27 results about "Automatic tuning" patented technology

This invention provides an automatic tuner for a loop antenna used in shortwave communication, belonging to the field of shortwave communication technology. This tuner aims to solve the problems of existing automatic tuners for loop antennas, such as difficulty in balancing versatility and high power performance, and poor high-voltage withstand capability. The tuner includes: a matching circuit, which includes couplers for shunting forward and reverse radio frequency signals; a tuning actuator, which includes an oil-immersed variable capacitor and a drive mechanism; and a main control circuit. Based on the signal shunted by the coupler, the main control circuit autonomously calculates the standing wave ratio (VSWR) and generates a control signal to control the matching circuit and tuning actuator independently of the radio transceiver, thus completing automatic tuning. This invention improves voltage withstand and power handling capability by using an oil-immersed variable capacitor, and achieves universal tuning through the autonomous control of the main control circuit, overcoming the shortcomings of existing technologies and offering the advantages of high power, high reliability, and strong versatility.

The utility model provides an automatic tunable middle wave sky tuning system adapting to complex environment, including transmitting system, excitation source, signal switching system, radio frequency sampling system, impedance sampling system, middle wave sky tuning network, middle wave antenna and automatic tuning system. The utility model discloses an automatic tunable middle wave sky tuning system adapting to complex environment reduces the signal attenuation between the machine room and the sky tuning room, reduces the influence (such as electromagnetic interference, temperature and humidity change) of environmental factors to the tuning accuracy simultaneously, adapts to complex environment, and automatically tunes the out-of-tune middle wave sky tuning network under complex environment, realizes the sky tuning system adaptation again, reaches the reliability of middle wave sky tuning system power transmission, and ensures the reliable operation of transmitting system.

This invention discloses an automated tuning method and system for large language model service configurations, belonging to the fields of artificial intelligence and cloud computing technology. The invention includes an offline analysis module for performance profiling of the large language model (LLM) service system under a finite configuration set and for collecting training data; an online search module for exploring and evaluating the performance of different configurations in the configuration space based on a Monte Carlo tree search algorithm; and a simulator module for receiving the configurations generated by the online search module and predicting the performance metrics of the configurations under a target workload based on a logical step simulation model. This invention effectively solves the problems of low search efficiency and poor performance of existing automatic tuning methods in existing large language model service frameworks (such as vLLM, SGLang, etc.) due to the huge number of configuration parameters (reaching a petabyte-scale search space) and the complex dependencies and mutual exclusion relationships between parameters.

This invention discloses an automatic tuning method and system for deep learning compilers based on loop-memory access features and reinforcement learning, belonging to the field of deep learning compiler optimization technology. This invention extracts loop and memory access features from the tensor intermediate representation of the deep learning compiler, fuses the memory access and loop features according to the loop axis, generating a loop-memory access fused feature sequence; and constructs a cost model based on a long short-term memory network, using the extracted loop-memory access fused feature sequence to achieve performance prediction; the performance prediction value is applied to the automatic tuning process of the deep learning compiler, while reinforcement learning guides candidate program selection and search direction adjustment. This invention significantly improves the automatic tuning efficiency of deep learning compilers by accurately extracting core features and optimizing the model structure, while maintaining prediction accuracy and reducing feature processing overhead.

The application discloses a sound automatic tuning method and system based on intelligent sensors, and relates to the technical field of sound control, comprising the following steps: collecting sound field data of each intelligent sensor in a target sound field and sound system tuning parameters, dividing each intelligent sensor into zones according to the response of each channel, and obtaining sound field results corresponding to each zone; comparing the sound field results with contrast results after tuning is completed, calculating the change and determining the change source; classifying and storing the change record, judging continuity and consistency, and generating an adjustment instruction; modifying the gain value, equalization parameter or delay parameter in the corresponding zone according to the adjustment instruction, and verifying the modified sound field results; the method can relatively reduce the dependence on artificial tuning judgment, and improve the continuity, stability and consistency of tuning decision under dynamic sound field conditions.

The application belongs to the technical field of automatic assembly, and discloses an automatic assembly system and method for large nuts of a water turbine ball valve, which comprises a mechanical arm, an end effector and an intelligent nut tightening control system. The end effector is integrated with an electromagnet adsorption mechanism, a servo-driven rotating mechanism, a floating compensation mechanism and a visual recognition mechanism. Different specifications of nuts are adaptively grabbed by the electromagnet, sub-millimeter-level accurate positioning is realized by using a laser galvanometer stereo camera, and the assembly posture is automatically adjusted in combination with the floating compensation mechanism. The intelligent control system adopts a phased tightening strategy, realizes real-time monitoring through a torque sensor, and has a double protection mechanism. The application realizes the full automation of the assembly of large nuts of a water turbine ball valve, has the advantages of high assembly efficiency, accurate positioning precision, strong adaptability, reliable quality and the like, solves the problems of high labor intensity, many safety hazards and poor quality consistency in manual assembly, and significantly improves the assembly quality and efficiency of key components of a hydropower station.

The application discloses an automatic calibration device and an automatic calibration method. The automatic calibration device comprises a verification head, a first driving assembly, a second driving assembly, a detection unit and a marking assembly. The verification head is provided with an internal tooth structure for cooperating with a gear on a crank box. The first driving assembly is connected with the verification head and used for driving the verification head to rotate. The second driving assembly is connected with the first driving assembly and used for driving the first driving assembly and the verification head to reciprocate along a set direction. The detection unit is used for detecting the moving stroke of a moving contact in a circuit breaker. The marking assembly is arranged on the first driving assembly and used for drawing a line at a predetermined position of the crank box when the moving stroke of the moving contact in the circuit breaker reaches a set stroke. The application can improve the detection efficiency and the detection accuracy. Moreover, the marking assembly can draw a line at the predetermined position of the crank box when the moving stroke of the moving contact reaches the set stroke, so that the assembly efficiency is improved, and the reference state of the moving side of the circuit breaker can be accurately marked.

The present application belongs to the technical field of vehicle audio processing, and particularly relates to a song style automatic tuning method based on an artificial intelligence model. The method comprises: obtaining audio to be tuned, vehicle acoustic data and user preference memory, the vehicle acoustic data comprising vehicle structural acoustic parameters and vehicle cabin response data; determining a song style feature vector based on the audio to be tuned, and determining a vehicle acoustic feature vector based on the vehicle acoustic data; generating a target preference memory in response to a user vehicle switching operation; generating a frequency band residual; and determining a target tuning parameter, which is used to control the output of a tuned audio by a vehicle audio processing chain. The present application can take into account the differences in song style, vehicle acoustics and user preferences, and improve the consistency, personalization and cross-vehicle continuity of the tuning result.

This application relates to a unified cross-template matrix multiplication automatic tuning method, apparatus, computer device, and storage medium for heterogeneous accelerators. The method acquires the hardware architecture information of the target heterogeneous accelerator and the geometric dimensions of the target matrix multiplication task. It maps raw tuning parameters from various matrix multiplication template libraries to a unified parameter semantic space, orthogonally decomposing the raw tuning parameters into a shared geometric layer parameter set and a specialized adaptation layer parameter set. A first-stage search is performed in the shared geometric layer parameter set to select a preset number of preferred geometric configurations. For each preferred geometric configuration, a second-stage search is performed in the corresponding specialized adaptation layer parameter set to obtain the corresponding optimal specialized parameter combination. The preferred geometric configurations and the corresponding optimal specialized parameter combinations are fused to generate the target parameter configuration. Based on the target parameter configuration, executable matrix multiplication kernel code adapted to the target heterogeneous accelerator is generated, improving automatic tuning efficiency.

This utility model relates to the field of lectern technology and discloses an automatic adjustment device for speaker presentations, including a radar measuring mechanism and an automatic adjustment control mechanism. The radar measuring mechanism is located on top of the automatic adjustment control mechanism and feeds back the detection results to the automatic adjustment control mechanism. The automatic adjustment control mechanism includes a standing platform, and a hydraulic assembly for controlling the raising and lowering of the standing platform is provided at the bottom of the standing platform. The hydraulic assembly includes a first support system, one side of which is connected to a second support system via a pressure-controlled solenoid valve. A hydraulic pump is provided on one side of the second support system. The pressure-controlled solenoid valve controls the raising and lowering of the first and second support systems by controlling the pressure of the hydraulic pump, thereby controlling the raising and lowering of the standing platform. This utility model can solve the problem in the prior art that the fixed microphone height and random speaker height make it impossible to adapt to the needs of speakers of different heights.

[Problem] To more suitably control the frequency of a carrier signal given to a mixer in a tuning process. [Solution] An automatic tuning device comprises: an output unit that outputs a tuning control voltage for adjusting the frequency of a carrier signal to be given to a mixer; an acquisition unit that acquires a tuning instruction voltage which indicates the intensity of a component of a prescribed frequency in an output signal of the mixer; a setting unit that sets a reference value on the basis of a prescribed range, which includes a first level that is a tuning control voltage corresponding to the maximum tuning instruction voltage, in a first period; a calculation unit that calculates an index value on the basis of the result of scanning of a first range including the first level, in a second period; and a determination unit that determines the output level of a tuning control voltage which is output by the output unit, on the basis of the result of comparing the index value and the reference value.

This utility model discloses a self-adjusting dual-wheel caster for AGVs, belonging to the field of AGV technology. It includes a mounting base with a bracket rotatably connected to its bottom. Casters are symmetrically arranged on both sides of the bracket. A movable groove is formed at the bottom of the bracket, housing an automatic adjustment mechanism for the tire tread to contact the ground. The two casters are respectively mounted on the left and right sides of the self-adjusting mechanism. An auxiliary adjustment mechanism for resetting the self-adjusting mechanism is provided above it. This utility model, through its movable groove and free adjustment mechanism, allows the casters to automatically adjust their angle according to load changes, ensuring the tire tread remains in effective contact with the ground, enhancing grip and driving stability. The auxiliary adjustment mechanism utilizes the elasticity of a spring to ensure the free adjustment mechanism quickly returns to its original position, compensating for the reset delay problem inherent in purely mechanical automatic adjustment. Combining manual intervention with automatic adjustment improves system reliability.

The application discloses a modal separation type MEMS gyroscope frequency difference automatic tuning system. An excitation signal generation module converts drive detection current generated by a drive detection electrode into drive voltage and applies the drive voltage to the drive detection electrode, realizes extraction, storage and generation of a drive frequency, generates a sinusoidal excitation signal with the same frequency as the drive frequency, and makes the detection resonator vibrate forcibly. A frequency difference self-tuning module represents forced motion amplitude of the detection resonator through amplitude-frequency response of the excitation signal, generates a tuning voltage through automatic gain control after difference with an adjustable reference value, feeds back the tuning voltage to the moving mass electrode, and tunes the frequency difference through electrostatic negative stiffness effect. The stable tuning feedback voltage is stored in a tuning voltage generation module, a fixed tuning feedback voltage is generated by a time sequence control unit to act on the moving mass electrode, and the frequency difference self-tuning module is disconnected, so that the frequency difference automatic tuning is completed. The frequency difference dispersion of the MEMS gyroscope in batch production is overcome.

The application provides a method and system for suppressing transient impact of a variable frequency compressor during startup, the method comprising: obtaining a startup condition of the variable frequency compressor; the startup condition comprising: cold startup and hot startup; the startup condition being determined by a variable frequency drive electronic control module, which is used to drive the variable frequency compressor; if the startup condition of the variable frequency compressor is cold startup, obtaining a first stator temperature of the variable frequency compressor; determining a startup parameter of the variable frequency compressor according to the first stator temperature; if the startup condition of the variable frequency compressor is hot startup, obtaining a second stator temperature of the variable frequency compressor; and determining a startup parameter of the variable frequency compressor according to the second stator temperature, so as to solve the problem that a variable frequency refrigerator cannot automatically adjust the positioning current, open-loop drag time and oil return speed running time of the compressor during startup according to the actual temperature of the compressor, thereby reducing the efficiency of the compressor.

The application discloses a multi-specification spring winding forming and pitch self-adaptive adjusting device and relates to the technical field of spring processing equipment. The device comprises a base and an external control module. A winding mechanism is arranged at one end of the base. A feeding mechanism is arranged at the other end of the base and on the feeding side of the winding mechanism. A pitch adjusting mechanism is arranged above the winding mechanism. The device has the advantages that the controller is arranged to control the servo motor to drive the adjusting wheel to move along the winding spindle in the axial direction in real time according to preset pitch parameters and the current winding spindle rotating speed, so that the spring wire is pushed away from the previous circle by a fixed distance when it winds a circle, pitch adjustment is realized, the photoelectric sensor is arranged to pass through the spring after forming, the specification of the wound spring is detected in real time, the moving speed of the adjusting wheel along the winding spindle is automatically adjusted, the self-adaptive adjustment of the spring winding is realized, the device does not need to be stopped for intervention, and the production efficiency and product precision are improved.

The application discloses a kind of high-efficiency automatic tuning methods for dynamic shape model reasoning, the tuning method includes: step one, one or more to be tuned tensor program is normalized and is processed, is converted into the standard form conforming to preset scheduling template;Step two, including standardization tensor program, program feature information and target hardware specification information is passed to tuning engine and optimization is started;Step three, tuning engine receives including the bucketing factor obtained by bucketing strategy and the TopK value indicating the optimal performance of the first K configuration parameters, allocates scheduling template and configuration parameter transmitter, cooperates, and optimizes each barrel corresponding subproblem, obtains the corresponding optimization kernel in each barrel range to complete automatic tuning.The application also discloses a tuning system for implementing the above method, which has a wide range of application scenarios.

This invention relates to the field of oilfield engineering technology, and particularly to an intelligent operation control method and system for pumping units based on negative power time ratio identification. The control method includes setting the operating parameters of the drive motor, starting the pumping unit at a set high speed; reading the set values ​​or calculating the duration of each stroke, statistically analyzing the duration of zero or negative power values, and calculating the proportion of zero or negative power in the stroke time; continuously collecting the proportion of zero or negative power in each stroke time, and determining whether the change in the proportion of zero or negative power in the stroke time over time is less than a cavitation threshold. If cavitation is determined, a speed control phase begins, with the motor running at low speed; after running for a specified duration, the motor controller increases the motor speed to run at high speed again; this process is repeated. The control method provided by this invention automatically adjusts the pumping unit operating parameters in real time, optimizes the pumping unit's operating conditions, extends the pump inspection cycle, reduces energy consumption, and saves manual labor.

The application relates to the technical field of artificial intelligence and reinforcement learning, in particular to a reinforcement learning training parameter automatic optimization system and method based on TensorBoard log driving and large model service, which comprises a data acquisition module, a log analysis module, a large model service module, a reinforcement learning parameter adjustment module, a feedback optimization module, a multi-modal log fusion analysis unit and a dynamic parameter adjustment strategy generation unit. The beneficial effects are that manual intervention is reduced through automatic parameter adjustment, and in the MuJoCo continuous control task, the parameter adjustment time is shortened from 72 hours of the traditional method to 24 hours, and the efficiency is improved by 3 times.

1. The name of the design product: die precision adjusting mechanism. 2. The use of the design product: for automatic adjustment of the lip gap of a slit extrusion die. 3. The design points of the design product: in shape. 4. The picture or photo that best shows the design points: perspective view.

This invention relates to the field of intelligent control, specifically to a motor control method, device, electronic device, and storage medium. Firstly, this application uses fuzzy processing to convert disturbances such as changes in bar stock specifications into fuzzy input quantities. Combined with fuzzy logic reasoning, a control strategy is dynamically generated. Compared to fixed-parameter PID control, this approach can quickly adapt to random disturbances, improving anti-interference capability and response speed. Secondly, a target matrix is ​​constructed using the original control parameter matrix and adjustment data to achieve automatic adjustment of motor parameters, eliminating the need for repeated trials by operators and reducing technical requirements. Finally, the precise target control parameter matrix ensures the motor operates according to the set parameters, stabilizing processing quality while improving the automation level and production efficiency of the production line.

The application discloses a large language model inference parameter automatic tuning method and system, comprising the following steps: a, receiving a user end tuning task to construct a task queue; b, calculating the required total number of GPUs for each experiment of a processing task, screening a target GPU set for each experiment according to GPU storage resource information, if successful, executing step d, otherwise, executing step c; c, judging whether the waiting time of the experiment is less than a task waiting time threshold, if yes, returning to step b after a preset time, otherwise, marking the current experiment as FAILED and executing step e; d, testing the experiment by using an LLM inference performance test tool and calculating a service level index score of the experiment according to a test result; e, releasing the GPU, judging whether the maximum number of experiments is reached, if yes, selecting an optimal experiment according to the service level index scores of all experiments of the current task, otherwise, returning to step b to execute the next experiment of the task. The application can effectively reduce the tuning time.

The application provides a helicopter flight management system radio automatic tuning method and device, the method comprises the following steps: S1, the flight management system acquires helicopter position information, the position information comprises longitude and latitude; S2, the flight management system searches a preset database, and determines whether the helicopter enters the coverage range of a second ground station except a first ground station; the first ground station is a current access ground station; the preset database stores the frequency and coverage range of each radio communication and navigation ground station; S3, if yes, the frequency of the airborne radio communication and navigation equipment is switched to the frequency of the second ground station. According to the current position of the helicopter, the frequency of the radio communication and navigation equipment can be automatically tuned, the operation automation degree in the flight process is greatly improved, and the operation load of the crew is effectively reduced.

This application relates to the field of electric grippers, and particularly to a high-precision self-calibrating small electric gripper, comprising a housing, a movable clamping plate at the lower end of the housing, a driving assembly between the clamping plate and the housing, a vacuum suction cup at the lower center of the housing, and a telescopic assembly between the vacuum suction cup and the housing, the telescopic assembly being mounted on the upper end of the housing. The rollers of this invention are rotatably connected to the clamping plate via a rotating shaft, and are aided by damping pads and rubber rings, enabling automatic adjustment and fit according to workpieces of different shapes, effectively avoiding uneven force and ensuring clamping stability. Driven by a driving source, the movement of the clamping plate can be precisely controlled to achieve accurate clamping of the workpiece. Furthermore, an infrared ranging sensor can accurately measure the upper position of the clamped item, further ensuring operational accuracy and enabling it to better adapt to the operational needs of workpieces of various materials and shapes.