16results about How to "Guaranteed detection stability" patented technology

The invention relates to the technical field of image data processing, and discloses an image data processing method in food surface tiny flaw detection, which comprises the following steps: acquiring an original gray level image as a reference mask matrix; performing morphological corrosion on the matrix by using the structural elements to generate a basic mark matrix; calculating a gradient amplitude matrix of the reference mask matrix; performing weighted superposition on the gradient amplitude to the basic mark matrix to execute gray gain compensation, and generating a self-adaptive mark matrix; by taking the reference mask matrix as a limit and the self-adaptive mark matrix as a seed, executing limited morphological reconstruction to generate a background fitting matrix; according to the method, the problem of frequency domain aliasing of natural textures and tiny flaws is effectively solved through a gain compensation mechanism based on gradient amplitudes, and precise detection of the tiny flaws is achieved while complex texture backgrounds are inhibited.

A spraying device for penetrant testing of rods includes a housing, a placement rack inside the housing, a storage slot on the placement rack, rubber wheels inside the storage slot, a cover plate on the storage slot, an electric push rod on the placement rack, a transverse frame on the housing, a moving slot on the transverse frame, a support frame inside the moving slot, a limit slot on the support frame, a spray can on the limit slot, and a drain port on the housing. The device uses rubber wheels to drive the rotation of the rods and the two-dimensional movement of the spray can, replacing manual rotation of the rods, reducing manual intervention and labor intensity, and is suitable for batch testing. The uniform rotation of the rods and the lateral and longitudinal movement of the spray can ensure that every part of the surface is covered with penetrant, avoiding the omissions of traditional manual spraying and improving the accuracy of the test results. The electric push rod can adjust the position of the placement rack to accommodate rods of different sizes.

A kind of identification calibration method and system suitable for medium voltage distribution network fault travelling wave, for the problems that existing technology is difficult to adapt to complex and varied wave front shape, calibration precision and robustness is insufficient, the present application first adopts medium voltage distribution network travelling wave signal and pre-processes, constructs five types of wave front shape vectors;Then design multi-dimensional classification criterion to divide wave front into five categories, realize online calibration of classification threshold value through threshold value dynamic updating mechanism;Finally, for different types of wave front, corresponding adaptive calibration method is used to complete accurate calibration.The present application can effectively solve the problems of wave front calibration lag, reflection wave misselection, strong noise interference and other problems, adapt to complex working conditions of medium voltage distribution network, and have excellent engineering landing performance.

The invention discloses a lightweight point cloud defect identification method and system for portable sensing equipment. Comprising the following steps: acquiring an original point cloud containing a scanning motion time sequence attribute, and constructing a space-time mixed neighborhood; the neighborhood features are input into a double-flow decoupling neural network, the network extracts geometric and motion features in parallel, and attention weights are generated by using the motion features to suppress motion artifacts in the geometric features; and performing density adaptive enhancement on the features after artifact suppression, and inputting the features into a classifier to identify defects. The method can effectively distinguish motion artifacts and real defects, reduces the false alarm rate, does not need high-computing-power reconstruction, can be deployed in portable equipment for real-time identification, can adapt to non-uniform sparse data, and improves the accuracy and stability of defect identification.

This application relates to the field of power equipment testing technology, and in particular to a method and apparatus for detecting the amount of dissolved gas degassing in oil. The method includes: acquiring oil sample information and operating condition information of the corresponding expandable gas collector; determining the pressure change data of the expandable gas collector based on the oil sample information and operating condition information; and calculating the total amount of dissolved gas degassing in the oil sample based on the pressure change data and a preset mapping relationship. This solves the problems of gas chromatography, which is cumbersome to operate, requires highly skilled personnel, has high instrument and consumable costs, and is time-consuming, making real-time monitoring impossible; and vacuum degassing, which requires strict vacuum levels, has high equipment purchase and maintenance costs, poor degassing effect stability, difficulty in guaranteeing detection accuracy, and is prone to oil sample contamination and complex operation.

The invention discloses an indoor personnel video behavior process identification system based on target detection and an MS-TCN + + algorithm, accurate detection and identification of a personnel behavior process in an indoor free scene are realized through cooperative processing of target detection, behavior triggering and time sequence identification, and the system utilizes adaptive switching of high and low frame intervals to realize accurate identification of the personnel behavior process in an indoor free scene. Details when behaviors occur are fully captured, and the calculation amount in a non-behavior period is remarkably reduced, so that automatic segmentation and positioning of behavior segments in a real-time video stream are realized; through key limb detection and cross-frame integration based on YOLOv7, continuity and detection stability of target tracking are guaranteed, and detection errors caused by shielding, posture changes and multi-person interaction are effectively coped with; and in combination with multi-scale time sequence modeling and inter-frame consistency constraint of the MS-TCN + + model, the system can extract behavior laws in long time sequence data and suppress the influence of noise frames, so that the recognition accuracy and robustness are improved while the real-time performance is ensured.

The utility model discloses a kind of mould processing surface quality detection devices, belong to mould detection technical field.It is equipped with detection table, the side surface of detection table is equipped with vertical rod, vertical rod is supported with drive table, the bottom surface of drive table has cavity, first driving part and second driving part are equipped in cavity, the bottom surface of drive table is equipped with protective box, first driving part drives protective box transverse movement, second driving part drives protective box longitudinal movement, the bottom surface of protective box is equipped with mounting groove, detection probe is embedded in mounting groove, the top surface of detection table has cavity, the inside of cavity is slidably equipped with placing plate, the inside of detection table is equipped with third driving part, third driving part drives placing plate vertical movement;The technical scheme of the present application can change the position of detection probe in horizontal plane and the height of the mold to be detected by the cooperation of the first driving part, the second driving part and the third driving part, so as to obtain clear and comprehensive image information of the mold to be detected, and further ensure the accuracy of the mold detection result.

The application provides an aluminum heat preservation nail bonding strength detection device. The detection device comprises a mounting base, a heat preservation nail, a heat preservation nail clamping assembly and a detection assembly. The mounting base is fixedly connected with a detection base at the top. The heat preservation nail is bonded to the detection base. A resistance increasing assembly is arranged on the heat preservation nail. The resistance increasing assembly comprises a resistance increasing ring, a clamping ring and a sleeve ring. The clamping assembly comprises a sliding frame and two clamping frames slidingly connected to the outer side of the heat preservation nail. The top of each clamping frame is fixedly connected with a first sliding block. The two clamping frames are slidingly connected to the sliding frame through the first sliding blocks at the top. The detection assembly is located at the top of the sliding frame and provides upward pulling force for the sliding frame and the clamping frame. The heat preservation nail is pasted on the detection base, and then a detachable leverage point is arranged on the outer side of the heat preservation nail, so that the clamping ring is conveniently pulled, and the pulling force is conveniently detected, thereby reducing the detection difficulty and increasing the detection convenience.

The utility model relates to a kind of power plant condensate regeneration resin conveying automatic identification device, and four sides of the fixed module setting is set on conveying pipeline, comprising: light source, resin conveying window, receiving device and acquisition module;Light source is set at the side of resin conveying window, for sending the transmission light of specific wavelength to resin conveying window;Receiving device is set at the other side of resin conveying window opposite with the side of light source setting, for receiving the optical signal formed after transmission light and resin interact after passing through resin conveying window, and optical signal is converted into electric signal after processing to optical signal;Acquisition module includes analog-digital conversion unit and signal transmission unit, and analog-digital conversion unit is used to collect current signal and is converted into digital signal;Signal transmission unit is used to transmit digital signal to digital signal analysis circuit;Four fixed modules are set around resin conveying window and install equidistantly with the four sides of resin conveying window. Corresponding system is also disclosed.

The invention discloses an identification and calibration method and system suitable for medium-voltage power distribution network fault traveling waves, and aims to solve the problems that in the prior art, complex and diversified traveling wave head forms are difficult to adapt, and calibration precision and robustness are insufficient, medium-voltage power distribution network traveling wave signals are firstly collected and preprocessed, and five types of wave head form vectors are constructed; designing a multi-dimensional classification criterion to divide wave heads into five classes, and realizing classification threshold online calibration through a threshold dynamic updating mechanism; and finally, aiming at different types of wave heads, adopting corresponding self-adaptive calibration methods to complete accurate calibration. The method can effectively solve the problems of wave head calibration lag, reflected wave mistaken selection, strong noise interference and the like, adapts to the complex working condition of the medium-voltage power distribution network, and has excellent engineering landing performance.

This invention discloses an automatic detection and sorting process for ball track dimensional defects within a ball cage shell, relating to the field of ball track dimensional defect detection. The process includes: acquiring an original contour point cloud dataset; obtaining a robust ball track contour dataset using random sampling circular arc fitting; establishing a spatial radial dimension compensation model to obtain the radial dimension correction value at the bottom of the ball track and the corrected ball track contour coordinates; obtaining the compensation value for the ball track arc radius based on a least-squares circle fitting algorithm; calculating the angular deviation and straightness error of each ball track to generate a multi-dimensional dimensional defect feature vector; and sorting the workpieces into qualified, reworkable, or scrap bins using a pneumatic sorting mechanism based on the comparison results. The advantages of this invention are: by combining anomaly point suppression and clamping posture calibration compensation, the influence of surface interference and clamping errors on measurement accuracy is eliminated, achieving high-precision automatic detection and reliable grading and sorting of ball track dimensions within the ball cage shell.

The utility model discloses a portable electrical equipment detection device, which relates to the field of electrical detection, and comprises a detection device body, the detection device body comprises a shell, a display screen, a detection circuit and an operation button arranged on the shell, the detection circuit is used for detecting various parameters of electrical equipment, the display screen is used for displaying detection results in real time, and the operation button is arranged on the shell. The device further comprises two holding pieces, an angle positioning assembly and a reset elastic unit, the two holding pieces are symmetrically arranged on the back face of the shell, rotating shafts are fixedly installed at the upper ends and the lower ends of the holding pieces, and the angle positioning assembly is arranged at the end of one rotating shaft. According to the portable electrical equipment detection device, the symmetrically-arranged holding pieces are matched with the rotating shaft, the clamping grooves and the angle positioning assembly, multi-angle adjustment is achieved, multiple holding postures such as palm holding, first web hooking or hand back insertion are supported, hand pressure is effectively dispersed, rigid locking is formed after the clamping parts are meshed with the circumferential clamping grooves of the cylindrical positioning pieces, and the clamping parts are clamped in the clamping grooves of the cylindrical positioning pieces. Accidental rotation during operation is avoided, and the detection stability is ensured.

This application provides a testing device for an ultrasonic transducer. The testing device includes a base, a cylinder, multiple testing modules, and a pressure assembly. The cylinder is mounted on the base. Multiple testing channels are connected to the same inner cavity. The first end of the ultrasonic transducer under test is connected to the inner cavity. Multiple testing modules are all connected to the same cylinder and connected to the testing channels. The output end of a cylinder is connected to the pressure assembly, which drives the pressure assembly to press against the second end of the ultrasonic transducer under test, thereby pressurizing the ultrasonic transducer under test. A pressure inlet valve inputs nitrogen under the action of an external nitrogen source. The nitrogen flows in the inner cavity and acts on the ultrasonic transducer under test, so that the ultrasonic transducer under test can be tested under dynamic pressure. It makes full use of the detection of temperature transmitters and pressure transmitters, and can perform multiple tests at different pressures without the need for multiple disassembly and assembly, thus improving the versatility and convenience of the ultrasonic transducer testing device.

The invention relates to the technical field of artificial intelligence and machine learning, and discloses an open environment-oriented new class adaptive adjustment method, which comprises the following steps: 1) when a new sample arrives, calculating an energy score and a consistency score; 2) utilizing the score quantile of a recent sample, and updating an energy and consistency threshold value through index moving average, so that the threshold value drifts adaptively along with the change of data distribution; 3) according to comparison between the sample scores and a threshold value, dividing the sample scores into a strong new class, a weak new class or a known class; 4) performing weighted accumulation on the new class evidence in the window according to time decay, when an accumulated value exceeds a preset threshold, formally judging that a new class appears, and immediately expanding the dimension of an output layer and randomly initializing a newly added weight so as to support the output of the new class; (5) an online updating stage: adopting a relatively low fixed learning rate for a known class, adopting an attenuation learning rate from high to low for a new class, and considering both old knowledge stability and new knowledge rapid absorption; according to the method, the continuous learning and generalization ability in the open environment is remarkably improved.

The application provides a kind of intelligent production line operation rate monitoring system in thin-film solar cell production, which solves the problem of poor glass yield statistics effect, etc., comprising a discharge table body, a transmission assembly is arranged on the discharge table body, a contact detection assembly is arranged between the transmission assembly and the discharge table body, an optical detection assembly opposite to the transmission assembly is installed above the discharge table body, a distribution assembly is arranged at the outlet of the discharge table body, and monitoring assemblies are respectively arranged at the inlet and outlet of the discharge table body. The application has the advantages of stable operation, good detection effect, etc.

The application discloses a kind of gas sensors based on the principle of proton exchange membrane fuel cell of asymmetric MEA structure, including asymmetric membrane electrode assembly, asymmetric membrane electrode assembly includes proton exchange membrane, anode catalytic layer and cathode catalytic layer being compounded in one side of proton exchange membrane and the other side of proton exchange membrane;Anode catalytic layer side away from proton exchange membrane and cathode catalytic layer side away from proton exchange membrane are covered with gas diffusion layer.The gas sensor is unified with porous nano gold as the basis to build anode and cathode differentiation catalytic system, solve the problem of low concentration detection sensitivity of traditional symmetric MEA structure, detection limit is high, while overcoming the defects of existing asymmetric MEA structure baseline current imbalance, large zero drift, high interface impedance, poor stability and complex preparation process, so that sensor has high detection sensitivity, high selectivity, excellent long-term stability and low baseline fluctuation, meet the high-precision, high-reliable detection needs of carbon monoxide and other gases.