7results about How to "Improve distribution speed" patented technology

The invention discloses a P2P (Peer-to-Peer)-based efficient extensible federated learning system construction method, equipment and a product. Firstly, a parameter server selects a client based on statistical efficiency, processing speed, distribution contribution and other historical information, and a client distributor assigns a special actuator to the selected client. The actuator operates in an isolated container environment, carries out real communication through a special network card, and distributes a global model by using a P2P communication library; the actuator sends the training configuration, the local data set and the global model to the computing node, and the computing node performs a plurality of rounds of local training and then returns model update to the actuator; the parameter server receives a model update of the actuator and generates a new global model. According to the method, model distribution is optimized through P2P communication, and a client selection scheme comprehensively considering various historical information is designed. In addition, the FedP2P constructs a high-fidelity simulation platform to accurately simulate calculation and communication behaviors in actual federation training, and then the effectiveness of scheme design is verified.

The invention discloses a full-automatic medicine dispensing machine which comprises a box body, a multi-layer medicine dispensing device and a display device, the multi-layer medicine dispensing device comprises an installation frame, a plurality of containing discs, a plurality of limiting assemblies and a plurality of buckles, and the installation frames are arranged in the box body; the plurality of placing trays are arranged on the mounting frame at equal intervals; the multiple limiting assemblies are movably arranged on one sides of the multiple containing discs correspondingly. The plurality of buckles are respectively inserted into the bottoms of the plurality of placing trays; the display device comprises a display assembly, and the display assembly is arranged on one side of the box body. Therefore, medicine distribution is greatly accelerated through automatic operation, it is guaranteed that patients obtain medicine in time, accurate placement and distribution are guaranteed through PLC accurate control and sensing monitoring, reliability is improved, operation management is facilitated through a friendly user interface, and service quality and patient satisfaction are improved.

The present disclosure relates to a memory management method, device, medium and electronic device, the method comprising: obtaining a target object of memory to be currently allocated in a target thread; determining a memory allocation strategy corresponding to the target object according to the references of the target object to other objects, the memory allocation strategy comprising a first strategy and a second strategy, the first strategy indicating that a memory of a corresponding size is allocated to the target object in an allocation buffer of a heap memory, and the second strategy indicating that a memory of a corresponding size is allocated to the target object in the heap memory; according to the memory allocation strategy corresponding to the target object, memory is allocated to the target object in a target allocation buffer corresponding to the target thread, or memory is allocated to the target object in the heap memory, the target allocation buffer being used to allocate memory to multiple objects of the target thread corresponding to the first strategy. The method can improve the speed of memory allocation, thereby improving the efficiency of memory management.

The application provides a high-altitude detection system of a wind speed and direction sensor, wherein the high-altitude detection system is installed on a transport operation platform, is lifted and stabilized on a wind speed and direction sensor operation plane through the platform, and is used for high-altitude real-time detection and / or calibration of the wind speed and direction sensor. The wind speed and direction sensor is usually installed at a high position, such as a wind speed and direction sensor on a catenary support post, which is installed at a position with a height of 4000 mm from a rail surface, and two wind speed and direction sensors are installed and fixed at each position; meanwhile, a railway line is relatively long, and even if real-time detection is carried out on site, it is very time-consuming and laborious. The application utilizes the extensibility of the operation platform at a high altitude and the convenience of long-distance transportation on the ground, meets the high-altitude detection needs of the wind speed and direction sensor, and unexpectedly solves the important practical problem of restricting the on-site detection of the wind speed and direction sensor.

The invention belongs to the technical field of meteorological data transmission, and particularly relates to a numerical weather forecast distribution device and method, and the device comprises a cloud communication module which is used for obtaining short-term and ultra-short-term numerical weather forecast data from a numerical weather forecast center; the equipment end communication module is used for collecting local actually-measured meteorological data; the data processing module is used for collecting and processing short-term and ultra-short-term numerical weather forecast data and actually measured meteorological data; and the data distribution module performs external communication by adopting a standard communication protocol and is used for distributing the processed numerical weather forecast data and the actually measured meteorological data to end equipment. According to the invention, the numerical weather forecast data can be conveniently and flexibly acquired at the user terminal, and the terminal equipment can conveniently carry out applications related to numerical weather forecast.

The utility model relates to a peripheral pump impeller which comprises a wheel rim and a wheel shaft integrally formed in the center of the wheel rim, blades are arranged on the two axial end faces of the wheel rim and extend in the radial direction of the wheel rim, the outer ends of the blades are flush with the outer edge of the wheel rim, and the outer edge of the wheel rim is flush with the outer edge of the wheel rim. And the angle between each blade and the rim deflects towards the rotating direction of the impeller. Aiming at the efficiency and energy consumption problems of the peripheral pump, by changing the shapes of the blades in the impeller and deflecting the angles of the blades towards the rotating directions of the blades, a structure similar to a paddle shape is formed, so that fluid can be guided to enter a flow channel of the peripheral pump more smoothly, the speed distribution of the fluid in the flow channel of the impeller is improved, the hydraulic loss is reduced, and the efficiency is improved; and the energy consumption is reduced.

A task scheduling and data interaction system and method for large-sample simulation includes a three-layer system of "center-region nodes". The center layer is used to batch allocate tasks to the most suitable regions according to the computational characteristics of the tasks. The region layer performs fine-grained allocation based on the real-time resource status of the nodes in the node layer. The computational results generated on different nodes flow to the distributed stream processing of the center layer to complete the pre-deployed analysis job. Data interaction between different regions in the region layer is carried out using the DDS data exchange bus. The flow of computational results to the distributed stream processing of the center layer is carried out using the DDS data exchange bus or a regional data aggregator. The node layer can automatically and seamlessly switch between microsecond-level "shared memory communication" and DDS-based real-time "network communication" depending on whether the models involved in the interactive data are located on the same physical node.