37results about How to "Reflect the importance" patented technology

The invention relates to a video description method and system based on an information loss function, and the method comprises the steps: obtaining a training video, and obtaining the semantic information of each frame of a set training video; Inputting the semantic information of the training video into an LSTM-combined hierarchical attention mechanism model to obtain character description of thetraining video; According to the importance of each word in the character description to the expression video content, performing loss weighting on the words to obtain an information loss function, and taking the information loss function as an objective function to perform back-propagation gradient optimization on the hierarchical attention mechanism model to obtain a video description model; Obtaining a to-be-described video, respectively inputting the to-be-described video into the target detection network, the convolutional neural network and the action recognition network to obtain a setof target features, overall features and motion features of each frame of the to-be-described video as semantic information of the to-be-described video, and inputting the semantic information into the video description model to obtain character description of the to-be-described video.

Accurate assessment of aircraft fuel consumption is an important means for reducing flight cost and controlling fuel emission of airlines. The invention provides an aircraft departure fuel consumptionassessment method based on flight data analysis. On the basis of the actual flight data of QAR, a function relationship between pneumatic parameters in a fuel consumption model in an off-site climbing stage is fitted, side wind is modeled and added into a fuel consumption evaluation model, and an improved kinematics model of an off-site continuous climbing operation mode is established. Through the method, the problem that aerodynamic parameters in a traditional fuel consumption model need to look up a static chart relation obtained by a wind tunnel test , extra fuel consumption caused by aircraft performance attenuation cannot be estimated, and the calculation result is inaccurate due to the fact that the influence of crosswind on fuel consumption is not involved are solved; and accuratecontrol over the fuel loading amount is achieved, the phenomenon of fuel burning caused by too much fuel filling amount before flight is avoided, and therefore the fuel cost is reduced.

The invention discloses a nonuniformity-based I / O access system in a virtual multicore environment and an optimizing method and relates to the field of computer virtualization. The system comprises a performance detection module, a thread binding module and a memory migration module, the performance detection module monitors hardware information of a virtual machine and a physical host in real time through a modified performance monitoring tool, the thread binding module judges whether a current system is with low load or high load according to the hardware information collected by the performance detection module and binds virtual machine threads on nodes with high load onto nodes with low load when the current system is with high load, and the memory migration module migrates related threads onto a node closest to a network adapter when the current system is with low load. An affinity optimizing model of I / O performance based on the virtual multicore environment is built, and real-time dynamic optimized placement strategies with high throughput and low delay are provided for the system so as to efficiently utilize multicore resources and performance of the high-performance network adapter, so that load of the system is lowered effectively.

The invention relates to the field of position prediction and recommendation, in particular to a user space-time similarity measuring method in interest point prediction and recommendation. The user space-time similarity measuring method in interest point prediction and recommendation includes the steps of preprocessing data; measuring space-time cooccurrence mode similarity; measuring user space-time similarity. The invention provides the user space-time similarity measuring method, the space-time cooccurrence mode and similarity thereof are defined through check-in time close degree and interest point position close degree, then the user space-time cooccurrence mode and similarity thereof are defined, user space-time similarity among users is defined based on the weight of interest point check-in frequency through the most similar space-time cooccurrence mode and similarity thereof, and the user space-time similarity measuring method is provided for a user-based collaborative filtering method in interest point prediction and recommendation.

The invention discloses a quick assessment method for site selection of an underground powerhouse cavern of a hydropower station, relates to the field of hydraulic and hydroelectric engineering, and provides an assessment method capable of objectively and quickly judging the site selection condition of an underground powerhouse cavern of a hydropower station. The method comprises the following steps: A, summarizing and classifying factors influencing site selection to obtain factor sets; B, endowing each factor set with a total score according to the influence degree on site selection, and distributing the total score of each factor set to factors of the factor set to obtain a total score of each factor; C, determining the actual score of each factor according to the difference between the actual condition and the ideal condition of each factor; and D, calculating the actual score sum of each factor, and judging whether the target address is suitable for building the underground powerhouse cavern according to the difference between the score sum and the total score. By adopting the method, numerous factors involved in site selection of the traditional underground powerhouse cavern are quantified and unified in the same assessment system for assessing via a large amount of existing engineering examples and numerical analysis results.

The invention discloses a groundwater environment quality evaluation method. An improved formula and further improved formulas have the same definite physical significance, namely when each single index is equal to 1, Pj=1; when each single index is equal to 2, Pj=2; a characteristic that each single index is synchronous with a comprehensive index is achieved. Moreover, the groundwater environment quality evaluation method has the advantages of simple form and convenience for calculation.

The invention relates to the field of information extraction. The invention discloses an unsupervised multi-model fusion extraction type text abstracting method. An existing extraction type text abstract technology is solved. In order to solve the problems that semantic information of sentences cannot be considered and extracted results cannot accurately and comprehensively describe article contents, the technical scheme of the method is as follows: a centrality text abstract model is trained and optimized in advance, and after optimization, a preprocessed to-be-processed document is calculated to obtain a first batch of abstracts summary1; the preprocessed to-be-processed document is calculated by using a semantic similarity capture model to obtain a second batch of abstract summaty2; thefirst batch of abstracts summary1 and the second batch of abstracts summary2 are fused to obtain a candidate abstract middle_summary2; the MMR algorithm is used to calculate the candidate abstract middle_summery so as to obtain a final abstract final_summery; and semantic understanding and analysis are carried out on text content in a multi-model fusion mode, position information of sentences isfully considered, the importance degree of each sentence can be accurately calculated, and the accuracy, flexibility and diversity of abstract results are improved.

The invention discloses a reverse-flow full-automatic battery grouping method and a reverse-flow full-automatic battery grouping system based on dynamic optimization. The battery grouping system comprises a battery performance detection device, a conveying device, and a main control computer. In the conveying device, an input conveyor belt and one side of an annular conveyor belt are parallel to and connected with each other, the annular conveyor belt and an output conveyor belt are perpendicular to and connected with each other, and the output conveyor belt and an outlet conveyor belt are perpendicular to and connected with each other. A baffle, a push-in cylinder and a photoelectric door are arranged at the tail end of the input conveyor belt. A push-out cylinder and a photoelectric door are arranged at the joint of the annular conveyor belt and the output conveyor belt, and a sensor is arranged beside the push-out cylinder. The distance between the push-out cylinder and the push-in cylinder is equal to the spacing between two adjacent batteries on the annular conveyor belt. The battery grouping method comprises the steps of battery input and grouping optimization calculation, battery pushing-out, and battery grouping output. By adopting the battery grouping method and the battery grouping system, real-time grouping optimization selection can be carried out according to the current state, and the quality and success rate of grouping are improved.

The invention discloses a data query method and device and a readable medium. The method comprises the following steps: carrying out word segmentation on a search string submitted by a user; aiming ateach keyword obtained after word segmentation, determining a word feature value of the keyword, a first association feature value used for representing a relationship between the keyword and the search string, a second association feature value used for representing a relationship between the keyword and an adjacent word of the keyword in the search string, a word feature value of an adjacent word of the keyword in the search string and a feature value of the search string; Respectively determining the weight of each keyword in the search string according to the word characteristic value of each keyword, the first association characteristic value, the second association characteristic value, the word characteristic value of the adjacent word and the characteristic value of the search string; And based on the weight of each keyword in the search string, determining a query keyword and performing query to obtain a query result of the search string, so that the weight of the obtained keyword can better conform to the actual importance degree of the keyword in the search string.

The invention relates to an optimization method for calculating the position of a tidal front based on an ROMS simulation result, and the method comprises the following steps: obtaining a formula forcalculating the tidal front under the condition of considering a surface layer wind friction effect, a bottom layer tidal current friction effect, a net heat flux effect and a buoyancy advection effect; calculating to obtain an action item A of the underlying tide to the peak; calculating to obtain an action item B of the surface wind field to the front; calculating to obtain an action item C of the net heat flux to the peak; calculating to obtain an action item M of the buoyancy advection to the sharp point, and performing back calculation processing on the calculation of the M items in orderto simplify the calculation process and reduce M errors at the same time; calculating to obtain the position Hfront of the tide front; subtracting the terrain water depth Hdepth output by the ROMS mode from the position depth Hforward of the tidal front to obtain a difference value delta H, and taking the place where delta H is zero as the real position of the tidal front in the ocean.

The invention provides a distributed manufacturing scheduling method for fixed-point output. The data field model is adopted to measure a task structure and an equipment distribution state, the task structure measurement reflects the influence of a subsequent process on an alternative process, and the equipment distribution state measurement reflects the local influence of equipment in an equipment network; By adopting a reverse scheduling method, starting from termination equipment and termination working procedures, equipment is reversely distributed for the working procedures, so that fixed-point output constraint of the task is guaranteed; Quantitative measurement of the established alternative processes and alternative devices can provide a basis for task scheduling, the scheduling process is simplified, and optimization of two aspects of tasks and device structures is considered.

The embodiment of the invention provides a safe driving index weight determination method and a device, and electronic equipment. The method comprises the steps that based on scores of a plurality ofdrivers on a plurality of safe driving indexes, calculating entropy values of the plurality of safe driving indexes respectively, and the entropy value of each safe driving index is used for representing the dispersion degree of the scores of the plurality of drivers on the safe driving indexes; and determining a comprehensive weight of the safe driving index based on the entropy of each safe driving index in the plurality of safe driving indexes, the comprehensive weight of each safe driving index being negatively correlated with the entropy of the safe driving index. The entropy value is obtained by counting the dispersion degree of scores of a plurality of drivers on a safe driving index; the information carried by the safe driving index and the relative advantages and disadvantages ofa plurality of drivers on the safe driving index can be quantitatively expressed, so that the comprehensive weight can objectively and accurately reflect the actual importance degree of the safe driving index.

The invention relates to a flow network key node identification method and device based on multi-attribute information fusion. The method comprises the following steps: constructing a node flow network according to a to-be-identified event; obtaining a maximum flow counting index, a minimum cut counting index and a maximum flow centrality index of each node in the node flow network; constructing an identification framework of node importance in the node flow network; respectively calculating basic probability assignment functions corresponding to the maximum flow counting index, the minimum cut counting index and the maximum flow centrality index in the identification framework; fusing the basic probability assignment functions corresponding to the maximum flow counting index, the minimumcut counting index and the maximum flow centrality index to obtain importance indexes of the nodes; and sorting the nodes in the node flow network according to the importance indexes, and determiningkey nodes in the node flow network according to a sorting result. By adopting the method, the accuracy of key node identification can be improved.

The present invention provides an artificial intelligence-based electronic map rendering method, device, electronic equipment, and computer-readable storage medium; the method includes: performing prediction processing based on feature vectors of geographic elements in the electronic map, so as to use Determine the rendering level associated with the geographic element in each rendering level; score the geographic elements associated with the same rendering level in the electronic map according to the corresponding feature vector to obtain a rendering score; calculate the rendering score of the geographic element Perform fusion processing with the rendering levels associated with the geographic elements to obtain rendering weights; sequentially render the geographic elements associated with the levels to be rendered according to the descending order of the rendering weights; wherein, the levels to be rendered are the multiple rendering levels A rendering level in the hierarchy that matches the current scale of the electronic map. Through the present invention, the rendering efficiency and precision of the electronic map can be improved.

The invention discloses a method and system for similarity recommendation of coronal mass ejection space environmental events. The velocity at the solar radius, the longitude and latitude coordinates of the source area of ​​the CME, the F10.7 index on the day when the CME erupted, and the average velocity, temperature, proton density, dynamic pressure, plasma β value and magnetic field components in three directions of the solar wind, After preprocessing, the characteristic parameter values ​​of 18 kinds of coronal mass ejection events are obtained; then input the pre-established and trained CME recommendation model to obtain historical similar CME events and predicted arrival times; the CME recommendation model is used to The weighting process weights the characteristic parameter values, and then performs distance matching with the historical events of the pre-established CME historical database to find the closest historical event.

The invention provides a distributed manufacturing scheduling method for fixed-point output. The data field model is adopted to measure a task structure and an equipment distribution state, the task structure measurement reflects the influence of a subsequent process on an alternative process, and the equipment distribution state measurement reflects the local influence of equipment in an equipment network; By adopting a reverse scheduling method, starting from termination equipment and termination working procedures, equipment is reversely distributed for the working procedures, so that fixed-point output constraint of the task is guaranteed; Quantitative measurement of the established alternative processes and alternative devices can provide a basis for task scheduling, the scheduling process is simplified, and optimization of two aspects of tasks and device structures is considered.

The present invention relates to a method for generating document summaries, which includes performing sentence segmentation on a document set to obtain a sentence set and expressing it with a vector space model, determining the similar sentences and the number of similar sentences corresponding to each sentence according to a preset similarity threshold, and calculating Get the corresponding importance score, obtain each sentence in the sentence set as the current processing sentence in turn, compare the number of similar sentences in the current processing sentence with the corresponding similar sentence numbers of all similar sentences in the current processing sentence, find the maximum value and The corresponding sentences are added to the diversity reference set, and then the diversity score and comprehensive score of each sentence are calculated, and finally all the sentences in the sentence set are sorted and screened to form a document summary. In addition, a device for generating document summaries is provided. The above method and device for generating document summaries comprehensively consider the internal information of the sentence and the global information in the document collection, and reduce the redundancy of the document summaries as a whole.

The invention discloses a coronal mass ejection space environment event similar recommendation method and system. The method comprises the following steps: collecting CPA, MPA, coronal mass ejection angular width, linear velocity, initial velocity, final velocity, velocities at 20 solar radiuses, latitude and longitude coordinates of a coronal mass ejection source area, an F10.7 index of a coronal mass ejection burst day, and average solar wind velocity, temperature, proton density and dynamic pressure and a plasma beta value and magnetic field components in three directions in a set time period; performing preprocessing to obtain characteristic parameter values of 18 coronal mass ejection events; inputting a pre-established and trained CME recommendation model to obtain historical similar CME events and predicted arrival time; employing the CME recommendation model for carrying out weighting processing on the characteristic parameter values according to the weight of each characteristic parameter, then carrying out distance matching with historical events of a pre-established CME historical database, and finding the nearest historical event.

The invention discloses a reverse-flow full-automatic battery grouping method and a reverse-flow full-automatic battery grouping system based on dynamic optimization. The battery grouping system comprises a battery performance detection device, a conveying device, and a main control computer. In the conveying device, an input conveyor belt and one side of an annular conveyor belt are parallel to and connected with each other, the annular conveyor belt and an output conveyor belt are perpendicular to and connected with each other, and the output conveyor belt and an outlet conveyor belt are perpendicular to and connected with each other. A baffle, a push-in cylinder and a photoelectric door are arranged at the tail end of the input conveyor belt. A push-out cylinder and a photoelectric door are arranged at the joint of the annular conveyor belt and the output conveyor belt, and a sensor is arranged beside the push-out cylinder. The distance between the push-out cylinder and the push-in cylinder is equal to the spacing between two adjacent batteries on the annular conveyor belt. The battery grouping method comprises the steps of battery input and grouping optimization calculation, battery pushing-out, and battery grouping output. By adopting the battery grouping method and the battery grouping system, real-time grouping optimization selection can be carried out according to the current state, and the quality and success rate of grouping are improved.

The invention relates to an APP humming composition system and a method thereof. The system comprises an APP client logging module, an APP client selection module, an audio data transmit-receive module, an audio data obtaining module, an audio data coordination module, an audio data storage module, an audio data synthesis module and an audio data finishing module. The APP humming composition system and the method thereof have the advantages that a song is divided into 5 parts according to the structure of the song, the song is fragmented according to the five parts and the five parts are recombined to complete an accompaniment, rules of popular music are summaries, universal chord blocks are created, the universal chord blocks are spliced at will, and a user can compose the composition by splicing the chord blocks and autonomously humming and enjoy the entertainment of song fragmentation and recombination of a new song; in addition, the style and the keynote of the song are selected, the APP humming composition system and the method thereof are simple and easy to understand, and the professional language is popular and visual, so that the user can directly make selections without understanding the meaning. By adopting the APP humming composition system and the method thereof, a common person can make a song of himself without the need of knowing musical instruments and arrangement, and the song of the user himself can be output by one time of humming.

The invention provides a method and a device for mining key pages of a website. The method includes: extracting a navigation link list from each page of the website, splitting each extracted navigation link list into link pairs, wherein each link pair is composed of two links at adjacent positions in the navigation link list; determining key link pairs from all the link pairs, and taking the pages corresponding to the key link pairs as the key pages in the website. By the means, recall rate and accuracy rate when the key pages of the website are mined can be increased.

The invention discloses a data value-based multi-copy heterogeneous storage method. The method comprises the steps of obtaining a dynamic value Vd of data according to a data access frequency, defining a static value Vs of the data according to a data type, and defining a subjective value Vu of the data according to a user demand; calculating a total value V (V=wd*Vd+ws*Vs+wu*Vu) of the data, wherein wd is a weight of the dynamic value, ws is a weight of the static value, wu is a weight of the subjective value, and the sum of wd, ws and wu is equal to 1; comparing the total value of the data with preset first and second thresholds, and dividing the data into high-value data, medium-value data and low-value data; and storing all copies of the low-value data in a disk, storing all copies ofthe high-value data in a solid state disk, storing a copy of the medium-value data in the solid state disk, and storing other copies in the disk. Proper storage policies are selected according to different data values, and the copies are stored in different storage mediums, thereby achieving optimization of system performance and cost.

Accurate assessment of aircraft fuel consumption is an important means for airlines to reduce flight costs and control fuel emissions. The present invention proposes an aircraft departure fuel consumption evaluation method based on flight data analysis. On the basis of QAR actual flight data, it fits the functional relationship between the aerodynamic parameters in the fuel consumption model during the departure climb stage, and models the crosswind And adding it to the fuel consumption evaluation model, the improved kinematics model of the departure continuous climb operation mode is established. Through the present invention, it is solved that the aerodynamic parameters in the traditional fuel consumption model need to consult the static chart relationship obtained by the wind tunnel test, and the extra fuel consumption caused by the aircraft performance attenuation cannot be estimated, and the influence of the crosswind on the fuel consumption is not involved, resulting in The technical problem of inaccurate calculation results realizes the precise control of fuel loading, avoids the phenomenon of "oil burning" caused by excessive refueling before flight, and reduces fuel costs.

The invention relates to an SSD storage data error processing method and device, equipment and a storage medium, and the method comprises the steps: obtaining error information of SSD NAND flash to complete error collection; performing data management and scheduling on the collected errors to obtain a management scheduling command; and processing the SSD storage data according to the management scheduling command. According to the method, the business layer and user layer characteristics of the data can be associated, the data importance and the system influence can be better reflected, and risk control and preprocessing can be carried out on the associated data as soon as possible, so that the loss or damage to core data can be better avoided, the scheduling with the main process of the storage equipment is well coordinated, and the scheduling efficiency is improved. And error data processing management is completed orderly.

The invention provides a spaceborne message scheduling method and device of a space-based data link, and belongs to the field of space-based data links. The method comprises that spaceborne messages of the space-based data link are received, and arranged in different queues according to the application priority of the messages; the extraction sequence and to-be-distributed resource corresponding to the messages are determined according to service types of the received messages; according to the determined extraction sequence, messages with the same service type are extracted from each queue; and according to the determined to-be-distributed resource, the extracted messages are scheduled. Different types of the messages in the queue are processed in a balanced way, the channel utilization rate is improved, the fairness among the queues is improved, and the total service capability of the network is improved.

The invention discloses a geofence grading method, a vehicle scheduling method and a server. The geo-fence grading method comprises the following steps: acquiring information of a geo-fence network, wherein the information of the geo-fence network comprises distances between different geo-fences in the geo-fence network; obtaining vehicle riding order data of the geo-fence network, wherein the vehicle riding order data comprises a source geo-fence and a target geo-fence corresponding to a vehicle riding order; and according to the vehicle riding order data and the information of the geo-fence network, determining a grade value of a geo-fence in the geo-fence network. According to the grading method, connectivity between geo-fences and geographic factors are introduced into the grading process of the geo-fences, so that the grading result can more accurately reflect the importance of the geo-fences in shared vehicle service.

The invention relates to a method for a nonlinear fuzzy matter element assessment of a carrier-based aircraft launching scheme. The method includes steps: firstly, establishing an assessment index system of the carrier-based aircraft launching scheme, determining assessment index weights of the carrier-based aircraft launching scheme based on an entropy method, calculating a fuzzy combined matterelement, obtaining the weights of a major assessment index and a minor assessment index, obtaining a fuzzy concentrated matter element of the launching scheme through resolution, obtaining a nonlinearassessment value, analyzing an assessment result of the nonlinear fuzzy matter element, and finally realizing nonlinear fuzzy matter element assessment of the carrier-based aircraft launching scheme.For a significant influence of a certain index of the carrier-based aircraft launching scheme, the defect of a conventional linear weighting method is avoided, the index weights are distinguished, primary indexes and secondary indexes are assessed by employing the nonlinear fuzzy matter element method, and the problem of difficult processing of the significant influence of the conventional linearassessment method on the certain index is solved. According to the method, the reliability and the validity of the assessment scheme are verified through a simulation experiment.

The invention relates to a circuit design and modeling method of a balanced high-temperature superconducting receiver, belonging to the technical fields of microwave and terahertz communication and high-temperature superconducting. Including circuit design of balanced high-temperature superconducting receiver and receiver modeling method; step 1, circuit design of balanced high-temperature superconducting receiver includes MgO chip circuit design and alumina PCB circuit design, and is connected by adhesive tape; MgO chip circuit design includes 3dB branch bridge network, impedance matching, Josephson junction with DC bias circuit, sector choke filter and fifth-order choke filter; alumina PCB circuit design includes design of bias tee circuit and intermediate frequency Signal mixing circuit; step 2, modeling of the balanced high-temperature superconducting receiver, including establishing a multi-port network interleaving model and obtaining analytical expressions of conversion gain and noise temperature of the superconducting receiver. The designed circuit has the advantages of small size, low cost, low noise, extremely wide IF bandwidth, high frequency upper limit, and low power requirement.

The invention relates to a flow network key node identification method and device based on multi-attribute information fusion. The method comprises the following steps: constructing a node flow network according to a to-be-identified event; obtaining a maximum flow counting index, a minimum cut counting index and a maximum flow centrality index of each node in the node flow network; constructing an identification framework of node importance in the node flow network; respectively calculating basic probability assignment functions corresponding to the maximum flow counting index, the minimum cut counting index and the maximum flow centrality index in the identification framework; fusing the basic probability assignment functions corresponding to the maximum flow counting index, the minimumcut counting index and the maximum flow centrality index to obtain importance indexes of the nodes; and sorting the nodes in the node flow network according to the importance indexes, and determiningkey nodes in the node flow network according to a sorting result. By adopting the method, the accuracy of key node identification can be improved.

The invention provides a communication record management method, which includes counting the communication times of each communication record, and sorting the communication records according to the communication times; selecting the communication records that need to be deleted according to the sorting results and clearing them; The number of times is cleared. The present invention also provides a mobile terminal used to support the communication record management method. Through the present invention, the communication records can be sorted according to the communication times, and the importance of the communication records can be intuitively reflected according to the sorting results, and then according to the importance The number of times corresponding to the degree, delete non-important communication records, so as to quickly clear communication records in batches.