8results about How to "Quick culling" patented technology

The utility model discloses a rhizome kind of traditional chinese medicine decoction piece spectrum detection grading system, including the body carrier, the upper end of body carrier is provided with the conveying structure, the upper end of conveying structure is provided with material guide structure, the upper end of conveying structure is provided with visual sorting system, and the bottom end of conveying structure is provided with the storage box side away from material guide structure side in parallel, and conveying structure includes the positioning frame fixed setting in the upper end of body carrier, and the inside transmission of positioning frame is provided with the conveyer belt. The utility model provides a rhizome kind of traditional chinese medicine decoction piece spectrum detection grading system, is placed in the top of conveyer belt through the conveyer belt after setting visual sorting system cooperation conveying structure, traditional chinese medicine slice can realize the continuous conveyance to traditional chinese medicine, and the spectrum detection of traditional chinese medicine can be carried out to the cooperation spectrum detection appearance of conveying, and if the medicine control electric push rod pushes the push plate and can be removed to the medicine fast if mouldy, so this can realize the accurate sorting of traditional chinese medicine.

This invention provides a backdoor defense method for federated learning based on gradient consistency and dual unsupervised clustering, comprising: obtaining gradient updates submitted by multiple clients participating in federated learning in the current training round; for each client, predicting the gradient update of the current round based on historical global model parameters and the client's historical gradient updates; calculating the difference between the actual gradient update and the predicted gradient update of each client to generate a malicious score for the client; adaptively determining the optimal number of clusters based on the malicious scores of all clients using the Gap statistic; performing dual unsupervised clustering detection with a coarse-to-fine approach based on the optimal number of clusters: first, removing clients with high malicious scores using K-means clustering, and then removing residual malicious clients from the remaining clients using DBSCAN clustering, thus selecting benign clients; and performing federated learning global model aggregation based on the gradient updates of the selected benign clients.

The utility model discloses a paper die cutting technical field's a kind of defective product paper automatic rejection die cutting device, to solve the problem of inconvenient rejection of defective product generated in the paper die cutting process in prior art. It includes paper feeding mechanism, cutting mechanism and material receiving mechanism, at least one group of sensors is provided between paper feeding mechanism and cutting mechanism, both sides of material receiving mechanism are equipped with mounting bracket, feed shaft is rotatably arranged between two mounting brackets, adhesive tape roll is installed on feed shaft, the outside of adhesive tape roll is adhesive surface, feed shaft can move in the vertical direction of mounting bracket, material receiving shaft is also installed between two mounting brackets, and material receiving roll is installed on material receiving shaft;The utility model is used to detect the position of defective product in paper die cutting work, and automatically rejects corresponding paper according to the position of defective product paper, since adaptive automation equipment, convenient operation, can avoid secondary damage, with excellent application prospect.

This invention discloses a method and system for testing the phase noise of a crystal oscillator. The method includes: acquiring continuous phase noise time-series data after the crystal oscillator is powered on, monitoring the noise reduction during the preheating process, and forming an initial convergence curve; iteratively approximating the initial convergence curve to construct a noise convergence model; discretizing the noise convergence model, calculating the first derivative sequence, and using continuous intervals with derivative values ​​close to zero as a candidate set of potential inflection point positions; performing threshold judgment on the candidate set of potential inflection point positions to obtain confirmed inflection point positions; extracting subsequent steady-state data segments from the confirmed inflection point positions for compliance verification; filtering and energy integration processing on the steady-state data segments to obtain a crystal oscillator verification queue; and determining whether the crystal oscillator verification queue meets a preset stability standard. If it does, the final phase noise performance result is obtained. The method of this invention can accurately determine phase noise under complex operating conditions.

This invention relates to a non-contact liquid level detection method and system, belonging to the field of industrial detection technology. The method includes: receiving a ranging command; simultaneously triggering a millimeter-wave radar sensor and an ultrasonic sensor to measure the liquid level in the same container, obtaining raw echo signals respectively; performing signal processing and liquid level calculation on the raw echo signal from the millimeter-wave radar sensor to obtain a first liquid level value from the millimeter-wave radar sensor, and performing signal processing and liquid level calculation on the raw echo signal from the ultrasonic sensor to obtain a second liquid level value from the ultrasonic sensor; and determining the liquid level in the container based on the first and second liquid level values. This method achieves rapid and efficient removal of invalid data, preventing contamination from flowing into subsequent data systems, and solves the problem that a single sensor cannot identify invalid data in complex engineering environments in the liquid industry.

This invention proposes an efficient K-short path calculation method and system for dynamic road networks, relating to the field of path planning technology. It involves obtaining multiple OD pairs from a directed graph of the road network, constructing a list, and initializing it. For each OD pair, a shortest path is calculated, and a set of deviation paths is constructed based on the shortest path. The shortest path and the set of deviation paths are stored as an initial K-short path set. When the edge weights of the road network structure are adjusted, potentially affected OD pairs are screened based on the changed edge attributes and the initial K-short path set. The initial K-short paths from the potentially affected OD pairs are obtained, and a substantial impact test is performed: if an initial K-short path does not contain any changed edges, and its path cost is less than the total cost of potential new paths constructed from the changed edges, then the path is retained; otherwise, the path is deleted. New paths are supplemented based on the set of deviation paths until K shortest paths are satisfied, thereby significantly reducing computational complexity and improving the efficiency and accuracy of path updates in dynamic scenarios.

The application discloses a kind of aluminum electrolytic capacitor nail flower visual detection and sorting method, to solve the problems of low efficiency, vulnerable to metal reflection interference and unable to effectively identify the inherent defects of nail flower in existing detection technology.The method comprises the following steps: system calibration and parameter presetting step;Through composite light source synchronous illumination and image acquisition step;Image preprocessing and nail flower region positioning step;Double-flow parallel analysis and feature extraction step, one flow is measured by sub-pixel technology Geometric dimension, the other flow is identified by convolutional neural network model Deep defect;Weighted decision fusion and comprehensive determination step, this step is based on the type and confidence of defect identification Dynamic selection of size tolerance threshold for fusion determination;And sorting execution and data management step.The application overcomes reflection by composite lighting scheme, combined with accurate size measurement and intelligent defect identification, significantly improves production quality and product reliability.

The application relates to the technical field of flow detection, and discloses a malicious flow detection method based on feature rules, which extracts malicious flow detection feature rules in a piece-by-piece mode for format checking and extracts relevant rule option information; constructs a rule divide-and-conquer chain according to the relevant rule option information, constructs a rule grouping for rules with the same divide-and-conquer chain path; selects part of rule options of each rule according to the rule grouping, and constructs a multi-mode intensive graph corresponding to the grouping; constructs a precise detection chain for each rule according to the multi-mode intensive graph; the detection performance of a preposed malicious flow detection engine based on feature rules is improved, the problem that the detection effect of a network security product is poor due to the insufficient performance of the preposed detection engine based on feature rules is solved, and the effects of discovering suspicious flow in a timely manner and removing normal flow more quickly are achieved.