12results about How to "Effective execution" patented technology

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training an action selection system using reinforcement learning techniques. In one aspect, a method includes, at each of a plurality of iterations: obtaining a batch of experiences, each experience tuple including: a first observation, an action, a second observation, and a reward; for each experience tuple, determining a state value for the second observation, including: processing the first observation using a policy neural network to generate an action score for each action in a set of possible actions; sampling a plurality of actions from the set of possible actions according to the action scores; processing the second observation using a Q neural network to generate a Q value for each sampled action; and determining the state value for the second observation; and using the state value to determine an update to current values of Q neural network parameters.

The invention provides a vehicle software upgrading method and device and a computer readable storage medium, whole vehicle software and hardware information of a target vehicle is obtained, a matched software upgrading package is determined by querying a target baseline, and the target baseline is a software configuration standard extracted from a factory bill of materials and determined for an upgrading target. Accurate matching for dynamically screening and deciding a unique correct upgrade package for a specific vehicle from massive software resources is realized, so that the problems of upgrade failure, function abnormality or system conflict and the like caused by mismatching of the software package and specific hardware configuration or software environment of the vehicle are fundamentally solved. And finally, the determined software upgrading package is deployed to the target vehicle, a closed loop of the whole upgrading process is completed, it is ensured that the accurately matched decision can be effectively executed, and therefore the success rate and reliability of software upgrading are remarkably improved.

The present disclosure provides an object detection method, an object detection device, an electronic device and a computer readable storage medium, which belong to the technical field of wireless communication. The method comprises: receiving a first echo signal of a first probe signal sent by a first base station to a detected object; analyzing the first echo signal to determine communication awareness information; eliminating offset information in the communication awareness information to determine intermediate data; performing feature decomposition based on the intermediate data to determine a signal transmission posture parameter of a second probe signal sent to the detected object; sending the second probe signal to the detected object according to the signal transmission posture parameter, and receiving a second echo signal of the second probe signal to determine an object detection result of the detected object according to the second echo signal. The present disclosure can accurately and effectively detect the detected object through multiple base stations.

This specification relates to a multi-roller system comprising two or more pairs of rollers, a battery can forming apparatus, and a forming method. The two or more pairs of rollers rotate around independent axes of rotation, each controlling its height variation. The two or more pairs of rollers have different cross-sectional shapes, and each of the two or more pairs of rollers applies pressure close to the battery can at equal intervals.

The invention relates to the technical field of experiment automation, in particular to an experiment execution method and device, computer equipment, a storage medium and a product. The method comprises the steps of obtaining experiment requirements fed back by a user; based on the experiment demand, determining an experiment formula corresponding to the experiment demand according to equipment configuration data of at least one piece of pre-stored candidate equipment; wherein the operation formula comprises target equipment selected from the candidate equipment and operation content corresponding to the target equipment; based on the experimental formula, performing experimental operation according to the target equipment to obtain an experimental result meeting experimental requirements; according to the method, efficient determination of the experiment formula is guaranteed, effective calling of the equipment is guaranteed, the error probability of the experiment equipment is reduced, then, effective execution of the experiment is guaranteed when the experiment operation is carried out according to the target equipment based on the experiment formula, and the success rate of experiment execution is increased.

The invention relates to a double-track business process management system considering infection disaster response business optimization and daily business continuity, a management method and a program, and the double-track business process management system considering infection disaster response business optimization and daily business continuity is characterized by comprising a provision module, a management module and a management module, current situation analysis, business influence analysis and danger analysis of the mechanism are implemented in stages, double-track business continuity process data of the mechanism are exported, and the double-track business continuity process data comprise disaster response business process data and daily business process data.

The application discloses a kind of south typical hilly terrain irrigation area crop soil moisture condition equipment layout method, comprising the following steps: obtaining south hilly terrain irrigation area remote sensing image, carry out semi-supervised classification in combination with field sampling results, interpret and form irrigation area crop structure distribution map A;According to irrigation area crop structure distribution map A, adopt field investigation and handling, form fine irrigation area crop structure distribution map B;South hilly terrain irrigation area crop is subdivided, and the distribution of each partition typical crop is counted;According to block crop planting importance order, according to primary classification, successively guarantee block crop soil moisture condition monitoring station site layout;According to soil moisture condition equipment layout distribution diagram, site location that meets layout requirement is found;Field detection station site communication signal, guarantee data transmission;The application focuses on soil moisture condition equipment layout work, forms differentiated results in the field of soil moisture condition site selection and planning, provides certain reference value for newly-built irrigation area soil moisture condition site selection.

The application provides a refrigerator. The refrigerator comprises an inner container, a mounting piece, a vacuum drawer, a magnetic field device and an electric connecting piece. The vacuum drawer comprises a drawer body and a cover. The cover and the drawer body are movably connected. The magnetic field device is arranged on the outer periphery of the vacuum drawer and is used for generating a magnetic field. The electric connecting piece comprises a power supply part and an electric connecting part. The power supply part is located in the storage chamber. The power supply part is arranged on the side of the mounting piece away from the opening of the inner container. The power supply part is used for electrically connecting with an external power supply. Part of the electric connecting part is located in the mounting cavity. The electric connecting part is arranged on the side of the vacuum drawer away from the opening of the inner container. The electric connecting part is electrically connected with the magnetic field device. The power supply part and the electric connecting part are used for butt joint in the pulling direction of the vacuum drawer. The power supply part and the electric connecting part are butt joint through the pulling of the vacuum drawer, so that the external power supply supplies power to the magnetic field device. The power supply structure is simplified through the butt joint.

The invention relates to the technical field of rare earth element detection, and provides a rare earth element detection method and system, and the method comprises the steps: obtaining the parameters of the size distribution degradation degree of aerosol droplets based on the physical wear degree of an atomizer under the current working condition; according to a preset association rule between the characteristics of the rare earth elements and the signal suppression degree and the parameters of the size distribution degradation degree of the aerosol droplets, determining the signal suppression amount of each to-be-detected rare earth element; according to the signal suppression amount, calculating and applying a corresponding correction weight factor, and correcting the original measurement signal value to obtain a corrected signal value; and using the corrected signal values to calculate the proportion between the pre-specified rare earth elements, and outputting the proportion to complete the detection of the rare earth elements. The method has the effect of improving the accuracy and reliability of rare earth element detection.

The invention relates to an instruction execution method and device, equipment, a storage medium and a program product. The method comprises the following steps: acquiring to-be-executed first function code information of the GPU; under the condition that the verification of the first function code information is passed, constructing a first function corresponding to a first instruction in the first function code information; wherein the first instruction is used for realizing a target function when being executed in a GPU environment, and the first function is used for realizing the target function when being executed in a non-GPU environment; and executing the first function to obtain an execution result of the target function. By adopting the method, effective interpretation and execution of the GPU instruction can be realized in an environment lacking GPU support.

The invention discloses a diffusion model single image generation method and system based on frequency domain guidance, and relates to the technical field of computer vision, and the method comprises the steps: obtaining a single training image, and defining a reference network based on a diffusion model; establishing a composite frequency domain constraint function; establishing an iterative structure refining strategy; and performing end-to-end training on the reference network based on the composite frequency domain constraint function, after the training is completed, performing reverse denoising sampling in combination with the iterative structure refining strategy, and generating a new image similar to the training image in distribution. Through dual guidance of frequency domain constraint in a training stage and structure refining in an image generation stage, on the premise of not changing a reference network core architecture, the structural integrity and visual fidelity of the generated image can be improved.

The invention relates to the technical field of computers, in particular to a task execution method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining a task intention of a to-be-processed task, and generating task state semantic information; in each iterative execution cycle, an execution action is generated based on the task intention and the task state semantic information, a result state is obtained after execution, and the task state semantic information is updated according to the result state; and determining to terminate or enter a next iteration execution cycle according to a result state and a task intention. Compared with the prior art, the execution action is generated based on the task intention and the semantic information of the current task state in each iteration execution period, and the semantic information of the task state is updated according to the execution action, the result state and the task intention, so that the execution action in the next iteration execution period can be dynamically adjusted according to the actual execution condition of the task; the problem that planning and execution are disjointed due to the fact that execution steps are solidified in advance is effectively avoided.