22results about How to "Reduce verification costs" patented technology

This application provides a method and apparatus for testing the inference consistency rate of a large language model, relating to the field of artificial intelligence technology. The method includes: sending preset test cases to an inference server, causing the inference server to execute baseline inference and comparison inference respectively, obtaining baseline inference results and comparison inference results; standardizing the baseline inference results and comparison inference results; generating inference precision and inference hit rate based on the standardized comparison inference results and standardized baseline inference results; and calculating the inference consistency rate of the large language model based on the inference precision and inference hit rate. The method and apparatus for testing the inference consistency rate of a large language model provided in this application achieve quantifiable, tolerant, and comparable evaluation of inference consistency, significantly improving the accuracy, interpretability, and adaptability of inference consistency verification.

The disclosure provides a design method for a connection sheet, a connection sheet, an energy storage device, and an electric device. The design method for the connection sheet includes: an area S of a welding region, connectable to a tab, of the connection sheet is obtained; the connection sheet is divided into a first effective region and a second effective region, where the first effective region is a region, having a regular shape, of the connection sheet, the second effective region is a region, connectable to a terminal post, of the connection sheet, the first effective region (10) is implemented as N first effective regions (10), N≥1, and multiple first effective regions (10) are connected on the same side of the second effective region (20); a gauge size of the first effective region is determined according to S= a*S1, a being in a range of 0.095~0.96, and S1 being an area of the first effective region. Thus, an area of the welding region, connectable to a tab, of the connection sheet satisfies the overcurrent requirement, and gauge size of the first effective region is reasonable, so that design redundancy may be avoided, verification times in a design process is reduced. Moreover, the gauge size of the connection sheet is reasonable, so that the production cost may be reduced.

The application discloses a kind of encapsulation glue selection methods and encapsulation structure for crystal screen film, and the selection method is established by the multi-dimensional performance index system with adhesion strength, light transmittance, weather resistance, yellowing index and modulus as core, and finally the encapsulation glue selection is completed by threshold preliminary screening, interface compatibility accelerated aging test, encapsulation interface stress simulation analysis through weighted quantitative comprehensive score;The encapsulation glue prepared by the selection method is used in the encapsulation structure of crystal screen film, including the layer-by-layer lamination of crystal screen film layer, encapsulation glue layer and cover layer.The application realizes the systematization and quantifiable selection of encapsulation glue for crystal screen film, solves the pain points of insufficient long-term reliability, interface failure and optical performance degradation caused by the experience selection of prior art, and can significantly improve the structural stability and service life of crystal screen film product, and is suitable for the encapsulation scene of crystal screen film of various high-end display devices.

This invention relates to the field of flowmeter testing technology and discloses a fault prediction method for trade metering flowmeters. The method includes: acquiring historical calibration data of the flowmeter under preset operating conditions; obtaining a valid flowmeter coefficient dataset after outlier removal; calculating the arithmetic mean and standard deviation of the dataset to establish an initial statistical control chart including a centerline, warning limits, and fault limits; introducing an exponentially weighted moving average algorithm to generate EWMA dynamic control limits; and real-time acquisition of the flowmeter's K coefficient, comparing it simultaneously with the initial statistical control chart and the EWMA dynamic control limits, and determining the flowmeter's operating status based on the different control intervals in which the K coefficient falls. This invention solves the problem of traditional methods being insensitive to small or gradual drifts in flowmeter metering performance by constructing a triple control logic combining static control limits and dynamic trend limits, thus achieving early warning and predictive maintenance before errors exceed limits.

The invention discloses a trust management method and device for heterogeneous devices in an edge computing network, a storage medium and a product, and the method comprises the steps: jointly submitting feature information such as a public key generated by a device unique identifier and a first timestamp to a block chain network to generate a feature record of the device, and when the device accesses the edge computing network, sending the feature information to the block chain network; carrying out legality verification by comparing equipment features with the feature records; equipment in the edge computing network communicates through the TCP message, and a signature and a second timestamp of the equipment are stored in a reserved option of the TCP message; and when it is monitored that the equipment executes a high-risk operation, a log is generated, a log full text and a Merkel tree are stored locally, a root node hash value and a key path node hash value are stored to the block chain network, and log auditing and verification are performed according to the hash values in the block chain network and the Merkel tree, so that trust management of the equipment is realized. According to the invention, the heterogeneous barrier of equipment is broken through, the authentication overhead is reduced, and decentralized trust management is realized.

This application relates to the field of Internet television technology, and provides a method, device, medium, and product for handling tariff disputes based on time slices. The method includes: responding to a tariff dispute handling request, acquiring multi-source business event records associated with a target object, constructing a standard event sequence based on the multi-source business event records, and generating billing time slices based on time nodes in the standard event sequence; determining the billing receivable data under each billing time slice based on the billing strategy corresponding to each billing time slice; comparing the billing receivable data and billed data under each billing time slice to determine the billing difference and the cause of the billing difference; and determining the tariff dispute handling method based on the billing difference and the cause of the billing difference. The time-slice-based tariff dispute handling method provided by this application can effectively reduce verification costs and achieve interpretable accounting of billing differences.

This invention discloses a smart contract execution method in data asset transactions, belonging to the field of smart contract execution technology. The method includes: based on multiple transaction nodes triggered during smart contract execution, a reputation value is obtained based on the historical transaction execution of each transaction node for hierarchical selection, resulting in a verification layer and an execution layer, each containing multiple transaction nodes; based on each transaction node in the verification layer, signature validity verification is performed, and the validity verification pass rate is compared with a first threshold; in response to a validity verification pass rate greater than or equal to the first threshold, based on the multiple transaction nodes in the execution layer, some transaction nodes are selected for cross-validation, and the cross-validation pass rate is compared with a second threshold; if the cross-validation pass rate is less than the second threshold, cross-validation is performed on all transaction nodes in the execution layer for transaction execution judgment; if the cross-validation pass rate is greater than or equal to the second threshold, the transaction is executed.

The application discloses a source code data coupling correctness analysis method and system based on low-layer requirements, evaluates the consistency of data coupling relations between low-layer requirement entries, and evaluates the correctness of source code data coupling based on low-layer requirements. The technical scheme is as follows: based on the traceability relation between source code and low-layer requirements, the data coupling relation based on low-layer requirements is quickly and reversely generated by using the result of source code data coupling analysis, which can be used as the criterion for the correctness of the data coupling relation between low-layer requirement entries, and in turn, can be used as the judgment benchmark for whether the source code data coupling relation conforms to the low-layer requirements, thereby bridging the correctness of the data coupling relation between source code and high-layer requirements. Meanwhile, the software defects, such as the missed low-layer requirement functions in software implementation and the extra introduced redundant codes, can be found.

The application discloses a kind of intelligent contract execution methods in data asset transaction, belong to intelligent contract execution technical field, comprising: according to the multiple transaction nodes triggered when intelligent contract execution, according to the credit value of each transaction node historical transaction execution judgment, to carry out hierarchical selection, obtain verification layer and execution layer, verification layer and execution layer all include multiple transaction nodes;Based on each transaction node of verification layer, the legality verification of signature is executed, and the legality verification pass rate is compared with the first threshold value;In response to the legality verification pass rate greater than or equal to the first threshold value, based on the multiple transaction nodes of execution layer, extract part of transaction node to execute cross-validation, and the cross-validation pass rate is compared with the second threshold value;If cross-validation pass rate is less than the second threshold value, then based on the cross-validation of all transaction nodes of execution layer, for transaction execution judgment;If cross-validation pass rate is greater than or equal to the second threshold value, then transaction execution.

This invention discloses a piecewise adaptive Kalman track tracking method and apparatus based on residual classification, relating to the field of radar track tracking technology. Addressing the problems of poor adaptability to non-stationary measurement noise environments, parameter dependence on human experience, and complex FPGA implementation in existing track tracking methods, this invention proposes introducing a residual classification decision and piecewise adaptive noise adjustment mechanism into the Kalman filter recursive framework. First, a residual vector is calculated based on the predicted state and measurement data. Then, a classification decision is made using the residual vector, the predicted covariance, and the measurement noise covariance of the previous cycle, outputting the measurement quality level. Next, the measurement noise covariance is updated piecewise adaptively based on the measurement quality level. Finally, the filter gain is calculated based on the updated measurement noise covariance and the predicted covariance to complete the correction and update of the state and covariance.

This invention discloses a simulation method for behavior analysis and intelligent risk early warning based on multimodal information fusion. First, a virtual 3D scene with precise physical properties and programmably controllable environmental parameters is constructed. Second, multimodal raw data streams are synchronously generated within the scene and processed using a physical degradation model based on set environmental parameters to generate data highly consistent with the real world. Third, the degraded data stream is fed into the intelligent system under test (SUT) in real time and automatically compared with the precise ground truth labels generated by the simulation, constructing a closed-loop simulation and performance evaluation system. This method can deeply verify the effectiveness of modules such as environmental perception, dynamic weight adjustment, and intelligent fusion decision-making within the SUT, and supports large-scale automated testing and parameter optimization, thereby systematically improving the system's environmental adaptability, recognition accuracy, and decision robustness under harsh conditions such as low light and high noise.

The application discloses a kind of chip PAD automation checking method and system, belong to integrated circuit verification technical field.The application is by constructing general test framework including configuration parameter and entity pin information, unified analysis and extraction are carried out to bottom data using automation script, to generate automatically interfacing bottom verification tool's intermediate test file.Subsequently, dynamic simulation or static formal analysis is executed based on the intermediate file by bottom verification tool, and the comparison of verification result is automatically completed.The application is compatible with two major application scenarios of conventional PAD function check and complex pin multiplexing connectivity check, and replaces the cumbersome process of traditional manual test case writing and manual path tracking with standardized data-driven flow, solves the problems of low efficiency and susceptibility to human error caused by traditional manual test case writing or check file, improves the efficiency and accuracy of verification, and significantly shortens the chip development cycle.

The invention discloses a diagnosis test method and device, computer equipment and a storage medium, and the method comprises the steps: setting a target virtual response model corresponding to a virtual vehicle communication interface module and a target control module in a diagnosis test system, and obtaining a diagnosis test instruction for the target control module when the diagnosis test instruction for the target control module is obtained; a diagnosis test instruction can be sent to a corresponding target virtual response model through a virtual vehicle communication interface module, and a target control module is simulated through the target virtual response model to respond to the diagnosis test instruction so as to obtain a corresponding control module operation result; the virtual vehicle communication interface module and the target virtual response model are arranged in the diagnostic test system, so that the diagnostic software is separated from the constraint of physical hardware and a real vehicle environment when carrying out diagnostic test on the electronic control unit, and diversified test scenes can be flexibly constructed according to diagnostic test requirements; the diagnosis test and verification cost is remarkably reduced, and the diagnosis test and verification efficiency of the automobile diagnosis software is greatly improved.

The embodiment of the invention provides a data processing method and device, a storage medium and a program product. The method comprises the steps of determining to-be-verified order information; obtaining a pre-constructed transaction database, and verifying the to-be-verified order information based on the transaction database; wherein the transaction database comprises a plurality of blending flow records, and the plurality of blending flow records are obtained by blending a marketing flow of a payment and liquidation network, user information and a transaction flow of an offline acquirer; the blending flow record represents an association relationship among the transaction data, the commodity information and the user information; in response to verification passing, performing anomaly detection on the to-be-verified order information based on the transaction database; and outputting a visual view including the verification result and the anomaly detection result.

This invention provides a method for evaluating the vibration reduction performance of an elastic ring-type pivot vibration suppression device. The elastic ring-type pivot vibration suppression device includes a coaxial rigid outer ring, an elastic ring, and a bearing outer ring arranged from the outside in. The method includes: obtaining the oil film force F at the design speed; and calculating the oil film radial force F based on the oil film force F. r Oil film tangential force F t According to formula F r / F t The vibration reduction performance value Kc at the design speed is calculated and compared with the vibration reduction performance evaluation index to evaluate the vibration reduction performance of the elastic ring support vibration suppression device at the design speed. The evaluation method designed in this invention can effectively evaluate the vibration reduction effect of the elastic ring support vibration suppression device design at an earlier stage, and can be used to guide the design of the elastic ring support vibration suppression device.

This invention provides a calibration system and method for a spin-free exchange-relaxation atomic magnetometer, belonging to the field of precision measurement technology. The calibration system includes: a magnetic field coil configured to generate a square-wave alternating magnetic field within a preset frequency range; an atomic gas chamber configured to provide a polarizable atomic medium, wherein, when the spin-free exchange-relaxation atomic magnetometer is in an unshielded state, the atomic medium undergoes Larmor precession under the influence of the square-wave alternating magnetic field to modulate the probe light of the spin-free exchange-relaxation atomic magnetometer, obtaining an optical modulation signal; a photodetector configured to convert the optical modulation signal into an electrical signal, wherein the electrical signal is used to characterize the amplitude-frequency characteristics corresponding to the frequency range; and a calibration module configured to perform peak detection on the electrical signal and determine the operating state of the spin-free exchange-relaxation atomic magnetometer based on the peak detection result.

The application discloses a kind of spinning body aircraft component partition aerodynamic load extraction methods, first calculate the full machine surface pressure and shear stress;Then carry out fuselage partition definition;Calculate fuselage partition aerodynamic load;Carry out rudder surface partition definition;Finally calculate rudder surface partition aerodynamic load.The application realizes the fine acquisition of local aerodynamic load of key position of spinning body aircraft, can significantly improve the precision and reliability of local load environment description, to accurately identify weak link of structure, guide optimization design, effectively avoid design redundancy or deficiency.

The utility model relates to the technical field of valve bodies, in particular to a valve body structure and an integrated valve. The valve body structure comprises a main body and a covering piece; the main body is provided with a connecting end; the covering piece can be selectively connected with the connecting end in a welded mode or detachably connected with the connecting end. The valve body structure is applied to the pile-up valve, a welding connection mode or a detachable connection mode can be selected according to requirements, when the detachable connection mode is adopted, verification and adjustment of products in the early stage of mass production can be facilitated, and after the mass production, the welding mode can be adopted, so that the production efficiency is improved, and the production cost is reduced. The efficiency and cost of product verification can be reduced, and meanwhile, the consistency of products in the early stage and after mass production can be improved.

The embodiment of the invention discloses a verification method and device for a vector retrieval chip based on an RRAM, equipment and a medium, and the verification method comprises the steps: inputting a first test vector into a preset RRAM simulation model, and obtaining a plurality of operation results outputted by the RRAM simulation model; for each operation result, if the operation result is not in the preset first interval, sending a first conductivity adjusting instruction to the preset simulation RRAM model, and returning to execute the operation of inputting the first test vector to the preset RRAM simulation model until each operation result output by the preset RRAM simulation model is in the preset target interval; and performing consistency comparison on the conductivity state change condition of the preset RRAM simulation model in the simulation test process and the conductivity control parameter value in the logic control circuit of the to-be-verified chip so as to perform function verification on the logic control circuit. By adopting the technical scheme, the reliability of the verification result is effectively improved.

The application provides an FPGA platform multi-chip interconnection verification system and method. The verification system comprises a plurality of programmable logic devices, each of which is in communication connection with at least one programmable logic device, and each programmable logic device is configured with a simulation logic module for simulating the function of a target chip. The verification system further comprises a host in communication connection with the plurality of programmable logic devices, the host being configured to perform data transmission with the plurality of programmable logic devices at the medium access control layer of the Ethernet through an Ethernet communication interface; and a plurality of inter-chip interconnection modules, each of which is in communication connection with the Ethernet communication interface in the corresponding programmable logic device, and the inter-chip interconnection module is configured to perform data transmission with at least one programmable logic device through the Ethernet communication interface and a serial-parallel transceiver unit of the programmable logic device in which the inter-chip interconnection module is located.

The invention relates to the field of artificial intelligence and automatic planning, in particular to a large-scale planning problem solving method based on general knowledge of a large language model, and the method comprises the steps: analyzing a PDDL domain file and an instance file, and obtaining an initial state, a target state and an action model; the target state is disassembled into sub-targets in a directed two-tuple form, a directed acyclic dependency graph is constructed, topological sorting is carried out, and an ordered sub-target sequence is obtained; recursively traversing a sequence to generate a sub-problem, calling a planner to solve, and optimizing through LLM two modes if the sub-problem fails or is overtime; the mode A selects actions from an executable action sequence to compress a search space, and the mode B predicts an intermediate state to split an intermediate sub-problem; and finally connecting the sub-plans in series in sequence to generate an overall plan. The invention aims to solve the problems that the traditional planning method is high in search overhead, the sub-target sequence is difficult to determine and the LLM output is unreliable.

This invention proposes a multi-operating-point electric propulsion beam control verification system, belonging to the field of aerospace electric propulsion technology. It solves the problem that the electrical coupling relationship between the shield, anode, and accelerator cannot be simultaneously simulated under non-ignition conditions, forcing beam control verification to rely on actual ignition. It includes an anode current AD acquisition module, which acquires the anode power supply current value and transmits it to a processor for processing via an internal communication module; an anode configurable load module, which configures the anode load resistance value based on the anode resistance configuration result generated by the processor, thereby adjusting the anode power supply voltage; a shield configurable load module, which configures the current of the shield load based on the shield current configuration result generated by the processor; an accelerator configurable load module, which configures the current of the acceleration load based on the acceleration current configuration result generated by the processor; and a storage module, which stores the anode power supply current value and each configuration result.