55 results about "Computational logic" patented technology

The application provides an automatic disk IO method based on structural mechanics CAE simulation software, comprising the following steps: S1, constructing a memory-disk cooperative computing framework, and dynamically monitoring a data access state in a simulation solving process; S2, automatically exchanging non-active core data structures to a high-performance disk according to a data access frequency, a life cycle and a memory pressure; S3, loading back to the memory on demand when subsequent calculation needs to access the data exchanged to the high-performance disk; wherein the core data structures comprise a stiffness matrix, shape function data, an intermediate result vector and derivative physical quantities, and the exchange process is transparent to the solver. According to the application, the exchange process is highly transparent to the core computing logic of the solver, is automatically managed by the system, and does not need user intervention, so that efficient utilization of memory resources and stable operation of super-large-scale model simulation are realized.

A method and system for ensuring consistency between a floating-point arithmetic algorithm verification prototype and hardware engineering, relating to the field of floating-point arithmetic, are disclosed. The method includes: obtaining the HDL source code of a hardware floating-point IP core; converting the sequential logic describing floating-point operations in the HDL source code into a behavioral-level combinational logic expression, which retains the computational logic of floating-point operations; compiling the converted HDL source code using an HDL compilation tool to generate a library file executable on a processor; calling function interfaces in the library file to perform floating-point operations in a prototype verification program running on the processor; and integrating a hardware floating-point IP core into the hardware system to implement floating-point operations, wherein the library file and the hardware floating-point IP core are based on the same HDL source code. Implementing this method can reduce the cost of ensuring consistency between the floating-point arithmetic algorithm verification prototype and hardware engineering.

Embodiments of the present application provide a data processing method, system, device, storage medium and program product, relating to the field of distribution. The method comprises: an edge computing node collecting transaction data in real time, extracting feature information of the transaction data, and generating a preliminary risk score based on the feature information through a lightweight model deployed in the edge computing node. The edge computing node receives a global model from a distributed parameter aggregation computing node, updates the parameters and computing logic of the lightweight model based on the global model, so as to update the calculation result of the preliminary risk score. The method of the present application increases the accuracy of abnormal transaction behavior identification and improves the data response speed.

This invention provides a multi-task AI intelligent agent cost estimation system and method adapted to hydropower engineering cost estimation scenarios, relating to the field of hydropower engineering cost estimation technology. It utilizes a visual task interaction module to input basic hydropower engineering information and multi-task instructions. A multi-task AI intelligent agent decouples the input basic hydropower engineering information and multi-task instructions to obtain a task list and task priorities. Based on the tasks, it calls the corresponding data resource sub-libraries, and according to the corresponding calculation logic and priorities, combines the data resource sub-libraries to complete multi-task calculations, outputting the calculation results for each task. This invention solves the problems of information silos, low efficiency, and high error rates in existing cost estimation systems and is applicable to hydropower engineering cost estimation.

The application provides a data processing method, an electronic device and a computer readable storage medium, wherein the method comprises: constructing a calculation model according to a calculation model definition request; determining a target calculation engine to be adapted according to a calculation logic feature of the calculation model, and acquiring a target engine plug-in corresponding to the target calculation engine from an engine plug-in library according to an identifier of the target calculation engine; running the calculation model according to a data processing instruction of a user; and the calculation model interacts with the target calculation engine through the target engine plug-in to execute the data processing instruction during running. The application uses the engine plug-in as an intermediate adaptation layer to shield the differences between different calculation engines, so that the same calculation model can run on multiple calculation engines without modification. In this way, not only is multi-engine compatibility achieved, but also the repeated development and maintenance costs caused by engine binding are significantly reduced, effectively solving the problem that the calculation model is difficult to adapt to multiple calculation engines in the prior art.

The application discloses a performance optimization method for a heavy machine tool hydrostatic guide rail system, comprising the following steps: S1. establishing a multi-objective optimization model: determining design variables of the hydrostatic guide rail system X , constructing an optimization objective function F(X) , and establishing an equality constraint based on force and torque balance H(X and an inequality constraint for preventing dry friction G(X) ; S2. constructing a multi-fidelity database: obtaining a high-fidelity data set through high-fidelity numerical model simulation, and obtaining a low-fidelity data set through low-fidelity numerical model simulation with the same calculation logic but a lower grid density; S3. constructing a data fusion prediction model: utilizing the high-fidelity data set and the low-fidelity data set, constructing a collaborative Kriging surrogate model based on an autoregressive structure, and outputting a predicted mean and a predicted variance; S4. generating a candidate solution population; S5. multi-criteria point selection; S6. simulating and updating a Pareto frontier solution set; and S7. repeating steps S3 to S6 until a preset convergence condition or a calculation budget is met, and outputting a final Pareto optimal solution set.

The present disclosure provides a model inference method and device and electronic equipment, relates to the technical field of computers, in particular to the technical fields of artificial intelligence, large models, front ends, data processing, memory and video memory resource management, and the like. The specific implementation scheme is as follows: in response to an inference request, model data of an inference model is obtained, and the model data is loaded into a memory buffer; wherein the model data includes model structure metadata and weight data; according to the model structure metadata that has been loaded into the memory buffer, video memory resources required by the inference model are created, and the weight data loaded into the memory buffer is written into the video memory resources; the weight data in the video memory resources is read by a graphics processing unit, and the calculation logic of the inference model is executed to obtain an inference result.

The application discloses a kind of method for calculating door and window legend thickness, device, computer equipment and storage medium, the method comprises: drawing target door and window component on target wall component;Based on the target two-way component relationship of target door and window component and target wall component, the target two-way component relationship of target door and window component and target wall component is established based on preset component association rule;Wherein, component association rule includes: by door and window component and the component identification of wall component attached to each other between it, the two-way component relationship of establishment of both;By surface layer component record the component identification of wall component attached to it, establish the one-way component relationship of both;Determine the thickness count value mode for calculating door and window legend thickness;Based on thickness count value mode and the component relationship associated with target wall component, call preset thickness calculation logic, calculate the door and window legend thickness of target door and window component.

The application provides a salary automatic payment system based on a smart contract, relates to the technical field of computer data processing, and first introduces an adaptive context anomaly detection mechanism based on group statistical characteristics, dynamically identifies anomalies by evaluating the relative position of data points in the group distribution, and effectively solves the high misjudgment and missed judgment problem of data cleaning in a complex correlation scene. On this basis, the scheme encapsulates the calculation process containing sensitive salary logic in an off-chain security sandbox, only generates a zero-knowledge proof of calculation without private information and a clear payment list. The on-chain smart contract does not need to touch the original data or the calculation model, and only needs to prove the validity through mathematical verification to confirm the correctness of the calculation result, thereby realizing the trustless verification of complex salary calculation logic and the automatic and accurate distribution of funds on the premise of completely shielding commercial secrets and personal privacy.

A motor starting method and a motor starting circuit are disclosed. The motor starting method includes: (1) obtaining the rotor current position; (2) asynchronously driving the motor at the current rotor position and the rotation direction to gradually increase the open-loop angle to a target angle and gradually increase the quadrature-axis reference current, wherein the open-loop angular increment gradually increases with time to implement smooth starting; and (3) motor entering a closed-loop control mode. By introducing the open-loop angle, open-loop angular increment, and angular addition increment, implementing the hardware starting circuit becomes easier, only one adder can complete starting, consuming only a few hardware, thereby releasing the CPU resources. This smooth starting approach reduces the “gear shifting” feeling, speeds up the starting process, and lowers the risk of inversion. The technique can be applied to other motors with shared computational logic to reduce costs.

This invention discloses a privacy-preserving analysis terminal and method for financial transaction records based on an edge computing gateway. Addressing the lack of logical error correction and data leakage privacy risks in existing financial bill recognition systems, this invention leverages a general heterogeneous computing architecture to construct an integrated NPU and TEE, and performs hardware-software collaborative processing logic through a high-speed collaborative channel. The method includes: the NPU performing OCR to extract text; the TEE replacing sensitive entities with encryption masks and mapping the text to SMT constraint expressions for logical verification; if a verification conflict occurs, a logical penalty term is calculated, triggering an NPU interrupt via the collaborative channel to call the operator dynamic update interface. While preventing redundant calculation of visual context features, the penalty term is injected as a bias into the decoding layer, dynamically adjusting the decoding weights and reconstructing the recognition result; finally, the desensitized data, based on homomorphic encryption to maintain the topology structure, is transmitted to the cloud for risk analysis and then restored locally. This invention achieves a closed loop between perception and deterministic logic, significantly improving feature extraction accuracy while ensuring privacy data remains within the data domain.

The application relates to the technical field of electric digital data processing, and discloses an electric digital processing and identification system for electric energy metering box data, which comprises a time sequence feature extraction unit, a dead zone logic unit, a self-referencing difference logic unit, a discrete logic mapping unit and a state machine matching unit.The time sequence feature extraction unit acquires voltage and current sequences and extracts zero-crossing time stamps, and calculates dynamic time differences.The dead zone logic unit compares time intervals with a dead zone time threshold, and outputs an effective time difference sequence.The self-referencing difference logic unit calculates a numerical difference value by using a first sample value stored in a register and a second sample value in a current period, and further determines a discrete difference parameter.The discrete logic mapping unit converts the parameter into a single-byte state code according to a mapping matrix.The state machine matching unit aggregates the state code to generate a sequence array, and outputs an alarm signal by calculating a logical distance.The application eliminates common-mode drift components generated by temperature drift and fundamental frequency fluctuation by using numerical hedging of adjacent periods, improves identification accuracy, reduces calculation power consumption, and ensures monitoring timeliness.

This invention provides a WebAssembly model protection system for browser-based secure inference. The system, through the collaborative work of a model parsing and storage module, a model conversion and compilation module, a security enhancement module, and a front-end API generation module, converts model files from different machine learning training frameworks into WebAssembly model files that can run independently in the browser. During the conversion process, it integrates and encapsulates model structure information, model weight information, and model computation logic, and enhances their security, achieving comprehensive security protection for the front-end model. This invention, by integrating model format conversion, WebAssembly compilation, and customized obfuscation technologies, constructs a highly secure and practical Web model protection system, significantly improving the resistance to reverse engineering of Web machine learning models, effectively preventing model theft and tampering risks, and is suitable for the secure deployment of AI applications on various browsers.

This invention relates to the field of electrical equipment condition monitoring technology, specifically disclosing a method and system for identifying the online status of electrical equipment. The method includes: first, establishing a parameterized feature model calibrated using measured data from the equipment's initial commissioning; then, substituting real-time operating parameters to generate dynamic benchmark feature values; calculating the deviation between the measured feature values ​​and the dynamic benchmark; determining the equipment status level and locating physical parameter deviation information based on a physical rule base; and finally, outputting standardized diagnostic results. This invention abandons the fixed threshold mode, achieving adaptive operating conditions for the benchmark value, eliminating the influence of operating condition fluctuations, significantly reducing the false alarm rate, and enabling interpretable alarms through physical parameter deviation location, providing precise guidance for maintenance. Deviation pattern recognition offsets individual equipment differences, enabling cross-equipment reuse of the same model. It employs lightweight computational logic, adapting to substation edge computing devices, resulting in low computational power consumption and meeting the low-cost, high-reliability application requirements of the power industry.

This invention discloses a low-level logic architecture for AI chips based on continuous proportional spectral computation. It abandons traditional binary 0 / 1 discrete computation and adopts the [0,1] continuous proportional range as the basic computational unit. Through core computational rules such as proportional superposition, proportional enhancement, proportional suppression, and proportional thresholding, a novel continuous logic computation system is constructed. This invention eliminates the loss of binary discretization from the bottom layer, achieves native processing of continuous signals, reduces computational levels and chip power consumption, and improves the generalization, interpretability, and decision quality of AI systems. It can be widely applied to AI large-scale model inference, edge intelligent devices, and multimodal data fusion scenarios.

This specification discloses a method and apparatus for fine-tuning a pre-trained model. The method fuses text feature vectors encoded by the data holder's encoding layer with a prompt text template corresponding to the target application scenario. Then, the model inputs the fused feature vectors into the network layer for logical inference. Finally, the loss between the logical inference result and the expected inference result labeled with the text feature vectors is calculated. Minimizing this loss is the optimization objective, and the text character feature vectors contained in the prompt text template, which serve as parameters for fine-tuning, are adjusted to complete the fine-tuning training of the LLM base model, resulting in an LLM service model corresponding to the target application scenario. This prevents the model from distinguishing between the text feature vectors and the prompt text template in the acquired fused feature vectors, thus avoiding the leakage of privacy information.

This invention relates to the field of traffic safety monitoring technology, specifically to a method and system for personnel positioning and safety alerts at highway construction sites. The method includes the following steps: spatiotemporal alignment based on raw distance data and visual features; coordinate transformation and error correction using a homography matrix to generate real-time position coordinates; inputting the real-time position coordinates into a filtering model for state estimation; calculating the velocity vector to construct a dynamic trajectory model; comparing the dynamic trajectory model with an electronic fence; and analyzing collision and intersection times to generate safety alert commands. In this invention, by fusing visual features and ranging sensor data to construct a dynamic vector model, using multi-point mapping logic to filter interference signals generated by mechanical obstruction, employing a trajectory prediction algorithm to calculate the intersection times of various moving targets to pinpoint dangerous areas, and directly driving various alarm feedback components based on localized computational logic, millisecond-level accurate alerts for dynamic risks in complex construction scenarios are achieved.

The application discloses a business experience validity verification and bidirectional enhancement method and device. The method comprises the following steps: receiving a business experience description in a natural language form; performing semantic matching on a business entity in the business experience description based on a database mode, generating a candidate field list if there is ambiguity, and determining a target field mapping in response to user selection; performing static validity checking on calculation logic in the business experience description based on the target field mapping, wherein the checking comprises compatibility checking of field data types and operators; if the checking is passed, explicitly injecting the target field mapping into the business experience description to generate an enhanced business experience, and writing the calculation logic to metadata annotations of corresponding fields in a database. The application solves the mapping error and logic non-executable problem caused by the separation of business experience and physical database, and improves the accuracy and maintainability of a ChatBI system.

This invention provides a bridge structure modeling method and apparatus based on graph data structures, belonging to the field of building information technology. The method includes: constructing a graph data structure for the bridge based on original engineering data; the graph data structure includes a set of nodes and a set of directed edges; the directed edges are associated with nodes in the node set that have dependencies and configured with rule functions; traversing along the dependency directions indicated by the directed edges, and using the rule functions to determine the component attribute data of component nodes and the rebar attribute data of rebar nodes; modeling the bridge to obtain a bridge structure model; and automatically updating the bridge structure model if the attribute data of the target nodes of the bridge structure model changes. This invention organizes nodes through a graph data structure and explicitly expresses mapping relationships and computational logic using directed edges configured with rule functions, thereby breaking the limitations of static parent-child relationships, effectively handling complex multi-directional dependencies between components and rebar, and achieving adaptive and linked updates when model parameters change.

A computing system is provided, including a processor configured to receive a standardized stabilizer instrument specification including an input Clifford unitary, an output Clifford unitary, and a plurality of stabilizer instrument bit matrices. The processor is further configured to receive a logical instrument input error correction code and a logical instrument output error correction code. The processor is further configured to compute a logical instrument specification based at least in part on the standardized stabilizer instrument specification, the logical instrument input error correction code, and the logical instrument output error correction code. The logical instrument specification includes a logical input Clifford unitary, a logical output Clifford unitary, a plurality of logical instrument bit matrices, and a logical instrument relabeling matrix. The processor is further configured to store the logical instrument specification in memory.

The process involves: acquiring a first software script; generating a reference fragment based on a first historical script whose similarity to the context information of the first parameter is not less than a first similarity threshold, and the attribute information corresponding to the first parameter; acquiring a first candidate fragment corresponding to the information of the end-to-end computation logic, wherein the end-to-end computation logic indicates the computation flow between the source table referenced by the second historical script and the target table generated based on the second historical script and the source table, and the second historical script refers to a historical script that references a data table related to the first parameter; converting the business metadata of the business to which the first parameter and the first software script belong to obtain a second candidate fragment; and determining a recommended fragment whose similarity to the reference fragment is not less than a second similarity threshold according to the similarity between the reference fragment and the first and second candidate fragments. The recommended fragments generated by this application not only meet user intent and business needs but also improve accuracy.

This disclosure relates to the field of electronic digital data processing technology, and in particular to a cross-platform performance-portable programming method and apparatus based on unified optimization primitives. The method includes: receiving unified source code, which contains computational logic code and hardware optimization primitives, wherein the hardware optimization primitives are attached to the computational logic code in the form of independent declarative primitive chains, and the hardware optimization primitives are used to describe optimization strategies adapted to different hardware architectures; parsing the unified source code to construct an extended AST, which contains semantic information of the computational logic code and optimization instructions of the hardware optimization primitives; and generating target source code adapted to the target hardware platform based on the extended AST. This disclosure achieves the goal of "write once, execute efficiently on multiple platforms" by decoupling the computational logic code from the hardware optimization primitives, not only realizing cross-platform performance portability but also significantly reducing the complexity and development cost of manual optimization for heterogeneous hardware.

The application relates to the technical field of adaptive filtering, and discloses a deviation compensation affine projection adaptive filter based on a logical distance criterion, wherein a desired signal vector, a noisy input signal vector and a matrix, and an adaptive weight vector are respectively constructed according to a desired signal sample value, a noisy input signal sample value and an adaptive weight value; an anti-impulse error square median is obtained according to an estimated error signal, an error signal variance estimation value is calculated, and an input noise variance estimation value is calculated in combination with a noisy input signal vector and an adaptive weight vector variance power estimation value; a derivative and a second derivative of a logical distance function are calculated according to an estimated error signal vector; a deviation compensation matrix is calculated by using the second derivative of the logical distance function, an intermediate matrix, a vector formed by a diagonal line of the intermediate matrix, a deviation compensation vector and the input noise variance estimation value, the adaptive weight value is updated, and adaptive filtering of a system output signal to be estimated is realized.

This invention relates to the field of electronic digital data processing technology, specifically to a method for edge data cleaning of industrial equipment integrated into a Manufacturing Execution System (MES). The method includes: acquiring the MES production configuration database through an edge computing unit to construct a dynamic data object context dictionary, and encapsulating the original sensor data objects into a standardized data stream; extracting statistical moment features and trend indicators within a time-series sliding window, and identifying noise types by combining dictionary logic constraint thresholds and classification models; dynamically matching and repairing algorithms to perform reconstruction based on the identification results, generating high-fidelity data; mapping the data to the overall equipment efficiency calculation logic of the MES, and iterating the noise classification weight parameters based on feedback deviation closed-loop iteration. This invention, through a business-driven dynamic dictionary and multi-dimensional feature classification and identification logic, achieves self-evolution of the cleaning strategy, significantly improving the processing accuracy and real-time performance of industrial data under complex operating conditions.

This invention discloses a personalized discount recommendation system based on big data user consumption and merchant habits, belonging to the field of computer information management technology. It includes a merchant service platform, as well as an operation management component and a data processing component. The operation management component acquires relevant operational data from the merchant service platform, cleans it, and categorizes it into interest-related data, discount-related data, and comprehensive data, which is then sent to the operation identification component. This invention constructs a full-link computational logic from basic user characteristics to discount adaptation and comprehensive decision-making, solving the common problems of inaccurate user profiling and weak scenario adaptation capabilities in traditional recommendation systems. It achieves the core objective of "pushing the right business discounts to the right users at the right time," providing a complete and implementable technical solution for personalized discount recommendations in multiple business scenarios, and promoting the development of this field from extensive push to refined, dynamic, and multi-objective balanced approaches.

This application provides a phased precision transformation method and system for Natural Language Query ABC. The phased precision transformation method includes the following steps: Step A identifies the object classes involved in the user's question based on the object class and relation class definitions of the ontology model layer and constructs an edge list of the object relation subgraph; Step B is executed in two phases sequentially: the first phase reads the business default rules from the CBC constraint pattern of the ontology and merges them with the user's explicit conditions, tracing the source of each condition through condition source annotation; the second phase determines the extraction fields of each object class according to the user's information requirements; Step C determines the calculation logic; and the outputs of Steps A, B, and C are assembled into an OQS intermediate representation using a node-edge separated JSON format. In the above technical solution, a standardized query decomposition method is established, and the probability of query illusion is reduced.