61results about How to "Improve decision accuracy" patented technology

The invention discloses a robot agent reinforcement learning training method in a complex scene, and the method comprises the steps: constructing a reinforcement learning algorithm environment and a state transition model which comprise a barrier, a dynamic target, environment constraints and other complex scenes and are suitable for a multi-agent system; an independent neural network is constructed for each robot agent by adopting a reinforcement learning algorithm; taking vectors of a given state, time sequence information and historical state information of the intelligent agent as input of the state transition model, and outputting a state transition probability and reward information of a prediction state; the action strategy and the action value output by the neural network are executed in the state transition model, and a new prediction state and reward information are generated; through adjusting hyper-parameters of the neural network, a large-scale distributed training architecture is used to carry out cooperative training on the decision network and the evaluation network, parallel learning and cooperative decision of multiple agents are realized, and performance and generalization ability of the agents in a complex scene are improved.

The invention discloses an intelligent ventilation system and method for a cage for livestock breeding, and belongs to the technical field of livestock breeding, and the system comprises an acquisition processing module, an environment modeling module, a change prediction module, an optimization decision module, a strategy simulation module, a storage safety module, an equipment control module, a monitoring feedback module, a user interaction module and an emergency response module. According to the invention, environmental conditions can be managed more actively, environmental parameters are prevented from deviating from an ideal state due to a hysteresis effect, the stress influence on animals is reduced, specific requirements of different livestock animals are met, the air quality and animal health level in an animal house are improved, the decision precision of the system is improved, and the economic benefit is increased. It is ensured that the cage environment is always kept in the most ideal state, improvement of air quality in a short term and minimization of long-term energy consumption are balanced, and the overall efficiency of the system is effectively optimized.

The invention relates to a data analysis method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the steps of obtaining a to-be-processed text, and performing word segmentation processing on the to-be-processed text to obtain a word segmentation data set; performing analysis processing on the word segmentation data in the word segmentation data set through a hybrid expert model, and determining a weight value corresponding to each expert network in the hybrid expert model based on an analysis processing result and a gating network; performing sorting processing and cumulative probability calculation on the weight values corresponding to the expert networks, and determining the number of the expert networks and a target expert network based on a cumulative probability calculation result; and based on a target expert network of the expert network number and a weight value corresponding to the target expert network, determining an analysis decision result of the word segmentation data until analysis decision results of all word segmentation data in the word segmentation data set are obtained, and determining a data analysis result of the to-be-processed text. By adopting the method, the overall decision-making accuracy of the hybrid expert model is improved.

The invention discloses an intelligent decision-making method and device based on service configuration, equipment and a medium, and the method comprises the steps: collecting real-time service data in a service system, employing a machine learning algorithm to carry out the learning recognition of noise data features, dynamically updating a data cleaning rule, and carrying out the recognition of the real-time service data. Cleaning the real-time business data according to the updated data cleaning rule to obtain standardized business data, obtaining a current business configuration model according to the business type, importing the standardized business data into the current business configuration model, and analyzing and processing the standardized business data through the current business configuration model to obtain the real-time business data. According to the method, the decision can be accurately adjusted according to the new situation, and the decision method is dynamically and flexibly configured based on the service data, so that new requirements brought by service changes are adaptively met.

The invention relates to the field of intelligent traffic management, and discloses an efficient road obstacle clearance rescue worker automatic scheduling method comprising the following steps: (a) deploying at least one message queue to cache rescue related data; (b) storing and querying rescue tasks and resource state data through a distributed database system; (c) collecting data from the rescue vehicle and the accident site in real time, and performing preprocessing; (d) asynchronously processing the collected data through a message queue according to the priority and the task type; (e) dynamically scheduling rescue resources by using a multi-factor decision algorithm; the message queue adopts a Kafka cluster and is configured with at least three partitions and at least two copies so as to ensure that a single server or partition fails. Through combination of a multi-factor decision scheduling algorithm and real-time traffic flow prediction, the system can dynamically adjust the scheduling route of the rescue vehicle, and the rescue vehicle is prevented from entering a congested road section, so that the time for arriving at an accident scene is remarkably shortened, and the overall rescue efficiency is improved.

The invention discloses a VVC coding unit rapid division method and system based on image frequency domain characteristics, and the method comprises the steps: employing a first image frequency domain characteristic, a first classification label and a first weight to train a group of binary classifiers based on a lightweight gradient lifting model, and obtaining a group of initial binary classifiers; performing feature selection through the group of initial binary classifiers, and training the group of initial binary classifiers by adopting the selected features to obtain a group of dimensionality reduction binary classifiers; performing parameter adjustment on the group of dimensionality reduction dichotomous classifiers to reduce the scale of the group of dimensionality reduction dichotomous classifiers to obtain a group of division decision classifiers; initializing online sample data, and deciding the division mode in each division decision classifier according to the online sample data to obtain a group of target division modes; and quickly partitioning the VVC coding unit according to a group of target partitioning modes. According to the invention, the division efficiency and decision accuracy of the coding units are improved, the video quality and code rate are ensured, and the coding time is shortened.

The invention discloses an action prediction method based on key points and imitation learning, a terminal and a storage medium, and relates to the technical field of intelligent equipment. According to the method, the thought of a world model is combined with imitation learning, scene representation more conforming to the environment context is learned through the environment dynamics prediction model, and the motion trail of the key point is predicted through the trained trajectory prediction model so as to assist the motion head module in motion decision making. The key point motion trail is used as auxiliary information, and the decision accuracy of the intelligent agent can be effectively improved. Firstly, the key points are not affected by environmental changes such as illumination, even if an object is partially shielded, the rest of key point tracks can still provide the same information, and therefore good robustness is achieved. And secondly, the motion trail of the key point and the motion trail of the intelligent mechanical device in the three-dimensional space have a certain mapping relation, so that the neural network can better fit the moving action of the intelligent mechanical device.

The invention discloses a business relationship mining method and device, computer equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps of: firstly, constructing an initial knowledge graph by intelligently analyzing service data and accurately identifying and extracting entities and mutual relationships in key service clues; and then, integrating the entities and the relationships into a knowledge graph of a professional field to form a target knowledge graph. Thirdly, identifying a highly-correlated entity group by utilizing a correlation analysis technology, quantifying correlation strength by adopting a common neighbor algorithm, and constructing a correlation strength matrix; and finally, constructing a weighted undirected graph based on the matrix, and carrying out deep clustering on the weighted undirected graph by fusing community discovery and a clustering algorithm to form a closely associated service clue cluster. The invention further relates to the technical field of the block chain, and the business data is stored on the block chain node. According to the method, the generalization of business relationship mining is enhanced, the maintenance cost is reduced, non-keyword related clues are accurately captured, and the business relationship is comprehensively revealed.

The invention relates to the technical field of security and protection monitoring, in particular to security and protection equipment and a monitoring method, and the method comprises the steps: obtaining the historical data of a disaster, carrying out the preprocessing of the data, carrying out the simulation, obtaining an ideal escape path, generating a data set training model through training data and the corresponding ideal escape path data, and carrying out the simulation. The verification data are input into the training models to verify the response speed and decision accuracy of the models, the training models with qualified decision accuracy are arranged in the security and protection equipment, a plurality of escape paths are automatically generated based on the data collected by the camera according to the disaster situation, and an instruction is sent to the signal receiver based on the single escape path; and the signal receiver performs feedback based on the received instruction. The security and protection monitoring system provided by the invention autonomously judges and gives a coping scheme to provide timely and accurate escape guidance for personnel evacuation in the case of a disaster, so that serious casualties caused by untimely personnel evacuation are avoided.

The invention provides an enterprise space appeal matching method based on an AI model, and the method comprises the following steps: collecting the multi-source sensing data of an enterprise space, and building a structured time-space information matrix through an access control, energy consumption, position and personnel label sensing terminal; constructing a causal relationship graph by using a structural causal modeling algorithm, and identifying key causal factors directly related to the spatial appeal; causal invariant characterization is extracted based on a multi-environment invariant risk minimization method, and an intervention effect model is established in combination with controllable intervention variables such as partition adjustment, capacity limitation and use strategies; generating an anti-fact sample of the candidate space unit by using an anti-fact reasoning mechanism, and calculating a matching adaptation score; establishing a matching optimization model according to constraint conditions such as topology, capacity, safety and moving lines; according to the method, high-precision and interpretable matching distribution under the conditions of complex space resources and multi-dimensional appeals is realized, and the method has the effects of flexible deployment, high matching stability and sustainable optimization.

The present invention discloses an agent distribution decision and inventory management optimization system based on GA-BP algorithm. By establishing a collaborative, forecasting and replenishment (CPFR) supply chain management model, it fundamentally realizes that suppliers, agents and the like fairly share relevant information on the supply chain. Each node enterprise on the supply chain shares the information on the supply chain on time, quickly and fairly, and negotiates according to its own capabilities and the information obtained to make a reasonable production or sales plan. In the present invention, by optimizing the weights through a genetic algorithm (GA), the obtained weights are applied to the BP network for training, effectively avoiding falling into a local optimal solution, and through a specific analysis of the enterprise's raw material safety inventory, the various factors affecting the safety inventory are understood in detail. Combined with the characteristics of the enterprise's raw material inventory, a good prediction effect is achieved, the unfavorable status of the enterprise's inventory management is improved, and the enterprise's profits are expanded.

The invention discloses a stock transaction decision support system driven by real-time big data analysis and a method thereof, and relates to the technical field of stock transaction, the system comprises a big data acquisition system, a big data transmission system, a big data analysis system, a transaction decision output system and a man-machine interaction system; the stock market information is acquired from different data sources through the big data acquisition system, the stock market information is preprocessed and then sent to the big data analysis system through the big data transmission system to be processed and analyzed, and then the transaction decision output system outputs the corresponding stock transaction decision according to the processing and analysis result. A decision framework based on big data driving is provided, data management is more efficient, data security is guaranteed, decision makers can be effectively helped to collect, analyze and mine stock market information, stock transaction decisions are supported based on big data analysis driving, decision accuracy is higher, and stock transaction efficiency is improved. And unnecessary loss of investors can be reduced to the greatest extent.

The invention relates to the technical field of artificial intelligence, in particular to a decision-making method and device based on multiple modes, equipment and a medium. The method comprises the steps of obtaining historical text data and historical image data used for decision making, obtaining a preset decision making model and a preset decision, determining a loss value of the preset decision making model generated by the preset decision making model according to the historical text data and the historical image data, obtaining feedback data corresponding to the preset decision making, and optimizing the preset decision making model according to the feedback data. And obtaining an optimized decision model, and generating a target decision according to the optimized decision model, the historical text data and the historical image data. According to the method and the device, the decision model is optimized according to the text data and the image data, so that the decision model can learn data in different modalities, the decision generated by the decision model can help investors to provide comprehensive market analysis, and the decision accuracy is improved.

The invention relates to the technical field of network security, in particular to a WEB application mimicry judgment system and method.The WEB application mimicry judgment method based on runtime data is characterized in that a parameter collection module is used for implanting instrumentation codes in key functions of WEB application programs and collecting runtime parameters of the WEB application programs in real time; the event conversion module is used for converting the acquired runtime parameters of the plurality of heterogeneous execution bodies into structured runtime events; the sequence alignment module is used for carrying out time sequence alignment on runtime events from different heterogeneous executors according to the business operation labels; and the risk judgment module is used for vectorizing the aligned text data of the runtime events, calculating a plurality of judgment factors and performing weighted fusion to generate a final confidence score, and performing feedback control to judge whether to trigger an execution body rotation operation or not according to the confidence score. According to the method, the WEB application runtime parameters are collected, through event conversion, sequence alignment and multi-factor weighted decision, the WEB application abnormal attack recognition accuracy is greatly improved, and the defect of traditional flow decision is effectively overcome.

The invention provides an event assessment and auxiliary decision-making method and system for an offshore operation platform, and the method comprises the steps: taking platform data in different complex operation scenes as input, employing different key data preprocessing methods, and obtaining event assessment key factors and relationships of the whole offshore platform; an evaluation analysis model is established to analyze and evaluate the whole platform event space, auxiliary decision support data is improved, visual display is carried out with different granularities or different abstract levels, and the function of accurately evaluating and deciding platform events is achieved. According to the method, the decision-making speed is increased by analyzing, concluding and integrating complicated data accumulated by subsystem equipment of the maritime satellite launching and recycling platform and carrying out classified analysis on evaluation data by utilizing an intelligent optimization genetic algorithm; an event evaluation model constructed based on a fusion algorithm is applied to assist the platform and personnel in performing remote diagnosis and decision-making on the operation of the offshore operation platform, so that risk threats caused by emergencies are reduced, and the operation efficiency of the platform is improved.

The present invention discloses a tunnel shield machine intelligent selection auxiliary model establishment and decision-making method, which belongs to the field of shield technology. The method uses an interval shield selection information map that integrates engineering geological longitudinal profiles, stratum physical information, shield selection information, and has undergone data expansion operations as a sample, and uses a real interval operation risk assessment map as a label to train a generative adversarial network, i.e., an image style transfer GAN model. Through a large amount of training, the model inputs the interval shield selection information map and can intelligently output the interval operation risk assessment map; through machine learning, the image style transfer GAN model internalizes the operation risk assessment information under various geological conditions and shield selection information, and displays the risk levels of different partitions within the range of the excavation face height through the interval operation risk assessment map, which is used as a reference for tunnel shield machine selection, thereby playing a role in assisting decision-making. The present invention solves the problem that the current selection is time-consuming and labor-intensive, and has low reliability by using expert experience.

The application discloses a kind of gas turbine combined cycle unit gas net heat value monitoring method, it is related to data monitoring field, comprising the following steps: step 1: by the acquisition equipment of deployment in preset monitoring point, obtains the direct heat value measurement data of gas;Synchronous acquisition index measurement data of gas turbine combustion chamber, and based on index measurement data back calculation, obtain the indirect heat value data of corresponding point;Step 2: identify the initial deviation fluctuation trend between indirect heat value data and direct heat value data of each point, and based on the safety threshold of current working mode, whether initial deviation fluctuation trend exceeds allowable range is judged;By fusing direct heat value measurement and indirect heat value data of multiple index back calculation, in combination with deep learning simulation preplay and verification acquisition instruction optimization, dynamically select optimal acquisition orientation and range, reduce the error introduced due to sensor position fixed or environmental interference, ensure that output data truly reflects gas heat value state.

The invention discloses a progressive upgrading method for an SDN (Software Defined Network). An SDN controller periodically acquires a historical traffic matrix of a hybrid network in a hybrid acquisition mode through management connection; performing order polynomial fitting on the traffic time sequence by using a least square method to generate a predicted traffic matrix; taking a prediction traffic matrix and a network topology as input, constructing an optimization model taking maximization of an SDN controllable traffic scale as a core target and taking upgrading budget and a link utilization rate as constraints, and efficiently solving and outputting an upgrading priority list of traditional nodes through a dual rising method of dual weights; and executing upgrading and evaluating the effect according to the list to form closed-loop optimization. According to the method, the characteristic of partial path controllability of the SDN is fully utilized, the contribution of node upgrading to the controllable flow is quantified, the SDN control range is rapidly expanded on the premise of ensuring the network performance, and the core value of progressive migration is improved.

The application provides a UAV group sensing method for city monitoring data time delay guarantee, comprising the following steps: a main process of a scheduling center creates a data time delay and a trajectory pool; the scheduling center starts a plurality of sub-processes for simulating city monitoring data generation and time delay updating; a UAV group collects simulated trajectories according to a current sensing strategy; the simulated trajectories are evaluated; the scheduling center optimizes the sensing strategy by using an asynchronous backtracking algorithm according to data of the data time delay and the trajectory pool, and synchronizes the optimized sensing strategy to each sub-process; until the UAV group sensing strategy no longer changes, the optimal sensing strategy is obtained; the UAV group is dynamically scheduled to guarantee city monitoring data time delay; the method is suitable for participating in a task sensitive to a time threshold, and therefore, compared with a traditional guarantee task, the UAV can be used to participate in city monitoring data time delay guarantee tasks to improve the completion satisfaction of the task.

The invention discloses an automatic decision-making method and system based on rule driving, and relates to the technical field of business decision-making automation, and the method comprises the steps: recording original decision-making data in a real decision-making scene, and extracting causal elements in the original decision-making data; creating a rule base based on causal elements of the original decision data; adding a factor element matching algorithm for the rule base, querying a rule graph corresponding to the input data based on the factor element matching algorithm, and outputting the rule graph; manual intervention is introduced to optimize the rule base, and it is ensured that each rule graph is put into a production environment after the accuracy of each rule graph reaches the standard; and the production environment outputs a final decision based on the optimized rule base. The decision-making efficiency can be improved, manual intervention is reduced through automatic processing, and the decision-making speed is increased; the interpretability and transparency are improved, the decision-making process is traceable, and audit and interpretation are facilitated; and the method is flexible and configurable, supports dynamic regulation and expansion of rules, and adapts to business change requirements.

The present invention discloses a reinforcement learning training method for a robot agent in complex scenarios. The method comprises constructing a reinforcement learning algorithm environment and a state transition model for a multi-agent system including complex scenarios such as obstacles, dynamic targets and environmental constraints, and adopting a reinforcement learning algorithm to construct an independent neural network for each robot agent. The vectors of the given state, timing information and historical state information of the agent are used as inputs of the state transition model, and the state transition probability and reward information of the predicted state are output. The action strategy and action value output by the neural network are executed in the state transition model to generate new predicted state and reward information. The hyperparameters of the neural network are adjusted, and the decision network and the evaluation network are collaboratively trained using a large-scale distributed training architecture to realize parallel learning and collaborative decision-making of multiple agents, thereby improving the performance and generalization ability of the agents in complex scenarios.

The invention relates to a personal risk management and control decision-making method. Target text data sent by a terminal and obtained through recognition and conversion based on user voice are obtained; performing decision suggestion processing, intention recognition processing and strategy deduction processing based on the target text data by adopting a personal risk management and control decision model to obtain a risk management and control decision result; and finally, the risk management and control decision result is sent to the terminal. According to the method, the user voice is acquired in real time and converted into the text, so that the request or the problem of the user can be quickly responded, and the information processing time is remarkably shortened. A personal risk management and control decision model is adopted, potential risks and user intentions can be identified through deep analysis of target text data, so that accurate decision suggestions are generated, and errors of subjective judgment are reduced.

The invention provides a hybrid power ship energy management method, device and equipment and a medium, and relates to the technical field of ship energy management, and the method comprises the steps: building a power system model of a series hybrid power ship; taking the ship speed data at the moment t and within a set range before the moment t as input data of a dual-channel CNN-LSTM prediction model, and outputting ship speed prediction data within the set range after the moment t through the dual-channel CNN-LSTM prediction model; determining a state variable and a control variable in a ship energy management process, and constructing a dynamic planning mathematical model according to the state variable, the control variable and the power system model; and substituting ship speed prediction data within a set range after the t moment into the dynamic planning mathematical model, and calculating to obtain an optimal ship energy management strategy. According to the method, the energy prediction model is more accurate, the energy use efficiency of the hybrid power ship is improved, and the energy management strategy is improved.

The present application relates to a method, device and robot for determining road width. The method comprises: first, obtaining a cost map, and then determining the front area and rear area of ​​the robot on the cost map according to the current position of the robot and the path pre-planned for the robot; then, extracting the road section in the front area and the road section in the rear area according to the path, the front area and the rear area pre-planned for the robot; finally, obtaining the width of the road section in the front area and the width of the road section in the rear area according to the cost map and the offline map. The width is closer to the true value, which improves the decision accuracy.

The invention provides an ultrasonic navigation action determination method and device, and relates to the technical field of ultrasonic navigation, and the method comprises the steps: training a discrete self-coding pre-training model by using a historical ultrasonic image, extracting discrete features from an input ultrasonic image, and carrying out image reconstruction based on the discrete features; a hidden space generation model is constructed and used for generating an ultrasonic image of a corresponding position after the ultrasonic navigation action is executed at the current position according to the input ultrasonic image of the current position and the ultrasonic navigation action; constructing an ultrasonic image quality evaluation model for outputting a quality evaluation result of the ultrasonic image according to the input ultrasonic image; training by using the hidden space generation model and the ultrasonic image quality evaluation model to obtain an ultrasonic navigation model; and inputting a current ultrasonic image obtained by ultrasonic treatment at the current moment into the ultrasonic navigation model to obtain a target ultrasonic navigation action.

This invention provides a multi-agent traceable decision-making system and method for environmental data mining, belonging to the interdisciplinary field of artificial intelligence and environmental informatics. The system includes: a task complexity assessment module, a multi-agent conditional collaboration module, an arbitrator triggering and adjudication module, an environmental knowledge graph module, a feedback learning optimization module, and a decision tracing and visualization module. The method includes the following steps: Step 1: Task complexity assessment and collaboration mode allocation; Step 2: Parallel processing and dynamic collaboration of two intelligent analysts; Step 3: Consistency check and arbitration triggering; Step 4: Arbitrator adjudication and weight integration; Step 5: Human expert correction and weight optimization feedback; Step 6: Decision tracing and weight visualization. This invention enables more accurate decision optimization, significantly improves the professionalism and accuracy of decision-making, and possesses the ability to proactively identify and prevent potential problems.

The application uploads the data obtained by the vehicle-mounted camera sensor of the unmanned mine car to the cloud server, carries out feature recognition on the cloud data through the cloud server, then selects the road curvature, the slope, whether there is an obstacle in the front road, whether there is a fork in the front road as the characteristic parameters according to the characteristics of the mine area to realize the data screening of the vehicle-mounted camera, so that the data input into the decision model of the unmanned mine car is reduced, and at the same time, the speed information of the unmanned mine car is referred to in the vehicle-mounted camera data screening process, and the screening weight is dynamically determined, so that the decision time of the decision model is reduced, and according to the characteristics of the mine area, the sensor data screening and the dynamic determination of the screening weight are considered, most of the interference garbage data is considered, the decision time is reduced, and the decision accuracy of the decision layer is improved.

The invention discloses an AI-driven dynamic decision and process linkage method and system, and the method comprises the following steps: S1, receiving business data of an enterprise system through an API interface, and carrying out the preprocessing of the business data; s2, inputting the preprocessed service data into a double-algorithm model, wherein the double-algorithm model outputs a path switching value according to the switching times, the switching time consumption and the switching deviation value of a historical process; s3, if the path switching value reaches a first preset value, judging a risk level according to the grading decision value to generate a flow adjustment decision instruction; and S4, generating a new flow through a flow adjustment decision instruction, calculating a flow matching degree, and feeding back the flow matching degree to the double-algorithm model, so that the double-algorithm model adjusts a path switching value according to the flow matching degree. According to the method, a double-algorithm model of the GBDT model and the RL model is adopted, and the advantages of the GBDT model in the aspects of feature mining and regression prediction and the characteristics of the RL model in the aspects of dynamic decision optimization and feedback learning are exerted.

This application discloses a system and method for matching new Chinese-style clothing based on virtual try-on and style matching, belonging to the field of e-commerce. This application obtains basic user data, behavioral data, feedback data, and information on currently purchased items from an e-commerce platform. Based on the multi-dimensional characteristics of the new Chinese-style clothing items purchased by the user, it maps the clothing wearing scenario space, evaluates the user's corresponding wearing scenario, and selects a set of new Chinese-style clothing candidates that meet the user's scenario needs from the e-commerce platform's product library. It generates matching effects between purchased items and candidate items based on virtual try-on, evaluates the matching of purchased items and candidate items, assesses the user's focus on clothing attributes based on historical user behavior and evaluation feedback data, and comprehensively scores candidate items based on the user's wearing scenario, matching evaluation, and focus evaluation, and makes recommendations based on the comprehensive score, thereby improving matching coordination and the accuracy of user decision-making.

This article provides a method, device, equipment and storage medium for determining the dirty state of the roadbed, wherein the method includes: obtaining a radar echo signal corresponding to the roadbed of the target mileage; determining a characterization value sequence of the roadbed dirty state according to the radar echo signal; determining a selected characterization value corresponding to the target mileage according to the characterization value sequence, wherein the selected characterization value is used to represent the characterization value sequence of the target mileage; associating the actual dirtiness rate corresponding to the position point of the selected characterization value in the target mileage with the selected characterization value to obtain a roadbed dirty model; determining the roadbed dirty state corresponding to the roadbed of the target mileage according to the characterization value sequence of the roadbed dirty state and the roadbed dirty model. This article can determine the roadbed dirty state with high accuracy and improve the decision-making accuracy during roadbed maintenance.