12results about How to "Adaptive adjustment" patented technology

The invention discloses a Transform large model training method based on cloud edge collaboration. The method comprises the following steps: S1, constructing a data set; s2, preprocessing the data set; s3, segmenting the preprocessed data set by using a data parallelism strategy, and performing distributed training by combining data parallelism and model parallelism technologies; s4, monitoring computing resources and network conditions of the edge equipment and the cloud server in real time, dynamically adjusting a data distribution strategy, and optimizing computing load distribution; s5, a multi-agent dominant actor-commentator algorithm is combined with a multi-arm tiger machine strategy, and the layered structure of the Transform model is adjusted in real time; s6, performing partial model reasoning tasks on the edge equipment in combination with the edge reasoning capability; s7, protecting data privacy by adopting a federal learning strategy and utilizing a differential privacy technology; and S8, performing hardware acceleration by utilizing heterogeneous computing resources. According to the method, a cloud edge cooperative computing and intelligent scheduling method is adopted, and large model training and reasoning with efficient privacy protection are realized.

The invention discloses a machine tool machining parameter optimization method and system based on digital twinning, and relates to the technical field of machine tool machining. The machine tool machining parameter optimization method based on digital twinning comprises the following six steps: generating a digital twinning model of a numerical control machine tool through a digital twinning technology, optimizing model parameters of the digital twinning model by adopting a parameter identification algorithm, establishing an initial BP neural network prediction model, and optimizing machining parameters of the numerical control machine tool according to the initial BP neural network prediction model. The genetic algorithm BP neural network prediction model is adopted for optimization and training, and the machining parameters of the working condition of the numerical control machine tool can be adaptively adjusted, so that the machining precision and the adaptability of the whole numerical control machine tool are improved. According to the machine tool machining parameter optimization system based on digital twinning, the control parameters of the numerical control machine tool can be adjusted by considering the real-time machining state according to actual requirements, and the flexibility of part machining control and the part machining effect can be improved.

The invention discloses a system and a method for voice mouth shape action matching of a dialogue robot, and relates to the technical field of voice matching, the method adopts a Bluetooth dual-channel technology, a time point array is transmitted to an Arduino mainboard of the dialogue robot through a first Bluetooth channel, a voice file is transmitted to sound equipment of the dialogue robot through a second Bluetooth channel, and the sound equipment of the dialogue robot is used for voice mouth shape action matching of the dialogue robot. High synchronization of voice playing and mechanical actions and synchronous transmission of voice files and time point arrays are achieved, close association between voice playing and mouth-shaped actions is ensured, and action lag caused by transmission delay or untimely hardware response is avoided. And each time point in the time point array can be accurately read and compared through a timer arranged in the Arduino mainboard, so that the steering engine can immediately respond at each set time point, and the opening and closing actions of the mouth of the conversation robot are driven. And when the time point arrives, an instruction is sent to control opening and closing of the mouth shape, so that the voice and the mouth shape action are highly consistent.

The invention relates to the technical field of pile foundation construction, in particular to a pile foundation air bag pressure sectional type inflation device and a construction method thereof.The device comprises columns installed at the bottom and the side of a pile foundation and pressure air bags arranged at the bottoms of the columns; the inflation main pipe penetrates through the top of the column body and is connected with the pressed air bag through a pressure supporting structure; the sectional type inflation structures are arranged in the side wall of the column body side by side and connected between the pressed air bag and the inflation main pipe, and the problems that in the prior art, the dynamic pressure problem faced by a pile foundation in the construction process is not effectively solved, and under most conditions, existing supporting measures cannot adapt to the change of soil layer stress in real time, and the construction cost is lowered are solved. The problems that the bearing capacity of a pile foundation is reduced due to the fact that the existing air inflation device is used for inflating the pile foundation, and when a traditional air inflation device deals with complex pressure changes, the traditional air inflation device can only achieve an overall air inflation state and lacks adaptive adjusting capacity for local stress changes, so that the pile foundation is prone to being affected by uneven stress, and the structure failure risk is increased are solved.

The present application discloses a movable cleaning device for the top surface of an I-beam for a steel structure, which belongs to the field of cleaning devices and is used for cleaning. The movable roller is arranged on a movable wheel frame, which is located on a longitudinal frame plate, and the longitudinal frame plate slides relatively with the transverse frame plate; the driving guide rod is hinged with a movable driving plate; the driving connecting rod is hinged with the telescopic end of the movable driving device; the movable driving device is located on the top plate, which is fixedly connected with the transverse frame plate, and a frame articulation seat is also fixedly connected on the top plate, and roller frames are respectively hinged with the two ends of the frame articulation seat through the frame articulation shaft, and the roller frame is connected with an electric brush roller at the end away from the top plate, and the roller frame is connected with a swing rod linkage rod at the top position of the top plate, and the swing rod linkage rod is hinged with the telescopic end of the swing rod driving device, and the bottom end of the swing rod driving device is hinged on the top plate. The present application can realize automatic cleaning of the top and bottom surfaces of the I-beam, and ensure cleaning efficiency and quality, and reduce the labor intensity of manual cleaning.

The invention relates to the technical field of adhesive tape laminating, and provides an adhesive tape laminating machine which comprises a bottom plate, a cutting unit and a rear plate are arranged at the top of the bottom plate, a laminating unit and a driving unit are arranged on one side of the rear plate, a conveying groove is formed in the bottom plate, and a conveying belt mechanism is arranged in the conveying groove; through use of a mechanical arm, a first roller, a first clamping plate and a second clamping plate, the first clamping plate and the second clamping plate slide in a first sliding groove of the mechanical arm, the first roller slides in a second sliding groove of the mechanical arm, and the distances, extending out of the mechanical arm, of the first clamping plate and the second clamping plate are adjusted to adapt to adhesive tapes with different widths; and through use of an elastic piece IV in the driving unit, a stabilizing rod can float up and down in a stabilizing ring, so that the attaching unit floats up and down, and self-adaptive adjustment can be performed on part of uneven surfaces.

A method for sharing offset between convolutional neurons by a lightweight deformable module belongs to the technical field of deep learning, and comprises the following steps: S1, designing a lightweight deformable module TDCN: introducing an offset field representation method, reducing the channel dimension of an offset domain from 2N to 2, N being the number of sampling points, no matter how large a convolution kernel is selected, and the channel dimension of the offset domain is reduced from 2N to 2N, and no matter how large the convolution kernel is selected, the channel dimension of the offset domain is reduced from 2N to 2N; the channel number of the offset domain is always 2; sharing offset through adjacent convolution operations, so that neurons of the convolution operations learn a proper offset, and establishing association between the neurons; s2, the TDCN is integrated into a Convnet system, and various different backbone networks are used; s3, the TDCN is integrated into the Faster R-CNN system, and various different backbone networks are used; and S4, the TDCN is integrated into the RetinaNet system, and various different backbone networks are used. According to the method, the problem of traditional deformable convolution is solved, meanwhile, the capacity of adaptively adjusting the convolution kernel to adapt to the geometric change of the object is reserved, and wider applicability is obtained.

The invention discloses a high-flow glue pressure stabilizing valve which comprises a base provided with a liquid inlet and a liquid outlet, a liquid inlet channel is arranged in the base, the lower end of the liquid inlet channel with the upper end communicated with the liquid inlet is communicated with a lower cavity, a plug is arranged on a second mounting seat located in the lower cavity and located on the side, close to the liquid inlet channel, of the second mounting seat, and a pressure stabilizing valve is arranged on the plug. A spring in a compressed state is vertically arranged between the other side of the second mounting seat and the supporting seat, a mounting hole communicated with the lower cavity is formed in the lower surface of the supporting seat, and the lower end of the spring is embedded into the mounting hole and is in extrusion contact with a threaded plug mounted in the mounting hole; the plug is in sealing fit with an outlet in the lower end of the liquid inlet channel. The pressure of fluid at the liquid outlet after pressure reduction and pressure stabilization can be adjusted, the requirement for fluid pressure adjustment is met, and the consistency and stability of the fluid pressure at the liquid outlet are met.

The application discloses a basalt fiber drawing adjusting method, system, device and medium, relates to the basalt fiber drawing control technical field, and comprises the following steps: obtaining the melt viscosity in the casting furnace; judging whether the melt viscosity is within the range of the preset viscosity threshold value, if yes, returning to obtaining the melt viscosity in the casting furnace, if not, continuously obtaining multiple drawing state images of the basalt fiber in unit time; according to the multiple drawing state images, identifying whether there is a drawing abnormal feature, if not, returning to obtaining the melt viscosity in the casting furnace, if yes, obtaining a first temperature adjusting parameter; wherein the first temperature adjusting parameter is used for adjusting the melt temperature in the casting furnace; the first temperature adjusting parameter is input into a viscosity-temperature initial curve of a drawing process to obtain a viscosity-temperature target curve, and the application has the advantages of self-adaptive adjustment and improved adjusting efficiency.

The present invention discloses a Transformer large model training method based on cloud-edge collaboration, comprising the following steps: S1, constructing a data set; S2, preprocessing the data set; S3, using a data parallel strategy to segment the preprocessed data set, and combining data parallelism and model parallelism technology to perform distributed training; S4, real-time monitoring of the computing resources and network conditions of edge devices and cloud servers, dynamically adjusting the data allocation strategy, and optimizing the computing load distribution; S5, using a multi-agent advantage actor-critic algorithm combined with a multi-armed bandit strategy to adjust the hierarchical structure of the Transformer model in real time; S6, combining edge reasoning capabilities on edge devices to perform partial model reasoning tasks; S7, adopting a federated learning strategy, and using differential privacy technology to protect data privacy; S8, using heterogeneous computing resources for hardware acceleration. The present invention adopts cloud-edge collaborative computing and intelligent scheduling methods to achieve large model training and reasoning with efficient privacy protection.

The application discloses a spiral propelling vehicle optimal rotating speed self-adaptive adjustment method, system and device, and belongs to the spiral propelling vehicle field. Considering the actual operation condition of the spiral propelling vehicle, the rotating speed of the motor is continuously increased from a low rotating speed, current information is collected, and the rotating speed of the motor is comprehensively judged to be increased or decreased according to the influence of the resistance and the slip ratio of the spiral propelling vehicle, the rotating speed of the motor is self-adaptively adjusted after the optimal rotating speed range is found, and the rotating speed of the motor is controlled to be in the optimal rotating speed range. The optimal rotating speed which makes the influence of the resistance and the slip ratio on the spiral propelling vehicle running to be minimum is found, the spiral propelling vehicle is kept to stably work in the optimal rotating speed range, and the rotating speed of the spiral propelling vehicle is self-adaptively adjusted according to the actual working condition.

The invention provides a bandwidth adjustment method and device of a special line, electronic equipment and a storage medium, and relates to the technical field of special line management. The method comprises the following steps: acquiring private line flow data and service flow data of a service system at the current moment; wherein the private line flow data comprises a private line bandwidth usage amount of the service system at the current moment, and the service flow data comprises historical private line bandwidth occupation amounts of a plurality of services executed by the service system at the current moment in a set historical time range adjacent to the current moment; based on a plurality of historical dedicated line bandwidth occupancy amounts included in the service flow data, predicting a total dedicated line bandwidth occupancy amount of the plurality of services at a next moment adjacent to the current moment; and determining a special line bandwidth adjustment strategy based on the total special line bandwidth occupation amount and the special line bandwidth usage amount included in the special line flow data, and adjusting a special line bandwidth threshold value of the service system based on the special line bandwidth adjustment strategy. In this way, the bandwidth threshold value of the private line can be adaptively adjusted.