88results about How to "Guaranteed difference" patented technology

The invention discloses a frame rate up-conversion motion estimation method and a frame rate up-conversion motion estimation system based on Kalman filtering. The method comprises the following steps: first, the parameters and the initial value of state of a Kalman filtering model are set to make the model accord with the actual system; then, the observed value of a motion vector is obtained through a strategy in which unidirectional motion estimation is carried out first and then a unidirectional motion vector is mapped to an interpolated frame; and finally, an observation vector is updated by a time-variant gain Kalman filtering method so as to obtain a more accurate motion vector. Based on the method, a frame rate up-conversion motion estimation hardware architecture based on Kalman filtering is put forward. High utilization rate and high throughput of the system are realized through an alternate block scanning sequence and two parallel data channels.

The invention discloses a display control method and device and belongs to the technical field of displaying icons of application programs. The display control method includes the following steps that before an application program is installed and an icon of the application program is displayed, the dominant tone of the icon is determined; according to the dominant tone of the icon, the corresponding background when the icon is displayed is determined; the icon and the determined background are displayed. Thus, the corresponding background when the icon of the application program is displayed is correlated with the dominant tone of the icon and otherness and coordination of the corresponding background when the icon of the application program is displayed are ensured. The problem that the corresponding background when the icon of the application program is displayed conforms to the tone of the icon in is solved. Resolution of the icon of the application program and visual experience of users is improved.

The invention relates to the field of pattern recognition, and discloses an unbalanced data classification method based on intuitive fuzzy integration and a system based on the method. The method comprises the following steps of: a) cleaning original data, and classifying original point-of-sale (POS) class samples according to intra-class positions to generate POS class artificial samples; b) training a base classifier by using different sample sets of inter-class approximate balance; c) converting the classification output equal utility of the base classifier into an intuitive fuzzy matrix; and d) integrating samples to be classified into the membership and the non-membership of the POS class and the negative (NEG) class by combining the weight of the base classifier, and making a classification decision. The invention has the advantages that: over learning is avoided by integrating over sampling and under sampling; the training samples of the base classifier are different, so that the difference of the base classifier is ensured; the base classifier is not specifically limited, so the method has good expandability; the intuitive fuzzy reasoning method quantitatively describes the uncertainty in classification so as to improve the performance of integrated learning; therefore, the system based on the method can better support the medical diagnosis decision and the like.

The invention discloses a graphic image recognition method and a patrol system used in patrol inspection of electric power equipment, which solves the problems that the prior electric power equipmentimage recognition method needs to manually label a large amount of sample data, which is time-consuming, labor-consuming, high economic cost and unsatisfactory recognition accuracy. The invention usesa pattern image recognition method used in patrol inspection of electric power equipment, Firstly, the image data of the same type of power equipment sample is collected, annotated sample data and unannotated sample data are collected, Secondly, the sample data is preprocessed, Then feature extraction of image data is performed, and the binary classifier model is trained based on semi-supervisedactive learning training. At last, that train model is used to classify, and the real-time image data to be detect is inputted to recognize, and the normal or abnormal image data is judged, and the normal and abnormal conditions of the power equipment are obtained; As that invention utilize sa large amount of unlabeled data, only a small amount of data is required to be label, which saves time andlabor, and achieves good recognition rate under the same condition.

The invention discloses a method and a device for constructing a GBDT (Gradient Boost Decision Tree) model, relates to the technical field of machine learning, and mainly aims to solve the problem oflow accuracy of the existing trained GBDT model. According to the main technical scheme, the method comprises the steps of obtaining a sample data set, wherein the sample data set comprises positive sample data with positive labels and unlabeled sample data without labels; training each regression tree of a GBDT model; constructing a positive sample training subset based on positive sample data inthe sample data set, sampling unmarked sample data in the sample data set to construct a negative sample training subset, and combining the positive sample training subset with the plurality of negative sample training subsets to obtain a training set of a current regression tree, training the current regression tree based on the training set of the current regression tree, and constructing a GBDT model according to each regression tree. The invention is used in the construction process of the gradient boosting decision tree.

The invention discloses a quadratic congruence equation-based method for resisting denial-of-service attacks of a wireless network, which belongs to the technical field of network safety, and mainly solves the problem that the DoS (Denial-of-Service) attacks exist in a wireless access authentication process of an IEEE (Institute of Electrical and Electronic Engineers) 802.11i protocol. The method is implemented by the following steps: an AP (Access Point) adds parameters used for constructing a puzzle into a beacon frame and broadcasts the beacon frame periodically by using the beacon frame in the IEEE 802.11i protocol; a client STA (Special Temporary Authority) acquires the beacon frame, extracts puzzle parameters from the beacon frame, constructs and solves the puzzle in combination with the current global parameters of the wireless network, adds the puzzle parameters and the solution into an authentication request message, and sends an authentication request message to the AP; and the AP receives an open system authentication message of the STA, and the AP sends the authentication request message to the STA to complete the association process when the solution of the puzzle is correct, or the request of the STA is terminated. When the method is adopted, the negotiation times increased to construct the puzzle in a traditional client puzzle scheme are reduced, and the negotiation efficiency and the DoS attach resistance of the wireless access authentication protocol are improved.

The invention relates to the field of blurred image type identification and parameter calculation, in particular to an image blurring type identifying and parameter setting method based on a fusion memory CNN (Convolutional Neural Network). The method includes the following steps: building a fusion memory network framework; setting an algorithm for each layer of the fusion memory network; getting network parameters through network training; and identifying the blurring type of an unknown image and setting the parameters. The method overcomes the problem that the network in the existing blurring identification has no independent memory function, and can improve the efficiency of image blurring type identification and parameter calculation.

The invention belongs to the field of big data prediction, and particularly provides a supply chain demand prediction method based on big data. The method comprises the following steps of: fusing a rule model and an algorithm model, Different data partitions and feature projects are constructed by using historical sales data of commodities; and two algorithms of a tree model and a linear model areadopted to construct a model for prediction, so that the difference of the model is ensured, and finally, the rule model and the algorithm model with relatively high difference degree and accurate prediction effect are fused based on a tree structure to obtain a final future sales volume prediction result. According to the method, long-term commodity sales can be accurately predicted, a data basis is provided for the supply chain, and key technical support is provided for establishing a global supply chain scheme for an enterprise.

The invention discloses a dynamic demand response solving method considering time-of-use pricing. The method comprises steps of adopting a fuzzy clustering algorithm to divide a load prediction curvein a scheduling period into a peak time period, a flat time period and a valley time period; determining corresponding points (p, q) on the basis of historical data fitting demand and price functionsaccording to the load mean value of each time period, and linearizing the functions near the points (p, q); according to the coefficients of the linear function in each time period, respectively calculating the elastic coefficients of the peak time period, the flat time period and the valley time period so as to establish a dynamic elastic coefficient matrix; establishing a dynamic demand responsemodel according to the dynamic elastic coefficient matrix; solving the dynamic demand response model so as to obtain the prices of the peak, flat and valley time periods in the scheduling period anda new load prediction curve formed after implementing the dynamic demand response. According to the invention, the technical problem that the optimal effect cannot be achieved when the demand side response is stimulated through the price change due to the fact that the fixed price elastic coefficient cannot dynamically reflect the relationship between the load characteristics and the time-of-use price in different scheduling periods is solved.

The present invention provides one kind of internal service system for hierarchial exchange network and its management and control method, and belongs to the field of computer network technology. The internal service system for hierarchial exchange network has system frame attached to the hierarchial exchange network, and each node domain has one or several service nodes with services of different types except exchange nodes. On one hand, the management and control inside the node domain is realized via service protocol; and on the other hand, the mutual operation between node domains is realized via the interaction of service message. Thus, the internal service system for hierarchial exchange network can provide various internal services. The present invention ensures the discrepancy and variety of service while raising the network interacting performance, and is favorable to service management and network safety.

The invention belongs to the technical field of neural networks, and particularly relates to a crowd counting method based on multi-scale feature fusion. The method mainly comprises the following steps: extracting feature maps of three scales from a backbone network, sending the feature maps into a feature fusion sub-network, and calculating a density map by using the fused feature maps so as to predict the number of crowds in the image, wherein the feature fusion sub-network is designed into three convolution network branches, each branch is identical in structure, adopts an attention fusionnetwork and is divided into two paths, each path is composed of a convolution layer, a normalization layer and an activation function, and the two paths are identical in input and different in outputchannel number and are a single channel and an N channel respectively; a single-channel branch learns the feature weight of a multi-channel output branch, the feature weight is multiplied by the output of a multi-channel output feature map, finally, the feature maps of three large branches are superposed, the feature maps are sent to a decoding module together to output an image density map, and the integral value of the density map is the number of people in the image. According to the invention, the people counting precision is improved.

The invention relates to a hybrid vehicle driver torque demand analyze method; the hybrid vehicle comprises an engine, a drive motor, a speed changer, and a high voltage battery assembly; the hybrid vehicle driver torque demand analyze method comprises a power system torque capability calculating method: calculating the maximum driving torque of a power source applied to a driving wheel end according to the engine present torque capability, the motor present torque capability, the power train transmission ratio, the transmission loss corresponding to the present transmission ratio, and the power train torque capacity. The driver torque demand analyze method can calculate and parse the present power system output torque according to the power system maximum output torque, the accelerator pedal opening degree and the speed information; when the accelerator pedal opening degree is at the maximum value, the torque demand is same with the power system maximum output torque.

The invention discloses a demodulator circuit used for the ultrahigh frequency radio frequency distinguishing chip, including a detecting circuit and an envelope shaping circuit, wherein the radio frequency signals received by the antenna pass through the envelope detecting circuit and the envelope shaping circuit in turn and output the S signals and the Sav signals to a comparator; the envelope shaping circuit includes a bleeder resistor, a diode, a filter resistance and a filter capacitance, wherein the bleeder resistor is connected between the signal output end of the envelope detecting circuit and the ground, the anode of the diode is connected with the signal output end of the envelope detecting circuit, while the cathode is connected with one end of the parallel branch constituted by the filter resistance and the filter capacitance, and the other end is connected with the ground, wherein the signal output end of the envelope detecting circuit is the S signal output end, while the junction of the parallel branch constituted by the filter resistance and the filter capacitance is the Sav signal output end. The invention realizes the coexistence of the low power dissipation and low cost of the demodulator circuit of the ultrahigh frequency radio frequency distinguishing chip.

The invention provides an edge self-adapting demosaicing method based on RS-SVM integration. The method comprises the implementation steps of (1) acquiring a mosaic image, (2) dividing the mosaic image into an edge area and a smooth area, (3) interpolating the smooth area through a signal correlation method, and (4) interpolating the edge area through an SVM integration method. In the step (4), an original sample set is firstly constructed on a chromatic aberration plane constructed through a color correlation and a chromatic aberration constant principle, then resampling is carried out on the original sample set through a Bagging method, resampled sample features are reduced through a rough set dynamic reduction algorithm, member support vector regression is trained through reduced samples, the chromatic aberration values of points to be interpolated are estimated through the trained member support vector regression, and finally a lost pixel value is calculated. The method can improve the edge features of the small edge area of the image, can restrain a pseudo-color effect or a sawtooth phenomenon, and can improve imaging quality.

The invention discloses a method for authenticating digital images of a digital camera. The method comprises the following steps: acquiring digital images shot by the digital camera, extracting the pattern noise of the digital camera by the acquired digital images, and authenticating the digital images on the basis of the pattern noise; wherein, the method is characterized in that the operation for acquiring the digital images comprises the following steps: selecting a plurality of independent random noise images, and displaying the random noise images on the same display in sequence; and then, shooting the displayed random noise images one by one by the digital camera to be detected, thereby acquiring the required plural digital images. The invention can greatly save time for taking pictures on the premise that the difference in image content is guaranteed, and improves the algorithm efficiency. The invention is applicable to the fields of digital forensics and image authentication as well as source detection, copyright protection, tamper-proofing and the like of digital pictures.

The invention relates to the field of learning-based neural networks, in particular to a structure of a deep memory convolution neural network and a construction method of the structure. The structure comprises a five-convolution layer clustering and dimension reduction containing convolution neural network structure, a deep memory neural network structure and a BP network structure. The invention also relates to the construction method of the network structure. According to the invention, the disadvantage of the expression ability of a complex function under the conditions of limited samples and limited calculation units can be overcome to a certain degree and efficiency of the current convolution neural network is improved.

The present invention discloses a method for the calibration and re-test of an LED beam splitter. The method includes the six steps of standard lamp fabrication, beam splitter parameter setting, single-bead correction, multi-bead confirmation, initial workpiece verification and timed re-test. According to the method of the invention, standard ED lamp beads are scientifically selected to correct a beam splitter on a production line, and therefore, the accuracy of beam splitting test data in subsequent production operation can be ensured, the qualification rate of products and the reliability of data can be improved, the number of the times of rework caused by test error can be reduced, and the improvement of production capacity and production efficiency can be favored; and the stability of the production operation of the beam splitter can be ensured, the differences of LED lamp beads in the same data classification interval can be decreased, and the consistency of the products can be enhanced.

The invention discloses a server selecting method and system, a recording server and measurement servers. The server selecting method includes the steps that the recording server receives a content distribution request sent by a client-side and forwards the content distribution request to at least one measurement server; the recording server receives and caches first lists returned by the measurement servers, wherein a content distribution server in each first list is selected by the corresponding measurement server through a CDN according to the preset rules; the recording server sends the first lists to the client-side. By the scheme, more content distribution servers with differences can be selected for the client-side, it is ensured that the content distribution servers with good distribution performance exist all the time when the client-side downloads content, and therefore after the client-side integrates server sources in self-adaption mode, the download speed of the client-side can be effectively guaranteed, and use experience of client-side users can be improved.

The invention relates to an industrialized production method of chaffy dish condensed soup. The industrialized production method comprises the following steps: (1) adding beef bones, chicken bones, beef and tallow into water for cooking and extraction, wherein the cooking temperature is 105-110 DEG C, the cooking time is 4-6 hours, and the pressure is 0.2-0.3 mpa; (2) standing for 2-3 hours; (3) carrying out vacuum concentration, wherein the concentration temperature is 55-58 DEG C, the concentration vacuum degree is -0.08 to -0.09 mpa; (4) then blending, homogenizing and filling, wherein the blending temperature is 85-88 DEG C, the blending time is 30-50 minutes, the homogenizing pressure is 15-25 Mpa, and the filling temperature is 60-75 DEG C; (5) warehousing at the temperature below 25 DEG C. According to the industrialized production method, raw materials are cheap and easily available, the industrialized production is facilitated, and the large-scale production can be realized.

The invention discloses a grouping test device for NAND Flash error rates. It is guaranteed that the products of the total erasing times and time intervals of different groups are equal from the wholeNAND Flash level, and the difference value of time points when each group finishes erasing is shortened; from the single group level, erasing operation is executed on the corresponding group every other time interval, and the difference value of erasing ending time points of different Blocks in the same group is shortened; from the single Block level, erasing operation is executed on the corresponding group every other time interval, and it is guaranteed that erasing operation is conducted on each Block at equal time intervals. Therefore, based on the three levels, the distribution uniformityof the erasing and writing process is improved, and the accuracy of error rate testing is improved. The invention further provides a grouping test method and device for the NANDFlash error rate and areadable storage medium, and the technical effect of the grouping test method and device corresponds to the technical effect of the device.

The method comprises: 1) sequentially generating the temporary user ID 'H-RNTI-TEMP'; 2) using the forward error correction coding to encode the temporary subscriber ID 'H-RNTI-TEMP' in order to generate the user ID 'H-RNTI'. The invention can avoid the error receiving of HS-SCCH in the HSDPA system.

The invention discloses a data recognition method and device, and relates to the technology of data processing. A label data set including label data samples is divided into a plurality of data subsets according to the difference of the label data samples, so that all recognizers conduct training according to the data subsets respectively, the difference of all the recognizers is guaranteed, hence, when data recognition is carried out on data to be recognized, given recognition results of the recognizers after training are obtained, then, a final recognition result of the data to be recognized is determined according to all the recognition results, and thus the accuracy of big data recognition is improved.

The invention discloses a multi-color fluorescence multiplex amplification kit for amplifying STRgene locuses of a human Y chromosome and application of the kit. Thirty STR gene locuses can be detected in a single amplification system. Compared with the prior art, the kit has the advantages that (1) the gene locuses included in the kit can guarantee the identity of the same family parting and the difference of different family partings due to the advantages high polymorphism, low mutation rate, reasonable frequency distribution and high individual identification; and (2) the kit is extremely high in detection material applicability and can detect various samples.

The invention discloses a multi-task neural network architecture searching method based on evolutionary computation, which comprises the following steps: firstly, initializing a population; evaluating the multi-task generalization abilities of individuals in the population; then randomly obtaining two chromosomes through a binary tournament selection algorithm; comparing the multi-task generalization performance of the two chromosomes; selecting the chromosome with better performance as a parent; then carrying out crossover and mutation operations on two parents to generating two children; evaluating the multi-task generalization performance of the children; then combining the children and the parents; carrying out environment selection according to an evaluation result; generating a new population; carrying out a new round of evolution until a predetermined termination condition is reached; and outputting the individual with the best multi-task generalization ability. According to the method, a genetic algorithm is used for optimizing the multi-task network model system structure, the neural network model suitable for multi-task learning can be automatically searched out without manual participation, and the cross-task information fusion capability of the multi-task network is improved.

The embodiment of the invention discloses a feature data processing method and device. The method comprises the following steps: determining outlier data in specified features of a sample set; Performing scaling processing on the outlier data in the sample set to obtain a scaled sample set, the scaled outlier data being greater than non-outlier data in specified features of the sample set before scaling; Performing clustering processing on the zoomed sample set; And based on the plurality of clusters subjected to clustering processing, respectively performing normalization processing on the specified feature data of the zoomed sample set in the specified feature interval corresponding to each cluster.

The invention discloses a rigid body generation method and device, equipment and a storage medium, and the method comprises the steps: generating at least three mark points as a current rigid body based on a given prior condition; if the given rigid body set is not null at present, determining the matching degree of the current rigid body and each generated rigid body in the rigid body set; and when each matching degree satisfies a preset condition, storing the current rigid body as a new generated rigid body in the rigid body set, and returning to execute a generation operation of a mark point until the total amount of the rigid bodies in the rigid body set reaches a preset threshold value. By adopting the technical scheme, the structural difference between the rigid bodies is ensured, and the tracking accuracy is improved.

The invention relates to a traditional Chinese medicinal preparation, and concretely relates to a blood stasis removing capsule containing Salvia miltiorrhiza and Radix Astragali, and a preparation method thereof. The capsule is composed of a traditional Chinese medicinal extract product and auxiliary materials. A weight ratio of the traditional Chinese medicinal extract product to the auxiliary materials is 10:0.2-0.8; the traditional Chinese medicinal extract product is prepared by using the following raw materials, by weight, 150 parts of Radix Astragali, 150 parts of Salvia miltiorrhiza, 75 parts of Radix Codonopsis, 75 parts of Chinese yam, 100 parts of Glabrous Greenbrier Rhizome, 75 parts of Chinese angelica, 150 parts of Millettia dielsiana, 75 parts of Semen Euryales, 150 parts of Herba Houttuyniae, 45 parts of Rhizoma Sparganii, 150 parts of Herba Patriniae, 15 parts of cinnamon, 45 parts of Rhizoma Atractylodis Macrocephalae, 22.5 parts of baked ginger, 45 parts of Fructus Toosendan, 75 parts of Sophora flavescens, 75 parts of scorpion, 45 parts of Rhizoma Curcumae, 75 parts of ground beetle and 75 parts of Corydalis tuber; and the auxiliary materials comprise, by weight, 5-10 parts of crosslinked sodium carboxymethyl cellulose, 1-3 parts of alpha-lactose and 2-6 parts of microcrystalline cellulose.

The invention discloses a one-key bid adjusting method based on engineering bidding, and a terminal, and the method comprises the steps: obtaining a target cost, an adjustment parameter, an adjustableitem and the lowest floating rate of the unit price of a material, carrying out the calculation to obtain the lowest cost and the lowest price of all adjustable elements when the adjustable elementsreach a reasonable lowest value, and enabling the adjustable item to comprise a plurality of adjustable elements; performing fitting indentation according to the current price and the lowest price inthe adjustment parameters until the fitted real-time cost and the target cost are within a preset difference value; and randomly obtaining adjustable elements to carry out random price adjustment until the real-time cost is equal to the target cost; according to the method, price adjustment can be carried out on the project conveniently and rapidly to achieve project bidding quotation required bya user, and meanwhile, rationality and difference of an adjusted scheme are ensured.

The invention provides a fine-grained entity classification method based on a diversified semantic attention model, and the method comprises the steps of obtaining a diversified attention segment sequence of a sentence based on a segmentation length and a step length; constructing a diversified semantic attention model, wherein the diversified semantic attention model comprises an attention graphprediction model and an attention feature integration model; constructing diversity constraints including attention graph constraints and attention fragment constraints, and determining a final loss function in combination with classification loss for training a diversified semantic attention model; and determining an attention graph corresponding to the diversified attention fragment sequence byusing the trained diversified semantic attention model, predicting a classification result of each time step for each fine-grained entity category in combination with a softmax network, and comprehensively obtaining a prediction result of the entities in the input sentence. According to the invention, the problem of low classification precision of fine-grained entities is solved.

The invention discloses a short-term wind power prediction method, which comprises the following steps: collecting wind power data, and dividing the wind power data into a training set and a test set; constructing a first fusion model and a second fusion model based on an SVM kernel function and a Stacking ensemble learning algorithm, and training the first fusion model by taking the training set as input to obtain a target training set; inputting the test set into the trained first fusion model to obtain a target test set; and inputting the target training set and the target test set into the second fusion model to obtain a wind power prediction result. According to the method, the output power of the wind power plant is predicted, the uncertainty risk can be reduced, better combined dispatching of the wind power generation system is achieved, and guarantee is provided for safety, stability and electric energy quality of an electric power system.