98 results about "Combination strategy" patented technology

The invention discloses a multi-knowledge-base scheduling routing method based on intention recognition and feedback optimization, and the method comprises the steps: S1, receiving multi-modal input from a user, carrying out the deep semantic analysis of the input content through natural language understanding and multi-modal analysis, and outputting a standardized intention label; s2, constructing a capability portrait for each knowledge base, and then generating a route matching score matrix according to a similarity relationship between the user intention tag and the capability portrait of each knowledge base; s3, screening out one or more most relevant candidate knowledge bases according to the scoring matrix and a set matching threshold value, and selecting a retrieval combination strategy to realize information recall; s4, information retrieval is executed based on the selected knowledge base, recall results and original user input are submitted to a large language model for content generation, and answers or suggestions are formed and output; and S5, user feedback collection and strategy optimization are carried out. According to the method, a more accurate, efficient and self-evolutionary knowledge base scheduling mechanism can be realized.

The invention discloses an intelligent item selection recommendation system of a live broadcast room based on AI digital people, relates to the technical field of intelligent item selection recommendation, and solves the technical problems of generating a dynamic and high-confidence matching scheme based on highly matched commodities and adjusting a combination strategy in combination with a real-time scene and user feedback. Through the multi-dimensional weighted calculation of the feature matching score, the crowd matching score and the scene association score, the hidden integrating degree of the commodity and the user demand is quantified, the highly matched commodity is screened out, the invalid recommendation is reduced, the comprehensive confidence degree of the commodity combination is calculated based on the association rule algorithm, the personalized matching scheme conforming to the real-time scene is generated, and the user experience is improved. A high-potential scheme is screened through a self-adaptive threshold value, the matching rationality and transformation potential are improved, a comprehensive feedback processing module is introduced, a user is supported to customize combined commodities, and optimization suggestions are provided through system evaluation; and for user change feedback, generating hidden collocation based on a user portrait, and realizing real-time iteration of a recommendation scheme.

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 self-service door opening method for a store, and relates to the technical field of self-service stores. The system carries out intelligent matching according to a pre-stored user file, calculates a similarity threshold value through an algorithm, judges that a user is a legal user if the matching degree exceeds a preset value, or rejects access; a dynamic combination strategy is adopted to allow an administrator to configure weights and combination modes of different fields according to the security requirements of the store, so that the user experience is optimized and the privacy leakage risk is reduced on the premise of ensuring the security; a background management system flexibly sets a weight value and combinatorial logic of each verification element, the matching efficiency is dynamically optimized based on a preset similarity threshold, if a comprehensive score after element combination reaches a security level score configured by the system, rapid door opening is automatically authorized, otherwise, secondary verification or manual auditing is triggered, and the safety of the system is ensured. The false rejection rate is reduced, the safety protection capability is improved, the verification modes are automatically switched according to regions or time periods, and the door opening speed is high when the operation is successful.

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.

The invention relates to the technical field of natural language processing, and provides a large language model event extraction method fusing label reconstruction and a multi-dimensional instruction set, which comprises the following steps of: 1, reconstructing a label; 2, constructing a multi-dimensional instruction set; constructing two different multi-dimensional instruction sets for the multi-dimensional instruction library by adopting a multi-dimensional layered alternate combination strategy through a second-level instruction architecture and a third-level instruction architecture; 3, fine adjustment of the model; taking the reconstructed document-level event text and the event record sequences in two different formats as input and output of fine tuning respectively, and performing fine tuning on an LLaMA-3. 2-1B model by using a multi-dimensional instruction set and adopting a LoRA technology; 4, event extraction; and extracting events by using a large model for fusing label reconstruction and multi-dimensional instruction set fine tuning. According to the method, the data labeling cost is reduced through label reconstruction, the semantic analysis capability of a large language model on financial field events is enhanced by utilizing a multi-dimensional hierarchical instruction set, and the financial field event extraction performance is improved.

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 application provides a kind of combination navigation method and system based on dual-antenna heading, comprising: INS and dual-antenna GNSS system are combined together, and utilize the proposed dual-antenna GNSS / INS combination navigation method, so that the system can output continuous stable high-precision carrier attitude information when GNSS signal is disturbed.In the method, a heading combination update strategy is proposed, and the continuous and stable output of the device heading information is completed by parallel execution of the solving process of dual-antenna orientation and the initialization process of IMU.Meanwhile, the sensor information is detected to judge whether the device is stationary, and if it is stationary, the IMU parameter calibration is carried out.According to whether the sensor is initialized, two different precision positioning and orientation results are obtained.The application combines the advantages of dual-antenna GNSS and INS, and applies the dual-antenna combination strategy, so that continuous and effective positioning and orientation results can be obtained without waiting for a long time, and has good practical value.

The invention belongs to the field of distributed version control, and relates to a Git multi-warehouse branch batch statistics and automatic management method and device, computer equipment and a storage medium, and the method comprises the steps: deploying a Shell warehouse branch management script at a terminal, and providing a consistent interaction interface for a user; obtaining a terminal warehouse path and a directory to be scanned under the warehouse path; traversing all Git warehouses under the catalogue to be scanned, acquiring a target branch and a reference branch in each Git warehouse, and comparing whether the target branch is merged with the reference branch or not; displaying the merging state of each warehouse and each target branch, classifying and listing Git warehouses with unmerged branches, and outputting a Git warehouse state statistical result; obtaining Git warehouse branches needing to be merged, configuring a merging strategy, and generating a merging plan; and executing a merging operation, and pushing a merging result to a remote warehouse. And the management efficiency of multi-Git warehouse branch merging can be greatly improved.

The invention discloses a flood storage and detention area optimization scheduling method, and relates to the flood storage scheduling and data processing technology, and the method comprises the steps: determining a plurality of flood storage and detention areas contained in a target drainage basin, and determining the coverage range of each flood storage and detention area; determining a plurality of flood storage heights in the coverage range of each flood storage and detention area, wherein different flood storage heights correspond to corresponding submerging losses; gate parameters of barrage bodies of all the flood storage and detention areas are determined, and gate-inundation loss association rules of the upper-level flood storage and detention areas and the lower-level flood storage and detention areas are established on the basis of the gate parameters; acquiring flood flow data of the target drainage basin; and based on the flood flow data and the gate-inundation loss association rule, generating a gate opening combination strategy to complete optimal scheduling of the flood storage and detention area. According to the method, the operation efficiency of different flood diversion requirements can be improved, the decision-making time when a disaster comes is shortened, the flood diversion effect of the flood storage and detention area is utilized to the maximum extent, and the drainage basin loss is greatly reduced.

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 relates to the technical field of electrical equipment diagnosis, in particular to an electrical equipment diagnosis research method based on an information fusion technology, which comprises the following steps: collecting multi-source sensor data of electrical equipment, and extracting fault features to construct feature vectors after normalization processing; performing primary diagnosis on the feature vector by using an improved BP neural network, and generating a primary fault recognition result as an evidence body; constructing a hybrid combination strategy fusing a Dempster rule and a fuzzy rule, and dynamically allocating two types of rule weights by calculating a conflict coefficient and an inconsistency coefficient between evidence bodies; the method effectively solves the problems that a traditional diagnosis model is slow in convergence and prone to falling into local optimum, the reliability of multi-source information fusion is remarkably improved through intelligent coordination of conflict evidences, and the fault diagnosis accuracy is improved. And a complete diagnosis closed loop from data perception to intelligent decision making is realized.

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.

The invention discloses a dynamic combination method and system for participation of a user group in an auxiliary service market, and a medium, and the method comprises the steps: solving an optimal adjustable power curve of the user group according to the adjustable power level, the duration time and the fluctuation rate with the comprehensive dynamic adjustment characteristic index of the maximum user group as a target, and obtaining the optimal adjustable power curve of the user group; a hierarchical matching user group dynamic combination strategy is adopted, an outer layer is matched with an optimal adjustable power curve of a pair to a user group, and a set pair with the strongest complementarity is screened out through complementary coefficient calculation; in the inner layer matching, a pairwise combination result of user groups in a set is obtained by calculating a complementary coefficient between individuals. Through pairwise dynamic combination of the user groups, the dynamic adjustment characteristics of the combined user groups are effectively improved, so that the user groups can play a greater role while obtaining greater benefits in an auxiliary service market.

The application discloses a cloth warehouse-out scheduling method and system based on multi-strategy cooperation, comprising the following steps: constructing a warehouse three-dimensional coordinate model, and establishing accurate mapping of warehouse position coding and three-dimensional coordinates; adopting an analytic hierarchy process-fuzzy comprehensive evaluation (AHP-FCE) model to quantize order priority, and adapting resource allocation in idle period and busy period through double-mode scheduling; planning the shortest picking path based on the Dijkstra algorithm; and dynamically deciding an order merging strategy based on a reinforcement learning deep Q network (DQN), so as to reduce invalid movement and resource waste. The application can significantly shorten order processing time, improve resource utilization, balance personnel efficiency, effectively relieve operation bottlenecks in cloth warehouse-out, and is suitable for various cloth storage scenes.

This invention discloses a method for designing optimal strains for product synthesis using an enzyme-constrained model combined with an evolutionary algorithm, comprising the following steps: Step S1, solving the enzyme-constrained model with the objective function of maximizing the specific growth rate of the strain to obtain the simulated metabolic flux of the wild-type strain; Step S2, performing dimensionality reduction analysis and labeling on genes directly related to enzyme synthesis within the model by solving a series of flux balance analysis problems with fixed biomass synthesis rates; Step S3, predicting the yield of single-target editing; Step S4, using a genetic algorithm to search for combined targets after dimensionality reduction, adjusting relevant parameters to obtain the optimal gene editing combination strategy; Step S5, repeating the experiment to verify the stability of the algorithm. This invention utilizes genetic algorithms and enzyme-constrained models to discover non-intuitive gene editing combination strategies, providing a new method for constructing efficient microbial cell factories for synthetic chemicals, and demonstrating the potential of applying heuristic algorithms to strain optimization design.

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 invention relates to the field of operation risk prevention and control, and discloses a power system supply and demand balance boundary regulation and control method, system and device and a medium, and the method comprises the steps: obtaining a network topology structure and system network parameters, and generating a power system operation state and a corresponding probability through a Monte Carlo simulation method; judging whether the system state needs to be subjected to load shedding or not, identifying the system state, converting the optimal power flow model into a low-dimensional optimization model, and establishing a supply and demand balance boundary model of the power system in each operation state; calculating the load loss amount of the power system based on the relationship between the actual operation state of the power system and the supply-demand balance boundary, and constructing an operation risk constraint of the power system; and solving the power system operation risk prevention and control model based on supply-demand balance boundary dynamic regulation and control, and generating a power system unit combination strategy to realize risk prevention and control. According to the method, the randomness challenge caused by high-proportion new energy access is effectively solved, and an efficient and feasible technical scheme is provided for safe and economical operation of a power system.

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.