20results about How to "Data support is good" patented technology

The invention relates to the technical field of unmanned aerial vehicle positioning, in particular to an unmanned aerial vehicle operation positioning system based on artificial intelligence, which comprises a flight positioning management and control platform, a flight information unit, a flight matching unit, a positioning matching unit, a yaw adjusting unit and a flight management unit, the linear distance between the target unmanned aerial vehicle and the current ground base station is collected and compared with the working range of the current ground base station, so that the current positioning mode of the target unmanned aerial vehicle is judged, the positioning mode is further analyzed, namely positioning scheme rationality processing and analysis are conducted on the positioning state information, and the positioning accuracy is improved. According to the method, whether a current target unmanned aerial vehicle flight positioning scheme is appropriate and reasonable or not is judged, improvement of target unmanned aerial vehicle flight positioning accuracy and reliability is facilitated, yaw interference matching feedback analysis is carried out on flight environment data in a progressive mode, then the yaw range is reasonably adjusted according to the yaw interference condition, and the target unmanned aerial vehicle flight positioning accuracy and reliability are improved. And data support is provided for follow-up target unmanned aerial vehicle flight positioning correction.

The invention discloses a training method and device for a federal learning decision tree model, and relates to the technical field of big data. A specific embodiment of the method comprises the following steps: acquiring first credit characteristic data and a credit label of each user in a plurality of users; determining a first dissimilarity matrix according to the first credit feature data; receiving a second dissimilarity matrix sent by the data provider; determining global perception sample selection probability distribution according to the first dissimilarity matrix and the second dissimilarity matrix; and then cyclically executing the following processes: training to obtain a decision tree model: determining a user core set of the current node according to global perception sample selection probability distribution, then determining each sub-node after the current node is split according to the user core set, and taking each sub-node as the current node. According to the embodiment, the accuracy of model prediction can be improved, the performance of the model is improved, the optimization efficiency of the model and the training speed of the model are improved, and the training cost of the model is reduced.

The application discloses a kind of natural gas network steady-state model solving method considering slack node and compressor. The method comprises the following steps: step S1, based on the flow conservation equation at each load node and slack node, establish natural gas network steady-state model;Step S2, the Jacobian matrix corresponding to the natural gas network steady-state model is constructed;Step S3, the pressure of each load node and slack node is obtained by iterative calculation to the natural gas network steady-state model;Step S4, based on the node pressure, the injection natural gas flow at reference node and slack node is calculated.The application can be used to analyze the natural network configured with slack node and compressor, and provide a basis for natural gas network planning and design and operation scheme analysis.

The application discloses a traffic abnormal event detection method based on multi-source data fusion, comprising: a single millimeter wave radar and multiple industrial cameras, target detection data is acquired for sensor characteristics; fusion information is obtained by associating multi-source information; a fusion target is tracked, a target trajectory is formed, the trajectory is smoothed and filtered, and a filtered target trajectory is formed; illegal event detection is realized by analyzing the target trajectory, the illegal events include intrusion, overspeed, underspeed, lane changing, illegal parking, reverse driving and line pressing; traffic state event detection is realized by analyzing the correlation between a lane level detection area and target data, the traffic state events include congestion, and event evidence is left for aligned event time, including event video and event photo; event reexamination is carried out for specific events through target trajectory anti-shake verification, event qualitative analysis and reporting are carried out. The application can effectively realize intelligent prevention and control of traffic safety risks, optimize traffic efficiency and reduce the probability of traffic accidents.

The invention discloses a man-machine interaction method and system based on computer vision. The method comprises the steps that student labels are pasted according to student scores and subject types, and score label information is obtained; the subject types are divided into general observation subjects and key observation subjects based on the learning direction of the to-be-detected class, whether the students have violation or not is detected when the to-be-detected class carries out course projection learning of the general observation subjects, if yes, marking is carried out, and when the to-be-detected class is detected to carry out course projection learning of the key observation subjects, marking is carried out. Determining class violation behaviors of key subjects according to the teaching content and the historical class conditions, detecting whether the students violate the class or not based on the class violation behaviors of the key subjects, and marking if the students violate the class; and detecting the learning input degree of the student with the label to obtain a target student learning input degree P, and if the target student learning input degree P does not exceed a preset student classroom input degree threshold, marking the student as a feature student. The method has the effect of improving the classroom learning man-machine interaction efficiency based on computer vision.

The invention discloses a wind turbine generator parallel auxiliary control method, system and device based on data, a medium and a program, and belongs to the technical field of new energy power. The method comprises the steps that a wind turbine generator controller PLC receives data of an external sensor of a wind turbine generator; preprocessing the received sensor data and an algorithm logic variable in a wind turbine generator PLC (Programmable Logic Controller) to obtain preprocessed data; extracting feature data based on the preprocessed data; a recommended set threshold value is calculated according to the feature data, the calculated set threshold value is sent to a wind turbine generator controller PLC, the wind turbine generator controller PLC carries out a new round of calculation according to the threshold value, and a control instruction CO is generated; and the wind turbine generator performs control motion according to the control instruction CO. According to the method, the control parameters and the logic judgment threshold are directly updated in real time through the actual unit operation data, and the occurrence of events of inaccurate control and greater risk caused by mismatching of the actual unit operation and the control parameters is avoided.

The present invention discloses a method and system for cross-regional data aggregation, collection and processing of a digital substation system, the method comprising: constructing a data collection engine; constructing a data resource pool, performing data processing, and uniformly storing and managing data; constructing a unified data access interface, the request interface input parameter data and callback interface return parameter data contained in the interface interaction data are both in JSON format string mode, and permission verification is performed during interaction; according to different data types and function types, four data interaction modes are provided: data query, data subscription (WebSocket, message queue mode) and data reverse push. The present invention focuses on data aggregation, forms a plant-level data base, provides a safe and reliable standardized data call interface, and improves the data interaction efficiency between applications; separates the underlying data from the application, provides convenient data support for other advanced applications, reduces the data integration cost brought by application expansion, and shortens the application construction cycle.

The invention discloses a real-time vibration monitoring method and system for a wind turbine generator, and belongs to the technical field of state monitoring of the wind turbine generator, and the method comprises the steps: collecting an original vibration signal and a rotation speed signal of a key part of a transmission chain of the wind turbine generator; caching the original vibration signal; performing feature analysis on the original vibration signal in combination with the rotating speed signal to obtain an order frequency spectrum feature of the original vibration signal, extracting a time domain feature of the original vibration signal, and returning necessary transmission data and / or selective transmission data based on a multi-state triggering return mechanism; the data to be transmitted comprises order frequency spectrum features and time domain features of the original vibration signals; the selective transmission data comprises an original vibration signal and a rotating speed signal of a key component of a transmission chain of the wind turbine generator; and receiving the necessarily transmitted data and / or the selectively transmitted data, and transmitting the data to the server. According to the invention, the real-time acquisition and conditional data transmission functions of the vibration signals of the transmission chain of the wind turbine generator are realized, so that the technical problem of high storage pressure of data acquired in real time is solved.

The application relates to the technical field of motor rotor temperature detection, in particular to a motor rotor temperature rise test structure. The motor rotor temperature rise test structure comprises a rotating shaft, a plurality of rotor cores, a rotor pressing plate, a connecting tool and a plurality of temperature sensors, the rotating shaft is provided with a first channel, the rotor core is provided with a plurality of axial through magnetic steel grooves and glue overflow grooves, the glue overflow grooves of adjacent rotor cores are partially overlapped in the axial direction to form a second channel, the rotor pressing plate is provided with a third channel arranged along the radial direction, one end of the third channel is communicated with the first channel, and the other end of the third channel is communicated with the second channel, the connecting tool is fixedly connected with the end of the rotating shaft, the connecting tool is provided with a temperature rise recorder, the plurality of temperature sensors are installed on different circumferential positions in the second channel, and the temperature sensors are connected with the temperature rise recorder. The test tool of the application is simple in structure, can accurately obtain the temperature rise of the rotor assembly under the working state without damaging the rotor assembly.

The application discloses a semi-supervised HRRP noise label filtering method based on early learning guidance, comprising the following steps: constructing an HRRP training data set and a deep neural network model; pre-training the deep neural network model; predicting the HRRP training data set by using the pre-trained network model and dividing the HRRP training data set into two data sets according to the confidence; constructing a known label data set and an unknown label data set to form a semi-supervised training data set and performing data enhancement; semi-supervised training the network model by using the enhanced data; generating the label of the unknown label data set by using the semi-supervised trained network model and fully supervised training the network model; and target recognition of the HRRP data by using the fully supervised trained network model. The application uses a generalized cross-entropy loss function in the early learning training stage of the model, proposes a data enhancement scheme suitable for HRRP, can better reduce the influence of noise labels on the model, and has better stability and implementability.

The present invention belongs to the technical field of drug intermediate analysis and detection, and provides an analysis and detection method for a hydrochloric acid cefuroxime intermediate cefuroxime precursor acid. The mobile phase of the analysis and detection method of the present invention is mainly composed of methanol and water, and tetrapentyl ammonium bromide and sodium dihydrogen phosphate are added, which can not only improve the solution stability of cefuroxime precursor acid, but also ensure the accuracy of the analysis and detection method results; and, the peak type, tailing factor, separation degree and theoretical plate number of the main component are significantly improved; and, impurity A, impurity B, impurity C and impurity D all have higher response values ​​and separation effects, and more unknown impurity peaks can be separated and detected. At the same time, the operation process of the analysis and detection method of the present invention is simple and convenient, the running time is short (within 30min), and it is suitable for rapid detection of industrial production; and the equipment requirements are not high, and the brand of the chromatographic column is not limited.

This invention provides a pseudo-color fusion method and system for images against a dark background based on polarization angle characteristics. The method includes: acquiring the outgoing light intensity of the target scene using a polarization camera, calculating the Stokes parameter image data of the target scene, and parsing the polarization information; mapping the synthesized light intensity I, linear polarization degree P, and polarization angle A to the brightness V, saturation S, and hue H in the HSV model, respectively, to obtain normalized synthesized light intensity, linear polarization degree stretching data, polarization angle preprocessing data, and hue matching parameters; processing the characteristic data of polarization angle A using a preset polarization robustness discrimination logic for the current observation point in the target scene to obtain polarization robustness discrimination parameters; judging the state of polarization noise based on the polarization robustness discrimination parameters and performing pseudo-color model mapping; and converting the pseudo-color HSV image into an RGB color representation image for display. This invention solves the technical problem of unsatisfactory pseudo-color fusion results caused by excessively dark background areas where the target is located.

This invention discloses a multi-scale fusion attention-based method and system for recognizing driving styles in commercial vehicles, belonging to the technical field of commercial vehicle driver driving style recognition. The method includes: collecting real-time multi-dimensional vehicle runtime sequence data output from the CAN bus of the target commercial vehicle; performing data preprocessing to obtain structured multi-channel time-series samples; using a multi-scale fusion attention recognition network to extract multi-scale parallel convolutional features from the structured multi-channel time-series samples, obtaining multi-scale time-series features that fuse multi-dimensional time-series information; performing channel attention weighting processing on the multi-scale time-series features that fuse multi-dimensional time-series information to obtain discriminative time-series features that enhance channel information; and classifying the discriminative time-series features that enhance channel information to output the corresponding commercial vehicle driving style classification result. This invention can achieve relatively efficient extraction and classification of driving behavior features, adapting to the complex actual operating conditions of commercial vehicles.

The present application belongs to the technical field of unmanned aerial vehicle search and rescue, and particularly relates to a method for calculating the same-route dual-aircraft cooperative flight interval for improving dual-aircraft search and rescue efficiency. First, the target effective search leg length is calculated according to the aircraft position, the aircraft searchable distance, the aircraft flight speed and the distance length information of the target being captured by two aircrafts, and then the target effective search leg length is converted into the aircraft cooperative interval. Second, the target positioning accuracy is calculated according to the aircraft position, the aircraft direction-finding error and the distance from the aircraft to the target, and then the target positioning accuracy is converted into the aircraft cooperative interval. Finally, the optimal aircraft cooperative interval general model satisfying the positioning accuracy requirement is constructed, and the dual-aircraft cooperative optimal interval is obtained by solving the highest total value of the dual-aircraft cooperative search target. The present application can solve the dual-aircraft cooperative flight optimal interval for different routes, increase the probability of the target being simultaneously searched by the cooperative aircrafts, and improve the dual-aircraft cooperative search target efficiency.

The invention provides a device for cold start of silver-loaded zeolite and a use method of the device, and belongs to the technical field of research on filtering characteristics of gaseous methyl iodide by a nuclear power station based on the silver-loaded zeolite. The problems that an existing cold-state starting scheme is not suitable for the silver-loaded zeolite, and the methyl iodide adsorption characteristic of the silver-loaded zeolite in the cold-state starting stage cannot be researched are solved. The device comprises a carrier gas system, a methyl iodide gas distribution system, a silver-loaded zeolite column assembly, a pipeline carrier gas flow path and a sampling and measuring system, the carrier gas system comprises a nitrogen source and a steam generator, the output end of the nitrogen source is connected with the input end of the steam generator, the methyl iodide gas distribution system comprises a gas cylinder, and the silver-loaded zeolite column assembly comprises a silver-loaded zeolite bed. Mixed gas output by the steam generator and methyl iodide gas output by the gas cylinder are mixed and conveyed to the silver-loaded zeolite column assembly through a main flow path of the pipeline carrier gas flow path, and the silver-loaded zeolite column assembly is connected with the sampling and measuring system. The device is mainly used for testing the adsorption characteristic of the silver-loaded zeolite.

The utility model discloses single crystal furnace main heater calcining equipment which comprises an equipment base, a side plate is integrally formed on the side of the equipment base, a top plate is integrally formed at the upper end of the side plate, six groups of main heaters are positioned in the equipment base, and a device seat is positioned at the bottoms of the main heaters. An observation window is formed in the edge side of the side plate, high-temperature-resistant glass is positioned on the inner side of the observation window, codes are arranged on the equipment base, and a heater supporting mechanism is fixed to the position, where the main heater is positioned, of the equipment base. According to the single crystal furnace main heater calcining equipment, six main heaters can be calcined at the same time, the use effect of the heaters can be verified in time by calcining in advance, the heating state, hot resistance, cold resistance and the like of the heaters can be monitored in time, the state of the heaters can be monitored in real time, the ash content of graphite pieces can be reduced, and the service life of the heaters is prolonged. And therefore, the blow-in ignition probability is greatly reduced.

This invention provides a method, electronic device, and storage medium for office space scene recognition. Multiple IoT devices are installed within the office space. The method includes: acquiring event information from the multiple IoT devices in real time; standardizing the event information of each IoT device into a standard event sequence according to a preset structure, the preset structure including timestamp, device identifier, event type code, event value, and event credibility; aligning each standard event sequence through a time window; constructing a semantic vector for the current moment based on the data in the time-aligned time window; and determining the current spatial state of the office space based on at least some parameter vectors in the current moment's semantic vector, wherein the multiple preset spatial states include at least the following spatial states: vacant, partially occupied, actively occupied, meeting preparation, meeting in progress, meeting adjournment, and nighttime energy saving. This method can improve scene recognition accuracy.

The invention discloses a hospital medical record information storage system with an automatic classification function, and relates to the field of hospital medical record management. Comprising a patient identity information acquisition module 1, a patient personal medical record establishing account 2, a medical record information acquisition module 3, an information classification module 4, a storage module 5, a display terminal 6, a medical record calling module 7, a calling personnel identity authentication unit 8 and a medical record index module 9. The medical record information of the patient is uploaded and stored, the patient is prevented from losing historical medical record data, data support is provided for later treatment, meanwhile, medical staff retrieve the personal medical record information of the patient based on the identity information of the patient, and the medical record information of the patient is retrieved by the medical staff based on the identity information of the patient. The patient needs to be subjected to identity verification to obtain patient case information, the privacy information of the patient is protected, the patient case information is prevented from being leaked, meanwhile, the patient case information is digitized through the system, case data support can be provided for communication, discussion and learning of a hospital, and the data utilization rate is increased.

The present application relates to the field of intelligent driving and intelligent networked electric vehicles, in particular to an ECAV multi-vehicle game lane-changing decision-making method based on inverse reinforcement learning, and the specific steps are as follows: S1, a non-cooperative game model of lane-changing decision-making in a multi-vehicle interaction scene is constructed; S2, the state information and road environment information of the lane-changing initiating vehicle and the surrounding associated vehicles are collected by using the vehicle-mounted sensor to construct a data set of the multi-vehicle interaction lane-changing scene; S3, the method of inverse reinforcement learning is used to construct the HDV revenue function with safety and traffic efficiency as the target; S4, the global optimal strategy is obtained and output; S5, the lane-changing decision-making is completed, the lane-changing decision-making method based on inverse reinforcement learning and game theory uses the IRL algorithm to construct the HDV revenue function, predicts the behavior decision-making of the HDV vehicle in the lane-changing process, adjusts the revenue parameters in the non-cooperative game based on the particle swarm optimization algorithm, and obtains the optimal lane-changing decision-making method.

The invention relates to the technical field of touch detection of touch screens, in particular to a touch screen touch position detection method and a game machine checking system. The method comprises the following steps: acquiring a moving track and a touch duration of a finger, a moving duration of a dragged object in the touch screen, a moving track of the dragged object and a capacitance signal in a process of touching the touch screen by the finger; according to the difference between the touch duration in the process of touching the touch screen each time and the moving duration of a dragged object in the touch screen and the difference between the moving track of the finger and the moving track of the dragged object, obtaining a random factor of touching the touch screen each time; based on the fluctuation characteristics of the capacitance signal, obtaining the finger contact degree of touching the touch screen each time; the response qualified rate is obtained by combining the random factor, the finger contact degree and the scanning frequency for detecting the touch position of the touch screen, and then whether the touch response precision of the touch screen is qualified or not is judged. According to the invention, the accuracy of the touch response precision detection result of the touch screen is improved, and the credibility of the touch position detection result is ensured.