430results about How to "Save manpower, material and financial resources" patented technology

The invention provides a hardware part defect detecting system and a method. The hardware part defect detecting system comprises an image acquisition module, a defect information extraction module, a artificial neural network training module, and a defect identification module; the image acquisition module is used for acquiring standard images of hardware parts and corresponding defect sample images, performing image gray-scale treatment and image de-noising processing, and obtaining standard images and defect sample image gray-scale treated images; the defect information extraction module is used for extracting defect characteristic values of defect sample images using a defect minimum bounding box position extraction unit and a defect characteristic extraction unit; the artificial neural network training module is used for training artificial neural network via BP algorithm and the defect characteristic values of the defect sample images. According to the hardware part defect detecting system, the trained artificial neural network is used for identifying the defect kinds and defect positions of the hardware parts to be detected, hardware part surface defect automatic detection is realized, hardware part surface defect detection efficiency is increased, and manpower, material resources, and financial resources are saved.

The invention belongs to the technical field of agricultural remote sensing, and particularly relates to a Landsat remote sensing data-based multi-cloud rain and fog region rice identification method.The method includes: distinguishing rice from other obvious ground objects, and obtaining a potential rice planting area in a remote sensing image; based on ground measured data, extracting the relationship between the spectral indexes of the rice sample pixel data in the transplanting stage; establishing a vegetation index threshold value model for rice identification by combining rice plantingregion division data and rice potential transplanting period data, extracting rice planting areas in remote sensing images step by step in a segmented mode, combining the planting areas of different rice types, and obtaining a final rice planting region distribution diagram rapidly and accurately. Compared with an existing method for extracting the rice planting area through a remote sensing technology, the method is mainly suitable for southwest China regions with cloudy rain and fog. According to the invention, the actual planting condition can be truly reflected, and the planting area extraction precision is improved.

The invention relates to a quality evaluation model for natural ageing of redried strips and a building method thereof and belongs to the technical field of tobacco processing and quality evaluation. The building method includes: according to factors influencing quality changes during natural ageing of tobacco, determining a major quality evaluation index, performing screening by means of quantitative characteristic analysis, factor analysis, linear correlation and canonical correlation analysis to obtain quality evaluation indexes highly related to tobacco sensory evaluation quality, subjecting the quality evaluation indexes to factor analysis to obtain factor score coefficient matrixes, subjecting scores of common factors and the sensory evaluation quality to principal component regression to obtain a standard equation, substituting the score coefficient matrixes obtained by factor analysis, to the standard equation so as to obtain the equality evaluation model for the natural ageing of the redried strips. The method and the model are scientific and effective and good in stability, and index detection related is simple and fast and has small errors.

The invention relates to electro-fusion welding of a polyethylene pipeline, and aims to provide an electro-fusion welding machine for realizing welding quality monitoring through a wireless network and a method for realizing electro-fusion welding quality monitoring through the wireless network. The electro-fusion welding machine comprises a microprocessor connected with an electro-fusion welding machine main body and an identity authentication module, and an abnormal condition alarm module and a wireless data receiving and transmitting module which are connected with the microprocessor respectively, wherein a quality monitoring program is arranged in the microprocessor. The method comprises the following steps: identity authentication and qualification affirmation of a welder, welding engineering project management, welding parameter management, pre-examination before welding, control of the welding quality and the like. Welding quality management of an electro-fusion welding joint of the polyethylene pipeline, including pipeline joint positioning, welder qualification verification, welding data storage, welding quality verification and the like can be realized; and welder qualification, level management and welding engineering project management can be realized.

The embodiment of the invention provides an intelligent video identifying method for cheat in a test. The method comprises the following steps of: firstly, dividing all pixel points in an examination room monitoring video into foreground pixel points and background pixel points through establishing and updating a (GMM) Gaussian mixture model aiming at the examination room monitoring video; extracting sports pixel characteristic values according to foreground images and extracting the characteristic value of a testee area in the examination room as the judging data of a decision tree in the next step; and training the decision tree by using a series of sample legends to obtain a preset threshold value of each node and probability values of final layer nodes corresponding to different cheating accidents and judging the corresponding cheating accidents according to the data of the decision tree. Through the technical scheme, the cheating behaviors and nonfeasance phenomena of supervisorsin the examination room monitoring video can be automatically detected, and staff is freed from staring the video constantly to search cheating accidents, therefore labor, materials and financial resources are saved greatly, and the working efficiency is improved.

The invention discloses an epidemic infectious disease virus field visualization method based on space-time trajectory data. The method comprises the following steps: S1, airspread infectious diseaseurban diffusion modeling; s2, collecting, cleaning and processing space-time trajectory data and infection attack data of virus carrying crowds; s3, obtaining position semantic information of a viruscarrier; s4, calculating the virus distribution density by taking the virus carrier as the position center; s5, mapping the time, space and virus distribution density data of the virus carrying population; and S6, constructing an epidemic infectious disease virus field cloud platform based on temporal GIS. The invention further provides an epidemic infectious disease virus field visualization system based on the space-time trajectory data. According to the method, space-time trajectory data and infection attack data of virus carrying crowds are processed into a virus distribution density map with the virus carrier as the position center, spatial position semantic information is added and mapped to a temporal GIS, and an epidemic infectious disease virus field is visually processed.

The invention discloses a survey method for an audience rating of digital television users. The method is characterized in that the method comprises the following steps that: a data collection step in which a data collection module is internally arranged on a middleware of a digital set-top box and is responsible for collecting user viewing behavior data; a data transmission step in which the digital set-top box is connected with an Internet network by a built-in Bluetooth or an externally connected network card and transmits the data to a server having a public network IP; a data storage process in which the server receives the data and then the data are parsed and stored to a database; a data analysis process in which the server carries out automatic analysis and statistics on the data; and a data presentation process in which a client invokes a server to carry out calculation and analyses on all indexes of the audience rating and present data of data analysis results of the reviewing behaviors. According to the invention, a set-top box with two-way transmission is supported; and user viewing behavior data is transmitted from an Internet or a metropolitan area network to a server; therefore, compared with the traditional survey mode, the survey method provided in the invention has advantages of high real-time performance, high accuracy and low cost and the like; moreover, lots of manpowers, material resources and financial resources are saved.

The invention discloses a regional carbon reserve space pattern monitoring system and method based on remote sensing data. The system comprises a remote sensing data processing module, an actual measurement data processing module and a data analysis and output module. The output ends of the remote sensing data processing module and the actual measurement data processing module are connected with the input end of the data analysis and output module; satellite remote sensing data, precipitation, vegetation coverage rate, surface temperature, land utilization data, light data, DEM data and meteorological data are used to establish a model to monitor the spatial-temporal change of regional carbon reserves, and the defect of long research period of large-scale regional carbon reserves is overcome. The monitoring system also has the advantages of easily available source data, high computer operation efficiency, low cost, strong timeliness and the like.

The invention discloses an external insulation natural direct accumulation testing method for a high-voltage power transmission line. The method comprises the following steps of: performing mesh division on a dirt test area on a Google Earth visible geographic information platform to divide the whole area into a plurality of meshes of 1 to 5 kilometers; then importing power transmission line tower coordinates in the dirt test area into the Google Earth visible geographic information platform; and importing the conventional dirt level distribution graph and terrain and physiognomy distribution graphs in the dirt test area into the Google Earth visible geographic information platform, determining a polygonal comprehensive attribute area, determining a natural direct accumulation test point of the high-voltage power transmission line, and calculating full-surface equivalent salt density and full-surface equivalent gray density. The method for selecting the external insulation natural direct accumulation test point of the high-voltage power transmission line has the advantages of low cost, simple and feasible operation and easy acceptance and mastering, and is convenient for popularization.

The invention discloses a submarine pipeline flange underwater abutting joint installation method. A main flange connecting device is installed on the abutting joint end of a submarine pipeline A, an auxiliary flange connecting device is installed on the abutting joint end of a submarine pipeline B, the submarine pipeline A to be butted and the submarine pipeline B to be butted are connected together through the main flange connecting device and the auxiliary flange connecting device, the negative influence of the marine condition on underwater flange connection can be removed, the influence of ship shake on the movement of the submarine pipeline A and the submarine pipeline B can be removed, underwater adjusting steps of a diver can be reduced, the diver can adjust the positions of the submarine pipeline A and the submarine pipeline B conveniently, the underwater operation difficulty degree of the diver can be reduced, the underwater operation time of the diver can be shortened, the case that flanges get close to each other in an aligned manner can be ensured, the force applied to a gasket of each flange is consistent, uniform deformation is generated, and the leakproofness of the flanges can be improved. The flange connection quality can be improved, the operating time of the support ship can be saved, the construction period of the whole fleet can be shortened, the working efficiency is improved, and the installation cost is saved.

The invention relates to a system for realizing TV news live broadcast by using a wireless communication network, which is a TV news live broadcast system for transmitting an IP data packet to a specified address to receive and broadcast through the wireless communication network and the Internet. The system comprises a mobile terminal with video / audio acquisition function, a media server for receiving video / audio data stream and a recording / broadcasting server for providing an SDI broadcast signal. Compared with the prior art, the system has the advantages of replacing most of the functions of a satellite news gathering truck, breaking through the limitations of time and territories to ensure that journalists can arrive the scene in time and perform video live broadcast in spite of the place and time of a news event, lowering down the threshold for viewers to participate in TV media communication which has an obvious advantage to ensure that the viewers can conveniently and quickly participate in live programs by using mobile phones, embodying the advantage of video communication of TV media and the interactivity of participating in the programs by the viewers in real time, along with low broadcast cost and a small number of limitations.

The invention relates to a glass cleaner, in particular to an obstacle surmountable safety intelligent glass cleaner comprising a master control component, a cleaning component, a main fixing plate, a main suction mechanism, a travelling mechanism, a lifting mechanism, a split suction fixing mechanism, a traction winding mechanism, a power supply component and a detecting component. The master control component, the cleaning component, the main fixing plate, the main suction mechanism, the travelling mechanism, the lifting mechanism, the split suction fixing mechanism, the traction winding mechanism, the power supply component and the detecting component are all mounted on the main fixing plate. The control output end of the detecting component and the control output end of the power supply component are both connected with the control input end of the master control component. The cleaning component, the main suction mechanism, the travelling mechanism, the lifting mechanism, the split suction fixing mechanism and the traction winding mechanism are separately connected with the control output end of the master control component. The obstacle surmountable safety intelligent glass cleaner is capable of automatically surmounting raised structures such as door and window frames; as safety measures are taken, the cleaner is safer to use; with dual-unit magnetic suction cleaning and single-unit vacuum suction cleaning, the cleaner is convenient to use, is suitable for domestic glass cleaning and urban high-rise building glass exterior wall cleaning and is widely applicable.

The invention relates to an automatically-moving three-dimensional seedbed system with the adjustable light transmittance. The automatically-moving three-dimensional seedbed system comprises a three-dimensional frame body, linear guide rails, seedbed conveying units, single seedbeds and sensing and positioning devices. The linear guide rails are arranged on the three-dimensional frame body in a layered mode through supporting plates. Each seedbed conveying unit comprises two side plates arranged in parallel, wherein the distance between the two side plates is regulated according to the depth by which corresponding interval insertion plate inner plates are inserted into corresponding interval insertion plate outer plates, wheels used for rolling to convey the corresponding single seedbed are arranged on the inner sides of the side plates, and the side plates are connected to the corresponding linear guide rail in a sliding mode through slide blocks. The sensing and positioning devices are arranged on the three-dimensional frame body, the seedbed conveying units and the single seedbeds to position the single seedbeds. Compared with the prior art, the seedbeds can be moved to the assigned seedbed positions, the positions of the seedbeds can be automatically adjusted according to the growth condition of plants, the illumination receiving rate of the seedbeds is increased, and the automatically-moving three-dimensional seedbed system conforms to the tendency that traditional greenhouses are developed towards large-scale plant factories.

The invention provides a miniature piezoelectric impedance device for online health monitoring. A piezoelectric ceramic wafer sensor is stuck on a to-be-measured structure, an impedance converter generates an excitation signal, the excitation signal is applied to the to-be-measured structure through the piezoelectric ceramic wafer sensor, and a feedback signal is collected and sent to a system control module; when the defect of the to-be-measured structure drives the piezoelectric ceramic wafer sensor to deform, a sampling signal returned by the piezoelectric ceramic wafer sensor changes, andthe system control module judges the health status of the to-be-measured structure according to a set threshold value. The device and the method in the invention have the characteristics of simple structure, light weight, small volume, low price, no pollution, easy operation, good stability, strong environmental tolerance, wide use structure range and the like, and online and continuous monitoringcan be realized.

The invention discloses a GRU codec training method. The GRU codec training method comprises: segmenting a sample audio into a plurality of sub-sample audio segments; obtaining sample Fbank filter characteristics of the plurality of sub-sample audio segments; inputting the sample Fbank filter features into a to-be-trained GRU encoder to obtain a sample feature vector with a fixed length; inputtingthe sample feature vectors into a to-be-trained GRU decoder to obtain corresponding sample word vector embedding and sample hidden layer vectors; generating corresponding reference sample word vectorembedding according to the sample annotation statement corresponding to the sample audio; generating word-level cross entropy loss according to the sample word vector embedding and the reference sample word vector embedding; and optimizing and adjusting network parameters of the to-be-trained GRU decoder and the to-be-trained GRU encoder at least based on the word-level cross entropy loss. According to the method, automatic generation of the character abstract according to the audio becomes possible, manpower, material resources and financial resources are saved, and efficiency and accuracy are greatly improved.

The invention discloses an optimization method of electrode design parameters of a lithium ion battery, which relates to the field of internal structure design of the lithium ion battery. The electrode design parameters can include electrode thickness, porosity, particle size of an active material, compaction density and surface density, etc. The specific optimization steps comprise (1) selectingthe battery to be optimized, and measure the electrode design parameters that can be actually measured; (2) establishing the electrochemistry of lithium ion batteries according to the measured and estimated parameters, and using the thermally coupled model to adjust the estimated parameters through experimental verification. (3) in order to maximize the energy density and the product of energy density and power density, using two optimization methods to obtain the optimized electrode design parameters. The method of the invention can select the optimized electrode design parameters in the battery design stage, reduce the development cost, improve the energy density and the power density of the battery, and provide a guiding basis for the electrode design.

The invention discloses a flue gas desulfurization wastewater treatment process for a coal-fired boiler. The process comprises the steps of:mixing and collecting the overflow obtained by treating desulfurization wastewater with a hydrocyclone and an underflow filtration squeezing filtrate in a wastewater buffer tank, and introducing the mixture to a flocculation reactor to complete a dosing and flocculation reaction process; precipitating the wastewater after flocculation reaction into a clarification concentration tank for a period of time, and using a supernatant as desulfurization process water or sending the supernatant into a factory public drainage system; dewatering the sludge at the bottom of the clarification concentration tank, introducing the dewatering liquid into a waste water buffer tank or returning to a water inlet pipe of the clarification concentration tank, and treating the dewatering solid residue and solid residue from underflow dewatering by hydrocyclone. The outlet water quality from the process meets the requirements of DL / T997-2006 standard. The process only needs to add a solid powder treatment agent, and doesn't need drug dissolution preparation; the pH value of the outlet water doesn't need to be adjusted; and the process has the advantages of simple process flow, less equipment investment, simple control system, and great improvement on the operation stability of the desulfurization wastewater treatment system.

The invention discloses an automatic control device of constant soil water potential. The device mainly utilizes a semi permeable material and solution with constant water potential to control a certain range of the soil water potential to be constant at a specific value. The device comprises a constant hydraulic pressure water supply device, an automatic water inlet control device, a semi permeable water control device and a micro porous container; the constant hydraulic pressure water supply device can supply water out at a constant hydraulic pressure, and the automatic water inlet control device can automatically control carrying capacity transported to the constant water potential solution, thus ensuring the water potential is constant; the semi permeable water control device employs the semi permeable material as main functional parts, which can realize water exchange between the solution with constant water potential and the closed water body; a plurality of holes with micro hole diameter are distributed at the surface of the micro porous container, water can pass through the holes but with comparatively low air inlet value, thus realizing the water exchange between the closed water body and the soil. The invention can effectively control constant soil water potential, and is widely applicable to the agriculture and scientific research field.

The invention discloses a method for identifying excited and antagonistic actions of an organic estrogen. By utilizing the molecular dynamic method, the binding process between an organic substance and an estrogen receptor is computed and analyzed; and based on the ligand binding mode and the change of the estrogen ligand binding area conformation, particularly the change of the spiral position of H12 in the LBD structure of ER, whether the organic substance is the excitant or antagon is identified. The method has the advantages of effectively identifying the pseudo / anti-estrogen action, having simple method and low cost, saving a large amount of manpower, material resources and financial resources and providing a basic tool and a shortcut for the large-scaled implementation of the poisonous chemicals environment safety evaluation(REACH).

The invention relates to an automatized tide type irrigation device matched with a mobile seedling bed. The automatized tide type irrigation device matched with the mobile seedling bed comprises a mobile seedling bed frame, a single mobile seedling bed connected onto the mobile seedling bed frame in a sliding mode and a seedling tray arranged on the single mobile seedling bed and further comprises a reciprocating motion unit, an irrigation spray head, a connecting top plate, an irrigation pipeline, a storage pond and a magnetic valve. The storage pond is arranged at the middle-lower portion of the mobile seedling bed frame, one end of the irrigation pipeline is inserted into the storage pond, the other end of the irrigation pipeline is connected with the irrigation spray head, the reciprocating motion unit is connected with the irrigation spray head through the connecting top plate and drives the irrigation spray head to move vertically so as to enable the irrigation spray head and the seedling tray to be connected or separated, and the magnetic valve is arranged inside the irrigation pipeline and controls liquid feeding or liquid return of the irrigation pipeline so as to perform tide type irrigation to the seedling tray. Compared with the prior art, the automatized tide type irrigation device matched with the mobile seedling bed is low in cost, simple, convenient, compact in mechanism and long in service life, reduces transmission of diseases and insect pests inside a greenhouse, basically achieves closed automatic regulation and control and achieves the energy-saving, environmental-friendly, accurate and efficient purposes.

The invention relates to a method for identifying the ubiquitin-like modification sites of proteins. The above specific enrichment labeling identification method comprises the following steps: carrying out enzymatic hydrolysis of purified ubiquitin-like modification proteins, enriching the ubiquitin-like modification enzyme digestion peptide fragments through a specific ubiquitin-like antibody, protecting the non-modification lysine amino residues in the enriched peptide fragments through utilizing chemical labeling, carrying out a deubiquinaning enzyme catalysis reaction to expose modified lysine on a substrate peptide fragment sequence, analyzing the peptide segment sequences through a liquid chromatograph-mass spectrometer, and matching through protein mass spectrum data analysis software to obtain the chemically modified peptide fragment sequence, wherein lysine sites containing free amino groups in the finally obtained sequence are the ubiquitin-like modification sites. Compared with common gene mutation methods for obtaining the sites at present, the method provided by the invention has an advantage that information about the ubiquitin-like modification sites of proteins can be rapidly, efficiently and accurately obtained without a tedious gene mutation technology.

The invention discloses an immersive driving behavior correction method and system based on a driving simulation platform. The correction method comprises: a background sets a driving accident tendency questionnaire and lowa testing by a display device, accident tendency types, anger values and riskiness of drivers are collected, and driver classification is carried out; the background collects driving operation information at a preset driving task by a driving simulation platform and determines illegal events of the drivers based on the driving operation information; immersive education is carried out on illegal events not qualified in first driving testing and second driving testing is carried out; and immersive education is carried out on illegal events not qualified in the second driving testing and third driving testing is carried out; and then driver evaluation reports are outputted. Besides, the correction system is composed of a background, a display device and a driving simulation platform; and the background is connected to the display device and the driving simulation platform respectively. Therefore, problems of the low educational effect of illegal driving behaviors and high frequency of the traffic accidents at present are solved.

The invention provides a semantic Web services system and applications thereof in the logistics field, and the system comprises a registration center, a service requester client and a service provider client, and the registration center, the service requester client and the service provider client are connected with one another, the services system is based on semantic Web, and the registration center, the service requester client and the service provider client are interactive with one another by a built service body. The invention solves a plurality of problems of difficult service discovery, very low service discovery efficiency, high consumption of human, material and financial resources and the like in the current third party logistics services, realizes the automatic discovery and automatic combination of services, greatly improves the efficiency of logistics service discovery and saves mass human, material and financial resources.

The invention discloses an intelligent normal pool level optimal selection method based on genetic neural network models. The method includes the steps of firstly, determining high-low limit of normal pool level; secondly, providing a normal pool level elevation threshold, acquiring area and volume index to form a draft selection scheme, and repeating the second step to form a comparison and selection scheme set; thirdly, selecting index evaluation factors, and selecting evident difference indexes to participate in model calculation; fourthly, forming network learning samples according to required sample number, and normalizing training samples and test samples; fifthly determining a BP network learning structure and initial genetic algorithm population; sixthly, using genetic algorithm to optimize neural network weights and thresholds; seventhly, fine adjusting BP neural network weights, and using built models to evaluate to-be-selected schemes. The method is significantly practical in the field of hydraulic and hydroelectric engineering, the influence of subjective factors is reduced effectively, and objectivity of index weight determination is increased.

The invention discloses a fitness training guide system on the basis of somatic games. The fitness training guide system is characterized by comprising a remote guide service platform, a fitness guide client, somatic game equipment, audio and video signal control switching equipment and display equipment. The fitness guide client is used for acquiring health information and fitness requirements of users; the remote guide service platform is used for providing somatic game fitness guide schemes for the users; the somatic game equipment is used for outputting somatic game projects; the audio and video signal control switching equipment is used for actively switching the fitness training guide schemes and somatic game operation display interfaces and outputting the fitness training guide schemes and the somatic game operation display interfaces to the display equipment. The fitness training guide system has the advantages that the fitness training guide system can access third-party somatic game equipment, the somatic games are combined with fitness guide, and accordingly the personalized fitness requirements can be met.

The invention discloses a determination method for stress between electrode particles during a lithium ion battery discharge process based on an electrochemical-mechanical coupling model, and relatesto the field of computation of the lithium ion battery stress. The method builds a three-dimensional model on a lithium ion battery electrode particle scale, and specifically comprises the following steps of: 1, selecting positive and genitive electrode materials, acquiring electrochemical and mechanical initial parameters, and building the electrochemical-mechanical coupling model on a three-dimensional particle scale; 2, computing a solid phase lithium concentration according to the electrochemical reaction of a battery and the Fick law; and 3, coupling the lithium concentration computed inthe step 2 into a stress model, wherein the expansion rate of the electrode particles is a function of the solid phase lithium concentration, and then computing the stress of the negative electrode particles. According to the determination method provided by the invention, the stress suffered by the electrode particles of the lithium battery during the discharge process can be computed, and then the certain guidance basis is provided for cracking and crushing of the electrode material.

The invention discloses an overhead cable vibration testing device and a testing method. The device comprises an overhead cable test bench, a fiber grating sensor array, a vibration table, a fiber grating CCD(Charge Coupled Device) demodulation device and a PC(Personal Computer). The surface of an overhead cable on the overhead cable test bench is adhered with the fiber grating sensor array in theaxial direction. The vibration table is located below the overhead cable test bench. The bench top is hinged to the overhead cable. A control port is connected with the PC. The optical interface of the fiber grating CCD demodulation device is connected with the fiber grating sensor array. A data transmission port is connected with the PC. According to the overhead cable vibration test device andthe test method, the vibration of the overhead cable under a real environment is simulated through a vibration table excitation, a range adjustment and the initial tension of the cable, so as to systematically analyze the vibration characteristics of the overhead cable under various complex environments; by optimizing the test method and focusing on monitoring parts where faults are susceptible tohappen, the ultimate vibration parameter on the overhead cable can be accurately grabbed; and compared with an online cable monitoring scheme, the device is convenient to adjust, a high-altitude operation is avoided, and high scientific research value and operability are achieved.

The invention discloses a multifunctional carrier based on cytogenic vesicle in body fluid, a preparation method and an application. The method comprises the following steps: (1) collecting human body fluid, performing gradient ultra-centrifugation, removing supernatant and using sterile PBS heavy suspension for precipitating and (2) adding the cytogenic membrane vesicle from the body fluid, ultra-small luminescent material and electroporation buffer solution into an electric shock cup, uniformly mixing, placing onto an electroporation instrument and treating, placing under room temperature and repairing after the ending of electroporation, performing ultra-centrifugation and then using sterile PBS heavy suspension for precipitating, thereby acquiring the multifunctional carrier based on the cytogenic membrane vesicle from the body fluid and marked with the ultra-small luminescent material. The carrier is used for preparing a chemotherapeutic drug or gene therapeutic drug for treating or preventing tumor, has high biocompatibility, is convenient to carry various therapeutic drugs, is capable of effectively restraining tumor growth, is suitable for diagnosis and treatment of various innocent and malignant tumors and especially supplies a choice for the diagnosis and treatment of deep tumors in skull, thoracic cavity and abdominal cavity.

The invention discloses a light module test method which includes the following steps: conducting test of at least one test item on at least one light module to be tested and with unique serial number under multiple different environment temperatures, correlating environment temperature, serial number of the light module to be detected and test data of each test item in current test, storing the data in a local database in index table mode, enabling the test data of each test item to utilize the serial number of the light module to be tested as index and the serial number of the light module to be tested to utilize the environment temperature as index, searching and judging which light modules to be tested under each environment do not finish all test items according to an index table stored in the local database and continue testing on the light modules to be tested and not finishing all the test items when the test is suspended and restarted. The light module test method does not require testing the light modules finishing all tests, avoids repeated test and improves test efficiency.

The invention provides a multi-decoder full convolutional neural network and a corresponding microscopic structure identification method thereof, the multi-decoder full convolutional neural network isused for semantic segmentation of ceramic matrix composite material perform XCT slices, and an initialization method, a class balance method, a training method and the like of a network structure aredetermined. According to the method, the problem of edge blur or content information loss in the feature map caused by different decoders is solved, the accuracy of microscomic structure recognitionis improved by combining deep learning and semantic segmentation tasks, and the robustness of an algorithm is enhanced. The multi-decoder full convolutional neural network provided by the invention can be used for ceramic matrix composite preforms of different weaving types, including 2.5-dimensional weaving, three-dimensional four-direction weaving structures and other all types, the applicationrange is wide, and identified microscopic structures comprise fiber bundles (warp yarns and weft yarns), holes and matrixes.