218results about How to "Make it easier to get" patented technology

The invention discloses a tree species identification method based on transfer learning, comprising the following steps: S1, collecting tree images and making tree species image data sets; S2, data enhancement is performed on the original data set image to expand the number of pictures; S3, acquiring more than one pre-training model based on convolution neural network trained on large image dataset, S4, training the pre-training model with tree species image dataset, optimizing one or several full connection layers of the pre-training model in the training process, thereby training a pluralityof classifiers based on convolution neural network; Test the accuracy of each classifier and select the classifier with the highest accuracy. S5 performs tree species identification using the classifier selected in step S4, thereby obtaining an identification result. The invention recognizes tree species based on the transfer learning mode, so that a model with good generalization ability can betrained under the condition of limited sample size of tree species, and the tree species recognition rate under the complex background is greatly improved.

The invention relates to a road traffic OD information collection system for license plate recognition data, which comprises three subsystems of a roadside data collection subsystem, a roadside communication subsystem and a license plate recognition collection and processing platform subsystem; high-speed communication, transmission and data invocation between the data collection subsystem and the license plate recognition subsystem are finished by utilizing the roadside communication subsystem. The processing method comprises the following working steps from step 1 to step 7: storing and uploading the license plate recognition information of an external field collection section; secondary cleaning and secondary recognition; tracing and updating vehicle routes in real time; processing single-vehicle track tracing and revising single-vehicle OD errors caused by errors generated during the license plate information collection and transmission; counting the number of specific vehicles with the same types, the same starting points and destinations and in the same period when passing through the destinations; classifying and storing the revised OD values according to the information of the statistics periods, the vehicle types, the destinations and the like; and displaying an OD distribution matrix in forms of reports and graphs.

The invention discloses a method and system for acquiring indoor navigation data based on an intelligent mobile terminal. The method comprises the steps: the intelligent mobile terminal constructs an indoor road network model and generates a collecting optimal path; the intelligent mobile terminal collects sensor data during user's walking along the collecting optimal path and detects user's stride and identification turning; a dead reckoning method is used to perform position labeling, and position labeling results are corrected through video image data collected by a camera; corrected position results are used to perform position labeling on data obtained by other sensors except the camera and perform labeling on floor positions; indoor information points and landmark data are extracted, and an indoor texture image library and a landmark database are established. According to the method and system for acquiring indoor navigation data based on the intelligent mobile terminal, an indoor road network is constructed automatically, a smartphone is utilized to collect multisource sensor information on navigation routes, a position fingerprint database and a landmark database are provided for indoor pedestrians, and basic data are provided for indoor positioning and route guidance.

The invention provides a flight dynamic information display method, system and equipment and a storage medium. The method comprises the steps that flight information is acquired; a flight display interface with a flight information area is provided, and a take-off information bar, a stopping information bar and an arrival information bar are displayed in the flight information area along the timeaxis in sequence; current flight information is acquired, and flight information which is displayed in the flight information area is updated; whether or not a current flight has a new event of landing preparation or course reversal is judged. According to the flight dynamic information display method, based on the time dimension, all flight states of take-off, arrival, stopping, landing preparation and course reversal in the process are arrayed in sequence, an airport gate, a check counter, a baggage turntable, an exit and predicted take-off / arrival time are integrated, and according to the flying progress, a user can conduct quickly position and focus on the flight dynamic information of an airport according to a used scene; stopping, landing preparation and course reversal data is supplemented, more sufficient data information is provided on a complex flight scene, and the method helps the user to successfully complete travel.

The invention provides a floating seal ring with a composite working face. The floating seal ring with the composite working face comprises an upper sealing ring body and a lower sealing ring body which are symmetrical in structure. The sealing ring bodies are made of metal materials, a circle of groove for arranging a sealing ring is formed in the outer side of each sealing ring body, a circle of wear-resistant belt assembling groove is formed in the working face of each sealing ring body, the width 12 of the wear-resistant belt assembling groove is 25%-75% of the total width 11 of the working face, a wear-resistant belt is fixedly assembled in the wear-resistant belt assembling groove, the wear-resistant belt is made of aluminum oxide ceramic materials or other engineering ceramic materials, the friction surface of the wear-resistant belt and the working faces of the sealing ring bodies are located on the same plane, or the friction surface of the wear-resistant belt protrudes out of the working faces of the sealing ring bodies. The floating seal ring with the composite working face can effectively improve the wear resistant degree of the working face, prolong the service life, and is easy to machine and low in production cost.

The invention discloses an overhead line fault probability prediction method based on a comprehensive state health degree. The overhead line fault probability prediction method is characterized by comprising calculation of four factors of an overhead line, wherein the factors comprise aging, health degree, external weather and operation conditions. On the basis of data completeness, an algorithm can be used for calculating the fault probabilities of all lines at a time at each monitoring interval, early warning can be conducted on the lines with the high fault probabilities, the parts with many deduction items are displayed to serve as early warning reasons, and the health degree evaluation method is digitalized in rule and does not depend on evaluation after manual observation any more. The data acquisition difficulty is low, the information is comprehensive, the evaluation interval is small, and the risk change condition of the line can be continuously reflected; and the prediction result is more persuasive, and the expansibility of the proportional risk model is stronger.

The present invention discloses a parameter identification method for a linear model of a pump turbine. The method comprises the following steps: coupling an approximate elastic water hammer model ofa water diversion system, an IEEE six-parameter model of the pump turbine, a generator motor system model and a PID governor model to obtain a linear model of a pumped storage unit speed control system under the power generation condition, and constructing a linear mapping model of the pumped storage unit speed control system; determining the upper and lower boundaries of to-be-identified parameters of the pumped storage unit speed control system, and obtaining parameter identification sample data; using a BP nerve network method to train the parameter identification sample data, and establishing a BP neural network parameter identification model; and taking running measured data of the pumped storage unit speed control system as the input of the BP neural network parameter identificationmodel, and solving to obtain a parameter identification result of the pumped storage unit speed control system. According to the method disclosed by the present invention, not only linear model parameters can be determined easily and the accuracy of the data is relatively ideal, but also the method has the advantages of less difficulty in obtaining samples and better flexibility in selecting samples.

The invention discloses a method for recognizing and diagnosing a HVAC system gradual fault based on the deep learning. The method uses the operational data of the health system to train the principalcomponent analysis (PCA) and the cyclic neural network (RNN) model, uses the PCA model to recognize the system fault, and uses the RNN model to diagnose the system change process and the gradual fault, thereby judging the specific fault of the system. Since the RNN belongs to the deep neural network, and has the characteristic of history information storing, the model can well fit a highly nonlinear time-varying system similar to HVAC. The method has high precision for early gradual fault diagnosis, and can accurately predict parameters of a unsteady system; and the data acquisition difficulty is greatly reduced. The method has a good effect on the gradual fault diagnosis, and is a feasible and high-precision gradual fault recognizing and diagnosing method.

The invention discloses an automatic segmentation method of a dangerous organ based on a convolution neural network, belonging to the technical field of image processing. The method comprises the following steps: S1, acquiring patient CT image data and corresponding labeling data; S2, preprocessing the CT graphic data and the corresponding labeling data; 3, establish a 3D convolution neural network model, inputting data block, and obtaining a prediction result image output by that model; S4: optimizing the prediction result image output by the 3D convolution neural network model. The present invention is only based on CT image data, has small difficulty in obtaining original data and wide application range, and can realize automatic segmentation of dangerous organs in CT image, and the segmentation process does not require manual interference, effectively improves segmentation efficiency and segmentation result precision, increases post-processing operation, and further optimizes the segmentation result.

The invention provides a face changing method and device, electronic equipment and a storage medium. The method comprises the following steps: determining a source face image and a target face image; and inputting the source face image and the target face image into a face generator to obtain a face changing image output by the face generator, wherein the face generator is obtained by adversarial training with the face discriminator based on the sample image pair, the face discriminator is used for discriminating the authenticity of the sample image pair and the sample face-changing image thereof, and the sample face-changing image is obtained by the face generator by changing the face based on the sample image pair. According to the method and device, the electronic equipment and the storage medium provided by the invention, the face generator obtained by training in combination with the face discriminator has relatively high generalization performance, can adapt to any face image and has no identity limitation. Moreover, the training sample of the face generator does not need a large number of face images with the same identity, thereby effectively reducing the sample obtaining difficulty and obtaining cost, and further improving the application range of the face changing method.

The invention discloses a mechanical arm action learning method and system based on third-person imitation learning. The method and system are used for automatic control of a mechanical arm so that the mechanical arm can automatically learn how to complete a corresponding control task by watching a third-party demonstration. According to the method and system, samples exist in a video form, and the situation that a large number of sensors are needed to be used obtaining state information is avoided; an image difference method is used in a discriminator module so that the discriminator module can ignore the appearance and the environment background of a learning object, and then third-party demonstration data can be used for imitation learning; the sample acquisition cost is greatly reduced; a variational discriminator bottleneck is used in the discriminator module to restrain the discriminating accuracy of a discriminator on demonstration generated by the mechanical arm, and the training process of the discriminator module and a control strategy module is better balanced; and the demonstration action of a user can be quickly simulated, operation is simple and flexible, and the requirements for the environment and demonstrators are low.

The invention discloses a mapping method between an ontological schema and a relational database schema. The mapping method comprises the steps that an RS2O graph is established as an intermediate model of the ontological schema and the relational database schema, and the ontological schema and the relational database schema are respectively converted to an RS2O graph; in combination with the RS2O graph of the ontological schema and the RS2O graph of the relational database schema, a dual connected graph is established, and flip edges are added on mapping pairs in the dual connected graph to construct an induced propagation map; fixed point iteration calculation and artificial feedback are performed on the similarity of the mapping pairs, and matching is performed according to calculation results to obtain the mapping relations between relational database schema elements and ontological schema elements. Candidate matching relations are obtained through calculation, the matching relations are determined after artificial verification, and therefore labor cost is greatly lowered; while the obtaining difficulty and time cost of the mapping relations are lowered, it is effectively guaranteed that the correctness rate of the finally obtained mapping relations is always 100 percent. The mapping method is applicable to data exchange, data sharing and data integration between heterogeneous relational databases.

The invention relates to the technical field of blood pressure measurement, in particular to a photoplethysmography based noninvasive continuous blood pressure measurement method. The method comprises the following steps that 1, a volume pulse wave of a measuring object is collected, and a digital volume pulse wave is obtained; 2, the maximum amplitude Hp, minimum amplitude Ht and cardiac cycle T of the digital volume pulse wave are extracted, and then the pulse waveform area Aarea and pulse wave average amplitude Have of the digital volume pulse wave in one cardiac cycle T are obtained; 3, a second-order differential volume pulse wave is obtained, a first rise fulcrum P1 and a second rise fulcrum P2 in one cardiac cycle T of the second-order differential volume pulse wave are extracted, and then pulse wave translation time PWTT is obtained, wherein the pulse wave translation time is a time difference of the first rise fulcrum P1 and the second rise fulcrum P2; 4, a pulse-pressure difference DP is obtained; 5, systolic blood pressure SBP is obtained; 6, diastolic blood pressure is obtained. Accordingly, the measurement precision is high, and measurement is easy and convenient.

The invention provides a new energy power station model verification method based on variable impedance and a controlled alternating-current voltage source. The model verification method comprises steps of acquiring actual-measurement voltage, active power and reactive power amplitude of integration points of a wind power plant and a photovoltaic power station; setting amplitude and phase angle of the controlled alternating-current voltage source as fixed values; according to the voltage and active / reactive power amplitude of integration points of the wind power plant and the photovoltaic power station, calculating resistance and reactance values of the variable impedance or setting the resistance and reactance values of the variable impedance to be fixed value so as to calculate the voltage amplitude and the phase angle value of the controlled alternating-current voltage source, and using the variable impedance and the controlled alternating-current voltage source in which the real-measurement data is input to acquire precise value of an external power grid; acquiring active / reactive power values of the integration points via digital simulation calculation, comparing the active / reactive power values with the real-measurement data, and judging that the wind power plant and the photovoltaic power station models are correct if the active / reactive power values is consistent with the real-measurement data, and judging that there are errors if not; and further shrinking the actual-measurement equivalent points. The model verification method can be used for positioning specific positions of error sources of the wind power plant and the photovoltaic power station models.

The invention discloses a production technology for producing sintered wall bricks or building blocks with glaze on surfaces from construction waste and belongs to the technical field of recycling of construction waste for brick production. Green bricks are produced from bricks, concrete blocks, viscous spoil and other construction waste as well as garden waste, trees, plant straw, rice chaff, sawdust or waste in paper mills, sugar refineries and coffee plants and other mixtures of plant materials, and molded green bricks (green building blocks) or dried and preheated green bricks (green building blocks) or sintered bricks (building bricks) are glazed and fired. With the adoption of the method, the construction waste, the garden waste and industrial waste are utilized effectively, the recycling efficiency of waste is improved, glaze application is easier due to addition of plant materials, glaze can permeate micropores in surfaces of the bricks (building blocks), glaze layers and the bricks are bonded more closely and firmly, and the probability of falling of glaze in the later period is far lower than that of existing exterior wall tiles.

The invention discloses a manufacturing method of a lamellar anisotropic rock similar material, and belongs to the technical field of rock surveying. Common cement, river sand, water and other raw materials are casted into a mould so as to form a mold, the lamellar anisotropic rock similar material is manufactured and is then cut into an international-standard cylinder sample, and a triaxial testis carried out. The manufacturing method is simple and fast, the materials are wide in source, low in cost and good in forming effect, and the rock parameters can be adjusted conveniently.

The invention discloses a low-voltage transformer area topology identification system based on characteristic current, and relates to the technical field of electric power. According to the system, branch terminals are installed on all power supply branches of a low-voltage transformer area, by means of the mode that a characteristic current is injected into one branch terminal and the characteristic current is identified by the other branch terminals, the topology logic of the power system of the low-voltage transformer area can be automatically identified, the automation and intelligence level is high, the acquisition difficulty of the power line topology of the low-voltage transformer area is greatly reduced, and the acquisition accuracy is improved.

The invention discloses an HTTPS flow identification method, an HTTPS flow identification device, a terminal device and a storage medium. The HTTPS flow identification method comprises the steps of: obtaining HTTPS flow data; calculating each XDR characteristic variable of the HTTPS flow data according to a pre-set calculation calibre, so that all XDR characteristic variables of the HTTPS flow data are obtained; and inputting all the XDR characteristic variables of the HTTPS flow data into a pre-created support vector machine classification model, so that the flow type corresponding to the HTTPS flow data is obtained, wherein the flow type is a picture flow type or a non-picture flow type; the support vector machine classification model is pre-created according to XDR characteristic variable data of pre-established sample flow data; and the sample flow data is data labelled as picture flow and non picture flow. By adoption of the embodiment of the invention, the data obtaining difficulty and the calculation complexity of picture flow identification of HTTPS flow can be reduced; and the identification cost is reduced.

The invention discloses a wind speed numerical forecasting correction method based on anomaly deviation, and a system thereof. The wind speed numerical forecasting correction method comprises the steps: obtaining historical real-time wind speed data and historical prediction wind speed data, and calculating the average value of the historical real-time wind speed data and the historical predictionwind speed data; selecting a time period to be predicted, and calculating a real-time wind speed average value, a real-time wind speed anomaly value, a predicted wind speed average value and a predicted wind speed anomaly value; selecting a certain moment point t in the to-be-predicted time period, calculating a real-time wind speed anomaly value and a predicted wind speed anomaly value of the moment point t, and calculating an anomaly deviation value of the moment point t according to the real-time wind speed anomaly value and the predicted wind speed anomaly value; calculating the anomaly deviation value of the to-be-measured time point t + 1 according to the anomaly deviation value of the time point t; and correcting the wind speed data at the moment t + 1 to be measured according to the predicted wind speed data at the moment t + 1 to be measured, the historical actual wind speed data average value, the historical predicted wind speed data average value and the anomaly deviation value at the moment t + 1 to be measured so as to provide information for forecasting the wind speed data.

The invention relates to a wind power plant / photovoltaic power station model verification method based on a quick response generator. The method comprises the following steps: acquiring a measured voltage value, a measured frequency value, a measured active power value and a measured reactive power value at a wind power plant / photovoltaic power station grid-connection point; realizing accurate equivalence of an external power grid by taking the measured voltage value and the measured frequency value as control reference values of an excitation and speed regulation system of the quick response generator; and acquiring an active power value and a reactive power value at the grid-connection point through digital simulation computing, comparing the acquired active power value and reactive power value with the measured active power value and reactive power value, proving that a wind power plant / photovoltaic power station model is accurate if the acquired active power value and reactive power value are consistent with the measured active power value and reactive power value, otherwise indicating an error. Along with further contraction of a measured equivalent point, the method can be used for positioning the specific position of an error source of the wind power plant / photovoltaic power station model.

The invention discloses a high-speed railway passenger travel time-varying demand prediction method based on equal-intensity distribution. The method comprises the steps of counting high-speed railwaypassenger travel related data from a railway ticketing system; constructing a train taking scheme subset corresponding to ticket purchase time stage division according to the passenger travel relateddata, and determining a flow suction interval of each train taking scheme in the train taking scheme subset; distributing the passenger flow of each train taking scheme into the flow suction intervalof each train taking scheme according to the idea of equal-intensity passenger flow distribution; and obtaining the passenger travel time-varying demand by superimposing the passenger flow distributed to each flow suction interval. The passenger travel related data required in the invention is low in obtaining difficult and obtaining cost, the predicted result is high in precision, and the predicted result provides necessary basic data support for the operation organization optimization of the high-speed railway.

The invention provides a lake and reservoir water ecological health evaluation technical method, and belongs to the technical field of ecological environment protection. According to the method, firstly, eutrophication, macroinvertebrate and floating algae indexes of a lake and a reservoir are measured; then by the method disclosed by the invention, a comprehensive nutritional status index, the macroinvertebrate index and the floating algae index are calculated and are subjected to normalization processing, and a water ecological health index is calculated; and finally, according to a water ecological health index grading standard divided by the method, water ecological health statuses of the lake and the reservoir are evaluated. According to the method disclosed by the invention, a small amount of indexes are adopted, the indexes include a great information quantity, the adopted indexes, particularly the biological indexes are low in acquisition difficulty and low in requirement for professional knowledge, and the method is convenient to implement and apply in a large range. By applying the water ecological health evaluation technical method established by the invention, comprehensive evaluation on lake and reservoir water ecological health can be implemented from the points of water quality and biology.

The invention provides a hyperspectral image acquisition method based on a very few of optimization selection wavebands. The method includes the steps that firstly, high-definition images with a very few of optimization selection wavebands are obtained; then low-definition hyperspectral image sets with ground feature classes similar to those of the obtained high-definition images are selected, dictionary pre-training is carried out, and an over-complete spectrum dictionary is obtained; then the high-definition images with a very few of optimization selection wavebands are subjected to sparse representation under the nothing but negative constraint condition, and a sparse representation coefficient is obtained; finally, hyperspectral images with high definition are obtained through the dictionary and the spare coefficient. The obtained high-definition images with a very few of wavebands contain hyperspectral important waveband information, and it is possible to obtain high-quality spectrums; dependence on hyperspectral images under the same scene is avoided, and the data acquisition difficulty is reduced; the defects of end member decomposition on the descriptive power of spectral characteristics of the hyperspectral images are overcome, and the precision of obtained spectrums and the obtaining effectiveness are effectively improved.

The invention belongs to the technical field of codes and a method and tool for automatically filling code samples by using JavaFX. The method comprises the following steps of: accessing different uniform resource identifiers, positioning elements in an HTML document, crawling text contents positioned in the elements, and obtaining problem information and application program interface information;accessing a problem detailed information page through a crawler, extracting a use scene, establishing mapping of an application program interface type, a use scene and a code sample, and processing original data into data meeting document requirements; and operating a document object by using a program, predefining a combination form of document templates and data through code logic, and finallytaking out the stored data, embedding the data into the document, and storing and generating a final application program interface document.

The invention discloses a three-dimensional indoor model retrieval method based on deep learning. The method comprises the steps of establishing a model feature database; performing main body detection on the input two-dimensional image, and segmenting a main body image; performing feature vector extraction on each segmented main body image by using a feature extraction network; carrying out indoor model standard rendering graph retrieval in a model feature database by calculating the similarity of the feature vectors, and obtaining all similar indoor model standard rendering graphs; and sorting the similar indoor model standard rendering graphs according to the size sequence of the similarity, and selecting the indoor model standard rendering graph with relatively high similarity as a retrieval result to be output and displayed. The three-dimensional indoor model retrieval method based on deep learning has good distinguishing capacity for similar model individuals, meanwhile, the acquisition difficulty of a retrieval input source is reduced, and retrieval of the three-dimensional indoor model can be efficiently and accurately achieved.

The invention discloses a music score generation method and device based on audio file identification and terminal equipment. The music score generation method comprises the following steps: acquiringan audio file to be converted and converting the audio file to be converted into a spectrogram to be identified; performing feature extraction on the spectrogram to be identified according to the spatial density distribution features of the points in the spectrogram to be identified to obtain a feature vector to be converted; and inputting the to-be-converted feature vector into a preset music score conversion model, so that the music score conversion model generates a music score corresponding to the to-be-converted audio file according to the to-be-converted feature vector. By implementingthe embodiment of the invention, the corresponding music score can be automatically generated, and the difficulty of obtaining the music score by a user is reduced.

The invention provides a battery SOH evaluation model construction method and a battery SOH value evaluation method, and relates to the technical field of battery evaluation, and the method comprises the steps: obtaining a first training data set and a second training data set; wherein the first training data set comprises a plurality of first battery data samples which are not labeled with SOH labels, the second training data set comprises a plurality of second battery data samples which are labeled with SOH labels, and the number of samples of the second training data set is far smaller than the number of samples of the first training data set; based on sample intervals among the first battery data samples in the first training data set, determining a plurality of sample pairs from the first training data set, and training a pre-constructed similarity model by using each sample pair; and constructing a battery SOH evaluation model according to the similarity model obtained by training and / or the second training data set. According to the invention, the difficulty and cost of battery SOH evaluation can be effectively reduced.

The invention relates to an image processing method, an image processing device and a storage medium. The method comprises the steps of obtaining a to-be-processed image; selecting at least one category based on category information output by the to-be-processed image through the image classifier, and determining a thermodynamic diagram of each category in the at least one category based on the category information; for each category of thermodynamic diagrams in the at least one category of thermodynamic diagrams, determining a first positioning frame set corresponding to the target object inthe to-be-processed image; determining a second positioning frame set of the to-be-processed image according to an unsupervised target detection algorithm; and determining a target positioning frame set in the to-be-processed image according to the first positioning frame set and the second positioning frame set, the target positioning frame being used for representing the position of the target object in the to-be-processed image. Through the technical scheme of the invention, the target detection algorithm and the deep learning algorithm are combined, the data acquisition difficulty is low,the data annotation amount is small, and the position of the target object in the to-be-processed image can be quickly and accurately determined.

The embodiment of the invention relates to a method and device for generating a photoplethysmography signal. The method comprises the following steps: according to a preset red light pixel threshold range, sequentially carrying out single-frame red and green channel data calculation on all skin surface video data frame images of a skin surface video data frame image sequence to generate first redand green light digital signals; carrying out signal band-pass filtering on the first red and green light digital signals to generate second red and green light digital signals; acquiring a signal judgment identifier; judging whether to carry out signal maximum frequency difference judgment processing on the second red light digital signal and the second green light digital signal according to thesignal judgment identifier and generating a first judgment result; when the first judgment result is an up-to-standard signal identifier, performing signal-to-noise ratio judgment on the second red and / or green light digital signals to generate a second judgment result; and generating a PPG signal when the second judgment result is an up-to-standard signal identifier.

The invention relates to a tourism industry user portrait construction method and device and computer equipment. The method comprises the steps that product browsing record information of a target user is acquired, the product browsing record information comprises browsing record information of browsed products, the target user is a tourism industry user, and the browsed products are tourism products; Determining a first weight value according to the product browsing record information, wherein the first weight value is the weight value of the browsed product; Classifying the browsed productsaccording to the product labels of the browsed products to obtain browsed product sets corresponding to the product labels; Determining a second weight value according to the first weight value, wherein the second weight value is the weight value of the product label corresponding to each browsed product set; And constructing a target user portrait of the target user according to the second weightvalue. By adopting the method, the data collection difficulty can be lowered, and the pertinence of the user portrait can be improved.