85 results about "Evaluation strategy" patented technology

The invention provides a voice activity detection and wake-up method and device, and relates to the technical field of machine learning speech recognition. The method includes the steps of acquiring voice activity detection data and wake-up data, and performing Fbank feature extraction on the voice activity detection data and wake-up data to obtain voice Fbank feature data; inputting the voice Fbank feature data to a binary neural network model to obtain binarized neural network output result data; and according to a preset backend evaluation strategy, processing the binarized neural network output result data, determining a voice start position and a voice end position of the voice activity detection data, and detecting wake-up word data in the wake-up data. The system framework of the invention can be applied to voice activity detection and voice wake-up technologies at the same time, and can implement accurate, fast, low-delay, small-model and low-power voice activity detection technologies and voice wake-up technologies.

The invention provides a traditional Chinese medicine health management system and method. A traditional Chinese medicine health management platform is established, data is analyzed by using a predetermined screening model through data acqusition, and the health state of an informant is automatically evaluated to generate a nursing scheme. Intelligent / individual health real-time monitoring can be realized, a reliable health prediction and evaluation strategy is provided, and a user is helped to timely and effectively learn the physical situation, discover the precursors of some patients, improve the self-health consciousness and prevent the diseases and realize self-health.

The invention, which relates to the digital image processing field, provides a visual-saliency-based random image region quality evaluation strategy, so that an evaluation result meets the subjective perception of the human being. According to the technical scheme, the method comprises: (1), Gaussian filtering is carried out on a reference image IR and a distorted image ID respectively and the filtered image RGBs are converted into Lab color space units; (2), for different image sizes r*c, N windows with random sizes are generated randomly; (3), visual saliency calculation is carried out the reference image and the distorted image respectively by using a visual-saliency extraction algorithm and a visual saliency similarity matrix in each random window is obtained; (4), weighting and integration are carried out on the obtained similarity matrixes of all random windows. The method is mainly applied to an image processing occasion.

The invention relates to a fighting detecting method based on stereoscopic vision motion field analysis. The method is as follows: an object human body is extracted according to the motion vector field and the depth map of the object; by taking the motion vector field as the main characteristic, the entropy of a cumulative histogram within a certain time is calculated out based on a block statistics cumulative motion vector direction histogram; a motion intensity degree evaluation strategy is planned according to the depth information of the object by calculating the average motion vector intensity in the block; and finally, the disorder and intensity degree evaluation scores are combined to obtain the probability of fighting behavior, through the time and space accumulation, when the probability of the fighting behavior is continuously greater than a threshold value, then the fighting behavior is confirmed. The fighting detecting method can filter the motion vector calculation errors caused by scene illumination variation and object blocking under the monitoring environment. The intensity degree judgment is more reasonable according to the scene depth information, and the robust detection result is obtained after the analysis of the time and space distribution by establishing the fighting probability model.

A trusted network access controlling method based upon tri-element peer authentication comprises: Firstly initializing creditability collectors and a creditability verifier; then carrying out a tri-element peer authentication protocol among a network access requester, a network access controller and an authentication strategy server in a network access control layer to realize bi-directional user authentication between the access requester and the access controller; When authentication is successful or the locale strategy requires to carry out a when a platform creditability evaluation process, the TNC terminal, TNC server and evaluation strategy server in a trusted platform evaluation layer performing the tri-element peer authentication protocol to realize bi-directional platform creditability authentication between the access requester and the access controller; Finally the access requester and the access controller controlling ports according to the recommendation generated by the TNAC client terminal and the TNAC service terminal. The invention solves the technical problem about poor expandability in background, and further solves the problem about complex key negotiation and relatively low safety.

The invention discloses a scheme for the effective evaluation of online brokers of Over-The-Counter (OTC) traded financial assets. The techniques disclosed evaluate online brokers on the basis of a broker-score that is derived from a pricing-score and an execution-score correspondent to each online broker. Pricing-score and execution-score are further derived from pricing data and execution data collected from trading stations / platforms supported by the online brokers. Such data collection is performed with the aid of custom Expert Advisor (EA) modules developed and installed specifically for a given trading platform, e.g. MetaTrader4 (MT4). Evaluation results are generalized across all traders as well as customized according to the trading patterns of an individual trader. The results are presented to the traders as a ranked list of brokers recommended by the evaluation scheme.

The invention relates to a method and a device for processing a user network behavior. The method comprises steps: a network operation behavior of the user is acquired; a first evaluation strategy which is built in advance is adopted to carry out first scoring on the network operation behavior of the user; if the scoring result meets a preset condition for the first evaluation strategy, the user is judged to be a suspicious user; and a set strategy is adopted for the user for malicious attack. Malicious behaviors in a network ad, network activity service, network application service and the like can be effectively intercepted, and losses of an application developer and a promoter can be reduced. Further, as for the user who is judged to be a suspicious user, a second evaluation strategy which is built in advance is adopted to carry out second scoring, normal users are selected, the malicious user judgment accuracy is improved, and the wrong attach possibility is reduced.

The invention proposes a low-orbit satellite short data communication channel agile distribution method, and designs an idle channel evaluation strategy based on the sampling power judgment; and meanwhile, prior measurement data of a target frequency band and the ground network frequency distribution condition are fully considered, the idle degrees of alternative channels are sorted, and channel agile distribution and switching are carried out on low-orbit satellites. Compared with a traditional channel fixed distribution and access method, the low-orbit satellite short data communication channel agile distribution method provided by the invention improves the spectrum application efficiency of the satellite short data communication, effectively avoids the interference of a system itself and the interference to other systems, and solves the problem of scarcity of frequency resources; and the low-orbit satellite short data communication channel agile distribution method can effectivelyrealize multi-system compatibility under limited spectrum resources, improve the anti-interference performance and frequency use efficiency of a satellite communication system, and has the characteristics of being simple in process, high in flexibility and high in feasibility.

The invention discloses a heuristic power distribution network reconstructing method based on a minimum spanning tree and is aimed to realize the minimum active power loss and solve the power distribution network reconstructing problem through combining the minimum spanning tree algorithm with a heuristic method. The solving process includes that using the relationship between voltages of two sides of a branch as a weight, and performing preliminary optimization through the minimum spanning tree algorithm to obtain an initial solution to enable the subsequent optimization to be independent of the initial structure of a network; the secondary optimization includes that using a new candidate switch evaluation strategy to evaluate candidate switches, obtaining an optimized reconstructing structure through a little of candidate switches to be evaluated, and correcting to obtain the optimal solution. The heuristic power distribution network reconstructing method based on the minimum spanning tree does not depend on the initial structure of the network, has high reconstructing efficiency and is suitable for different sizes of power distribution networks.

Memristor-based nano-crossbar computing is a revolutionary computing paradigm that does away with the traditional Von Neumann architectural separation of memory and computation units. The computation of Boolean formulas using memristor circuits has been a subject of several recent investigations. Crossbar computing, in general, has also been a topic of active interest, but sneak paths have posed a hurdle in the design of pervasive general-purpose crossbar computing paradigms. Various embodiments are disclosed which demonstrate that sneak paths in nano-crossbar computing can be exploited to design a Boolean-formula evaluation strategy. Such nano-crossbar designs are also an effective approach for synthesizing high performance customized arithmetic and logic circuits.

The invention discloses a large-scale reaction kettle distributed fault diagnosis method based on a sensor network. Each sensor node unit mainly comprises a detection module, a communication module and a calculation module. The method comprises the following technical steps that first, a cascade model of a large-scale reaction kettle system is built; second, a distributed fault diagnosis device based on the sensor network is built, and the information interaction relation between sensor nodes is determined; third, optimal diagnosis device parameters are designed, and the performance of the fault diagnosis device is optimized; fourth, a fault evaluation strategy is designed, and fault diagnosis is carried out according to difference signals generated by the diagnosis device. The fault diagnosis method has the advantages that power consumption and cost are low, and a distribution mode is adopted; a high degree of spatial freedom is achieved in the configuration process, maintenance and extension are convenient, the characteristics that large-scale chemical industry production is large in scale and continuous are met, and the problems that in industry application, a traditional centralized fault diagnosis scheme has constraints in layout and wiring and is poor in reliability are ameliorated.

The invention discloses an employee evaluation method and system. The method comprises the following steps of: obtaining identity information of a to-be-evaluated employee; obtaining workload completed by the employee from a constructed evaluation database by taking the identity information as a query condition; identifying work types included in the workload according to an identifier of each working matter in the workload; calculating evaluation data generated by the employee under each working type according to an evaluation strategy of each working type and evaluation parameters of each working matter under the working type; and obtaining an evaluation result of the employee according to the evaluation data of each working type. According to the method and device, the evaluation data, generated under a plurality of working types, of employees is considered in the evaluation process, so that the enthusiasm of the employees for completing different working types can be better stimulated, benefit is brought to stimulate the maximum potential of the employee and higher earnings can be brought to the companies.

The invention provides an online dynamic task planning terminal and planning method for medium and low orbit micro satellites. The terminal of the invention adopts an online task planning module of aparallel processing structure, and when the satellite maintains communication with a ground measurement and control station, the communication and coordination among multiple satellites, and dynamic task planning are achieved in combination with a Beidou short message transceiver. The method of the invention adopts the form of an auction contract network, and comprises: processing auction rights,selecting a dominant satellite, and selecting the current auction task. The dominant satellite performs bidding on member satellites for the task. A bidder performs task bidding according to an own state. A tenderee completes bidding evaluation according to a bidding evaluation strategy and selects a contract satellite. Contract establishment, and process monitoring and contract information updateare completed, and the contract is terminated after the task is completed. By adoption of the online dynamic task planning terminal and planning method provided by the invention, the online dynamic task distribution and rational planning of a plurality of medium and low orbit micro satellites can be achieved, the efficiency is relatively high, the planning result is feasible, and it has practicalsignificance for improving the on-orbit autonomy and joint planning capability of the micro satellites.

The invention discloses a coal mine water burst predicting method based on a long-short-time memory neural network. The coal mine water burst predicting method introduces the long-short-time memory neural network into coal mine water burst prediction. Firstly, a feature selection method based on a Wrapper evaluation strategy is adopted to preprocess data, extract feature data and eliminate the influence of redundancy features on a follow-up prediction algorithm; by adopting an MSRA initializing method, a weight matrix is initialized into Gaussian distribution with the mean value of 0 and the variance of 2 / (input number), so that the prediction method has more reasonable initializing weight, and the convergence rate of the method is improved; an LSTM method is adopted to learn the change law of dynamic water burst data and the influence of the law on water burst, the method is prevented from overfitting by using a Dropout technology in the learning process. With increase of iteration number, the weight matrix of the prediction method is constantly updated, and accordingly the precision, stability and robustness of the prediction method are improved.

The invention provides a self-adaptive multi-object evolution method adopting constraint cloud workflow scheduling. The overall detection and local mining capability of the multi-object evolution method can be improved. The multi-object evolution method comprises the steps that S1, the evolution states of populations in the evolution process are detected according to the number of Pareto solutions and Pareto entropies, and corresponding individual evaluation strategy processing constraint conditions are self-adaptively utilized to sort individuals in the populations according to the detected evolution states of the populations in the evolution process, wherein a constraint violation processing method is adopted to process the constraint conditions in individual evaluation strategies; S2, according to individual sorting results, individuals are selected from the populations to perform genetic manipulation, and sub-populations are obtained, wherein genetic manipulation parameters are self-adaptively adjusted according to the evolution states of the populations in the evolution process during genetic manipulation. The self-adaptive multi-object evolution method is suitable for solving the multi-object evolution problem having constraints and can be applied to the technical field of workflow scheduling in a cloud computing environment.

The invention discloses an inspection image quality evaluation method and system based on saliency detection, and the method comprises the steps: carrying out the preprocessing of an original inspection image through a saliency detection algorithm based on the combination of a global part and a local part for a to-be-evaluated inspection image, and obtaining a saliency map of the inspection image;and based on the saliency map, using an edge description-based algorithm to calculate and obtain two indexes for describing the image ambiguity, and obtaining a quality score of the whole image according to the ambiguity indexes. In addition, the convolutional neural network algorithm based on deep learning adopts a partitioning evaluation strategy to calculate and obtain another quality scoringresult of the image; and finally, comprehensively processing evaluation results of the two algorithms by adopting a weighted fusion strategy to obtain a final score of the image quality.

The invention discloses a system for evaluating service quality of a power ICT network. The system comprises a network management unit, a service management unit, a strategy formulation unit and an evaluation unit, wherein the network management unit is used for acquiring performance parameters of the network; the service management unit is used for acquiring service data and formulating ranks ofuser services and service quality indexes; the strategy formulation unit is used for formulating an evaluation strategy and setting an evaluation reference value and a weight value; and the evaluation unit is used for evaluating the performance parameters of the network and the service data according to the evaluation strategy, the evaluation reference value and the weight value so as to offer anevaluation result for the reliability and the service quality of the network. By providing the system for evaluating the service quality of the power ICT network, accurate objective basis is providedfor network planning and service management of the power ICT network so as to promote the development of the power ICT network and the production and operation of the power.

The invention belongs to the technical field of machine learning, and discloses an adaptive information feedback particle swarm robot path selecting method based on multiple subgroups and a system thereof. Multiple diversity unified defining manners are used for a diversity evaluation strategy, thereby evaluating group diversity from multiple angles. A multiple-subgroup dividing strategy is performed on a particle group through comprehensive analysis of adaptability and spatial angle. Presetting of subgroup number is not required. A wide area searching effect in a dynamic optimizing process can be improved. A sparse processing strategy is utilized for adjusting a distribution condition in the subgroup so that the particles in the subgroup can be far from local central particles in a certain probability, thereby executing a real-time exploration process in a local area and improving diversity level in the subgroup. The adaptive information feedback particle swarm robot path selecting method and the system thereby facilitate fine exploration for a local optimal value and wide-range searching in a global range. Through result analysis on a testing problem, the method and the system according to the invention can be effectively applied in problems of multidirectional dynamic optimization and robot global path planning and further have advantages of simple and applicable method and certain use value.

The invention provides a power transmission device state evaluation and risk analysis method and device based on multidimensional data and relates to the technical field of meteorological and electric power technology application. The method comprises steps of: acquiring and storing information data of a PMS production management system, information data of a power transmission and transformation monitoring system and information data of a meteorological bureau system through multidimensional heterogeneous data access; carrying out data preprocessing on the information data of the power transmission and transformation monitoring system and generating online monitoring effective data of a power transmission circuit; determining a power transmission device evaluation index according to the information data of the PMS production management system and the online monitoring effective data of the power transmission circuit, and determining an evaluation state corresponding to the power transmission device evaluation index according to a preset state evaluation strategy; and determining a risk index of the power transmission device according to the information data of the meteorological bureau system, the asset value of the power transmission device, the asset loss value of the power transmission device and the average fault rate of the power transmission device. According to the invention, the multidimensional data is comprehensively considered.

The invention discloses a variable-pitch adaptive control method and system for a wind generating set. According to the type of the wind generating set and the running condition of the set site, a corresponding relation between the wind speed and the pitch angle in a pitch angle limit table and an interval of the pitch angle optimization are determined; a performance evaluation strategy of the wind generating set is designed, the performance of the unit under different pitch angle parameters are compared, so that the pitch angle parameters corresponding to the optimal performance of the set are found to serve as an optimal parameter of the set; and the interval is optimized for the pitch angle, an optimal pitch angle parameter value is found out by continuously narrowing the range of the optimization according to the tracking iteration, the control parameters are guaranteed to be in the optimal position, and self-adaptive control is achieved. According to the method and system, the problem of local stall of blades of the wind generating set can be well solved, the power curve of the unit is optimized, the generating capacity of the unit is greatly improved, meanwhile, the problem that the current parameter optimization needs to be debugged on site is solved, and the time and the labor cost are greatly saved.

The invention discloses an online state evaluation system and method for a substation power supply. The system comprises a substation power supply system information acquisition subsystem of a transformer substation end and a power supply system online state evaluation main system of a dispatching end. The subsystems realize the construction of an equipment model of the substation power supply system for the substation end station and the collection and uploading of operation information. The main system of the dispatching end uses a substation end power supply system model and real-time operation information uploaded by the subsystem. Depending on an on-line state evaluation strategy of the substation power supply system, the system achieves the online monitoring of the operation state ofthe substation power supply system for the substation end and the online state evaluation of the operation condition of the system, achieves the online monitoring and evaluation of the power supply system for the role-shaped transformer station oriented to the operation personnel and professional operation and maintenance personnel, and improves the level of safe operation and risk early warningof the transformer station.

The invention discloses a recovery health effect evaluation strategy. The recovery health effect evaluation strategy comprises the following steps of a, measuring all kinds of factors, which include blood pressure, blood fat, blood sugar, weight monitoring and heart rate, of a user; b, surveying life styles, which include smoking, physiological emotion, motion, dietary habit, sleeping and family history, of the user; c, grading each factor and life style, distributing weighing factors to each factor and life style, and figuring up to obtain a comprehensive mark, wherein the weight monitoring comprises a BMI (Body Mass Index) and a waist-hip ratio. The recovery health effect evaluation strategy has the following beneficial effects that all kinds of objective signs, which influence people, and all kinds of factors of subjective self-perception are comprehensively considered; the body health development is visualized through values, and the power for causing the user to keep exercising is activated; the test is simple and easy, and is easy to operate.

The invention discloses a distribution communication network cross-domain protection network oriented to quantum protection communication service. A centralized control architecture of ubiquitous optical access network for cross-domain protection is proposed by introducing software-defined networking technology in the distribution communication network which is composed of optical access network carrying quantum secret key distribution. The cross-domain protection function module is further designed and the corresponding protocol is extended. Then, based on this architecture, a global evaluation strategy is designed, which integrates the ubiquitous access resources and optical convergence layer resources to make decisions for the resource allocation of protection links. Finally, the interactive flow of cross-domain protection is designed. Compared with the traditional access network, it can achieve a more robust network form and provide effective service protection and network performance.

The invention provides a terminal and a system for prompting a safety state of a user terminal and an implementation method. The implementation method comprises the steps of storing an evaluation result of an evaluated terminal in a local database of the terminal; acquiring state information related to a safety index of the terminal; analyzing the state information of the terminal according to a preset safety evaluation strategy to generate an evaluation result; comparing the evaluation result of the terminal with those of other terminals in the local database, and acquiring ranking information of the terminal according to a preset ranking mode; displaying the evaluation result and the ranking information of the terminal on a screen. According to the implementation method, the current state information of the terminal can be analyzed according to the preset safety evaluation strategy so as to generate the safety state evaluation result, and the safety state evaluation result can be compared with those of other terminals so as to acquire the ranking information; the ranking information is finally sent to the terminal for displaying, so that a user can directly learn about the safety state of the terminal; the application is humanized, and the experience result of the user is greatly enhanced.

The invention belongs to the optimized control field and discloses a multi-strategy particle swarm optimization method and system for solving a permutation flowshop scheduling problem. The multi-strategy particle swarm optimization method comprises steps of constructing a diversified evaluation strategy, measuring population diversity of particle colonies according to an information entropy mode through subintervals divided by gravitational values, directing alternative execution of local exploration and global searching during an optimizing process, during global optimal particle selection which borrows an ant-city selection strategy in a particle speed-displacement model, calculating a probability of each particle choosing another particle as a global optimal particle, determining a global optimal particle which is following during a particle moving process according to the probability, designing a set variation, and guiding the particle swarm to jump out of a local optimal solution area to perform particle swarm global searching. The set variation operation of the multi-strategy particle swarm optimization method can better help the particle to escape which falls into the local optimal region, prevents a premature convergent phenomenon, can perform continuous optimizing in a global range and can obtain a better result.

The embodiment of the invention discloses a method of evaluating advertisement conversion rate prediction models, servers and a computer-readable storage medium. The method includes: receiving an advertisement conversion rate prediction model calling request, randomly calling a new-version advertisement conversion rate prediction model or an old-version advertisement conversion rate prediction model to calculate predicted conversion rates of candidate advertisements, and recording model calling identifiers corresponding to the candidate advertisements; sending the predicted conversion rates ofthe candidate advertisements to an advertisement platform; calculating first actual advertisement conversion rates, which corresponds to the new-version advertisement conversion rate prediction model, and second actual advertisement conversion rates, which corresponds to the old-version advertisement conversion rate prediction model, according to model calling identifiers corresponding to alreadyexposed advertisements; and carrying out advantage / disadvantage evaluation on the new-version advertisement conversion rate prediction model and the old-version advertisement conversion rate prediction model according to the first actual advertisement conversion rates, the second actual advertisement conversion rates and a preset evaluation strategy. The embodiment of the invention can improve accuracy of advantage / disadvantage evaluation of the advertisement conversion rate prediction models.

The present invention discloses a dispatching operation information patrol inspection method and system, and belongs to the electric power dispatching automation technology field. The method comprises the steps of obtaining the power grid real-time operation data from an OPEN3000 dispatching automation monitoring system; according to the obtained power grid real-time operation data, establishing a transformer substation model of the connection modes of all power plants in a power grid and the equipment in a transformer substation and the equipment connection sequences; pre-processing the obtained power grid real-time operation data to obtain the effective measurement data; according to the effective measurement data and the switch on-off data in the transformer substation, implementing an alarm evaluation strategy; utilizing the alarm evaluation strategy to evaluate the alarm information, and then classifying the alarm information according to different alarm types to display in an alarm page. According to the present invention, by pre-processing the acquired measurement data, a lot of frequent invalid alarm information is shielded, the number of the alarm information that the monitoring personnel need to check is reduced substantially, so that the monitoring personnel can determine and process accurately, rapidly and appropriately.

The invention discloses a content-oriented trust evaluation method in a vehicle-mounted ad hoc network environment. Historical trust evaluation results can be learned through the strategy adjustment module to dynamically determine trust evaluation strategies under different situations; in addition, the trust evaluation module obtains the optimal evaluation strategy through the trust evaluation request and the trust evaluation request related information, and the evaluation accuracy of 90% or above can still be kept under the condition that the proportion of false information transmitted in thenetwork exceeds 50%. Therefore, the influence of false information in the node decision-making process is reduced, and the driving safety and comfort are improved.

The technology disclosed proposes a novel asynchronous evaluation strategy (AES) that increases throughput of evolutionary algorithms by continuously maintaining a queue of K individuals ready to be sent to the worker nodes for evaluation and evolving the next generation once a fraction Mi of the K individuals have been evaluated by the worker nodes, where Mi<<K. A suitable value for Mi is determined experimentally, balancing diversity and efficiency. The technology disclosed is extended to coevolution of deep neural network supermodules and blueprints in the form of AES for cooperative evolution of deep neural networks (CoDeepNEAT-AES). Applied to image captioning domain, a threefold speedup is observed on 200 graphics processing unit (GPU) worker nodes, demonstrating that the disclosed AES and CoDeepNEAT-AES are promising techniques for evolving complex systems with long and variable evaluation times.

The invention discloses an indoor air quality evaluating method, and belongs to the field of air purification. The method includes the steps that air quality parameters monitored by a sensor in a to-be-evaluated space are obtained; according to all air evaluation parameters, evaluation coefficients of air quality grades corresponding to the air evaluation parameters and the upper limit of evaluation sections of the air quality grades are determined in a database; according to the air evaluation parameters, the evaluation coefficients and the upper limit of the evaluation sections of the air quality grades, air quality coefficients are calculated; according to the air quality coefficients and preset evaluation strategies, target air quality grades corresponding to the air quality coefficients are determined; and an air quality evaluating report containing the target air quality grades is generated. The invention further provides an air quality evaluating system. According to the scheme, multiple air quality parameter values are used in a combined manner to evaluate the air quality, and the evaluation results are more close to the actual air quality condition and are more accurate.