502 results about "Data vector" patented technology

The invention provides a medical auxiliary diagnosis method and system based on time sequence and semantic weighting, and belongs to the technical field of medical information processing. Constructing the preprocessed current and historical medical record information of the patient into patient medical record data, and inputting the patient medical record data into a pre-trained medical language model to generate a preliminary diagnosis result; the comprehensive weight of the patient medical record data vector sequence is calculated through a time decay function and the content correlation weight, a medical record fusion vector is obtained through fusion, and the medical record fusion vector and the preliminary diagnosis result are spliced into a query vector; searching candidate fragments in a clinical guide knowledge base based on the query vector, calculating semantic evidence scores and coverage scores based on patient medical record data and tokens of the candidate fragments, obtaining sorting probabilities of the candidate fragments in combination with metadata prior scores of the candidate fragments, and screening out target guide fragments; and the preliminary diagnosis result and the target guide fragment are fused to generate final auxiliary diagnosis and treatment information, so that auxiliary diagnosis and treatment suggestions with traceability, verifiability and authoritative basis are provided.

The invention discloses a special large language model for a solid waste cementing material and an innovative hypothesis generation method thereof, which realize accurate understanding and reasoning of information such as material composition, process parameters, mechanical properties and the like by constructing a special knowledge graph in the field of materials, combining experimental data vector retrieval and fusing a mixed retrieval enhanced generation technology, so that the innovative hypothesis generation of the solid waste cementing material is realized. And a scientific reasoning module is used for deducing a material design rule to generate a potential innovation hypothesis. The method effectively breaks through the problems of long development period and low innovation efficiency of traditional materials, and is suitable for the fields of low-carbon building material development and solid waste high-valued utilization.

The invention relates to a construction method and system of an intelligent question-answering system based on a planning knowledge graph library. The method comprises the following steps: acquiring multi-source space planning data in real time; carrying out fusion on the planning data; vectorizing the fused planning data; a plurality of preset retrieval enhancement models are modularized, and interaction of vectorized planning data among the modules is realized through a message queue; through a natural language processing method and a message queue method, constructing a plurality of knowledge maps which can be accessed by the preset retrieval enhancement model; training a pre-training model through the knowledge graph; in response to a user question, retrieving an enhancement model through dynamic routing selection, and retrieving entities and relationships in the knowledge graph; and based on the retrieval result, generating a questioning result through a pre-training model. According to the method, intelligent question answering, scheme generation and evaluation of space planning are realized through the knowledge graph and the pre-training model constructed by the multi-source data, the working efficiency of the space planning is improved, and the manual workload and the error rate are reduced.

The invention relates to a structured text analysis method and system based on title recognition and hierarchical abstracts, and the method comprises the steps: receiving a text, and obtaining sample data of a training language model; the vector database can perform vector dimension adjustment on the data of the stored text. The vector database takes vectors as basic storage units, and converts unstructured data into high-dimensional vectors through an embedding technology. And calling the data storage type information of the title text data set to determine the data storage type of the title text data set matched with the text storage mode. And obtaining a structure type of sample data of each text by utilizing a semantic segmentation script. And segmenting the document according to the identified chapter titles, and subdividing each chapter into a plurality of logic paragraphs to form structured text block units. The multi-level abstract generation model matched with the target model can fully mine level information and semantic association in the document. Therefore, the effect of analyzing the structured text by the intelligent question-answering system is improved.

The invention discloses a music stave sentiment classification method and system based on multi-level distillation, and belongs to the technical field of music information processing. The method comprises the steps of firstly collecting music stave data and converting the data into stave data vectors, then performing feature extraction by using a long short-term memory network, then constructing a teacher network and a student network for knowledge distillation, and realizing multi-level knowledge transmission through temperature scaling, KL divergence loss and mask feature distillation. And finally, training a lightweight classification model to complete sentiment classification. The knowledge distillation technology is creatively applied to staff sentiment classification, the classification accuracy is effectively improved through an online multi-level distillation mode, and the technical problems that a traditional method lacks semantic information and a self-supervised model is not suitable for sentiment tasks are solved. The method has the main advantages of high classification precision, light model weight, capability of effectively capturing music emotion features and the like.

The invention discloses a digital twin intelligent test run system based on a marine medium-speed diesel engine and a monitoring method, relates to the technical field of monitoring, and is used for solving the problems of degeneration identification lag and low emission early warning precision of an oil injector. The system comprises a multi-source heterogeneous data preprocessing module and a running state intelligent diagnosis module. The multi-dimensional data acquisition module is used for acquiring in-cylinder pressure, crankshaft torsional vibration and transient air-fuel ratio signals to construct multi-dimensional data vectors, and the multi-dimensional data acquisition module is used for analyzing combustion fluctuation characteristics based on an attention mechanism LSTM (Long Short Term Memory) network, identifying early deterioration of an oil injector and outputting an oil injection consistency coefficient and a fault type identifier. The method comprises the steps that according to an oil injection consistency coefficient, electromagnetic valve response delay and combustion parameter correlation are tracked, a degradation trend is predicted, and emission early warning is generated; and according to prediction and early warning results, fuel injection compensation parameters are dynamically adjusted in the digital twinborn model, a recursive least square method is adopted to identify combustion parameters and feed the combustion parameters back to an electric control unit, and self-adaptive optimization and closed-loop control in the test run stage are achieved.

The invention discloses an intelligent Internet of Things home energy management system, and relates to the technical field of intelligent homes, the system comprises a multi-dimensional data sensing module which uses a sensor and an intelligent equipment interface to collect human physiological data, obtains environment and external data through an environment sensor and a weather API, and sends the environment and external data to a cloud server; historical physiological records, environmental regulation logs and user feedback data are called, timestamps and space coordinate information are added to all kinds of data at the same time, and the data are converted into data vectors and uploaded to a system database; through the multi-dimensional data sensing module, human body physiological data, environmental data and external data can be comprehensively collected, and a physiological state prediction model and a health risk identification model exclusive to a user are constructed in combination with historical records and user feedback, so that the personalized level of health management is improved, and the health risk of the user is improved. And through a chaos embedding biological rhythm prediction formula and a dynamic Bayesian health risk assessment formula, accurate prediction of the physiological state of the user and identification of the health risk are realized.

The invention belongs to the technical field of intelligent early warning, and particularly relates to an intelligent early warning method for a DMF waste liquid purification and recovery control platform, and the method comprises the steps: carrying out the principal component analysis of long-period historical data, and constructing a principal component transformation matrix of a static reference model; for a moment to be diagnosed, calculating a reconstruction value by using the static reference model to obtain a residual vector, carrying out eigenvalue decomposition on a covariance matrix of a residual matrix of a sliding time window, calculating a drift coherence index according to the distribution of drift eigenvalues, modulating a drift principal component vector of the sliding time window by combining the residual vector, and carrying out diagnosis on the moment to be diagnosed; and obtaining a drift compensation vector, superposing the drift compensation vector with a reconstruction value of a real-time data vector at a to-be-diagnosed moment to obtain an adaptive reconstruction value at the to-be-diagnosed moment, calculating a reconstruction error, comparing the reconstruction error with a fault alarm threshold, judging whether a fault exists at the to-be-diagnosed moment, and performing early warning. According to the invention, the early warning accuracy and robustness are improved.

The embodiment of the invention provides a data vectorization acceleration method and system, and the method comprises the steps: writing an optimizer rule into an optimizer, and carrying out the vectorization acceleration according to the optimizer rule, respectively fusing respective original operators in the first operator rule and the second operator rule with respective matched replacement operators of the first operator rule and the second operator rule to obtain a vectorized fusion operator; and converting the vectorization fusion operator into a vectorization execution operator according to the conversion logic, so that the vectorization execution operator calls a vectorization execution engine through a local interface JNI bridge to accelerate data vectorization. Therefore, through the embodiment of the invention, the problem that the calculation efficiency is relatively low in a large-scale data processing scene due to the fact that an existing Flink framework cannot fully utilize an instruction set and vectorization hardware resources in a batch processing mode can be solved.

The invention discloses an index data prediction method based on a time sequence large model and a related device, and belongs to the technical field of time sequence prediction. The method comprises the following steps: acquiring time sequence data and preprocessing the time sequence data to obtain a time sequence data D-dimensional vector; aligning the D-dimensional vector of the time sequence data through a cross attention mechanism to obtain an aligned time sequence data vector; generating a cue word with related information according to the D-dimensional vector of the time sequence data; converting the cue word into a D-dimensional vector, splicing the D-dimensional vector with the aligned time sequence data vector, and inputting the spliced vector into a pre-established large language model; and outputting predicted text data through the large language model, and converting and combining the text data to obtain a complete time sequence prediction data sequence. Under the practical constraints of insufficient computing power and insufficient fine adjustment data, efficient and accurate time sequence prediction is realized only through transformation of a large model, and the applicability of a time sequence prediction technology in different scenes is expanded.

The invention discloses a bank intelligent operation knowledge base implementation method and system based on a large language model and RAG, and the method comprises the steps: firstly extracting business terms from structured, semi-structured and unstructured multi-source heterogeneous multi-modal data of a bank, and then constructing a business dictionary containing term similarity data and an ontology tree; quantitatively calculating the release mechanism weight, the timeliness coefficient and the punishment association degree of the supervision clauses to obtain authority values, and sorting the authority values; then using a double-constraint loss function to finely adjust the pre-trained large language model as a vector representation model, vectorizing service data, and storing the vectorized service data into a vector database; and finally, aiming at user query, retrieving in a vector database, rearranging in combination with a supervision clause sorting result, and generating a business response after compliance verification. According to the method, the bank unstructured knowledge processing capability and retrieval precision can be improved, and efficient intelligent operation is realized.

The invention provides an SQL (Structured Query Language) statement generation system, a multi-source knowledge base cluster module comprises a plurality of vector libraries, each vector library is used for storing a data vector of a data type corresponding to the vector library, a query understanding module can analyze and process a natural language query problem input by a user after receiving the natural language query problem, and the query understanding module can obtain a query result. According to the technical scheme, the vector engine module can perform collaborative retrieval on at least one vector library according to the processed query problem to obtain the target data information, multiple data types are respectively stored in different vector libraries, so that only the vector libraries related to the processed query problem need to be retrieved, and the retrieval efficiency is improved. According to the method, the retrieval accuracy is improved, multiple vector libraries correspond to multiple data vectors of different data types, and the accuracy of the generated SQL statements can be improved.

The invention is suitable for the technical field of intelligent wearable equipment, and provides a multi-modal physiological data fusion analysis method and device and intelligent wearable equipment. According to the multi-modal physiological data fusion analysis method, the various asynchronous original body data collected by the intelligent wearable device are subjected to preprocessing corresponding to the data types, and basic optimization of various data quality is ensured. The method comprises the steps of preprocessing various data, then performing space-time alignment fusion on the preprocessed various data, generating a space-time synchronous multi-dimensional fusion data vector, effectively overcoming the problem of time sequence misalignment caused by sampling frequency difference of different sensors, and reducing time sequence correlation errors caused by a traditional simple interpolation method. On the basis, a plurality of data features are extracted and fused to obtain a fused feature vector, so that the integrity and consistency of features dependent on subsequent analysis are guaranteed. And data features of different dimensions in the fusion feature vector are further independently derived, so that the state of each physiological index can be analyzed in a targeted manner.

The invention provides a database relationship intelligent discovery and ER graph construction method and a self-evolution learning method, and belongs to the technical field of ER graph construction, and the database relationship intelligent discovery and ER graph construction method comprises the steps of S1, metadata collection and preprocessing; s2, vectorizing domain sensing metadata; s3, recalling a multi-strategy candidate relationship; s4, performing LLM structured reasoning and verification; s5, carrying out data sampling inspection and cross validation; s6, hierarchical negative sample self-evolution learning; and S7, generating and visualizing an ER graph. According to the invention, a layered negative sample self-evolution learning mechanism and the structured reasoning ability of LLM are organically combined, and LoRA rapid fine tuning and a multi-stage verification process are supplemented. The defects of accuracy, generalization ability, cold start efficiency, reasoning transparency, automation integration and the like in the aspect of discovery of the primary and foreign key relationship of the relational database in the prior art are effectively overcome, and a more advanced and practical solution is provided.

A method for calculating domain relevance scores for responses generated by LLMs is disclosed. The method includes receiving a response generated by LLM corresponding to user query. The user query is associated with a domain. The method further includes splitting the response into a plurality of response chunks using a splitting technique. The method further includes generating a plurality of response vector embeddings based on the plurality of response chunks using at least one sentence transformer. The method further includes computing a plurality of cosine distances between the plurality of response vector embeddings and a corresponding plurality of training data vector embeddings, wherein the plurality of training data vector embeddings corresponds to domain-specific training data of the LLM. The method further includes calculating a domain relevance score corresponding to the response, based on a sum of the plurality of cosine distances and a number of the plurality of chunks.

The invention discloses a data processing device, a data processing method, a chip and electronic equipment, and relates to the technical field of data processing. The data processing apparatus includes: a storage module configured to store data; the calculation module is coupled to the storage module and is configured to access an original data vector stored in the storage module, and the original data vector comprises m data elements; splitting the original data vector based on the target k value to obtain a plurality of first data vectors; performing odd-even merging sorting on the data elements in the first data vectors to obtain a plurality of first ordered data vectors; and merging and sorting the data elements in the plurality of first ordered data vectors to obtain top-k data elements in the m data elements. According to the data processing device, the data processing method, the chip and the electronic equipment provided by the invention, the top-k data elements can be obtained without a special hardware sorting network.

The embodiment disclosed by the application belongs to the technical field of computing, and particularly relates to a method and device for performing stencil calculation and a computing device. The method comprises the following steps: acquiring a plurality of node data included in a calculation region, and grouping the plurality of node data into a data vector. According to at least one calculation condition included in the stencil calculation, the plurality of node data are subjected to judgment processing, and a judgment result corresponding to each calculation condition is obtained for each node data. For each calculation condition, the judgment result corresponding to the calculation condition is grouped into a judgment result vector, and the judgment result vector is obtained. Based on the data vector and the judgment result vector corresponding to each calculation condition, a calculation operation corresponding to each calculation condition in the stencil calculation is performed, and a calculation result of the stencil calculation is obtained. The application can convert the stencil calculation into vector calculation, and can improve the efficiency of the stencil calculation.

The invention relates to the technical field of education big data analysis, in particular to a student information intelligent management system based on big data, comprising a data convergence module used for converging multi-source heterogeneous behavior data streams of target students in campus Internet of Things and interaction logs of the target students in a digital learning platform in real time to obtain an original time sequence data vector set; and the feature generation module is used for carrying out cross-modal time domain alignment and wavelet coherent transformation on the original time sequence data vector set so as to generate a multi-scale spatial-temporal feature map representing the individual learning and living states of the students. According to the invention, global network time protocol synchronization timestamp synchronization and Kalman filtering interpolation are introduced through the data aggregation module, and the feature generation module adopts remoley wavelet transform and wavelet coherence spectrum analysis, so that the phase locking relationship of online interaction and offline behaviors of students on a time-frequency domain can be quantified; therefore, the learning and living states of the students can be more comprehensively described.

The invention provides a multi-mode cross data fusion scheduling method for a flywheel and battery hybrid energy storage system, and the method comprises the steps: collecting multi-mode signals in parallel, and fusing and packaging the multi-mode signals into a multi-mode data vector with a unified time sequence. Preprocessing and normalizing the multi-mode data vector to obtain a preprocessing sequence, and extracting a multi-dimensional grid-connected feature from the preprocessing sequence; and constructing a feature association graph based on the multi-dimensional grid-connected features, and fusing node risks of the feature association graph through a linear diffusion model to determine a grid-connected readiness score. And mapping the grid connection preparation degree score to a preset interval to determine a slope starting time window. And performing slope parameterization processing based on the slope starting time window to construct a power step sequence, and generating a plurality of control packets for issuing instructions according to power steps according to the power step sequence. According to the method, by generating a progressive grid-connected power slope, stage injection of short-circuit current is achieved, and one-time large current impact is avoided.

The invention relates to the field of Internet of Things data processing, in particular to a hydropower station IoT real-time data acquisition and uplink method based on edge computing, the core technology comprises data acquisition and preliminary screening processing of received data, and the screening processing method comprises a threshold detection algorithm and a moving average anomaly detection algorithm. Cleaning the data through a quartile distance judgment algorithm; judging and calculating the state of the equipment, and calculating the credibility of the data vector to obtain the running state of the equipment; data packaging and uploading: generating a data uploading block chain collected by a digital fingerprint combination through dual hash; the oracle machine establishes a contract rule to monitor the uploaded data in real time, and pushes abnormal information to the early warning platform when the data is abnormal, so that real-time early warning is realized; through the three-level architecture of edge calculation, the block chain and the oracle machine, on the premise of ensuring data credibility, the decision-making precision and the response speed are improved, and the three pain points of uncredibility, high delay and many false alarms in the industrial monitoring field are solved.

The invention discloses a data government method and system based on a large government affair model, and belongs to the technical field of deep learning, and the method comprises the steps: government affair data collection: obtaining basic data from a data source for implementing data government, the basic data comprising historical data and current data; government affair data treatment: preprocessing the collected basic data, and integrating the data to obtain a data sequence; a platform adaptively selects a preset deep learning framework according to the government affair scene features, and a staged parameter optimization strategy is combined to train and generate a government affair decision model with a domain cognitive ability; and intelligent analysis: deploying a semantic understanding engine, converting the unstructured government affair text into data vector representation which can be analyzed by a machine, and outputting a data analysis result through a multi-modal feature fusion technology. According to the invention, the data management difficulty in a government affair scene can be reduced, and deep transformation of intelligent government affairs and establishment of an intelligent decision support system are promoted.

The invention relates to the technical field of motor rotor production monitoring, in particular to an intelligent monitoring method and system for motor rotor production based on the Internet of Things. The method comprises the following steps: collecting equipment operation data of motor rotor production equipment; obtaining a reference model and a health data subset based on the health multi-dimensional sequence data set; obtaining a multi-dimensional mean vector based on the average condition of the multi-dimensional data vectors in the sliding time window; acquiring a working condition fluctuation index; determining the corresponding sliding time window as a working condition stable window based on the working condition fluctuation index; constructing a temporary model of each working condition stable window, obtaining average path similarity, and calculating a drift quantitative index of data in each working condition stable window; and when the drift quantitative index is greater than a preset drift threshold value, an updated isolated forest model is constructed, abnormity monitoring is performed on the multi-dimensional data vector, an abnormity score is calculated, and early warning information is sent, so that the intelligent level and the operation and maintenance efficiency of motor rotor production monitoring are improved.

The invention provides a vehicle-mounted process optimization method and device, electronic equipment and a computer storage medium, relates to the technical field of automobiles, and is applied to a vehicle control system. The method comprises the following steps: determining a vehicle panoramic data vector based on multi-dimensional vehicle running state panoramic data; inputting the vehicle panoramic data vector into an intelligent scene recognition model, outputting a vehicle operation activation scene and querying a corresponding target vehicle-mounted resource allocation strategy to schedule vehicle operation resources and process operation modes; by fusing multi-dimensional vehicle running state panoramic data and utilizing an intelligent model to actively identify a driving scene, resource scheduling is converted from static and universal to dynamic and accurate; according to the method, the system resources are finely distributed according to the real-time scene, the real-time performance and the fluency of the key tasks are effectively guaranteed, the processing capacity of the system in the multi-task concurrent scene is optimized, and therefore the response speed, the resource utilization efficiency and the overall user experience of the vehicle-mounted system are remarkably improved.

The invention relates to the technical field of data processing, in particular to a collaborative recommendation method, device and system based on shared healthy diet and a medium, and the method comprises the steps: obtaining health data and diet demand data of each user in a family; generating a family health data vector based on the health data of each user, generating a family diet demand vector based on the diet demand data of each user, and constructing a knowledge graph based on the family health data vector and the family diet demand vector; determining a conflict function value between the members based on the difference of each diet demand component in the diet demand vector and the corresponding weight; forming a member pair set by the member pairs of which the conflict function values are positive numbers, constructing an objective function based on the diet demand vector of each member and the member pair set, and solving the objective function to obtain a unified diet scheme of the family; according to the invention, a family shared diet scheme which can meet individual demands and balance conflicts can be obtained.

The invention discloses a Chinese vocabulary entry translation method and device, electronic equipment and a storage medium. The method comprises the following steps: performing vector coding on a Chinese vocabulary entry to be translated to obtain a historical data query vector and a terminology query vector; determining a historical data query index according to the historical data query vector and the historical translation data vector index, and determining a terminology query index according to the terminology query vector and the terminology vector index; determining a historical translation data example according to the historical data query index and the historical translation data bank, and determining a professional term example according to the professional term query index and the professional term bank; splicing the historical translation data examples, the terminology examples and the to-be-translated Chinese vocabulary entries to a prompt template to obtain a target prompt project; and inputting the target prompt project into the target large model for translation, and outputting an entry translation result. The method can improve the accuracy and speciality of entry translation, and can be widely applied to the technical field of computers.

The invention discloses a color consistency control method in a bamboo kitchen ware carbonization process, and belongs to the technical field of bamboo processing. The method comprises the following steps: acquiring a multi-dimensional spectral data vector of the surface of the bamboo kitchen ware; determining a multi-dimensional change rate vector of the spectral data; respectively generating a first candidate control signal and a second candidate control signal based on the spectral data and the change rate vector; determining a system dynamic state index based on the change rate vector; generating a dynamic weighting coefficient according to the dynamic state index of the system; and performing weighted fusion on the first candidate control signal and the second candidate control signal by using a dynamic weighting coefficient to generate a final heat regulation control signal. According to the invention, the method achieves the self-adaptive optimal control of all working conditions through the construction of a dual-mode decision fusion composite control system, guarantees the quick response in a dynamic stage, guarantees the smooth convergence in a steady-state stage, and fundamentally improves the thermal inertia problem of the system.

A collaborative management authority rule optimization system based on community discovery comprises a feature extraction module, an organization role screening module and a data attribute screening and authority distribution control module. Generating a standardized user feature vector and a data vector through structured coding and semantic modeling; the organization role screening module constructs a user collaboration graph based on a community division algorithm, and reasones user optimal role mapping by using a community portrait model; the data attribute screening module is combined with a data classification tree and a structure dependency relationship extraction algorithm to recognize a semantic category and upstream and downstream dependency chains of target data; and the permission allocation control module performs reasoning in combination with a permission knowledge base according to roles, data classification and a dependency relationship, and finally generates a permission allocation table, so that dynamic permission configuration and strategy matching oriented to a collaborative scene are realized. Through automatic role mapping, data semantic association analysis and authority inheritance reasoning, accurate authority distribution and real-time strategy adjustment in a ship collaboration scene are realized, so that data security and collaboration efficiency are improved, and core support is provided for digital transformation of the ship industry.

The invention provides a marketing service method and system based on multi-modal generation type artificial intelligence. The method comprises the following steps: firstly, collecting multi-modal data related to a marketing target and carrying out vectorization; then constructing a joint embedding space containing marketing concept anchor points, aligning data vectors with related concept anchor points through a calibration mechanism, and dynamically updating the anchor points according to a new data track; related integrated vector clusters are identified in the space according to marketing targets, and an optimal commodity combination meeting budget constraints is searched in the clusters by adopting an optimization algorithm; and finally, inputting the commodity combination and the semantic core of the vector cluster into a language model as a context, and generating a marketing scheme matched with the target customer group. According to the method, the vector representation of the fused multi-modal data is realized, and the fused representation can be subjected to structured alignment with predefined marketing domain knowledge.

The invention discloses a data layout optimization method and device applied to NPU code compilation and a medium. The method comprises the steps that an intermediate representation IR input by an upper layer is split into an operation type OP and an operand, and the operand input by the upper layer is divided into logic data and a logic mask; converting the logic data of the upper layer into data vector representation of the abstraction layer, and converting the logic mask of the upper layer into mask vector representation of the abstraction layer; obtaining a corresponding bottom hardware instruction capability according to an operation type OP input by an upper layer, and performing legalization operation on the operation type OP; and performing legalization processing on the converted abstract data vector type representation and mask vector representation according to the target hardware capability, performing instruction mapping on the processed abstract instruction according to the underlying hardware pair operation type OP, and generating a target LLVM instruction compatible with the underlying hardware. Register layout abstraction and conversion can be carried out on the NPU code under mixed precision input, and the utilization efficiency of the NPU bottom layer register is improved.

The embodiment of the invention discloses a charging pile power distribution method and device based on artificial intelligence and a medium, and relates to the technical field of electric vehicle charging, and the method comprises the steps: collecting multi-dimensional charging data of a target charging pile cluster; performing data fusion based on the multi-dimensional charging data to generate a fusion data vector, extracting local features in the fusion data vector by using a convolutional neural network, and extracting time sequence features by using a gating cycle unit to generate a depth feature vector; based on the depth feature vector, a multi-objective optimization function is constructed, and the multi-objective optimization function comprises a charging efficiency objective, a power grid load fluctuation objective, a user satisfaction objective and a battery life objective; and solving the multi-objective optimization function by using a particle swarm optimization algorithm to obtain a power distribution parameter of each charging pile, and sending the power distribution parameters to a corresponding charging pile controller so as to adjust the output power of the charging piles and realize power distribution of the charging piles.