26results about How to "Speed balance" patented technology

The invention provides a target tracking method and system based on multi-scale features and a collinear attention module. The method comprises the following steps: constructing a feature fusion network model; taking the template image and the search area image as input features of a single-residual multi-scale block to obtain output features of a single-residual multi-scale feature block; in the template branch, processing the output features of the single-residual multi-scale feature block of the template branch to obtain final self-attention; pre-training the feature fusion network model in combination with the final self-attention; taking the template image and the search area image as input features of a pre-trained feature fusion network model to obtain new output features of a single-residual multi-scale feature block, and further obtaining a maximum similarity score; and obtaining a prediction result according to the maximum similarity score. According to the method, the advantages of single-residual multi-scale features and collinear constraint attention are fully combined, so that better feature fusion is achieved, redundant calculation is reduced, and the balance between performance and speed is optimized.

The invention provides a programmable AES (Advanced Encryption Standard) encryption module hardware accelerator, an instruction set and an operation method.The accelerator is connected with a bus, carries out data interaction with the bus and carries out AES encryption and decryption on data, the accelerator comprises a master device and a slave device which are connected with the bus, the main equipment is used for receiving and transmitting a data carrying instruction; the slave device communicates with other modules on the bus and receives the data carrying instruction and data of the master device; the slave device is connected with the decoder through an instruction stack and transmits an instruction to the decoder; the decoder decodes the instruction transmitted from the slave device and sends the decoded instruction to the calculation module or the carrying module connected with the decoder for corresponding operation or data transmission; the instruction set is used for controlling the accelerator; the operation method is realized by adopting the accelerator.

The invention provides a segmented cleaning photovoltaic cleaning robot and an application method, and belongs to the technical field of photovoltaic cleaning, the segmented cleaning photovoltaic cleaning robot comprises a walking module, a water supply assembly, a crawling track, a crawling module and a cleaning module; the two walking modules are symmetrically connected to the two ends of the crawling rail, and the water supply assembly is connected to the top end of one walking module. The crawling module is connected to the crawling track; the cleaning module is connected below the crawling module; the crawling module comprises a driving belt wheel, a driven belt wheel and a belt connected with the driving belt wheel and the driven belt wheel, and the belt abuts against the side wall of the crawling rail. The cleaning module is small in size, the gravity needing to be overcome during work is small, and then the power needed by work of the robot is small; as the cleaning module is light in weight, when the robot moves from top to bottom, the speed of the robot can be controlled only through small friction force; the friction force between the driving module and the side face of the crawling support is increased through a belt, the speed of the crawling module is balanced, and the cleaning effect is guaranteed.

The application discloses a physical information cycle network W-VSP velocity inversion method based on an imaging domain depth feature constraint, is applied to the field of seismic data processing and geophysical exploration, and aims at the non-convexity of a target function and extreme sensitivity of data quality of traditional data domain full waveform inversion; first, an RTM imaging network of PIRNN is established, a physical process of sound wave propagation is embedded into a recurrent neural network, forward propagation is realized to obtain simulation records by wave equation forward modeling, and a receiver-depth imaging domain gather is obtained by cross-correlation imaging conditions, and space-time wave field gradient information is recorded; then, a U-Net network is used for trace gather feature identification in the RTM prestack imaging domain, and a curved event skeleton is automatically picked up; then, a hybrid loss function fusing data domain residual and imaging domain flatness is constructed to realize representation of the trace gather feature; finally, space-time field gradient information of wave field propagation recorded by the PIRNN is used for backward propagation, and an adaptive gradient fusion algorithm is introduced to balance macroscopic constraints in the imaging domain and data domain detail updating, so that velocity inversion is realized.

The WPIoU loss function construction method and the YOLOv8 detection method belong to the field of image information detection technology, and solve the problem that the existing YOLOv8n cannot realize accurate detection of traffic images in complex environments. The following steps are included: Step S1, after processing the WIoUv1 loss function with gradient gain, the WIoUv3 loss function is obtained; Step S2, the PIoU loss function customizes an adaptive penalty factor; Step S3, after improving the WIoUv3 loss function described in Step S1 based on the penalty factor described in Step S2, the WPIoU loss function is obtained; Step S4, after further improving the WPIoU loss function described in Step S3 based on the non-monotonic attention function of the PIOUv2 loss function, the final WPIoU loss function is obtained.

The application discloses a kind of based on image data's track area abnormal detection method and device of machine tool, it is related to image recognition technical field.The method includes: using image recognition algorithm, introduce multi-objective optimization algorithm, build machine tool target detection model, and connect to candidate region segmentation engine and space-time correlation filter engine;Video stream of track area is collected, and pretreatment is carried out, and the image data of continuous frame is obtained;The image data of continuous frame is input to candidate region segmentation engine, and the ROI region of each frame of image data is obtained;ROI region is input to machine tool target detection model, and the detection frame of machine tool target is obtained;The detection frame of machine tool target is input to space-time correlation filter engine, identifies stationary effective target, and carries out abnormal detection to stationary effective target.The application solves the problems of low detection accuracy, difficult model deployment and poor anti-interference ability in the prior art.

The application discloses a software research and development security access control implementation method, device and equipment and a storage medium. The method comprises the following steps: the function test and the interface test are divided into multiple short period windows; test traffic is captured at the end of each short period window, and the test traffic is subjected to deduplication processing; the deduplicated test traffic is subjected to playback attack test to obtain a test result; the test result is subjected to access control detection at a first security checkpoint to obtain a first security scanning result; when the detection time of the access control detection exceeds a first preset time and the access control detection is not completed, the test result is subjected to access control detection at a second security checkpoint according to a second preset time to obtain a second security scanning result; and the test result, the first security scanning result and the second security scanning result are reported. According to the method, the application with long security test time does not need to block the process, and efficient cooperation of full-process automatic security test and agile delivery is realized.

The invention discloses a self-adaptive image transmission planning method and system based on image content and downlink transmission bandwidth, and belongs to the technical field of image processing and task planning. The method comprises the following steps: obtaining the size and resolution of an original image; a region-of-interest image and a non-region-of-interest image are generated by extracting a region-of-interest in an original image, and the position of the region-of-interest in the original image is acquired; wherein the resolution of the region-of-interest image is the resolution of the original image, and the resolution of the non-region-of-interest image is generated based on a downlink transmission bandwidth; and transmitting the region-of-interest image, the non-region-of-interest image, the size of the original image and the position of the region-of-interest in the original image to a second device via a downlink, so that the second device recovers the original image. According to the invention, the transmission pressure of the ground station can be effectively reduced, and redundant data are significantly reduced.

A mixed precision inference method, device, equipment, computer program product and computer readable storage medium for a deep neural network are provided. The method divides the deep neural network into functional modules, determines the precision level of each module by switching the precision of each module and comparing the changes in the output indicators, and based on the precision level, conditionally inserts a data type conversion operation on the data flow path to avoid unnecessary overhead. Finally, the calculation is performed according to the level of each module, and the precision conversion is completed according to the conversion operation. The method balances the precision and speed at the module granularity, conditionally inserts the conversion to greatly reduce the type conversion overhead, and guarantees efficient execution and framework compatibility through static planning, thereby maximizing the performance benefits of low-precision calculation while ensuring the output precision.

A crack detection method based on a high-resolution and large-receptive-field convolutional network comprises the following steps: S01, shooting a crack image on a traffic facility by using a camera, labeling by using a pixel-level label, and constructing a crack data set; s02, constructing an HRLRF-Net model based on a high resolution and a large receptive field; s03, training the HRLRF-Net model by using the crack data set; and S04, carrying out crack detection by using the trained HRLRF-Net model. According to the method, multi-scale feature extraction of a large receptive field is realized through an HRLRF-Net model; through a multi-resolution feature extraction and fusion method, the problems of large calculation amount and insufficient feature expression ability of a high-resolution network are solved, and the accuracy and robustness of crack detection are improved. The crack detector is especially suitable for detecting cracks of different sizes, and especially has advantages in detection of ultra-thin and long cracks, so that the crack detector is suitable for structural health detection of wall surfaces, roads and bridges.

The application discloses a ranging method, comprising: obtaining a mapping relationship between serial numbers of actual data boxes and serial numbers of preset data boxes, wherein each actual data box is mapped to at least one preset data box; obtaining ranging data, assigning storage space to the actual data boxes and storing the ranging data; determining a target data box and a serial number of the target data box according to data distribution of the ranging data in the actual data boxes; determining a target value of a preset data box corresponding to the target data box according to the mapping relationship between the serial numbers of the actual data boxes and the serial numbers of the preset data boxes and according to a preset algorithm; and determining a distance of a measured target according to the target value of the preset data box and a width of the preset data box. The application also provides a ranging system, which can reduce storage space.

The invention relates to the technical field of cable hot melting control, and discloses a high-voltage cable hot melting control method, a high-voltage cable hot melting control device, high-voltage cable hot melting control equipment and a storage medium. Generating the planned hot melting operation time of each hot melting operation point in the planned execution time table, and generating a hot melting control parameter set of the hot melting operation points by taking the time limit of the hot melting operation task as a limiting condition and taking the highest hot melting operation quality as a target, in this way, the hot melting equipment is driven to execute the hot melting operation task of each hot melting operation point by adjusting the parameters of the integral time constant, the differential time constant or the integral amplitude limiting, and the hot melting operation quality is improved as much as possible while the deadline requirement of the hot melting operation task is met. Meanwhile, by continuously correcting the plan execution time table, the influence of geographical and air temperature environmental factors on the deadline is solved, and the execution rationality and scene adaptability of the hot melting operation task are improved.

The invention discloses a rapid unsupervised lexicon construction method, system and device in the government affair service field and a medium, and the method comprises the steps: cleaning corpora in the government affair service field to obtain a corpus; counting word segments in the corpus; calculating mutual information entropies of all word segments, and adding the word segments with the mutual information entropies higher than a threshold value into the initial word bank; performing word segmentation on the corpus in the government affair service field in the corpus based on the initial lexicon, and storing segmented words in a dictionary; the dictionary is traversed, non-compliant words are deleted, and words with semantic value are reserved to form an unsupervised word bank. According to the method, the accuracy and efficiency of unsupervised word bank construction are improved, and the method has a wide application prospect.

This invention relates to the field of integrated circuits, specifically to a time residual processing circuit, a correlation multiple sampling method, an SS-ADC circuit, and a chip. The time residual processing circuit extracts the residual signal from the quantization result based on the output of the comparator in the SS-ADC and completes the quantization of the time residual. The time residual processing circuit consists of a D flip-flop, an NOR gate, two resistors of the same value, four NMOS transistors, two capacitors of different values, and a comparator. A charging and discharging channel formed by the two capacitors is used to perform a rate-multiplication of the pulse width of the residual signal. The proposed correlation multiple sampling method and its circuit adjust the ramp signal type and the number of repeated samplings based on the light intensity of the input signal during the pixel signal quantization stage, and quantize the time residual of the process to correct the comparator output. This invention solves the problems of insufficient accuracy and high power consumption of the SS-ADC used in existing CIS circuits.

The embodiment of the present invention discloses a method for analyzing the dynamic characteristics of a DC power system based on hybrid time scale modeling, which relates to the technical field of power electronic systems for DC power distribution and can improve the efficiency of digital electromagnetic transient analysis. The present invention includes: establishing an EMT subsystem and a DP subsystem of a DC power system; importing the established EMT subsystem into FPGA1, and importing the established DP subsystem into FPGA2; establishing an interface calculation module according to the external characteristics of the EMT subsystem and the DP subsystem, and deploying it on FPGA3; after initialization, the EMT subsystem performs n single-step analysis and the DP subsystem performs one single-step analysis until the interaction moment is reached; and outputting the analysis results to a terminal device.

This invention discloses a fermentation process for Bacillus subtilis, comprising: adding Bacillus subtilis seed culture to a fermenter containing a culture medium, and real-time collecting the real-time temperature, real-time dissolved oxygen content, and real-time pH of the culture medium; determining a first introduction rate of carbon dioxide gas in response to a real-time pH greater than 7; determining a second introduction rate of air based on the real-time dissolved oxygen content; determining a first introduction temperature of carbon dioxide gas and air based on the real-time temperature, the first introduction rate, and the second introduction rate; determining a first stirring speed of the stirrer based on the first and second introduction rates; introducing the first introduction rate of carbon dioxide gas and the second introduction rate of air into the culture medium after heating or cooling to the first introduction temperature, and controlling the stirrer to stir at the first stirring speed; determining that fermentation is complete when the content of the target product in the culture medium reaches a threshold. This invention simplifies the fermentation environment adjustment steps and reduces resource waste.

The present application discloses a microwave cavity filter system with automatic tuning function, which relates to the field of cavity filters and realizes adaptive control of signal frequency fluctuations through multi-level tuning strategies. The system includes a detection module, a control module, and first, second and third tuning units. The detection module monitors the quality indicators such as frequency fluctuation and signal-to-noise ratio of the signal in real time and generates fluctuation information; the control module selects the corresponding tuning unit according to the fluctuation amplitude based on the fluctuation information. The first tuning unit performs a small compensation, and the second tuning unit expands the bandwidth to accommodate medium fluctuations; when the signal frequency deviation trend is obvious or the signal-to-noise ratio is lower than the threshold, the system uses the third tuning unit to synchronously adjust the center frequency and bandwidth and reset the frequency range. In addition, the system sets the tuning parameters in advance through trend prediction, and dynamically monitors the signal-to-noise ratio during the bandwidth adjustment process, and gradually shrinks the bandwidth to ensure signal quality.

The utility model relates to electric fishing reel technical field especially electric fishing reel's reduction mechanism. Its technical scheme includes: protection shell, the left side of protection shell is provided with motor, the output of motor extends to the inside of protection shell and is provided with driving wheel, the left side inner wall of protection shell is provided with internal tooth ring, the left side inner wall of protection shell is provided with ring plate, the right side of ring plate is provided with a plurality of connecting rods. The utility model discloses the meshing setting of driving wheel and from the runner, the meshing setting of from the runner and internal tooth ring, can carry out the deceleration to the output of protection shell, makes the rotating speed of the rotating plate to drive the rotating rod to be in the appropriate rotation, thereby can reach the balance of the rotating speed and the pulling force of the winding cylinder through the reduction of rotating speed, the increase of torque, satisfy the actual demand of fishing scene, through the setting of pulley one to the wire block, it is convenient to arrange the fishing line body, and it is convenient to make the fishing line body winding uniform.

The present invention provides a binocular stereo matching method based on an attention cascade cost body neural network, belonging to the field of image processing technology. The present invention uses a binocular stereo data set for training and establishes an efficient stereo matching model. Among them, multi-scale feature extraction and cost body construction enrich the receptive field and improve the processing effect on repeated structures, weak textures or transparent objects and occluded areas. At the same time, the present invention introduces and improves the attention cascade cost body and cost body fusion method to reduce the redundant information in the cost body and the amount of network calculation, and maintains the balance of the amount of calculation, the amount of parameters and the speed.

This application relates to the field of electronic atomization equipment technology, specifically an atomizing core assembly, including an atomizing tube and a heating element disposed within the atomizing tube. The atomizing tube has a first liquid inlet. The atomizing core assembly also includes a first tube body, which is sleeved outside the atomizing tube, with a first gap forming between the first tube body and the atomizing tube. This application also relates to an atomizer and an aerosol generating device. The technical solution provided by this application can generate a siphon effect on the aerosol matrix after the heating element consumes the aerosol matrix. This ensures that the supply rate of the aerosol matrix to the heating element and the atomization rate of the aerosol matrix by the heating element remain balanced, resulting in a stable taste experience for the user. Simultaneously, the aerosol matrix in the liquid cup can be completely absorbed through siphoning, eliminating the need to add aerosol matrix when the liquid level drops to a specific position, preventing the heating element from dry burning and the core from clogging, and extending the service life of the atomizing core assembly.

The invention provides a method and a system for adjusting carrier frequency points, carrier bandwidth and the like. Performing multi-stage compression optimization on the original model through deep learning pruning or knowledge distillation to obtain a model after multi-stage compression optimization; and the like. The sensor calibration method and system have the capability of being deployed in the sensor terminal device, the base station and the cloud server respectively, application in multiple types of sensor devices and multiple scenes of the sensor devices is achieved, and simple and efficient inspection of calibration results is achieved through multi-stage cooperative calibration.

The utility model relates to an electric drive axle which is used for an electric truck, the electric drive axle is provided with a main drive motor, a speed reducer assembly and a differential mechanism assembly, the speed reducer assembly is provided with a first-level speed reduction level, a first-gear transmission mechanism, a second-gear transmission mechanism, a third-gear transmission mechanism, a first gear shifting mechanism and a second gear shifting mechanism, the first speed reduction stage is composed of a first speed reduction driving gear and a first speed reduction driven gear which are meshed with each other, the first speed reduction driving gear is fixedly supported on a motor shaft, and the first-gear transmission mechanism is composed of a planetary gear reducer, a first driving gear and a first driven gear. The second-gear transmission mechanism is composed of a planet carrier of the planetary gear reducer, a first driving gear and a first driven gear, and the third-gear transmission mechanism is composed of a second driving gear and a second driven gear, the torque of the main drive motor is transmitted to the differential mechanism assembly through the first-stage speed reduction stage and the first-gear transmission mechanism or the second-gear transmission mechanism or the third-gear transmission mechanism.

This invention discloses a method and system for analyzing the operation of a CAD pathfinding algorithm connection program, belonging to the field of program operation analysis technology. The method includes: constructing a static complexity scoring function; optimizing the initial working parameters of the CAD pathfinding algorithm connection program based on the static complexity score of the target executable file; constructing a runtime pressure scoring function; determining the first execution path of the CAD pathfinding algorithm connection program; optimizing and adjusting the working parameters of the CAD pathfinding algorithm connection program according to a preset parameter tuning order; and determining the second execution path of the CAD pathfinding algorithm connection program. This invention enables the pathfinding connection program to maintain a stable and efficient execution state under drawings of varying complexity, improving the accuracy and drawing efficiency of wiring results, effectively balancing system resource consumption and program response speed, and ensuring smoother and more reliable connection operations in complex design tasks.

The application claims a feature-enhanced lightweight network FGNet face expression recognition method, belonging to the technical field of pattern recognition. The method comprises the following steps: firstly, aiming at the problem that the deep convolutional neural network is too complex, a lightweight network model FGNet based on the Keras neural network framework is designed, and channel slicing and channel shuffling are added to the backbone; secondly, a lightweight and efficient attention module is designed to extract different scale features and output rich multi-scale information by using the channel and spatial feature extraction capability of the network attention module, so as to improve the performance of the network model; then, aiming at the problem of uneven sample categories of the face expression database, an adaptive class weight is proposed, and is weighted to the self-defined loss function, so as to further improve the accuracy of each expression recognition; finally, a face expression recognition system platform is designed and demonstrated, and end-to-end face expression recognition is realized.

This invention provides an artificial intelligence algorithm acceleration system and method for real-time junction temperature prediction of IGBTs, including a parameter acquisition module, an artificial intelligence algorithm acceleration module, and a host computer. The parameter acquisition module is used to acquire the saturated on-state collector-emitter voltage VCE(Sat), collector current IC, and substrate temperature TC during IGBT operation. The artificial intelligence algorithm acceleration module is based on a Zynq7000 series FPGA to build a SoC architecture, integrating a dual-core ARM Cortex-A9 processor and a dedicated neural network processor (NPU). The host computer is used to receive the junction temperature prediction results and raw acquired parameters transmitted by the artificial intelligence algorithm acceleration module, enabling real-time data display, historical data storage, junction temperature anomaly alarms, and model parameter calibration. This invention aims to solve the high latency, high communication cost, and data privacy issues of cloud computing solutions through collaborative optimization design of algorithms and hardware, while simultaneously overcoming the bottlenecks in computing power, energy efficiency, and accuracy of existing edge computing platforms.

The invention discloses an automatic fritillaria identification system and method based on hierarchical deep learning, and relates to the technical field of intelligent detection of traditional Chinese medicinal materials. The system comprises a hardware integration and image acquisition module, an image processing and target positioning module and a grading identification algorithm module. Wherein ConvNeXt-is adopted for the category pairs of the Songbei and the Pingbei, which are relatively low in similarity; carrying out classification and identification through a Tiny network; the method comprises the following steps of: performing classification identification by adopting a double-branch ConvNeXt model aiming at furnace shell and illite type pairs with relatively high similarity, respectively extracting local microscopic texture and overall macroscopic morphological characteristics by adopting texture branches and morphological branches which are constructed in parallel, and performing optimization by adopting a joint loss function fusing tuple marginal penalty terms so as to enhance the inter-class discrimination capability. According to the invention, full-process automation from sample feeding to identification result output is realized, the identification precision is high, the speed is high, and the method is suitable for rapid nondestructive batch detection in the circulation link of traditional Chinese medicinal materials.