58results about How to "Avoiding Combinatorial Explosion Problems" patented technology

The invention relates to a method for reconstructing a power distribution network based on double-ant-colony optimization. The method comprises the steps of extracting a looped network, initializing parameters, searching for double ant colonies, decoding a path where each set of ants walks through, obtaining the reconstruction scheme of the power distribution network, conducting looped network detection, island detection and constraint condition detection, judging feasibility of the scheme, judging whether the number of two sets of the ants reaches the ant colony scale or not, judging whether the frequency of information exchange is achieved or not, conducting information exchange, judging whether the terminal condition is met or not, and ending and obtaining the optimum power distribution network reconstruction scheme. The deep searching method is adopted, the reconstruction of the looped network of the power distribution network on a looped network upper interconnection switch and a section switch is determined, digits participating in switch reconstruction are reduced, the problem that when the switch combinatorial dimension in the power distribution network is higher, the searching recovery scheme faces combinatorial explosion is effectively solved, the calculated amount is reduced, and the optimum speed is improved.

The invention discloses an opinion mining method based on ensemble learning, which comprises the following steps: selecting a group of element classifier group according to the trained element classifier, and integrating the prediction results of the element classifier group into a sample of a trained ensemble classifier, thereby obtaining the ensemble classifier, and predicting the sample through the ensemble learning of the ensemble classifier. The method selects an optimum element classifier group to carry out a first prediction to the opinion, carries out a secondary study and prediction on the ensemble classifier trained by the prediction results, and generates the final opinion mining result. The classifying results are complementary in diversity; the classifying effect of the method is better than the best single classifier; the generalization ability of the whole classifying model is enhanced, so that the accuracy of the opinion mining is further improved.

The invention relates to a method for identifying damage to a large structure, in particular to a method applicable to identifying damage to a space grid structure. The method includes performing substructure-oriented damage location for the integral structure as a research object, in other words, finely dividing the space grid structure into substructures according to a formation rule of the space grid structure, and identifying possibly damaged substructures by a probabilistic neural network; and performing node-oriented damage location for each integral substructure as a research object within a reduced range, locating damage to specific bars of integral nodes as research objects and determining damage degrees of the specific bars. The method has the advantages that the method is applicable to identifying damage to large structures such as bridges, power transmission towers and high-rise buildings, and is particularly applicable to identifying damage to long-span space grid structures with numerous nodes and bars, the structure of the neural network is simplified, the nonlinear mapping ability and the damage identifying efficiency are improved, and the method has certain engineering practical value.

The present invention provides a cycle-optimized waste collection and transfer planning method and system. Through area division, area path planning, transfer station setting, parking point setting, and path re-planning, according to the garbage points that need to be cleared and transported in an area and the generated garbage Quantity, the present invention utilizes the cost optimization principle, calculates the model and the quantity of the garbage removal vehicle required for completing the garbage removal task, plans the specification, quantity and site selection of the transfer station that needs to be built, and calculates the number of transfer vehicles required, And arrange the daily route planning of each garbage removal vehicle and transfer vehicle. The present invention can solve the problem of garbage removal and transfer planning in the region within the range of optimal cost, and can measure and calculate the overall operating cost.

The invention relates to a method and a device for finding hot videos based on user query logs in real time. The method comprises the steps of: firstly, carrying out word segmentation on user video query logs in a certain period of time to obtain atomic words; then, counting the occurrence number of each atomic word in the user video query logs in the certain period of time and the occurrence number of any two atomic words presenting in the same user query simultaneously; calculating the association degree of any two atomic words in the user video query logs by adopting a pointwise mutual information (PMI) method according to the obtained values of the occurrence number, merging any two atomic words with the association degree exceeding a certain threshold value into a compound word and placing the compound word into a compound word list; and finally, sorting the compound words in a descending manner, wherein the compound words ranking in the front are taken as keywords for finding the hot videos in real time according to a certain proportion finally.

The invention discloses a millimeter wave radar target tracking method in a complex traffic environment. The method is suitable for tracking a radar target in the complex traffic environment. According to the method, a measurement selection mode and associated event generation conditions in a traditional JPDA algorithm are improved, so that the algorithm becomes simple, the calculation amount is greatly reduced, retention of an effective flight path is increased, the possibility that the flight path is a false alarm is lower, and meanwhile, tracking stability is also improved. The method mainly comprises the following steps: 1) updating the track state in a radar target library in real time; 2) generating a confirmation matrix according to the flight path and new measurement; 3) judging whether the flight path is associated with measurement or not through the confirmation matrix, updating the life state of the successfully associated flight path, not tracking the flight path, the lifestate Lt of which is smaller than or equal to 0, and continuously tracking the flight path, the life state Lt of which is larger than 0; 4) generating an incidence matrix according to the continuouslytracked track and the measurement, and calculating incidence probability, and 5) dynamically estimating the motion state of the track.

The invention discloses a self-improved recovery strategy method for complex network local destruction based on improved reinforcement learning, so as to solve the problem of recovery strategy generation when a complex network is subjected to cluster maintenance. The method comprises the following steps: 1, according to local destruction information, a complex network cluster maintenance state matrix is built; 2, based on an initial cluster maintenance state, a complex network adjacency matrix is generated; 3, based on a neural network model, a cluster prior maintenance state transition probability and a maintenance strategy value are predicted; 4, based on a Monte Carlo tree search algorithm, cluster maintenance strategy solution space is traversed, and the global best maintenance actionat present time is selected; 5, based on changes of the cluster maintenance state, the complex network adjacency matrix is updated; 6, based on the cluster maintenance state and the adjacency matrix,the recovery degree of the complex network is calculated and checked; 7, based on reinforcement learning experiment parameters, neural network parameters are trained; and 8, based on a series of bestmaintenance actions during a recovery strategy self improvement process, a complete maintenance recovery scheme is generated.

The application discloses a base station capacity expansion reconstruction scheme design method and related equipment. The speed, the efficiency and the accuracy of a to-be-designed base station scheme design are improved. The method disclosed by the embodiment of the invention comprises the following steps: acquiring the existing network base station information, to-be-designed base station information and a historic scheme set; performing data settlement on the existing network base station information and the to-be-designed base station information to obtain a to-be-used feature; performingalgorithm training learning on the historic scheme set to obtain a scheme evolution path model and a device connection knowledge graph; selecting candidate scheme sub-sets from the historic scheme set through a reverse subtraction algorithm according to the to-be-used feature; performing recommendation level sorting processing on all candidate schemes in the candidate scheme sub-sets according tothe scheme evolution path model to obtain a recommendation result; performing self-perfecting processing on the imperfect candidate scheme according to the device connection knowledge graph when theimperfect candidate scheme is existent in the recommendation result, thereby obtaining the to-be-designed base station scheme set.

The invention discloses a method for polymorphic and dynamic reliability assessment for a shield cutterhead and a drive system. The method comprises the following steps of firstly, establishing a dynamic fault trees for failure of an earth pressure balance shield machine cutterhead and a drive system; secondly, determining a failure rule for a polymorphic and dynamic gate and extending an established dynamic fault tree model to polymorphism; and lastly, establishing a dynamic Bayesian network model for system failure for a dynamic Bayesian network transformation rule according to the dynamic fault tree and determining system reliability. According to the method, a problem that the dynamic reliability analysis of the shield cutterhead and the drive system cannot consider polymorphism, dynamics and interdependency simultaneously is solved, the dynamic reasoning can be carried out according to the statistical data for failure of basic events, and the more efficient and accurate decision information is provided for fault maintenance and repair of a shield machine.

The invention discloses a method for mining a rare mode of multi-dimensional time sequence data. Massive multi-dimensional time sequence data sets are divided into one-dimensional time sequence data sets, and the minimum rare mode is calculated for the time sequence data of each dimension by utilizing an AMRSPM algorithm; the generated minimum rare modes serve as a pattern string dictionary, and all modes are searched out for each dimension of time sequence data through a character string matching algorithm Brute-Fore or a KMP algorithm; rare sub-sequences in the multi-dimensional time sequence data are mined through clustering (AP) operation.

A wind power real-time high precision prediction method based on an adaptive neuro-fuzzy inference system belongs to the wind power technical field. The method is characterized in that firstly, an experiment scheme is designed to collect and process data, establishing a multi-step rolling prediction mode, then constructing a wind power prediction model based on an adaptive neuro-fuzzy inference system (ANFIS), and finally evaluating the prediction precision. In this way, a wind power multi-step rolling real-time prediction method that is based on data, has a self-learning capability and can meet the requirement of real-time prediction precision is achieved. The wind power real-time high precision prediction method based on an adaptive neuro-fuzzy inference system provides a prediction result that is quite close to a real value and is effective, is high in prediction precision, and is highly operable.

The invention discloses a command stream based three-dimensional character animation synthesis method. The method comprises the steps of firstly performing classified description on character action models according to moving states, whole body postures and upper body postures; prefabricating specific three-dimensional animation files according to a combination relation table of action classification; building a distributed client / server network model, and formulating protocol specifications for datagram transmission; and searching the combination relation table of action combination by a client after parsing a command data stream transmitted by a server, acquiring relevant action files, and synthesizing a new animation sequence after performing motion blending and performing interpolation on the location and the orientation of a character. According to the invention, automatic switching between difference actions of virtual soldiers in a virtual battlefield environment can be reproduced authentically, smooth interpolation update of the position and the orientation of virtual characters is supported at the same time, a problem of combination explosion between the tedious command control process and actions is avoided, action switching is performed smoothly and naturally, and the method can be widely applied to various virtual battlefields. Therefore, the command stream based three-dimensional character animation synthesis method has a certain application value.

The invention relates to a minimum cost test case generation method based on an EFSM model. Based on an expansibility finite state machine and a set division theory, a migration coverage algorithm isdesigned, and a migration sequence set is obtained; And abstract models are given to different positions of the two migration sequences, analysis is carried out according to the abstract models, and then a reduction merging algorithm is designed to obtain a minimum test sequence set. Converting the minimum test sequence set to obtain an effective test path set, and finding out a shortest path froman initial state to a certain state by adopting reverse thinking in a conversion process; According to the method, the migration coverage criterion is met, invalid test paths and combination explosion problems can be avoided, the code coverage rate can be ensured, the number of the test paths in the set reaches the minimum, and the balance between the test cost and the test efficiency is ensured;The method is not only limited to an EFSM model, but also applicable to other models capable of being abstracted into directed graphs, and particularly remarkable in effect of solving a large complexnetwork graph.

The invention belongs to the field of industrial control safety technology and in particular relates to an efficient industrial control protocol fuzz test method, comprising the following steps: analyzing a protocol format, varying a single weak point in a protocol by adopting a strategy in a variation strategy set Mut, generating an original one-dimensional deformed sample set, sending to to-be-tested equipment as a test sample set TDB, detecting equipment exception record feedback information, calculating a fitness function, selecting a parent sample set FDB based on a roulette strategy, performing crossover and variation on individuals in the parent sample set FDB in pairs and generating a new son sample set SDB, and taking the parent sample set FDB and the son sample set SDB as a testsample set TDB for a test in a new round; and repeating by using the new test sample set TDB until dimensionality is saturated. The method provided by the invention generates a multi-dimensional testcase based on multiple strategy variances of fields at the same time, has diversity and can raise a multi-contact exception caused by adoption of different strategy variances on different fields.

The invention discloses a method for constructing a cascading Bayesian network for solving a combinatorial explosion problem. The method comprises two core parts: A) a method for constructing the topological structure of a cascading Bayesian network by a system access, and B) an intermediate node probability parameter setting method of the cascading Bayesian network. Firstly, the construction of the topological structure of the Bayesian network is finished, wherein the topological structure of the Bayesian network is consistent with a fault cascading configuration in engineering practice, then, the setting of each node condition probability table of the Bayesian network is finished, and finally, the construction of the cascading Bayesian network is jointly finished. When the cascading Bayesian network finishes being constructed, any existing Bayesian network reasoning technology can be applied to carry out reasoning calculation on the cascading Bayesian network so as to obtain system reliability. While a reliability calculation result of the system can be guaranteed to be correct, the probability parameters in the network are effectively reduced, specifically, the number of the probability parameters on each access can be lowered to a linear order from an exponential order, calculation efficiency is improved, and a combinatorial explosion problem is solved.

The invention discloses a multi-target tracking data association method and system, and the method comprises the steps: enabling a known starting time period track of a target measurement point to serve as a reinforcement learning training process according to the multi-target tracking data association characteristics, generating random clutters around the known measurement point in one step, andenabling the clutter points and the known measurement point to serve as radar collection measurement points; screening out candidate measurement points from the measurement points according to a tracking door, performing data association on all the candidate measurement points according to the matching degree and the position distribution rule by utilizing motion matching and reinforcement learning according to the motion characteristics of the target, and training an experience matrix of a reinforcement learning data association model according to an association result checked by one-step known measurement points; and performing data association on the track points of the target entering the clutter area in combination with motion matching, and continuing to optimize the empirical matrixaccording to an association result until track association is completed. The problems of low correct association rate, high calculation complexity and the like are solved, the correct association rateis improved, and the calculation complexity is reduced.

The invention provides an intelligent game decision-making method and system for multi-unmanned aerial vehicle task distribution in a countermeasure environment. The method comprises the steps of determining a plurality of second monitoring strategies of a monitoring party according to the same monitoring area in first monitoring strategies of the monitoring party and attack strategies of an attacker; screening a target attack strategy from the attack strategy on the basis of a non-discovery probability of a strategy pair comprising a first monitoring strategy and an attack strategy; and finally, screening a target monitoring strategy from the second monitoring strategies to monitor the attacker based on the second monitoring strategies, the target attack strategy and the monitoring discovery probability. According to the technical scheme, a stackelberg game model is used for solving, so that the technical problem that an effective patrol strategy cannot be made on the premise that theattacker can observe an implemented patrol strategy to discover the attacker with the maximum probability is solved; and meanwhile, the monitoring strategies of the monitoring party are processed, and the strategies of the attacker are screened, so that the calculation amount is reduced, and the calculation efficiency is improved.

The invention discloses a multi-target tracking data association method and system. The method comprises: regarding a track of a target measurement point in a known starting time period as a reinforced learning training process according to a multi-target tracking data association characteristic, generating random noise around a one-step known measurement point, wherein both a clutter point and the known measurement point are regarded as radar acquisition measurement points; screening out candidate measurement points and a target motion characteristic from the measurement points according to atracking gate, performing data association on all candidate measurement points by using sport matching and reinforced learning according to a matching rate and a position distribution rule, checkingan association result by using the one-step known measurement point, and training an experience matrix of a reinforced learning model; and performing, according to the trained experience matrix in combination with sport matching, data association on a track point that the target enters a clutter area, and continuing to optimize the experience matrix by using the association result until track association is completed. Problems in the prior art that a correct association rate is relatively low and calculation complexity is relatively high are resolved, so that a correct association rate is improved, and calculation complexity is reduced.

Disclosed are a hot video real-time finding method and device based on user query logs. The method comprises the steps of firstly, conducting word segmentation on the user video query logs in a period of time and obtaining atomic words; then, counting times of each atomic word occurs in the user video query logs in the period of time and times of any two of the atomic words occur in the same user query at the same time; according to a obtained time value, adopting a pointwise mutual information (PMI) method to calculate the correlation degree between any two of the atomic words in the user query logs, and merging any two of the atomic words of which the correlation degree is larger than a threshold value into a compound word and putting the compound word into a compound word list; finally, descendingly sorting the compound words, and at last, using the compound words sorted in the front as key words of hot video real-time finding according to a certain proportion.

The invention discloses a method for mining Granger causality between visibility multi-dimensional spatio-temporal data, belonging to the technical field of data mining. Firstly, the Granger causalityanalysis is used to extract the causal features of visibility from some sample data, and then classify all the data according to different administrative districts of Beijing Municipality, Fc causality measurement factor is used to determine the strength of the influence relationship between different regions, Finally, an improved spatio-temporal Granger Lasso algorithm is used to train the causality model, so that the Granger causality scores of different regions, different influence factors and visibility can be obtained, and the qualitative and quantitative analysis of influence factors can be realized.

The invention relates to an object data processing method and device, computer equipment and a storage medium. The method relates to a decision tree model in the field of machine learning, and comprises the following steps: converting object behavior data corresponding to each portrait dimension into a strategy effect corresponding to a target strategy according to the proportion of a first hit category belonging to the implementation of the target strategy on a sample object in the object behavior data corresponding to each portrait dimension of the object portrait data; inputting the strategy effect corresponding to each portrait dimension into a decision tree model, and training the decision tree model by taking strategy effect optimization corresponding to portrait dimension combinations represented by nodes in the decision tree model as a target to obtain a group division decision tree; and determining an optimal subdivision group corresponding to the target strategy based on thestrategy effect corresponding to each subdivision group represented from the root node to the leaf node of the group division decision tree. By adopting the method, the efficiency of mining the subdivided group strategy result from the test result can be improved.

An automatic driving decision-making system based on a knowledge graph comprises a knowledge graph database and a decision reinforcement learning module. The knowledge graph database maps the driving knowledge on the Internet into a triple form, and knowledge expression and reasoning are performed by using the driving knowledge graph, so that massive knowledge can be managed in a classified manner, the time spent in a traditional rule-case matching process can be shortened, and the real-time performance of knowledge retrieval is improved. The knowledge graph database obtains a driving scene, provides expert experience for the decision reinforcement learning module through a driving scene experience sample stored in the knowledge graph database, outputs a high-confidence-coefficient driving strategy to the decision module, guides the decision module to adapt to a complex and changeable traffic environment, guarantees the safety of a vehicle, and improves the safety of the vehicle. And meanwhile, the interpretability of the automatic driving decision information is realized through the knowledge graph of the knowledge graph, so that the credibility of the automatic driving decision system is improved, and the credibility of passengers on automatic driving vehicles is improved.

The invention relates to a lithium battery pack system reliability optimization design method based on a multi-physical field simulation and response surface analysis method. According to the method,a redundancy design scheme is formulated according to the size and reliability requirements of a lithium battery pack physical model; a response surface experiment scheme is designed by determining abattery arrangement mode and to-be-optimized design parameters; the physical process of a system is subjected to simulation analysis by establishing a lithium battery pack multi-physical-field model;the reliability of the system is evaluated and analyzed by constructing a lithium battery pack polymorphic system reliability model and a randomness model based on multi-physical field coupling; and based on all design schemes and analysis results thereof, a response surface is constructed, and the reliability optimization design work of a lithium battery pack system is finished by utilizing a response surface analysis method. According to the method, a multi-physical-field simulation technology, a system reliability method, a random uncertainty method and a response surface analysis method are fused, the physical process can be scientifically and accurately described, and the optimal redundancy and layout design scheme can be efficiently obtained.

The invention provides an optimal design method for bending rigidity ply sequence of symmetrical composite laminate, which comprises the following steps: screening a sensitivity laying angle at a certain specific laying position through a linear and nonlinear sensitivity detection technology, and sequentially replacing an existing laying angle with the sensitivity laying angle from a middle surface position to an external position for optimization. The method can effectively solve the problem of combined explosion of the equal-thickness composite laminated plate laying sequence design space, and the purposes of greatly improving the bending rigidity and improving the design efficiency are achieved.

The invention discloses a method for constructing a TCM syndrome differentiation medical theory model, comprising: S1. Constructing a syndrome-syndrome-element relationship S set according to the mapping relationship between syndromes and syndrome elements; S2. Performing a flute on the S set Carl product operation to obtain a plurality of evidence element combinations; S3. Generate corresponding evidence element relationship combinations according to the multiple evidence element combinations, and form multiple evidence element directed graphs according to the multiple evidence element combinations and their corresponding evidence element relationship combinations; S4. Using the ISO-R rule to screen out the ISO-R directed graph from the element-directed graph; S5. Outputting the ISO-R directed graph in the form of a graph. The present invention uses graphic form to output the pathogenesis, compared with the four-character syndrome type obtained by the traditional model, the content of the pathogenesis is comprehensively and intuitively explained; and the Cartesian product according to the mapping relationship between syndromes and syndrome elements The result of the operation and the relationship between the certificate elements generate a certificate element directed graph, and then use the ISO-R rule to screen the certificate element directed graph, which solves the problem of certificate element combination explosion in the traditional certificate element dialectical model.

The invention discloses a cluster system preventive maintenance method based on deep reinforcement learning, and solves the preventive maintenance problem of a cluster system in a long-term operation process. The method comprises the following steps: 1, establishing a residual life state matrix of a single system-unit cluster of a cluster system according to a degradation state; 2, evaluating the reliability level of the cluster system based on the residual life state matrix of the single system-unit cluster of the cluster system; 3, designing a neural network to predict the prior maintenance probability and the prior maintenance strategy value of the single system-unit cluster of the cluster system; 4, constructing a preventive maintenance strategy solution algorithm architecture, traversing a preventive maintenance strategy solution space, and selecting a series of optimal maintenance actions; 5, calculating the reliability of the cluster system based on the change of the residual life state of the cluster, and then checking the recovery degree of the cluster system; 6, generating a complete preventive strategy by the optimal maintenance actions stored in the preventive maintenance strategy solving process.

The invention relates to a three-phase unbalanced treatment method based on linear programming. The method comprises the steps that an objective function of three-phase unbalanced treatment is constructed; a current equation constraint of three-phase unbalanced treatment is constructed; a phase unique constraint of three-phase unbalanced treatment is constructed; a phase non-adjustable constraintof three-phase unbalanced treatment is constructed; a phase-integral non-separable constraint of three-phase unbalanced treatment is constructed; a maximum phase switching number constraint of three-phase unbalanced treatment is constructed; a three-phase unbalance treatment model based on linear programming is formed to acquire a three-phase unbalanced treatment plan. The method can provide support for power system operators to manage three-phase imbalances and reduce line losses.

The invention provides a flight control actuation system configuration method based on the constraint satisfaction problem. The method comprises the following steps: S1 variable modeling; S2 value domain modeling; S3 constraint set modeling; and S4 constraint satisfaction problem solving.