55 results about "Combination strategy" patented technology

The invention discloses a cooperative control and disturbance suppression method for a multi-axis servo system, and relates to the technical field of cooperative and disturbance control of servo systems, position and speed signals of each servo axis are acquired in real time, a cooperative error vector is calculated based on a graph theory model, the vector reflects the deviation between the state of each axis and an expected cooperative trajectory, and the cooperative control and disturbance suppression of the multi-axis servo system are realized. And synchronous movement is ensured. The collaborative error is used for generating a preliminary control signal, and a proportional, differential and integral term combination strategy is adopted to quickly correct the error and eliminate the steady-state error. Meanwhile, disturbance vectors are estimated through an observer, and external interference or internal changes are compensated. And after the preliminary control signal and the disturbance compensation are added, amplitude limiting processing is carried out to prevent saturation of the driver. The control parameters are adjusted through an off-line optimization algorithm to minimize errors and disturbance influences, so that the tracking precision, the anti-interference capability and the adaptability of the multi-axis servo system are improved, and the method is suitable for the fields of high-precision manufacturing, robot control and the like.

The invention relates to the technical field of natural language processing, and provides a multi-granularity text compression method and system for large-model multi-round dialogues, and the method comprises the steps: carrying out the semantic coding of historical question and answer pairs and a current dialogue question through a pre-training model, obtaining a semantic vector, generating a historical dialogue semantic representation through a self-attention mechanism, and carrying out the semantic representation of the historical dialogue. A semantic vector of a current dialogue problem is combined, a correlation enhancement vector is obtained through a target attention mechanism, weighted fusion is carried out on the correlation enhancement vector and historical dialogue semantic representation, a fusion vector is obtained, a feature vector is generated through a maximum pooling and average pooling combination strategy, clustering is carried out, positive and negative sample pairs are constructed, and semantic similarity loss calculation is carried out. Iteratively training the classification model; and based on the trained classification model, screening question and answer pairs related to the current dialogue question, and generating a structured abstract. Redundant content is remarkably reduced while key semantic information is reserved, and efficient context representation adaptive to large model reasoning is formed.

The invention discloses a table collaborative editing method and system, and relates to the technical field of artificial intelligence, and the method comprises the following steps: during multi-user collaborative editing, receiving user editing operations, packaging each user editing operation into an independent operation sequence, summarizing all the operation sequences, and performing unified operation on the operation sequences, the operation sequence comprises content information and parameter information of user editing behaviors; performing operation conflict detection based on the operation sequence, and if an operation conflict is detected, dividing the conflict into a direct conflict, a structure conflict, a style conflict and a dependency conflict through a conflict classifier; corresponding strategies are matched and executed for different conflict types through the intelligent conflict solver, the strategies comprise the automatic merging strategy, the user selection strategy and the delay processing strategy, and the problems that in the prior art, cell merging intelligent verification cannot be achieved, and conflict solving cannot be achieved during multi-user collaborative editing are solved are solved.

The invention relates to a tail end constant-pressure water supply strategy adjusting method and system based on AI, and belongs to the technical field of secondary water supply.The constant-pressure water supply strategy adjusting method comprises the steps that the flow of a header pipe and real-time health state data of each running pump are periodically collected; according to the header pipe flow, the dynamic demand lift is calculated, and all feasible pump set combination strategies are generated in combination with the dynamic demand lift and the real-time health state data; the score of each candidate strategy is calculated, the candidate strategy with the highest score is used as a pump set scheduling strategy, the pump with the most matched performance is selected from candidate pump sets corresponding to the pump set scheduling strategy to serve as a frequency conversion main guide pump, and the other pumps serve as frequency conversion auxiliary pumps; according to the set minimum service pressure and the dynamic demand lift, the target pressure of a pump outlet is calculated; and according to the target pressure and the actually-collected pump outlet pressure, the rotating speed of the variable-frequency main guide pump is adjusted. The method has the beneficial effects that the power of the pump set is matched with the actual requirement, so that the operation cost is reduced.

The invention discloses a system for improving item recommendation accuracy based on space-time diffusion and information bottleneck. The system comprises the following modules: a normalization processing module, a space aggregation module, a time aggregation module, a time diffusion module, a space diffusion module, an information bottleneck module and a prediction module. The system can effectively model an error behavior which is not solved by a current next basket recommendation technology, and through an innovative spatial diffusion module, the system does not depend on strong association among commodities or historical preferences of users any more, but focuses on learning of article spatial combinatorial logic in a shopping basket, so that the system is more intelligent. Therefore, the system can identify the targeted and hidden combination strategy of the user, thereby improving the understanding of the real purchase intention of the user, and achieving the purpose of improving the article recommendation accuracy.

The invention discloses a service combination method based on a colored Petri net and a deep reinforcement learning agent. The method comprises the following steps: S1, constructing a colored Petri net model for manufacturing a service combination system; s2, constructing a deep reinforcement learning environment based on the model, and defining a state space, an action space and a reward function; s3, interacting the intelligent agent with the environment, dynamically generating an action mask by using a Petri net model to constrain an action space, and training the intelligent agent to converge by using a near-end strategy optimization algorithm; and S4, taking the strategy obtained by training as an optimal service combination strategy, and triggering transition in the Petri network to generate a final manufacturing service combination scheme. According to the method, the formalized modeling capability of the colored Petri network and the self-adaptive decision-making capability of deep reinforcement learning are deeply coupled, so that the problem that in a dynamic complex industrial interconnection environment, the formalized modeling capability of the colored Petri network and the self-adaptive decision-making capability of the deep reinforcement learning are influenced is solved; the problems of low decision-making efficiency, poor adaptive ability and opaque optimization process caused by huge decision-making space and complex logic constraints in a manufacturing service combination method are solved.

The embodiment of the invention provides a strategy determination method and device. The method comprises the following steps: acquiring real-time index data and a real-time prediction load of a Serverless function; determining a plurality of target optimization functions according to the real-time index data and the real-time prediction load; the multiple target optimization functions comprise a first target function and a second target function, the first target function is used for minimizing the cold start delay of the function, and the second target function is used for minimizing the resource cost of the function; the candidate solution of each target optimization function indicates a combined strategy of a resource allocation strategy and a cold start strategy; a plurality of target optimization functions are processed by utilizing a multi-target evolutionary algorithm, and a target combination strategy is screened out from numerous candidate solutions by simulating the principle of survival of victims and inferior and continuous optimization of populations in a biological evolution process, so that the resource cost of the functions is reduced on the basis of ensuring cold start low delay of the functions; and the resource allocation efficiency is improved.

The present application relates to a kind of multi-stage self-defined combination strategy warehouse picking optimization method, device and medium, wherein, method includes collecting order data to be picked and preprocessing generation structured data set;Support user self-defined configuration wave division dimension and weight, complete order clustering and generate picking wave with unique identification;Divide the whole process of picking into four core stages, provide each stage strategy selection and combination interface, generate compatible multi-stage strategy combination scheme;Correlation scheme and wave task are executed in stages, real-time acquisition of field data and dynamically correct abnormal strategy;After completing picking, the core index is counted, report is generated and review output optimization suggestion, the present application solves the problem that traditional picking strategy is poor in adaptability, low efficiency, can be flexibly adapted to multiple warehouse scene, improve picking efficiency and accuracy, reduce resource consumption, and the practical value is remarkable.

The invention discloses an intelligent flight route planning method and system based on big data. The method comprises the steps of multi-source flight route data acquisition, obstacle cluster construction, high-density obstacle area identification, hierarchical planning cost function design, flight planning route parameter optimization and intelligent flight route planning. The invention relates to the technical field of data processing, in particular to an intelligent flight route planning method and system based on big data, according to the scheme, a multi-dimensional high-density obstacle area recognition mechanism is innovatively proposed, a hierarchical planning cost function collaborative technical scheme is designed, the flight safety is enhanced, and the obstacle avoidance efficiency is improved; introduction of a density perception initialization strategy and a double-layer particle ball merging strategy is innovatively proposed, a clustering algorithm is improved, and clustering accuracy and stability are improved. An optimization algorithm is improved by adopting a self-adaptive exploration strategy and a quadratic interpolation strategy, the efficiency of route parameter optimization and the quality of an optimal solution are improved, and accurate optimization of flight planning route parameters is realized.

A video recommendation method fusing social information comprises the steps that firstly, a user-video interaction graph and a social graph are constructed through video media platform data, different combination strategies of a diffusion model and graph convolution operation are applied in a denoising priority path and a structure priority path respectively, and collaborative perception and fusion of user embedding and structure information are achieved; secondly, aligning user embedding output by the double-track denoising path by adopting comparative learning, and enhancing robustness; secondly, performing deep coding on the user-video interaction graph and the social graph by applying a double-graph neural network, comprehensively capturing user preferences and embedding videos; and finally, for noise possibly introduced by multi-module fusion, deep denoising is executed to purify user embedding again. According to the method, guidance of structural information is introduced in the diffusion denoising process, and more robust user embedding is obtained by using a comparative learning strategy, so that a structural perception high-fidelity social denoising task is completed, and personalized video recommendation is realized on a video media platform.

The invention discloses a social governance event management method based on multi-source data fusion, and relates to the technical field of social governance and public services. The four modules are respectively a multi-source heterogeneous data fusion and feature standardization module, an event intelligent assignment decision engine module, a repeated event identification and combination strategy module and a whole-process closed-loop management and treatment efficiency quantitative evaluation module, and the multi-source heterogeneous data fusion and feature standardization module comprises feature extraction and feature fusion. According to the social governance event management method based on multi-source data fusion, a heterogeneous event feature vector fusion method based on deep learning is adopted, specifically, a text is vectorized through BERT / Word2Vec, and multi-source heterogeneous data is fused into a unified event feature vector in combination with a weighted summation method, so that information islands formed by multi-source data splitting are avoided, and the social governance event management efficiency is improved. And the integration of the social governance event management data is improved.

The invention discloses a multi-type product combined packaging method and system based on visual quality inspection and related equipment. The method comprises the steps that visual quality inspection is conducted on each product; packaging the plurality of products passing the quality inspection one by one; analyzing the uploaded package combination strategy, and determining all target product types and the quantity specification corresponding to each target product type; and according to each target product type and the corresponding quantity specification, calling each packaged product, and packaging each called product to form a commodity. Therefore, a full-link quality inspection packaging system including product quality inspection screening, packaging and commodity obtaining by combining a plurality of products according to requirements can be formed, and the quality stability and production pertinence of the commodities are ensured. Meanwhile, commodities can be customized and constructed according to a package combination strategy, the overall response speed of multi-type product combination package is improved while the defective rate of production is reduced, and the application universality of the method is improved.

This invention belongs to the field of resource allocation and discloses an emergency resource demand prediction method and apparatus based on WGAN-Stacking. The method includes: acquiring meteorological, geographical, and power grid data; generating fault samples using a Wasserstein generative adversarial network; quantifying the distribution difference between the generated fault samples and real samples based on FID distance and kernel Inception distance to screen target samples whose sample quality reaches a preset threshold, with the quantified distribution difference used to characterize the sample quality of the fault samples; inputting the target samples into a pre-constructed resource demand prediction model to predict the emergency resource demand in different regions; the resource demand prediction model adopts a Stacking ensemble learning framework, using RF, GBDT, and SVM as heterogeneous base learners and FNN as a meta-learner, learning the optimal nonlinear combination strategy between the prediction results of each base learner; and allocating emergency resources according to the prediction results output by the resource demand prediction model.

The invention provides a novel general design method of a bispecific antibody CH3 interface engineering strategy, which can efficiently screen an effective mutation combination strategy from massive CH3-CH3 asymmetric mutations so as to improve the purity of a heterodimer compound in a bispecific antibody production process. The core of the method is that a CH3 interface engineering strategy design technology driven by artificial intelligence is adopted, a machine learning model and a multi-dimensional mutation combination filtering and screening system are integrated, virtual screening is carried out on massive CH3-CH3 asymmetric mutations, and a small number of mutation combinations with high reliability are obtained; and through wet experiment verification, a new effective mutation combination strategy is finally determined and obtained. Benefited from accurate screening of artificial intelligence, the verification range of the wet experiment is greatly reduced, the research and development period can be remarkably shortened, and the experiment cost is reduced. The invention also provides a mutation combination strategy obtained by screening and verification through the design method, and the problem of purity control of the bispecific antibody heavy chain heterodimer can be effectively solved by adopting the strategy.

The invention provides a resource scheduling method and device, a medium and a program product. The resource scheduling method provided by the embodiment of the invention comprises the following steps: acquiring a plurality of resource requests; a feature tuple of each resource request is extracted, the feature tuple comprises at least one feature item, and each feature item represents a standardization result of one feature of the resource request; and based on the feature tuple, determining a merging strategy of the plurality of resource requests, the merging strategy being used for indicating a resource scheduling mode of the plurality of resource requests.

This invention discloses an energy-saving pump system based on adaptive flow regulation, relating to the field of energy-saving control technology for fluid transport equipment. The system comprises the following components: constructing a distributed control network, establishing a distributed multi-objective optimization model based on load forecasting, an improved intelligent optimization algorithm, executing distributed collaborative optimization, and outputting optimized control commands. By constructing a multi-objective optimization model that integrates operating energy consumption costs, equipment wear costs, and start-stop penalties, combined with a dynamic adaptive penalty function and an intelligent optimization algorithm based on the optimal efficiency point, this invention significantly reduces the total energy consumption of the system while ensuring accurate fulfillment of flow requirements. Furthermore, by dynamically correcting the pump performance model and adaptively adjusting the pump combination strategy, it avoids long-term deviation of the pump from the optimal efficiency point, reduces equipment wear index, extends expected service life, and significantly reduces the total life-cycle maintenance cost.

This invention relates to the field of energy-saving technology for distribution transformers, and discloses a method and system for optimizing energy saving in distribution transformers based on intelligent switching technology. The method uses intelligent terminals deployed on the low-voltage side of the distribution transformer to accurately collect real-time operating parameters such as load current, voltage deviation rate, temperature rise rate, and power factor of multiple adjacent transformers. Based on these parameters, a dynamic feature tensor reflecting the collaborative operating status of the transformer cluster is constructed, and an adaptive sliding window algorithm is used to perform piecewise normalization processing in the time dimension. The normalized tensor is input into an operating mode decision model based on a deep dynamic graph neural network, and the output is a decision vector containing the optimal switching combination strategy. This vector includes key information such as single-unit operating identifiers and parallel operating identifiers. A control command sequence is generated based on the decision vector and sent to the circuit breaker actuators of each transformer through the distributed communication protocol of the intelligent terminal, realizing automatic switching of operating modes.

The application discloses a content recommendation scene-based experiment strategy evaluation method, device and medium, comprising: creating opposite first and second shunt paths for first and second virtual recommendation requests; generating a first recommendation combination strategy based on N recommendation strategies associated with N path nodes in the first shunt path, and generating a second recommendation combination strategy based on N strategies associated with N path nodes in the second shunt path; obtaining a first recommendation result list by processing the first virtual recommendation request based on the first recommendation combination strategy, and obtaining a second recommendation result list by processing the second virtual recommendation request based on the second recommendation combination strategy; obtaining a target recommendation result list by combining the first and second recommendation result lists; and evaluating the effect difference of different recommendation strategies based on user feedback data of the target recommendation result list. The application can improve the AB test accuracy in the content recommendation scene.

The invention relates to a receipt merging method and device, computer equipment, a storage medium and a program product. The method comprises the following steps: acquiring a to-be-merged document and a document state corresponding to the to-be-merged document; determining a target merging strategy from a plurality of candidate merging strategies based on the to-be-merged documents and the document states corresponding to the to-be-merged documents; determining a target merged document corresponding to the to-be-merged document based on a merged document matching condition and a merged document limiting condition in the target merging strategy; and merging the to-be-merged receipts into the target merged receipts based on receipt merging conditions in the target merging strategy to obtain updated target merged receipts. By adopting the method, the document merging accuracy can be improved.

The invention relates to the technical field of optimization control of a fire coal pulverizing system of a thermal power plant, and discloses a combination strategy optimization method and system for multiple coal mills, and the method comprises the following steps: S1, building a multi-dimensional dynamic sensing system of the operation states of the coal mills, coal type adaptability parameters, barrel vibration spectrum characteristics and outlet pulverized coal fineness distribution of each coal mill and instantaneous load fluctuation data of a driving motor are obtained; and S2, constructing a comprehensive performance portrait of the coal mill based on the historical operation efficiency and the equipment health degree, wherein the performance portrait at least comprises an equipment efficiency attenuation trend, a key component wear early warning level and a historical average fault-free operation duration. By establishing a multi-dimensional dynamic sensing system and a comprehensive performance portrait, refined, digital and prospective management of the running state of the coal mill is realized, so that formulation of a combined strategy is changed from experience-based driving to data-based driving, and scientificity and accuracy of decision making are remarkably improved.

The invention relates to a multi-source data fusion building space lease intelligent recommendation method and a related device. The method comprises the following steps: acquiring a multi-source data set corresponding to a current building object, wherein the multi-source data set comprises building space information, Internet of Things perception information and lease demand information; performing knowledge fusion and state mapping on various types of data according to a scene association relationship among the various types of data in the multi-source data set to obtain a state label corresponding to each building space in the current building object; and in the lease scene where the current building object is located, obtaining an optimal lease combination strategy for allocating corresponding tenants to each building space according to the space function layout and the rent calculation condition jointly reflected by the state labels corresponding to each building space. By adopting the method, the overall relevance of the rental decision and the utilization efficiency of the building space can be improved.

The invention relates to the technical field of household appliances, and discloses a household appliance coordinated regulation and control system and method adaptive to environmental parameters, and the method comprises the steps: collecting the environmental parameters, human behavior states and equipment energy consumption data through a multi-mode sensing network after the system is started, fusing the urban microclimate prediction information, and generating a standardized multi-dimensional fusion state vector; according to the environmental field digital portrait and the evolution path thereof, mapping a regulation and control target into an optimal household appliance service combination strategy through a functional semantic equivalent replacement mechanism, and generating a semantic-level collaborative regulation and control instruction set; the method comprises the following steps: constructing a full-chain technical system of a multi-modal sensing network, neural implicit field environment modeling, functional semantic equivalent replacement, human body thermal comfort feedforward prediction, sound field cooperative control and federated learning closed-loop optimization; the intelligent cooperative control transition from traditional'responsive, homogeneous and equipment-centered 'household appliance control to'prospective, spatialized and human-centered' household appliance control is realized.

The present application relates to the technical field of structured travel commodity combination, and provides a combined travel commodity recommendation scheme generation method based on a retrieval enhanced large model, which comprises extracting user preference information and combination strategy from user data by using a large model, filtering commodity information according to the user preference information and combination strategy, integrating service information according to the user data and commodity information, selecting a semantic recall scoring mode or a feature scoring mode according to the service information and user data, and obtaining a candidate commodity list from the filtered commodity set; combining the candidate commodity list through the large model to generate a travel commodity combination meeting the user demand; and generating a graphic commodity recommendation scheme according to the travel commodity combination. The present application significantly improves the automation and intelligence level of the travel commodity combination, reduces the artificial operation cost, improves the combination adoption rate and user click rate, and has good commercial promotion value and application prospect.

The application discloses a game product play generation method and a terminal, which classifies and collects common elements of sweet points in case materials; determines a first combination strategy from preset sweet point combination strategies according to environment composition data of a target product, each sweet point combination strategy is adapted to at least one virtual scene, and each sweet point combination strategy indicates that the probability of adaptation of a sweet point common element to the sweet point combination strategy is higher than the probability of adaptation to other sweet point combination strategies; and generates a game play adapted to the target product based on the environment composition data of the target product and the first combination strategy. The application ensures the pertinence and efficiency of play adjustment by dynamically selecting the most suitable sweet point combination strategy; the method can intelligently adjust according to the environment composition data of the target product, ensure that the generated play can be highly matched with the actual environment of the target game, and further improve the overall experience of the game.

The present application relates to the technical field of big data analysis, in particular to a carbon emission monitoring system for a smart logistics park based on big data analysis, which comprises a cargo correlation analysis module for extracting item number and quantity information in an order according to order flow data of the smart logistics park, counting co-occurrence frequency of different items under the same order number, and generating a high-frequency co-occurrence cargo combination list. In the present application, by calculating distance increments of independent orders and combined orders on traversal paths, and combining time urgency and synchronization of deadlines to construct a wave carbon compatibility numerical matrix, the optimal combination strategy can be locked under the premise of meeting the time requirement, which closely combines static storage space optimization and dynamic job path planning, reduces invalid driving distance and repeated lifting times of the handling equipment, reduces energy consumption of logistics operations, and realizes active control of carbon emissions from the source.

The invention discloses a method and a system for generating and retrieving a searchable data asset label based on a multi-modal large model. Relates to the technical field of data asset label generation and retrieval. The method comprises the steps of obtaining different types of data assets and performing preprocessing to obtain multi-modal data; constructing a lightweight multi-modal model, and training the lightweight multi-modal model by adopting a knowledge distillation and fine tuning combination strategy; inputting the multi-modal data into the trained model to obtain different key information; generating corresponding labels according to different key information, and constructing an associated data asset association network according to the generated labels; and performing three-level mixed retrieval according to the associated data asset association network to obtain a final retrieval result. According to the method, diversified and high-correlation retrieval results are intelligently expanded by mining semantic association among the tags, and the discovery precision of data assets and the transaction matching efficiency are greatly improved on the premise of guaranteeing data privacy.

The application relates to the technical field of cloud computing and cloud service management, in particular to a cloud service adaptive combination method based on a Decision Transformer and a decision support system; the method unifies heterogeneous service data into hierarchical service vectors through hierarchical tokenization and service representation construction; a multi-scale interaction feature extraction and semantic enhancement mechanism is constructed to capture the semantic and behavior dependence between cloud services at different granularities to generate enhanced service representation and compatibility matrix; the state, action and target are jointly embedded into a Decision Transformer model to generate a cloud service combination strategy sequence; through adaptive combination optimization and backtracking adjustment mechanism, the execution feedback and environmental changes are dynamically evaluated, and multi-instance knowledge migration and online strategy generalization are combined to realize cross-instance and cross-scene decision support. The application realizes continuous output of an optimal service combination scheme in a dynamic cloud environment.

The application discloses a kind of blockchain health degree test method and its device, storage medium, program product, first determine the target blockchain to be tested and the multiple blockchain nodes corresponding to target blockchain, determine the running state of multiple blockchain nodes when running target blockchain, when multiple blockchain nodes include the first type node of online state and the second type node of offline state, determine the combination strategy of online health degree test sub-strategy and offline health degree test sub-strategy as target health degree test strategy, according to target health degree test strategy, obtain the multidimensional comprehensive service parameter between multiple blockchain nodes, and according to multidimensional comprehensive service parameter, determine the running health degree of target blockchain and generate health degree test result.The embodiment of the application can more accurately detect the running health degree of blockchain.The embodiment of the application can be applied to, for example, blockchain performance test, blockchain performance detection and other various scenes of blockchain technology.

This invention relates to the field of computer system performance and security testing technology, specifically a computer performance and security testing method and system. Specifically, it involves: acquiring and standardizing performance indicators, behavioral characteristics, and security event data of the target system in real time; using a Gaussian mixture model to cluster the dimensionality-reduced behavioral characteristics to establish normal behavior pattern clusters; combining a Hidden Markov Model to perform time-series analysis of the behavioral sequences to identify system state transition paths; employing a weighted combination strategy to inject normal and attack-type loads, and adjusting abnormal behavior chains through a perturbation function; constructing a directed weighted association between abnormal events and performance indicators based on an event-driven causal performance graph, quantifying the system risk index, and triggering alarms; outputting performance bottlenecks, security vulnerabilities, and optimization suggestions, and updating the knowledge base to optimize subsequent models. This invention achieves deep coupling of performance and security testing, intelligent scheduling of dynamic loads, and interpretable location of risk paths.