16results about How to "Lower development threshold" patented technology

The invention provides a neural network multi-objective optimization and FPGA hardware acceleration collaborative design method, and belongs to the field of deep learning model compression and hardware collaborative design. The method comprises the following steps: constructing a joint optimization space containing a neural network compression parameter and an FPGA hardware design parameter; a multi-target Bayesian optimization search strategy is adopted, iterative search is carried out in the joint optimization space, model precision, FPGA resource occupation and reasoning delay are synchronously optimized, and optimal candidate configuration is obtained; matching the compressed network structure with the FPGA parallel architecture by using a hardware-perceived pruning and quantification strategy; a multi-task performance prediction model is adopted to quickly predict the precision, resource occupation and delay of the optimal candidate configuration so as to accelerate the search process; according to the optimal configuration, a hardware accelerator code facing the target FPGA is automatically generated, and integration and implementation are completed. According to the method, collaborative optimization of neural network compression and hardware design is achieved, FPGA resource occupation can be remarkably reduced, the reasoning speed can be increased, and meanwhile the model precision is kept.

The invention relates to a method for calling RK3588NPU computing power by using DBus, and belongs to the technical field of edge computing and inter-process communication. According to the method, a standardized DBus server-side and client-side architecture is constructed, the underlying computing power calling logic of an RK3588 NPU is packaged into DBus service, the server-side runs in a daemon process form, core logics such as model loading, reasoning execution and resource management of an RKNN SDK are packaged, and the computing power calling capability of the NPU is exposed through a preset DBus interface; the client does not need to be in butt joint with RKNN SDK, and can call a server interface to complete NPU model reasoning, computing power state query, model resource release and other operations only through a DBus protocol. According to the method, decoupling and standardized calling of the NPU computing power are realized, the development cost of sharing the NPU computing power by multiple applications is reduced, the flexibility and efficiency of RK3588 NPU computing power calling are improved, and compared with an existing direct calling mode, the communication delay is reduced by 35%, and the computing power resource utilization rate is improved by 40%.

The invention discloses a heterogeneous environment prediction model training and deployment method and system, and relates to the field of prediction model training. In order to overcome the defects that an existing model training and deployment method is large in development difficulty, low in deployment efficiency and lack of unified management of multiple versions, an integrated development environment and an integrated development framework are used, and multiple pre-training models and operators are built in; carrying out data preprocessing and exclusive model building, obtaining training data to carry out model training, and generating a training model; all training models are arranged in a model center for unified management; aPI service development is carried out, a unified model algorithm reasoning framework is adopted, self-defined service development is carried out, and self-defined AI service is generated; the AI service is audited and published to be online, and the online AI service is used by an external application; based on distributed task scheduling and service arrangement, AI services are scheduled and run on a service running platform. The method is mainly used for establishing the prediction model for the heterogeneous environment.

This application relates to the field of rail transit technology, and in particular to a software and hardware development method and system for an inspection robot. The method includes: based on user-inputted task scenario requirements, calling multiple 3D hardware models from a physical modular hardware library to obtain a 3D simulation robot corresponding to the task requirements; based on the 3D simulation robot, the task scenario requirements, and the physical attributes of each 3D hardware model in the 3D simulation robot, calling multiple standard functional nodes from a pre-established node library to construct a task logic flowchart; and based on the task logic flowchart, performing operational simulation and logic verification on the 3D simulation robot to obtain a task execution strategy; and based on the task execution strategy, compiling and packaging the task logic flowchart to obtain executable code for the user to drive the physical robot to perform operations. This significantly improves the development efficiency of inspection robots while lowering the development threshold.

This invention provides a privacy-preserving intelligent agent cross-domain knowledge hybrid retrieval method and system, relating to the fields of artificial intelligence and data security technology. The method includes: responding to a query request initiated by a cloud-based intelligent agent, parsing the privacy tags embedded in the request workflow definition; and concurrently executing the following two operations through an asynchronous task scheduler: sending the public subquery to a cloud-based knowledge retrieval service, and routing the private subquery to a privacy computing agent client deployed in the target local environment; the privacy computing agent client executing the private subquery in a local private knowledge base; and at a cloud aggregation point, aggregating the decrypted ciphertext knowledge fragment set and the public knowledge fragment set into unified knowledge data, and returning it to the intelligent agent. This invention achieves efficient, secure, and integrated retrieval and utilization of distributed heterogeneous knowledge bases by a cloud-based intelligent agent while ensuring that the original privacy data does not leave the local security boundary.

The invention discloses a configuration file generation method and device, equipment, a storage medium and a program product, and the method comprises the steps: receiving a configuration task which comprises configuration intention information; determining a target configuration scene and execution parameters based on the configuration intention information; the execution parameters comprise at least one of an execution object, an execution action, a dialogue scene and an intention corresponding to the dialogue scene; calling at least one corresponding basic template file from a preset template library based on the target configuration scene; the basic template file comprises a dialogue scene template file, an instruction template file and an intention-instruction mapping template file; and filling corresponding fields of the basic template file with the execution parameters, and determining a target configuration file. According to the method, the problems of high development threshold, long period, difficult maintenance, inflexible configuration and the like in the prior art can be solved, and efficient and accurate generation of the configuration file is realized.

The invention discloses a mathematics accompanying online learning system and a method thereof, and belongs to the technical field of online education. Aiming at the technical bottlenecks of high course development threshold, insufficient interactivity, low individuation degree and weak family-school coordination of the existing mathematics learning platform, the system constructs a dynamic course management-embedded mathematics experiment-intelligent learning engine-family-school coordination monitoring four-layer architecture. Through supporting HTML / JS / CSS interactive course packet low-threshold uploading and real-time rendering, GeoGebrar / network drawing board and other mathematical experiment tools are integrated, AI algorithm is combined to analyze learning behavior data and dynamically plan a personalized path, and a data visual chemical condition monitoring system is synchronously established. According to the invention, a teaching mode of'learning while practicing 'and'learning from middle school' is realized, the course development cost of teachers is reduced, the understanding depth of mathematics abstract concepts of students is improved, a teaching closed loop is strengthened through a family-school cooperation mechanism, and an intelligent, interactive and personalized solution is provided for mathematics online education.

The invention provides a power intelligent training and pushing all-in-one machine platform, and relates to the technical field of power AI training and pushing integration, and the platform comprises a data collection unit which is used for collecting an original operation data set of power equipment through a multi-modal data collection module; the data processing unit is used for processing the original operation data set and outputting a power equipment structured training data set and a real-time reasoning data set; the instruction generation unit is used for training and reasoning an improved deep echo state network training engine based on the data set, and generating a power equipment operation state evaluation result and an operation control instruction; and the equipment regulation and control unit is used for displaying the operation state evaluation result through the visual data interaction module, and transmitting an operation control instruction to the control terminal based on a communication link to complete closed-loop regulation and control, so that training and pushing integrated efficient cooperation can be realized, the development threshold is reduced, the computing power utilization rate and the operation and maintenance efficiency are improved, and the development cost is reduced. Intelligent upgrading of power core services is supported, and application stability and reliability are enhanced.

The invention provides an inter-system integration method and device based on configured zero codes, a graphical arrangement front end is constructed by adopting frameworks such as Vue, a back-end execution engine is constructed based on Spring Boot / Cloud, and a complete nodal process configuration and execution system is formed. According to the system, JSON Schema metadata is generated by automatically collecting and analyzing a Feign interface, and a user is supported to assemble integrated logic in a zero-code mode by dragging prefabricated function nodes and connecting lines; and the background models the structured JSON into a directed acyclic graph for topological sorting and static verification based on types and rules, and finally serializes the structured JSON to configure persistent storage. According to the method, visual and configured construction of integrated logic is realized, and the development threshold and the coding quantity are greatly reduced; and through built-in verification and structured storage, the reliability, maintainability and traceability of the process are remarkably improved, so that business personnel or common developers can rapidly and flexibly arrange and reliably operate the enterprise-level integrated process.

The application discloses a JSBSim dynamics model plug-in for UE4 cluster simulation, belongs to the technical field of unmanned aerial vehicle cluster flight simulation, and comprises an initialization module, a data interaction layer module, a simulation cycle module and a UE4 cluster simulation system module; the initialization module is responsible for constructing a simulation basic environment; the data interaction layer module realizes bidirectional data flow; the simulation cycle module drives a simulation process at a fixed frequency; the UE4 cluster simulation system module takes an actor as a physical carrier, converts coordinates of data output by JSBSim into coordinates and rotation values in UE4 local coordinates, updates position and posture information of the actor in real time, and presents simulation results in a visualized manner; the JSBSim dynamics model plug-in for UE4 cluster simulation provided by the application solves the problems of low dynamics simulation precision of an existing cluster simulation system and the problem that an official plug-in of JSBSim only supports UE5.

The invention discloses a multifunctional mobile network dialing management method and system. The method comprises the following steps: pre-constructing a module information base for storing associated information of a mobile network module; detecting equipment corresponding to an externally accessed mobile network module, acquiring equipment identification information of the equipment, matching the corresponding mobile network module in the module information base based on the equipment identification information, if the matching succeeds, initializing a communication link with the mobile network module, and binding a corresponding performance function; unified interface operation of different mobile network modules is realized; configuration information of the mobile network module is loaded for initialization, and the state of the SIM card is detected; after the SIM card is ready, setting a target network mode; waiting for network registration, and if registration succeeds, configuring network access parameters and executing dial-up connection; if the dial-up connection is successful, the network configuration information is acquired and the network connection is established, and the method has the characteristics of wide compatibility, high reliability and simplified configuration.

The application provides an automated digital integrated circuit design system based on multi-agent and LLM, comprising: a requirement analysis module receiving user natural language input and generating a structured design specification; a code generation module generating an intelligent agent through a code generation intelligent agent and a test platform intelligent agent, respectively generating synthesizable RTL code and corresponding test platform code based on the structured design specification; a code execution and simulation module compiling and simulating the RTL code and the test platform code by calling a simulation tool according to the generated code and outputting a simulation log; a debugging module analyzing the simulation log, identifying error types and code positions, generating a minimized corrected hardware language description code, and re-entering the code execution and simulation module for verification until the simulation passes, forming a closed-loop iterative optimization process. The application realizes full automation from natural language requirements to simulation verification based on a multi-agent collaboration framework.

The invention relates to a quantum state encoder, in particular to a high-precision quantum state encoder based on a nested interferometer, which is characterized in that an unbalanced Mach-Zehnder interferometer is positioned on an anticlockwise path of a Sagnac interferometer and consists of a second beam splitter BS2 and a third beam splitter BS3, and two arms of the interferometer have unbalanced optical paths and are used for introducing fixed time delay; dividing a single light pulse into a morning time bin and a evening time bin in a time domain; the at least one phase modulator is located in the Sagnac interferometer, under the driving of the integrated control system, independent phase modulation is carried out on the passing light pulses, a target quantum state is directly synthesized by designing modulation phase combinations, and meanwhile a decoy state is formed; the integrated control system is used for generating a voltage pulse with a corresponding time sequence and amplitude according to a target quantum state to be generated and a decoy level in each pulse period, and directly driving the phase modulator; according to the method, the defects of discrete structure, high quantum bit error rate and poor expandability can be overcome.

This application discloses a method, apparatus, system, and electronic device for generating executable files, belonging to the field of visual inspection technology. The executable file is applied to a visual inspection system. The method includes: acquiring user input to obtain the target detection task of the visual inspection system; based on the target detection task, obtaining the detection information to be executed by the visual inspection system, including the object to be inspected, the items to be inspected, and the detection constraints; based on the detection information to be executed, the tool library and the detection process template library of the visual inspection system, obtaining multiple execution tools and the relationships between them; and generating an executable file based on the preset configuration information and relationships corresponding to each execution tool. This can significantly reduce the development threshold, improve construction efficiency, reduce reliance on expert experience, and achieve standardization and automation of the inspection process.

This invention discloses an automatic generation and fault location system for avionics equipment test plans. It includes a data layer comprising an ICD information database, an equipment interface information database, and a test bench resource database, and establishing a three-dimensional association table of equipment interface pins, ICD signals, and test bench resource channels; a physical layer including various standard avionics bus resource cards, general-purpose instruments, and the actual avionics equipment under test connected via adapters / junction boxes, responsible for the excitation and acquisition of actual signals; a service layer providing a test case execution engine and a fault diagnosis engine, using the three-dimensional association table to decouple test logic from physical hardware, and achieving refined fault location through reverse tracing using the three-dimensional association table; and an application layer providing a no-code visual test program editing and execution module. This invention enables automatic generation of test plans, flexible definition of test logic, hardware-independent execution of test cases, and rapid and refined fault location.

The application discloses a kind of visual report design method and system, it is related to Web front-end development technical field, including: user creates dynamic report through Web end, configures medical data source and associated field, to establish the import mapping channel of medical data and report unit;Design report appearance style and static content, the static content includes table header, cell constraint;According to the associated field, report appearance style and static content identify form scene, activate medical scene enhancement function, according to the medical data source content feature, carry out page column design or conditional rendering adaptive setting, output design report.Solve the medical data report form system development cost, research and development standard is not consistent, there is product quality level, product flexibility, poor reusability, weak integration capability technical problem.It is reached in reducing development threshold, while, it guarantees the uniformity of research and development standard, and product flexibility and reusability technical effect.