16results about How to "Effective balance" patented technology

The invention discloses a voiceprint recognition system attack resisting method and device, equipment and a medium, by constructing a shadow voice sample set and a substitution model, dependence on target speaker data and query capability is reduced, and privacy risk and query overhead are reduced; meanwhile, two types of substitution models are trained based on an information theory method, so that the mobility and stability of the models are improved, and the generalization ability is enhanced; besides, the attacker model adopts a specific architecture and combines an alternate training strategy, so that attack loss and prediction probability difference are optimized, imperceptibility and attack effectiveness are balanced, and tone quality reduction or attack failure is avoided; and finally, through systematic training and optimization, a more uniform evaluation caliber is expected to be provided, and more reliable guidance is provided for engineering landing and risk evaluation.

The invention relates to the technical field of intelligent specimen management methods, in particular to an intelligent specimen management method and system for medical examination, and the method comprises the following steps: building a specimen storage information database; environment parameters of the specimens in the transportation process are updated in real time; based on the updated environmental parameters, intelligently predicting the transportation risk of the specimen by using a progressive risk prediction algorithm, and verifying whether the specimen meets the detection requirements or not; state detection is carried out on the specimens meeting the detection requirements, image information and / or video information of the specimens are / is obtained and used for further inspection of the specimens, intelligent assessment can be carried out on the transportation risk of the specimens through a progressive risk prediction algorithm, multi-level risk assessment is achieved, and the risk assessment efficiency is improved. Particularly, through superposition and complementation of a self-adaptive nonlinear risk function, risk curvature analysis, entropy weight optimization and other methods, the algorithm still keeps high adaptability and foresight under different environment changes.

The utility model relates to the technical field of shoe sole components, in particular to a 3D printing foot arch supporting structure which comprises a shoe sole and a supporting component, and the supporting component is installed at the foot arch of the shoe sole. Wherein the supporting component comprises frames and a lattice structure, the lattice structure is formed by stacking a plurality of hollow polygonal parts, the sides of the lattice structure are connected with the frames, the frames are installed on the two sides of the arch of the shoe sole, and the lattice structure is installed on the plane of the arch of the shoe sole. The supporting component is arranged on the arch portion of the sole, the frame on the supporting component abuts against the two sides of the arch, the inner side and the outer side of the sole can be effectively balanced, imbalance caused by uneven stress of the inner side and the outer side is prevented, the sole is better stabilized, and the torsion resistance is achieved; the weight can be effectively reduced, enough arch support can be provided, sweat accumulation can be reduced through the holes of the lattice structure, the wearing comfort is improved, and the practicability is improved.

The invention provides a water supply pipe network pressure measuring point optimization method based on structured deep clustering, and relates to the technical field of pipe network pressure monitoring, the method obtains high-dimensional feature representation of original data through an SDCNs model, removes redundant information in the original data, then fuses high-dimensional features and structural features in the data, and finally optimizes the pressure measuring points of the water supply pipe network. The SDCNs learn the structural features of the data while learning the high-dimensional features of the data, so that the SDCNs can obtain a good clustering effect on the pressure monitoring point arrangement problem, initial populations of a traditional GA algorithm and a PSO algorithm are optimized based on the good clustering effect, a better pressure monitoring point arrangement scheme is obtained, and the pressure monitoring point arrangement efficiency is improved. The SDCNs can be effectively applied to the arrangement scheme of the pressure monitoring points of the water supply pipe network, and the space positions and the number of the monitoring points are optimized, so that the leakage monitoring capability of the pipe network is improved, and the balance of economy and effectiveness of a monitoring system is realized.

The invention provides a continuous embedding method of an increasing knowledge graph based on gradient decomposition, which comprises the following steps of: dividing a time sequence knowledge graph into a series of snapshot sub-graphs according to a time sequence to form a continuously increasing knowledge graph; dividing the overall loss of the embedding model into two different losses on an existing triple and an increment triple, and converting a continuous embedding task into a constraint optimization problem; according to the constraint optimization problem, constructing a continuous embedding algorithm suitable for the embedding model; inputting each snapshot sub-graph of the training set into the embedding model in sequence, and carrying out incremental training by adopting a continuous embedding algorithm; and inputting the test set into the trained embedding model to obtain the final average accuracy of link prediction. According to the method, a loss function of a knowledge graph embedding model is converted into a constraint optimization problem, and a new method is provided for promoting increment representation learning of increasing the knowledge graph; according to the invention, the learning efficiency and the updating effectiveness can be balanced; and the method has strong continuous embedding capability.

The application relates to a kind of unmanned aerial vehicle dynamic path planning methods based on hybrid intelligent optimization, introduce multi-stage hybrid intelligent optimization algorithm, on the basis of static path planning based on particle swarm optimization, consider the dynamic target condition of real-time change, when unmanned aerial vehicle gradually approaches target and needs to carry out dynamic path adjustment, utilize A* algorithm mechanism to give more accurate tracking path planning scheme, so that unmanned aerial vehicle can dynamically adjust path, effectively coordinate and complete task in overall task execution. The application comprehensively considers various complex multi-constraint conditions and coordination relationship, solves the multi-unmanned aerial vehicle dynamic target tracking path planning problem, so that unmanned aerial vehicle can quickly respond when finding out the latest position of target, dynamically and real-timely adjust the track of unmanned aerial vehicle.

The application provides a compound containing a fluorene group and an organic electroluminescent device thereof. The compound of the application contains at least one or more deuterium atoms in the structure, has higher electron mobility, and can effectively enhance the electron transport capacity of the organic electroluminescent device. The compound of the application can be applied to a light-emitting layer and an electron transport region, can effectively reduce the driving voltage of the organic electroluminescent device, improve the light-emitting efficiency, and can significantly increase the service life of the device, thereby enhancing the durability of the device. It can be widely applied to the field of information display technology, such as mobile phones, tablet computers, televisions, wearable devices, VR, vehicle displays and tail lights, etc.

The utility model discloses a river bank wave wall structure, relates to the field of river channel wave walls, and aims to solve the problems of small greening area and poor functionality of the conventional wave wall, a protective wall is arranged, an outer side wall extending out of a river bank retaining wall is formed, and the outer side wall is connected with an inner side wall extending into a bank, so that a planting area is formed in a surrounding manner, and the greening area is enlarged. Meanwhile, the height difference between the outer side wall and the inner side wall is utilized, the wave-resistant effect of the outer side wall is achieved, the inner side wall is provided with a stool serving as a rest area, the function of the wave-resistant wall structure is increased, and the requirements of parks, greening facilities and other facilities for the rest area are met.

The invention provides a hoisting device convenient to maintain, and relates to the field of hoisting devices, the hoisting device convenient to maintain is characterized in that the hoisting device comprises a mounting box, a three-stage gear assembly and a brake disc, the appearance of the mounting box is a cuboid, a cavity is formed in the mounting box, an upper hook is fixed at the upper end of the mounting box, and an electrical assembly is fixedly connected with the mounting box; the three-stage gear assembly is located in the mounting box, a chain connecting base is arranged on the side, close to the electrical assembly, of the three-stage gear assembly and is in transmission connection with the three-stage gear assembly through a gear set, a chain is connected into the chain connecting base in a wound mode, and a lower hook is fixed to the lower end of the chain. A gear box is arranged on the side, away from the electrical assembly, of the third-stage gear assembly, one end of the gear box is provided with a bearing and a rotating shaft and is in transmission connection with the third-stage gear assembly, and the two ends of the connecting shaft are connected with the gear box and the rotating shaft of the motor respectively. The device has the advantages that the device is convenient to maintain and use, and meanwhile, the cargo transporting and lifting effect is improved.

The present disclosure provides a screw compressor and an air conditioner. The screw compressor includes: a compressor housing, having a compression part installation space, a drive part installation space, a compressor inlet and a compressor outlet; a compression part, which is arranged in the compression part installation space, and is used to compress the gas at the compressor inlet and output it from the compressor outlet, including a first compression unit and a second compression unit; and a drive part, which is arranged in the drive part installation space, including a motor rotor and a motor stator which is relatively fixed to the compressor housing and relatively rotatably sleeved on the outer periphery of the motor rotor; wherein the motor rotor is fixedly sleeved on the rotating shaft of the first male rotor, and the rotating shaft of the second male rotor is coaxially and integrally arranged with the rotating shaft of the first male rotor to form a single rotating shaft. The axial forces of the two male rotors of the screw compressor can be effectively balanced directly through the single rotating shaft, and the transmission loss can be effectively reduced, thereby improving the energy efficiency and reliability of the screw compressor.

The application provides a LoRa modulation method based on three-dimensional constellation assistance, and belongs to the field of signal processing. By dividing the information bit stream into a frequency index bit stream and a constellation mapping bit stream, and respectively using them to generate a three-dimensional orthogonal LoRa base signal and perform three-dimensional constellation mapping, a new three-dimensional signal space is constructed for joint modulation. This core design enables more information bits to be transmitted simultaneously within a symbol period, significantly improving the data rate of the system. More importantly, compared with the traditional two-dimensional constellation, the three-dimensional constellation can provide a larger minimum Euclidean distance between symbols under the same modulation order. This fundamental improvement in geometric characteristics directly translates into stronger anti-interference capability and lower bit error rate of the signal in noise and fading channels. Therefore, the method effectively improves the spectral efficiency of the LoRa system while enhancing its transmission reliability, achieving a synergistic improvement in performance.

The invention relates to the technical field of sports equipment, in particular to novel noctilucent sports equipment and facility, and solves the problems that the existing sandbag for running training is loaded by filling sandstone or bulk materials, the load balance cannot be ensured, the running load training effect is poor, training in the dark and at night is inconvenient, and the training cost is high. The running weight-bearing equipment comprises an equipment balance facility and running weight-bearing equipment, the running weight-bearing equipment is installed on the upper end face of the equipment balance facility, and the running weight-bearing equipment comprises a balance weight installation mechanism, a friction adjusting mechanism and a noctilucent protection mechanism. And a friction adjusting mechanism is mounted on the inner side of the bottom end of the counterweight mounting mechanism. According to the weight-bearing running training device, the sandbag is modified into the weight-bearing running device, the load weight and the weight-bearing balance can be adjusted according to needs, the weight-bearing running training effect is effectively improved, double light supplementing operation can be conveniently conducted through noctilucence and vibration light emitting, the training atmosphere is improved, and collision of athletes is avoided.

This invention provides a multimodal fusion system and method based on neural networks. The fusion method specifically includes: (1) frequency filtering, which introduces a frequency filtering module before multimodal signal fusion to suppress the effects of high-frequency noise or texture artifacts in multimodal data that vary depending on the modality, thereby enabling the attention layer to focus on semantically meaningful content; (2) multimodal cross-attention fusion, which extends the early multimodal attention fusion mechanism by integrating cross-modal attention mechanisms and fine-grained hierarchical attention mechanisms through a cross-modal cross-attention fusion module to support more powerful cross-modal feature exchange. This invention can effectively improve the robustness of multimodal object detection under various datasets and sensor conditions. By combining frequency domain denoising with cross-modal perceptual attention, it can reduce noise from each modality and promote the effective exchange of complementary features between different modalities.

This utility model relates to a lifting mechanism for a welding robot, belonging to the technical field of welding robot technology. It includes a welding robotic arm and a base plate. A lifting structure and a force-bearing structure are provided on the top of the base plate. The lifting structure includes an electric slide fixedly connected to the top of the base plate. A support platform is fixedly connected to the movable shaft of the electric slide, and a control console is fixedly connected to the top of the support platform. The force-bearing structure includes a limiting frame fixedly connected to the right side of the electric slide. This lifting mechanism for the welding robot, through the action of the electric slide and the support platform, drives the control console and the welding robotic arm to move up and down. This not only achieves flexible adjustment of the welding height but also improves the adaptability and precision of welding operations. Adjusting the threaded rod adjusts the height of the force-bearing plate, and the limiting frame effectively limits the limiting plate, ensuring positioning accuracy during the lifting process and avoiding operational errors caused by mechanical shaking or offset.

The invention discloses an organic electroluminescent compound which is used as a bipolar host material to be applied to a luminescent layer of an OLED (Organic Light Emitting Diode) device, and can effectively balance holes and electrons and form a relatively wide current carrier composite region at the same time, so that the luminous efficiency of the device is greatly improved.