537results about How to "Avoid workload" patented technology

The invention provides a text classification method and device based on transfer learning, and the method comprises the steps: S1, employing an unlabeled text to train a BERT model, and obtaining a pre-trained word representation BERT model; S2, filtering links, forwarding symbols and user name contents in the to-be-classified text; S3, inputting the filtered text into the word representation BERTmodel trained in the step S1 to obtain a semantic file of the text; and S4, inputting the semantic file of the text into a convolutional neural network for processing to obtain a category label of the sentence in the semantic file. According to the method, transfer learning is used for text classification, and a BERT word representation model trained based on large-scale unlabeled corpora is provided. The word representation model has universality, does not depend on the specific text field, and can also be used for other tasks such as entity extraction and sentiment analysis.

An invasion monitor method based on computer vision and used for the field of intelligent video monitor is disclosed; the invention in advance collects video images of a plurality of frames through a monitor camera in order to establish a scene model on the basis of non-parameter statistics. For the new video images, the scene model is used to detect prospect points, and a motion target is obtained by further processing the prospect points through morphologic filter. Afterwards, weighting histogram characteristics are extracted from the motion target and a histogram matching method is applied to tracing the motion target to obtain the motion track of the target. The method judges whether the target track enters into the set warning region in order to decide whether to sends the alarm. The invasion monitor method adopts the computer vision technology and realizes the automatic detection and alarming by utilizing the video information acquired by the computer process monitor camera, therefore the invention is simple in structure and convenient in implementing as well as has the extensive market prospect and application value.

The invention discloses a concentrated multienzyme cleaning solution and a preparation method thereof, wherein the concentrated multienzyme cleaning solution is prepared from the following raw materials in parts by weight: 1 to 5 parts of organic preservative, 200 to 400 parts of surfactant, 20 to 65 parts of enzyme stabilizer, 40 to 80 parts of propylene glycol, 30 to 80 parts of triethanolamine, 10 to 60 parts of sodium sulfite, 0.1 to 5 parts of calcium chloride, 20 to 60 parts of sodium citrate, 10 to 65 parts of citric acid, 0.1 to 1 part of antifoaming agent, 30 to 80 parts of bio-enzyme, and appropriate deionized water; and the method for preparing the concentrated multienzyme cleaning solution comprises the following steps: dissolving the preservative into the propylene glycol, then adding the surfactant, the triethanolamine and the antifoaming agent into the obtained mixture so as to obtain a solution A; heating the deionized water, and dissolving the enzyme stabilizer, the sodium sulfite, the calcium chloride, the sodium citrate and the citric acid into the deionized water, then cooling the obtained mixture so as to obtain a solution bio-enzyme B; and finally, mixing the solution A and the solution B so as to obtain the finished product of the concentrated multienzyme cleaning solution. The concentrated multienzyme cleaning solution and the preparation method thereof have the advantages: the enzyme stabilizer is adopted to effectively protect the activity of the bio-enzyme, and used with the surfactant and the preservative, so that the obtained finished product is stable in performance, long in storage time and strong in capacity of decomposing biological tissues.

The invention relates to a data sharing method and system under the cloud computing environment. The method comprises the user data encryption storage step and the user data extraction step. The user data encryption storage step comprises the particular steps that a data processing server sets up data index keyword information for user data and generates a matched data ciphertext index and a matched data extraction token, the user data are stored in a cloud storage platform after being encrypted, the data ciphertext index and a data storage address are stored in a data extraction server, and the data extraction token is transmitted to a user. The user data extraction step comprises the particular steps that after receiving the data extraction token, the data extraction server verifies the legality of the data extraction token, extracts data from the cloud storage platform according to the data storage address, and the data are returned to the user by the data extraction server after being decrypted by the data processing server. According to the data sharing method and system, the data storage safety under the cloud computing environment is ensured, and meanwhile the problem of user data sharing is conveniently solved.

The invention discloses a method and a device for generating a configuration monitoring object based on graph mode integration. The method comprises the following steps of: pre-creating a graph template library and a data template library for the configuration monitoring object based on the graph mode integration; associating a graph template with a data template according to the classified configuration monitoring object, and generating a graph mode template based on the graph mode integration; acquiring the graph template which corresponds to the configuration monitoring object to be generated and the data template which is associated with the graph template from the stored graph mode template; and editing the graph template and the data template which is associated with the graph template according to the configuration monitoring object to be generated, and generating the configuration monitoring object based on the graph mode integration. By adoption of a graph mode integration design method, repeated and complicated creation and association operation can be avoided, and the generating efficiency of the configuration monitoring object and the reliability of the generated configuration monitoring object are improved.

The invention discloses a decentralized distributed rendering method and system. The system comprises multiple client sides (100), a main server (200) and multiple rendering servers (300), wherein each client side (100) is used for presenting a rendering task to the main server (200) and uploading a rendering task material file to the main server (200), the main server (200) is used for dividing each rendering task into multiple sub-tasks, monitoring the working states of the multiple rendering servers (300) and allocating the sub-rendering tasks to a fixed number of idler rendering servers (300) and providing the material file downloading service, and the main server (200) is further used for reading a destination file through a destination file storage path of a database (205) after judging that both the sub-task allocating and rendering work and rendering destination file uploading work are completed and feeding the destination file back to the client sides (100).

The invention relates to a production process of instant Chinese chestnut kernels and a product, which belong to the field of the deep processing of agricultural products. The production process comprises the following main technological processes: taking Chinese chestnuts with high water content, low reducing sugar, low starch and low fruit breaking rate as a raw material, processing instant Chinese chestnut kernels, sieving the raw material, classifying, removing outer peels, blanching, removing inner peels, quick-freezing, unfreezing, saccharifying, steaming, pasting, packaging and sterilizing. The instant Chinese chestnut kernels produced by the method have golden color, low fruit breaking rate and proper sweetness, are fragrant, glutinous and delicious and can be used as leisure foods for eating.

The invention provides a static analysis method of a smart phone application program. The static analysis method comprises the steps of receiving an uploaded application program sample; scanning an installation package of the application program sample and extracting specified feature information; matching the specified feature information with a feature code database of a cloud server; judging the safety level of the application program if matching is successful; scanning an API of the application program and judging the safety level of the application program if matching is unsuccessful. In addition, the invention further provides a static analysis system of the smart phone application program. According to the static analysis method and system, automatic detecting of the safety level of the application programs of an Android system and an IOS system can be achieved, the detection efficiency is high, and the accurate rate is high.

The invention relates to a system and a method for monitoring vibration of a circuit breaker, and in particular relates to an acceleration sensor-based system and an acceleration sensor-based method for monitoring vibration of a circuit breaker. The system comprises a resonance type wireless passive voice surface wave acceleration sensor, a reader, a semiconductor temperature sensor, a switching value state sensor and a host computer, wherein the acceleration sensor is arranged nearby a vibration source of a key part of the circuit breaker; the reader is arranged at a position not more than 2m away from the acceleration sensor; the semiconductor temperature sensor is arranged on the surface of the reader; the switching value state sensor is arranged on the circuit breaker; the reader is in communication with the host computer through a serial port and an Ethernet interface. The system disclosed by the invention adopts the wireless passive voice surface wave acceleration sensor to measure vibration signals of the circuit breaker, the monitored data are transmitted in a wireless mode, and the semiconductor temperature sensor is used for realizing temperature compensation of the acceleration sensor; the acceleration sensor has the advantages of being purely passive, free from maintenance, convenient to install and high in safety.

The invention discloses a delayed testing method and device and belongs to the technical field of communication. The delayed testing method comprises the steps of obtaining video data including screen display of a first terminal device and screen display of a second terminal device; obtaining time information displayed by a first terminal device screen and time information displayed by a second terminal device screen according to the video data; performing calculation according to the time information displayed by the first terminal device screen and the time information displayed by the second terminal device screen to obtain transmission delay between the first terminal device and the second terminal device. The large working amount and occurred matching errors due to the fact that the same-frame images are looked up in data messages of the first terminal device and the second terminal device are avoided, and the testing process is greatly simplified.

The invention discloses a residual error network-based detection method for four types of damage of a conveyer belt. The method comprises the steps of recording a conveyer belt running video; extracting picture samples from the video and performing preprocessing; dividing a training data set and a test data set; constructing a residual error network and performing training by using the training data set; performing assessment by using the test data set; and performing a series of processes of real-time detection on the conveyer belt by using the residual error network meeting the requirements,thereby real-time detection of the four types of the damage, including "tear", "scratch", "chamfered edges" and "holes", of the conveyer belt. According to the method, the detection of the four typesof the damage of the conveyer belt is realized through the video, so that complicated work due to use of a large amount of sensor detection apparatuses is avoided and the cost is reduced; and an identification network is constructed by using the residual error network, so that the network training precision is improved and the identification error rate is reduced.

The invention discloses a multi-protocol adaptation method and system applied to an intelligent building platform. The multi-protocol adaptation system comprises a receiving-transmitting allocation module, a protocol conversion module and a protocol interface module, wherein the protocol conversion module comprises a plurality of protocol conversion units for building control protocols; and the protocol interface module comprises a plurality of protocol interface units for the building control protocols. A downlink command adaptation process comprises the following steps that: the receiving-transmitting allocation module receives a standard format command of an application server, and sends the received standard format command to a corresponding protocol conversion unit; the corresponding protocol conversion unit converts the standard format command into a downlink command which corresponds to equipment, and sends the downlink command to the receiving-transmitting allocation module; and a corresponding protocol interface unit sends the downlink command to the corresponding equipment. An uplink command adaptation process comprises the following steps that: the receiving-transmitting allocation module receives an uplink command sent by a certain protocol interface unit, and sends the received uplink command to a corresponding protocol conversion unit; the corresponding protocol conversion unit resolves the uplink command, converts the uplink command into standard format data, and transmits the standard format data to the receiving-transmitting allocation module; and the receiving-transmitting allocation module uploads the uplink command in a standard format to the application server. The multi-protocol adaptation method and system support multi-protocol access, so that the influences of different access protocols on the application server are shielded, and changes stop at the adaptation system. Various access protocols can be made compatible easily, and complicated workload is avoided.

The method includes steps: calling user initiates a call; calling MSC/VLR obtains route number of IP from SCP/HLR; and establishing communication link between calling MSC/VLR and IP; executing service logic of voice value added service in virtual number category, IP receives number of virtual extension inputted from calling user through communication link between calling MSC/VLR and IP as well as sends the received number of virtual extension to SCP/HLR, which converts the number of virtual extension to real number of the called user; obtaining real number of the called user, calling MSC/VLR establishes communication link between calling MSC/VLR and called MSC/VLR based on obtained real number of the called user. The invention also discloses corresponding system for implementing voice value added service in virtual number category.

The invention discloses a method for intelligent and batch KKS encoding of a three-dimension design of a transformer substation. According to the method for intelligent and batch KKS encoding of the three-dimension design of the transformer substation, names of two-dimension symbols are standardized so that the similarity of the same type of devices can be guaranteed; sequential encoding is carried out on the two-dimension symbols with a bus rod naming method, and batch encoding of devices with the same type at the same interval is achieved; a first non-associated device which is encoded is selected, the sequence of batch encoding of non-associated devices is determined according to the sequence of distance between remaining devices and the first non-associated device, and then KKS encoding of the three-dimension design is achieved. The method for intelligent and batch KKS encoding of the three-dimension design of the transformer substation has the advantages that intelligent and batch KKS encoding is achieved, and accuracy and high efficiency are guaranteed.

The invention provides a QP (Quadratic Programming) unified and coordinated motion describing and programming method for a movable manipulator. The method comprises the following steps: (1) a QP unified and coordinated redundancy resolution scheme is established for the movable manipulator, the minimum performance index designed by the resolution scheme is a quadratic function of a combined velocity vector of a movable platform and the manipulator and is restricted by a unified description equation, a combined velocity limit and a combined angle limit of the movable platform and the manipulator; (2) the QP unified and coordinated redundancy resolution scheme in the step (1) is solved by a QP solver; and (3) the solvent result of the step (2) is sent to a lower-computer controller so as to drive the movable platform and the manipulator to simultaneously finish given tail-end tasks in a coordinated manner. The QP unified and coordinated motion describing and programming method has the characteristic that through unified describing and programming of the movable platform and the manipulator, both the movable platform and the manipulator of the movable manipulator are controlled to move simultaneously in a coordinated manner so as to finish given tail-end tasks.

The invention discloses a JSON format data query management system, aiming to improve the efficiency of querying JSON format data. In the system, a communication connection is established between an application client and an application server; a communication connection is established between the application server and a database server; the database server converts a data storage format of a data table in a database according to the JSON format required by the application client, so as to obtain a JSON format data view corresponding to the data table; the application client sends a JSON format data query request to the application server; the application server sends an SQL query statement to the database server; the database server acquires JSON format data displayed by the JSON format data view from the database according to the SQL query statement and sends the JSON format data to the application server; and the application client receives the JSON format data forwarded by the application server, and displays the JSON format data to a user.

The invention provides a self-buffering plunger. A spring is arranged in the plunger, which can achieve a self-buffering function of the plunger and can replace the conventional downhole buffers; a bottom guide rod and a baffle ring can fix a position of the spring, prevent the spring from sliding and falling off, and can support the spring; a guide head plays a role in guiding in falling of the plunger and can support the plunger body in a well bottom, and the guide rod is of a hollow structure; spring cylindrical pins can fix the baffle ring and the guide head. The plunger is mainly used for highly-deviated wells and horizontal wells, can overcome the problem that the conventional horizontal wells cause pollution, are high in cost, are large in workload, and are poor in fluid flowing effect due to the foam water drainage technology, can achieve the fact that horizontal gas wells can perform fluid flowing and gas production by using the plunger gas lift technology, can solve the difficult problem of fluid flowing and gas production of more than 1200 horizontal gas wells in a Changqing gas field, and can be used for the highly-deviated wells and vertical shafts, and can simplify downhole tools.

The invention discloses a vision-based container automatic joint and hoisting system and hoisting method. Image collecting equipment is installed at the vertex angle of the top of a hoisting platform and is used for conducting image collecting on joint corner fittings installed on containers; an image processing system conducts position identification on the joint corner fittings in images, according to the position relationship between the joint corner fittings and target joint positions, motion parameters for enabling joint parts of the hoisting platform to move to the joint positions are solved, and the motion parameters are conveyed to a control system; and the control system controls motion of the hoisting platform according to the received motion parameters, the position and the angle of the hoisting platform are adjusted accordingly until the joint parts of the hoisting platform are in butt joint with the joint corner fittings of the containers, and finally hoisting of the containers is achieved. By means of the vision-based container automatic joint and hoisting system and hoisting method, when the hoisting platform is in butt joint with the joint corner fittings of the containers, the defects of long butt joint hours and low efficiency which are caused in the prior art are overcome.

The invention relates to a device for monitoring foreign matters between a rail transit platform door and a train based on machine vision, wherein the device collects visible light images, infrared light images and depth images between the rail transit platform door and the train, integrates the visible light images, the infrared light images and the depth images to form three-dimensional space three-dimensional images, which can cover all targets in a shooting visual angle; events can be identified in the images by analyzing the data of the three-dimensional images, wherein the main events comprises: whether or not foreign objects exist in the track area, whether or not the train is at the station, the number of trains, the switch status of the platform door and the train door, the door seam of platform sliding door and train door, whether or not foreign objects inrude into the portion between the train and the platform door, the size and type of foreign objects, whether or not the foreign objects affects driving safety and the like; the device drives the foreign object alarm safety relay, and transmits the monitoring result and live real-time image in real time. The device can achieve no dead angle and full-area detection in an area, is not influenced by vehicle type, gap space size, vibration and environmental illumination, is suitable for underground, ground and ground platform environments, and is suitable for unmanned environments.

The invention provides a partial discharge multi-source fusion detection system and method for a high-voltage switch cabinet, and belongs to the technical field of electric power overhaul. The systemcomprises a sensor assembly, a signal collection and processing module connected with the sensor assembly, and a computer terminal connected with the signal collection and processing module; the sensor assembly comprises a UHF sensor, an AE sensor and a TEV sensor; the signal collection and processing module comprises an amplifier, a filter connected with the amplifier, and a USB data acquisitioncard connected with the filter. According to the system and method of the invention, ultrahigh frequency, ultrasonic wave and TEV partial discharge detection technologies are fused, the system and method have the advantages of high sensitivity, strong anti-interference capability and the like, can realize accurate positioning of discharge faults, obtain quantitative data, predict and accurately position internal discharge in advance, reduce the workload of an inspection department, and improve the working efficiency.

The embodiment of the invention discloses a current sharing control method, device and system for a multi-phase parallel system. The method comprises the following steps: detecting the output current of each phase of a power system; comparing the output current with preset current so as to determine that the phase with output current larger than the preset current is the phase required to be adjusted; and generating a current sharing control signal according to the preset current and the output current of the phase required to be adjusted, so as to control the output of the phase to control the delay conduction of a switch, thereby allowing the current output by the phase to be reduced. According to the method, the device and the system, the workload of selecting devices is avoided, which reduces the cost; and the current sharing control for the output current of the multi-phase parallel power system is realized, and the performance of the power system is improved.

The embodiment of the invention provides an automatic webpage testing method and system. The method comprises the following steps: loading a proxy script in a tested page, wherein the proxy script enumerates JS (Java Script) objects in the JS environment of the tested page; mapping the JS objects enumerated by the proxy script into the test case JS environment, wherein the test case triggers the page action of the tested page to execute testing logic according to the test case JS environment. After the embodiment of the invention is applied, the workload of element matching can be avoided, and the efficiency of forward testing is greatly improved, thereby reducing the testing cost.

The invention discloses an electromagnetic induction type automatic charging system and method for a robot. The electromagnetic induction type automatic charging system method for the robot includes a robot electrical energy receiving device arranged on the robot and a ground electric energy emitting device arranged on the ground, wherein the robot electrical energy receiving device includes a robot control center, a first wireless communication module, a battery, an electric energy receiving module and a receiving coil which is arranged at the bottom of the robot, and the ground electric energy emitting device includes a ground controller, a ground emitting coil, a second wireless communication module and an electric energy emitting module. When the electromagnetic induction non-contact automatic charging system for the robot provided by the invention is used, and an installed camera is utilized to automatically and perpendicularly align with the ground electric energy emitting device, and therefore, the working efficiency of the robot can be improved, the intelligence of the robot can be enhanced, production cost can be reduced, and charging efficiency can be improved. The charging method has high adaptability to adverse weather conditions and is also suitable for outdoor work of the robot.

The invention relates to a piloted driving vehicle training method based on a virtual environment and a depth double-Q network. The method comprises: constructing a vehicle racing track virtual environment based on Unity; establishing a communication connection between the Unity and a piloted driving vehicle model under a Websocket protocol; establishing a Markov model that represents a vehicle piloted driving process, transforming a depth double-Q network algorithm to train a piloted driving model; configuring an entity vehicle soft hardware environment, migrating a trained vehicle piloted driving model; and testing an entity vehicle in a real racing track to train a piloted driving operation of the model. According to the method, a virtual environment is used for training a model, to implement robustness of a piloted driving training algorithm model on a complex environment and road condition. According to the method, by means of the depth double-Q network algorithm, a problem of high train complexity caused by a large assessment value of a Q value is avoided, so that a training process is simple and fast. The method has advantages of strong robustness, a fast speed, low costs and the like. The method can be applied to training and learning of unpiloted operation in fields such as intelligent traffic, aerospace, robots and the like.

The invention relates to a CNN and model fusion based blood glucose prediction method for diabetes. The method includes the following steps: performing pre-processing, including the processing of vacancy values, the binarization of quantitative features and data conversion, on data; performing feature extraction on pre-processed data by utilizing a CNN; and performing model fusion on xgboost, catboost and linearRegression by utilizing a Stacking strategy. Catboost and xgboost models with high classification precision are adopted at a first layer, linearRegression with good robustness is adopted in a second layer model, so that the unification of prediction precision and model robustness can be realized, and stronger generalization abilities can be achieved. Thus, the blood glucose prediction problem of diabetes can be effectively solved, and compared with traditional blood glucose prediction methods, the method has significant improvement.