possibility degree that each node in the Lvi in the network is a Vi father node, and recording the possibility degree in a set PIvi; sorting the nodes in the Lvi in a descending order according to the associationstrength with the Vi to obtain a set SLvi, calculating the inflection point rate of each node in the SLvi, and recording the inflection point rates in a set BPvi; and finally, initializing a directedgraph G = {V, E}, sequentially considering each node Vi for n node sets V in the network, selecting p nodes with higher inflection point rates from BPvi to construct directed edges of the Vi, adding the directed edges into E, and returning G to a user. According to the method, the association degree and the influence relationship between the nodes can be quickly determined based on the collected infection state data, so that the latent influence relationship graph can be accurately deduced.

The present invention provides an air conditioner, which uses a plurality of human detection sensors to divide the area requiring air conditioning into a plurality of areas according to the positions of people in the area requiring air conditioning, and sets the frequency of each area according to the frequency of people. regional characteristics. Based on the response result of the human body detection sensor within the first predetermined time period, a judgment condition indicating human presence is set in each area characteristic, and whether there is a person in each area is estimated based on the set judgment condition. According to the response result of the human body detection sensor in the second specified time, when a person is detected in the first area, it is inferred that there is a person at a time interval shorter than that of the second area; Since no one is estimated at a long time interval, the time required for inferring presence in the first area is set shorter and the time required for inferring no one is set longer in the first area than in the second area.

The invention relates to a secret mark watermark extraction method and system based on mobile phone and image enhancement. The phenomenon that human eyes are insensitive to yellow is utilized, a secret mark watermark of which the gray scale is within a preset range is artificially generated on the color channel; the printed matter printed with the secret mark watermark cannot perceive the watermark pattern by human eyes; when the mobile phone is used for shooting the printed matter; a watermark extraction algorithm and an image enhancement algorithm are used; an image or coded information andother contents of a secret watermark can be displayed on a mobile phone screen in real time; and after the printed matter is copied, the mobile phone shoots the copied product again, the secret mark watermark image disappears or weakens due to copying, secret mark watermark coding information can be obviously changed, and the mobile phone can accurately, conveniently and quickly judge whether theprinted matter is a genuine plate or a duplicate plate through an identification algorithm, so that the practicability is high.

The invention discloses a formation temperature testing method and device. The formation temperature testing method comprises the following steps that drifting is carried out by using a gauge cutter,and the gauge cutter is lowered to a preset depth; a tester is lowered into a wellbore, wherein the tester is provided with a plurality of test points on the same cross section, the test points are uniformly arranged in the circumferential direction of the wellbore, and the test points are in contact with the inner wall of a casing in the wellbore; the tester is lifted, and in the process of lifting the tester, a fixed-point test is carried out at a predetermined interval distance; and temperature data of the test points in the tester are acquired, and the formation temperature is determined according to the temperature data. The formation temperature testing method and device can accurately obtain the formation temperature, so that better guidance is provided for field operation.

Present invention relates to whole number frequency multiplication rate error correction device and method for digit multimedia broadcast receiver. It contains multiplier for multiplication of received signal and modified integral multiple frequency error and clearing integral multiple frequency error, PRS generator for outputting standard PRS signal, part correlator for obtaining correlation numeric value, calculator, and corrector. Said invention utilizes partial correlation accurately concluding and clearing integral multiple frequency error, accurately to proceed frequency synchronism even if on vile receiving condition, using PRS signal to proceed modification to tiny signal timing offset, to raise whole ground wave DMB receiver performance.

The invention relates to the technical field of biology, and provides a kit, a compound system and application of the compound system in obtaining individual ages. The kit comprises an amplification primer used for amplifying at least one CpG site in 10 specific CpG sites in sample genome DNA and / or an extension primer used for obtaining the methylation rate of at least one CpG site in the 10 specific CpG sites, and the composite system comprises a genome DNA extraction system, an amplification detection system and a data acquisition system. The application is the application of the kit and the composite system in obtaining the age of an individual. The kit provided by the invention has the advantages of low cost, high speed and high accuracy for obtaining the methylation rate of 10 specific CpG loci in sample genome DNA, can be used for quickly and accurately deducing the age of an individual, is particularly suitable for being applied to the field of criminal investigation identification, and can quickly provide clues for case detection.

This invention relates to a presumption device of digital TV channel, which comprises a correlative value generating part, used for getting base belt signal and the signal during train; a correlative value maximum hunting part used in the pulse of the actual channel; the correlative value guidance generating part, used for outputting the correlative guidance corresponding to the CIR assigned range; the memory, used for outputting the self correlative value array of the train range signal; the computing part, used for computing the CIR assigned value with the correlative value radio guidance and the array of the self correlative value; the CIR separating part, used for removing the noise in the CIR assigned value.

Systems and methods for disease diagnosis that detect a plurality of biomarkers by receiving concentration values of the biomarkers, construct a training set using samples of the biomarkers, and correlate the biomarker concentration values to diagnose a disease.

The invention belongs to the field of graph theory learning and deep learning analysis, and particularly relates to a molecular regulation relation prediction method and system based on a knowledge graph, and the method comprises the following steps: (1), preparing a training data set; (2) constructing and training a directed graph model; (3) constructing and training a neural network; and (4) performing actual prediction. According to the method, the overall process design of the prediction method is improved, and the method comprises the following steps: firstly, carrying out optimal feature coding on a directed regulation relationship and each regulation molecule node (namely each regulation molecule instance) in a directed graph in combination with graph learning; the prediction capability of the directed graph is further enhanced by using a deep neural network framework, and the prediction accuracy of directed regulation and control is improved after iterative training.

The invention relates to a method for obtaining multi-text abstracts in a set of texts with the same subject. Firstly, the text is preprocessed, including word segmentation, stop word processing, feature selection, dimension reduction, etc.; in the next step, the processed feature words are used to construct a space A vector model is used to generate a distance matrix; then a method of sorting the sample density is added to the clustering method. The center vector is used as the center of the circle, and the average value of the distance of the eigenvalues ​​in the vector space is used as the radius to construct a circle. According to the similar content of the sorted texts in the circle The density of the samples generated by the degree to automatically determine the initial cluster center, so as to automatically find the number of potential corresponding sub-topic sets in the document collection; Training, scoring sentences, labeling, extracting central sentences from different subtopics as multi-text summaries, and finally, the method outputs the content of the summaries; improving the quality of multi-text summaries.

The invention provides a multiwindow-based target tracking method, belonging to the computer vision technology. The method comprises the following steps: 1) reading in the image of the initial position of the target, constructing N-numbered tracking windows according to the target characteristics and measuring the compensation di between each tracker and the initial position of the target X0=(x0, y0); 2) reading in the next frame image and computing the new position Yi of each tracker on the current image frame; 3) computing the weight Wi of all the trackers; 4) compensating the new positions Yi of all the trackers on the current image frame by the compensation di and obtaining the value shown in the specification; 5) computing the target position parameter E(X) with the weight Wi of the trackers and the value shown in the specification; and 6) initializing all the trackers according to the target position parameter, updating the reference windows of the trackers and continuously reading in the next frame image, wherein each window corresponds to one tracker. The tracking method can improve the tracking performance of the system under the complex conditions of shading, lighting variation, etc and has strong robustness, and the target tracking result is free from control by the result of a certain tracking window.

The invention relates to the technical field of subway AFC systems, and relates to an operation and maintenance management method of a subway AFC system based on the Internet of Things. An integratedInternet of Things framework of the subway AFC system is constructed. An operation monitoring device and a fault monitoring device are arranged at terminal equipment of a subway AFC. picture information of the visitor flow at the terminal equipment and Picture information of running state are obtained, the picture information is transmitted to a site AFC system; the station AFC system transmits the data to the central AFC system; wherein a visitor flow example picture and a terminal equipment fault example picture are stored in the central AFC system; a corresponding image processing and recognition algorithm is adopted; the inference information is transmitted to an operation and maintenance management module of the station AFC system; The operation and maintenance management module sendsout an action instruction, rapid exchange of information can be achieved through the integrated Internet of Things framework, timely response can be conducted according to different operation conditions, meanwhile, emergency terminal equipment can be started, timely evacuation of people flow is guaranteed, feedback and uploading can be conducted on the maintenance condition, and tracing management of the operation and maintenance state is facilitated.

The invention discloses an inter-domain aggregation path analysis method based on the characteristics of edge node request information. The method can use the edge node request information to detect whether there is an aggregation path in the network to be analyzed, and decompose the aggregation path to improve the inter-domain network topology. Accuracy of reconstruction. First, obtain the network topology structure with prefix aggregation; then, extract the inter-domain path identifier sequence from the edge node set, obtain the inter-domain path identifier from the ordered path prefix sequence for prefix extraction, and obtain the ordered path prefix sequence; finally, obtain the ordered path prefix sequence from the Obtain an ordered prefix sequence in the network, and compare it with the inter-domain path identifier sequence extracted by the edge node to obtain the aggregated prefix, identify and split the aggregated inter-domain path identifier prefix; repeat this process until there is no continuation in the network The split interdomain path identifier prefix. This method can effectively improve the accuracy of topology reconstruction between network domains and reduce the amount of information required for network topology inference.

The invention discloses a traffic bottleneck identification method in an urban traffic network, and relates to the technical field of traffic information control, and the method comprises the following steps: 1, constructing virtual road network information; 2, collecting traffic flow data in the virtual road network; 3, carrying out fusion processing on the collected data; 4, generating a virtualroad network PA back-stepping model by using a PA back-stepping technology; 5, establishing a virtual road network database based on the original road network database; 6, calculating the data in thevirtual road network database; 7, calculating and mining according to the data to obtain bottleneck points, and diagnosing real bottleneck points; 8, outputting reasons of the bottleneck points. According to the method, the road traffic flow information is monitored through the architecture virtual nodes and the virtual road sections, the real-time model is generated through the PA back-steppingtechnology, the road traffic flow information is simulated through the model, the bottleneck points are accurately deduced, and more accurate and scientific suggestions are provided for traffic jam treatment and optimization.

The invention provides a train flow calculation method and device, electronic equipment and a storage medium, and the method comprises the steps: a system determines a first target train entering a target marshalling station within three hours based on the target dynamic information of all trains in the system; and target traffic flow calculation of the first target train and / or the second target train in the target marshalling station are / is executed based on the target dynamic information of the first target train and / or the target dynamic information of the second target train parked in the target marshalling station. According to the invention, the target dynamic information is generated by acquiring the electronic transport unified data in real time, and the acquisition of the electronic transport unified data and the generation of the target dynamic information are uniformly completed in one system, so that the generated target dynamic information is ensured to be accurate; and target traffic flow calculation is accurately executed based on the target dynamic information of the first target train and / or the target dynamic information of the second target train, so that the problem of flow leakage and multiple flows in the traffic flow calculation process is solved.

The present invention relates to the field of computer vision, a weakly supervised image semantic segmentation method based on candidate regions and neighborhood classifiers, using candidate regions as the basic processing unit, and applying neighborhood classifiers to image semantic segmentation to obtain high-precision, less susceptible Weakly Supervised Image Semantic Segmentation Confused by Noisy Labels. As a key technology urgently needed in the current field of computer vision, the method of the invention can realize the label prediction of each pixel in the test image, and can obtain relatively high semantic segmentation accuracy of the weakly supervised image.

The invention discloses a multi-focus image fusion method based on a direction filter and a deconvolution neural network, which mainly solves the problems of low fusion precision and insufficient image indexes in the existing multi-focus image fusion method. Its implementation is as follows: select the multi-focus image data set to be fused; design the initial high-pass directional filter and low-pass filter for the data set; preset this filter group into the deconvolution neural network model; The network is trained by the method of constructing the image and the error of the input image; two images to be fused are input into the trained network, the feature maps of the two images are obtained respectively, and the feature maps of the two images are fused to obtain the fusion feature Figure; the fusion feature map and the specific filter bank are convolved and summed to obtain the final fusion image. The invention can improve the image fusion precision, and the filter group can be set in any size and direction, has high precision and wide application range, and can solve the problem of unclear images caused by out-of-focus multi-focus images.

The invention discloses a differential detection control plane fault diagnosis system and an implementation method thereof. The system comprises a data acquisition module, an online monitoring module,an offline fault diagnosis module and a network event replay engine. The implementation method comprises steps of acquiring code calling and state change information in the controller through the online monitoring module and the data acquisition module, and constructing a context-aware system behavior model; when the current system is not abnormal, extracting a system behavior model and storing the system behavior model in a reference library as a reference model; when the system finds out abnormity, finding out a system behavior model with a fault and a corresponding reference model according to the symptom of the fault; Difference comparison is carried out on the nodes from top to bottom, and abnormal nodes are found out; And then, carrying out static analysis at the abnormal nodes, finding out the source of the fault at the code level, confirming the source through a network event replay engine in a simulation environment, and submitting the source to an administrator as a fault diagnosis report. The fault source can be positioned, and the applicability is high.

The invention relates to the technical field of deep learning, and particularly discloses a cardiovascular and cerebrovascular disease recognition system based on an echo state network, which comprises a pulse signal sensor, a pulse signal processing module and a recognition module, wherein the pulse signal sensor is used for collecting fingertip pulse waveform curves of a normal human body, a human body with cardiovascular and cerebrovascular diseases and an unknown human body; the pulse signal processing module is used for converting the fingertip pulse waveform curves to obtain corresponding human body heart rate characteristic information; and the recognition module is used for training the echo state network according to the fingertip pulse waveform curves of the normal human body and the human body with the cardiovascular and cerebrovascular diseases, and training a connection weight from a hidden layer to an output layer so as to recognize the cardiovascular and cerebrovascular diseases according to the human body heart rate feature information corresponding to the fingertip pulse waveform curve of an unknown human body. The echo state network adopted by the system is simple in structure, rapid in training and accurate in inference, and recognition of cardiovascular and cerebrovascular diseases can be completed only through the fingertip pulse waveform curves of patients.

The invention relates to the technical field of biotechnology and genetics, and provides a primer composition, a kit and a method for detecting a Y-SNP haplogroup based on a next-generation sequencing technology and application of the primer composition, the kit and the method. According to the Chinese population haplogroup frequency, 381 Y-SNP loci are screened out by referring to the latest version of ISOGG website (Version: 15.58) in 2019, and a method for Chinese population haplogroup detection is designed by utilizing a next-generation sequencing platform and a multiple PCR amplicon capture technology. The method is high in sensitivity, good in specificity, high in applicability to case samples, good in detection capacity to degraded samples and mixture samples, capable of obtaining high-resolution haplogroup results, careful in haplogroup division and accurate in inference. The method can be applied to family investigation and paternal family source inference of unknown sample geographical sources, and a family source inference technology with high sensitivity, high resolution and strong system efficiency is provided for forensic medicine practice.

The invention provides a knowledge graph construction method and device, electronic equipment and a storage medium. The method comprises the steps of obtaining a plurality of entities to be classified and all HTML data corresponding to each entity; extracting description texts of the corresponding entities from the HTML data of the entities, and constructing a training data set based on the names of all the entities and the description texts of all the entities; training a model by using the training data set, and identifying the category to which each entity in a plurality of entities belongs based on the trained model; and constructing the knowledge graph based on the plurality of entities and the category to which each entity belongs. A large amount of type information presented in a network liquid level in an unstructured text form can be extracted, and a model with high accuracy and good effect is trained to classify entities. And based on massive data, constructing a comprehensive and deep knowledge graph which is formed by fusing or linking a plurality of libraries in an open field and a specific field.