21 results about "Local logic" patented technology

The invention relates to the field of data processing, and discloses an intelligent data processing system based on a large model, which is characterized in that hierarchical access is performed on structured, semi-structured and unstructured data with different sources, and asynchronous aggregation is performed on input streams to generate a data coupling set; semantic folding and noise overflow suppression are carried out on the data coupling set, redundant components are redistributed to a local logic domain, and convergent data intermediate mapping is formed; data intermediate mapping is imported into a cognitive base mapping layer of the large model, and semantic tensorization core representation is extracted through a feature overflow monitoring and parameter polarization control mechanism; executing multi-channel reforming processing based on the core representation, and generating an adjustable self-evolution characteristic channel set which is used for dynamically associating implicit relationships of different source data; a global reasoning field is constructed through a feature channel set, and a continuous adaptive loop is formed between data change and large model updating. The method has the advantage that the intelligent degree of data processing is improved.

The invention provides an intelligent gridding comprehensive anti-misoperation locking system for a power distribution network, and the system comprises an optimization module which achieves the hierarchical decoupling design of the system through a distributed communication middle platform, an intelligent logic analysis engine and a programmable execution control unit. The deployment module is used for modularly deploying functional components based on the micro-service architecture, reducing the system coupling degree by using an asynchronous message queue as a cross-component communication bus in combination with a domain event-driven mechanism, and completing local logic judgment at an edge computing node through an FPGA (Field Programmable Gate Array); the network reinforcement module is used for constructing a heterogeneous communication network and guaranteeing the communication quality through a dual-channel redundancy transmission technology; the intelligent decision-making module is used for forming a self-adaptive decision-making model based on reinforcement learning and a knowledge graph technology; and the digital twinning verification module is used for establishing a power grid virtual mapping system through a digital twinning technology. Cooperation among modules of the system can be balanced, instruction processing time delay is shortened, communication reliability is improved, and efficient implementation of an operation scheme is guaranteed.

The invention discloses a machine learning platform and method based on workflow, electronic equipment and a storage medium, and the platform comprises a management scheduling system, a workflow engine system and an operator runtime system. A user is allowed to construct a visual business process in a dragging and connecting mode, and each component corresponds to a specific AI algorithm processing unit written by Python; the platform supports the realization of multi-node business logic by using an operator concept by referring to a BPMN specification, automatically analyzes a flow chart designed by a user into a directed acyclic graph, converts the directed acyclic graph into an API calling sequence for a Python runtime system, disassembles a complex business system into a multi-branch and multi-step configurable process, and executes local logic by each node to realize multi-node business logic. The whole process is visually arranged, the development threshold is remarkably reduced, and the execution efficiency and the system flexibility are improved.

The application is suitable for the technical field of artificial intelligence security, and provides a jailbreaking security evaluation method and system based on dictionary guided sparse logic editing, which comprises the following steps: first, generating a static intensity vector; then, intercepting an original logit vector corresponding to each autoregressive decoding time step of a target large language model, and generating a dynamic gating signal based on the original logit vector; meanwhile, constructing a preliminary sparse intervention vector from the static intensity vector; and then, generating a modified adversarial logit vector based on the preliminary sparse intervention vector, so as to obtain response content for evaluating security. Through the application of extremely sparse local logic editing based on temperature scaling and trust region constraints in the vocabulary space, the high computing power overhead of traditional gradient attacks and the auxiliary model memory overhead of traditional logic intervention methods can be completely eliminated, and efficient, smooth, interpretable and robust large model jailbreaking security evaluation can be realized.

The invention discloses an agent application-oriented resource sharing method, and relates to the field of cloud computing. The scheme comprises offline and online two-stage processing: in the offline stage, constructing an agent dependency conflict graph, dividing a core and other agent sets and grouping, and determining candidate execution nodes through a convergence hash algorithm; in the online stage, a global scheduler selects a target node according to grouping information and node real-time loads, a local scheduler schedules a calculation sandbox to execute local logic, a communication sandbox calls a large language model, and a capacity sandbox is dynamically expanded and shrunk based on a preset strategy. The system correspondingly comprises an offline processing module, a global scheduler, a local scheduler and two types of sandboxes. Through calculation and communication decoupling, off-line grouping, double-layer scheduling and security isolation mechanisms, the problems of low resource utilization rate, high operation cost, insufficient concurrent bearing capacity and the like in the prior art are solved, and the resource reuse rate, the system stability and the throughput capacity are remarkably improved.

The invention belongs to the technical field of Internet of Things, and particularly relates to an Internet of Things-based muck recycling production line intelligent monitoring and regulation and control system, which comprises an Internet of Things sensing layer arranged in each process link of a muck recycling production line in a distributed manner and used for collecting material attribute data, equipment operation state data and environmental parameters in real time; the edge calculation layer is in communication connection with the Internet of Things sensing layer and is used for carrying out preprocessing, local logic judgment and real-time equipment control on the collected original data; according to the system, the production efficiency and the intelligent level are improved, real-time and accurate data collection and localized quick response of the whole process (feeding, crushing, screening and forming / aging) of the production line are achieved through cooperation of the Internet of Things sensing layer and the edge calculation layer, and the real-time and accurate data collection and localized quick response of the whole process of the production line are achieved in combination with a process model and big data analysis of a cloud platform. The system can automatically generate and execute a global optimization regulation and control instruction, dynamically adjust production parameters, and reduce manual intervention dependence.

The application provides a system and method for safe transmission of FC data through high-speed port, which comprises: a high-speed port for receiving data and generating a high-speed port recovery clock, and transmitting the received data to a data transceiver module in the high-speed port recovery clock domain; and the data transceiver module for converting the data into output data in the local logic clock domain; wherein the data transceiver module comprises a data storage area, a write control module and a read control module; the data storage area is composed of two clock domains: the write direction adopts the high-speed port recovery clock, and the input data is the high-speed port received data; and the read direction adopts the local logic clock, and is used for controlling the output of the data storage area to a subsequent logic module. The data storage area is configured with an overflow area, an underflow area and an equalization area, and the input and output enables of the fifo are accurately controlled through the data amount cached in the fifo and the current input and output fifo data stream content, so as to ensure the correct transmission of effective data and the normal work of the fifo.

The present application belongs to the technical field of enterprise-level backend service development and low-code platform, and particularly relates to a visual backend business flow execution method and system supporting hot updating and code generation, a storage medium and computer equipment.S1, a non-BPMN business flow is configured through a visual interface;S2, a DispatcherFlow executor instance is constructed; a dual-mode use mechanism is provided: a hot updating mode and a code generation mode.It is independent of Spring Integration in internal implementation, but is fully compatible with the programming paradigm of Spring Integration in interface and data model, supports visual embedding of Groovy script nodes for local logic customization, realizes hot updating in combination with dynamic management of Spring Bean, and can convert full-amount logic (including scripts) into standard Java source code during configuration.Development flexibility and production stability are taken into account.

The invention relates to the technical field of electric power safety real-time monitoring, and particularly discloses a grounding wire locking device state management system and method based on real-time monitoring. The system comprises an edge sensing and control terminal, a region convergence and decision node and a centralized management platform. According to the invention, a three-level decision-making architecture combining edge terminal local logic, regional node intelligent decision-making and centralized platform strategy management is constructed. The edge terminals have independent locking capability to guarantee basic safety, the regional nodes realize rapid regional cooperation and intelligent research and judgment, and the centralized platform is responsible for global optimization and learning evolution. The distribution and concentration combined architecture not only ensures the quick response and autonomous capability under the local fault condition, but also realizes the optimal configuration of global resources and the continuous accumulation of system intelligence, and enhances the overall robustness and intelligent level of the system.

The application relates to the field of data processing and discloses a data intelligent processing system based on a large model, which comprises hierarchical access to structured, semi-structured and unstructured data of different sources, asynchronous aggregation of input streams to generate a data coupling set, semantic folding and noise overflow suppression of the data coupling set, re-distribution of redundant components to a local logical domain to form a convergent data intermediate mapping, introduction of the data intermediate mapping into a cognitive base mapping layer of the large model, extraction of a semantic tensorized core representation through feature overflow monitoring and parameter polarization control mechanisms, multi-channel renormalization processing based on the core representation to generate an adjustable self-evolution feature channel set for dynamically associating the implied relationships of different source data, construction of a global reasoning field through the feature channel set to form a continuous self-adaptive loop between data changes and large model updates. The application has the advantage of improving the intelligent degree of data processing.

The application discloses an address remapping device, method and processor, and the device comprises: a first calculation module which takes the complement of the low L bits of an original logical address to obtain a complement result; a second calculation module which dynamically obtains an offset correction value which keeps the order of an output channel group according to the original logical address; a third calculation module which determines the base address offset across the input channel group; a fourth calculation module which performs summation and one addition operation on the complement result and the offset correction value, and extracts the low L bits to obtain a local logical address; and finally, an address composition module which adds the local logical address and the base address offset to generate a target logical address corresponding to the original logical address. In the memory access path, the application can efficiently realize the 180° rotation of the convolution kernel at the hardware level through additional processing of the address, so as to fundamentally eliminate the efficiency loss caused by the software rotation.

The invention relates to the technical field of intelligent lighting control of building automation, and particularly discloses an intelligent lighting control system based on multi-protocol dynamic switching and a control method thereof. The system comprises a lighting controller, a multi-protocol communication module, a protocol arbitration module, a unified state memory, a local logic execution engine and a man-machine interaction interface. According to the method, communication quality of each protocol channel is monitored in real time, a health degree score is calculated, a protocol switching process is dynamically triggered, cross-protocol state synchronization is realized by adopting unified state storage, local control is automatically degraded when a network is abnormal, and state backtracking is completed after the network is recovered. According to the invention, gateway-free intelligent cooperation of multiple protocols of a management layer and a field layer is realized, the real-time consistency of a control state is guaranteed, the reliability and fault-tolerant capability of the system are improved, the operation and maintenance process is simplified, and the system can be widely applied to illumination control scenes of high-end intelligent buildings and has remarkable economic and social benefits.

The invention provides a subarray control architecture for an ultrasonic chip, and the architecture comprises a digital control hub which is connected to a central digital IO interface, and is used for communicating with a host through the digital IO interface; the ultrasonic array comprises a plurality of sub-array groups distributed in a matrix, each sub-array group comprises a plurality of ultrasonic sub-arrays, and each ultrasonic sub-array is provided with a sub-array island module; wherein the plurality of ultrasonic sub-arrays in the same sub-array group are cascaded through the sub-array island modules and then are connected to the digital control hub. The sub-array island module is used for synchronizing input global control signals in a plurality of ultrasonic sub-arrays and sending the global control signals to the local logic module, so that beam synthesis in different ultrasonic sub-arrays such as receiving beam forming units is synchronized, signal distribution and time sequence control are optimized, and the accuracy of beam synthesis is improved. The global control signal integrity of the large-scale ultrasonic array chip is ensured, and the structure can significantly improve the beam forming precision and the signal-to-noise ratio.

The invention discloses an anti-misoperation method based on local logic, and belongs to the technical field of power systems, and the method comprises the following steps: configuring operation logic for all devices, the operation logic comprising operation logic itself and an operational expression of the operation logic itself obtained by performing logical operation on the states of other devices; obtaining an operation task and states of all devices at the current moment; executing the operation task, and performing identity verification on the equipment to be operated; if the identity verification is passed, logic verification is carried out based on the operation logic and the states of all the devices at the current moment; and if the logic verification is passed, operating the logic verification. According to the method, the current execution task can be verified through the local logic, and the efficiency and the flexibility of executing the operation task in the power system are greatly improved.

The invention relates to the technical field of rail transit automation control, in particular to a software-defined lightweight distributed rail transit station control system, which comprises a plug-in box, a power panel, a core board and a detachably installed software-defined peripheral interface board. The core board is connected with the software-defined peripheral interface board through a high-speed bus and a low-speed bus to realize real-time data transmission and configuration management, the software-defined peripheral interface board is integrated with a microprocessor, an interface circuit supports hot plug and local logic operation, a Lua script engine is arranged in the software-defined peripheral interface board, graphical or script programming is supported, and the software-defined peripheral interface board is connected with the software-defined peripheral interface board. The core board adopts an RISC-V architecture microprocessor, has the functions of board card registration, configuration issuing, health monitoring and redundancy scheduling, and supports automatic networking and second-level takeover of fault nodes. The system deploys an RT-Thread real-time operating system, has a multi-node instruction conflict resolution mechanism, arbitrates the control right according to the priority and the timestamp, and is suitable for multi-device cooperative control of the rail transit station.

The invention provides a privatized intelligent data processing method, device and equipment based on a storage and calculation separation architecture and a storage medium. When an NAS server receives a user task request, intention recognition is performed on the request, task metadata is extracted, rigid constraint judgment is performed according to dependency model information, and the task metadata is extracted; the tasks are divided into a first type of tasks which cannot be executed locally and a second type of tasks which can be executed locally. For the first class of tasks, the NAS server sends a magic packet to the computing terminal in a dormant state through a local area network to wake up the computing terminal, and after the computing terminal is started, a shared directory of the NAS is mapped into a local logic driver through a network file system protocol to establish mounting connection; and obtaining target data in a streaming reading mode to a memory, executing reasoning operation by utilizing the GPU, and writing back a result to the NAS after completing the reasoning operation. According to the scheme, the high-power-consumption computing terminal is awakened to work only when processing complex tasks, the high-power-consumption computing terminal is kept in a dormant state in other time, and meanwhile data is always stored at the NAS end.

The application relates to the technical field of electric power safety real-time monitoring, and particularly discloses a grounding wire locking device state management system and method based on real-time monitoring. The system comprises an edge perception and control terminal, a regional convergence and decision node and a centralized management platform. The application constructs a three-level decision architecture combining edge terminal local logic, regional node intelligent decision and centralized platform strategy management. The edge terminal has independent locking capacity to guarantee basic safety, the regional node realizes rapid regional cooperation and intelligent research and judgment, and the centralized platform is responsible for global optimization and learning evolution. This architecture combining distribution and centralization not only guarantees rapid response and autonomous ability under local fault conditions, but also realizes optimal allocation of global resources and continuous accumulation of system wisdom, thereby enhancing the overall robustness and intelligent level of the system.

The invention relates to the technical field of power system safety and stability control, in particular to a remote safety regulation and control method and system for a low-frequency and low-voltage load shedding device, electronic equipment and a medium. The method comprises the following steps: firstly, acquiring running state data of the low-frequency and low-voltage load shedding device and real-time running data of a power grid; based on the running state data and the real-time running data of the power grid, generating a standardized operation ticket of the low-frequency and low-voltage load shedding device through the cloud platform; the standardized operation order is transmitted to side equipment through a safety pipeline; and performing instruction verification on the received standardized operation order through the side equipment, and after the instruction verification is passed, executing local logic locking verification based on the real-time operation data of the power grid so as to generate a security permission allowing instruction issuing. By means of the mode, the problem that it is difficult to achieve efficient and reliable management of remote modification of the low-frequency and low-voltage load shedding device under the condition that safety is guaranteed in the prior art is solved, and the intelligent level and operation and maintenance efficiency of power grid safety control are improved.

The invention relates to the technical field of foundation detection, and discloses an automatic foundation load detection system integrating intelligent sensing and remote collaboration, which comprises a field sensing execution layer, an edge computing control layer and a cloud remote collaboration layer. The field sensing execution layer is responsible for executing physical loading and collecting environment parameters; the edge computing control layer serves as a local processing center, forms closed-loop control independent of a remote network according to local logic, and uploads data in real time; the cloud remote collaboration layer is used for visually monitoring and issuing an intervention instruction, and the edge computing control layer is configured to preferentially execute the intervention instruction. According to the invention, through a layered architecture and a dual-mode communication mechanism, in combination with dynamic attitude self-stabilization and intelligent data analysis functions, cooperation of local millisecond-level response and remote safety supervision is realized, the problems of equipment stability and foundation damage early warning under an unattended working condition are effectively solved, and the detection efficiency and safety are improved.

The present application relates to the technical field of computing power equipment thermal management, and provides a computing power cabinet coupling device with active thermal management and safety interlocking function and a control method, which comprises a current sampling unit, a processing unit, a flow regulating unit, a controllable switch unit and a position detection unit; the current sampling unit, the flow regulating unit, the controllable switch unit and the position detection unit are connected with the processing unit respectively; the processing unit is configured to: output a feedforward control signal to the flow regulating unit based on the current transient characteristics obtained by the current sampling unit; and control the controllable switch unit to perform a disconnection action in response to an abnormal insertion state detected by the position detection unit. The present application realizes feedforward regulation of thermal management through current transient characteristics, solves the problem of thermal inertia lag, and realizes millisecond-level active safety protection through local logic interlocking.

The invention discloses a storage device management method based on fault prediction and automatic recovery, which dynamically calculates health scores of storage modules and automatically switches the modules into a normal mode, a degradation mode and an isolation mode according to the scores. When the module enters the degradation mode, the controller divides the data into a plurality of data segments which can be independently moved, the data segments are sequentially migrated to the healthy module according to the risk sequence, and the local logic mapping table is updated in real time after each segment is migrated. In the migration process, the controller evaluates a health score in real time, and if the score rises to a preset threshold, the residual data segment migration and new data writing rate is adjusted until the module is recovered to a normal mode; if the health score is continuously lower than the threshold, the module is permanently switched to an isolation mode, writing is forbidden, and data logic is mapped to other health modules. By means of the method, active protection and automatic recovery of SSD data security can be achieved, the reliability of a storage system is improved, and the service life of the storage system is prolonged.