510results about How to "Improve correction efficiency" patented technology

The invention discloses an optimized bridge damage identification method based on a neural network. The optimized bridge damage identification method based on the neural network comprises the steps of S1, constructing sample data; S2, determining network topology; S3, conducting training and testing; S4, identifying damage, wherein real-time strain data of a bridge are input into a trained BP neural network, so that damage identification of the bridge is achieved, the real-time strain data of the bridge are obtained through sensors which are arranged optimally, the minimum number Ymin of unidentifiable models serves as an object function, and the arrangement positions of the sensors corresponding to Ymin is the optimal sensor arrangement. By the adoption of the optimized bridge damage identification method based on the neural network, various possible damage conditions of a structure can be identified to the maximum extent through the minimum number of sensors, and an identification result is high in precision and tends to be stable.

The invention discloses a double-shaft continuous rotation-based hybrid type platform inertial navigation system calibration method. The method comprises the following steps: (1) performing coarse alignment to obtain a coarse initial attitude matrix; (2) controlling an outer ring shaft and a platform body shaft of a platform system to rotate at the same angular speed, enabling an inner ring shaftto be under a locked state, and acquiring gyroscope and accelerometer output in the rotating process; (3) resolving an error state equation and an observation equation and calibrating gyroscope scalefactor error, gyroscope mounting error and accelerometer equivalent zero offset; (4) controlling an outer ring frame angle, an inner ring frame angle and a platform body frame angle to lock in a zerostate, rotating the platform body shaft by 180 degrees and locking, and acquiring output data of the gyroscope; (5) controlling the outer ring frame angle, the inner ring frame angle and the platformbody frame angle to lock in the zero state, rotating the outer ring shaft by 180 degrees and locking, and acquiring output data of the gyroscope; and (6) calculating the constant drift of the gyroscope. By the method, the calibration time is effectively shortened and the calibration efficiency is improved.

The invention provides a method for automatically correcting an identification error of repeated words in Chinese pronunciation identification. The method comprises the following steps of: (1) performing similarity matching on word confusion networks which are obtained after identification of each sentence, word groups in a word group library and intermediate identification results, and searching the repeated word groups, wherein each word confusion network is a set of all possible identification results and comprises an optimum identification result, namely the original optimum identification result, and the intermediate identification result which corresponds to each word in the optimum identification result, and the word group library comprises the word groups and the intermediate identification results which correspond to the word groups; (2) according to word group information which is obtained by searching, re-calculating a similar probability value and a word identification probability value; (3) according to a new probability value, sorting the word confusion networks according to the size of the probability value; and (4) replacing the optimum identification results and the intermediate identification results of the word confusion networks by using a sorting result. The method has the advantages that: by using experience knowledge in the corrected identification result, the identification error of the repeated words in the current identification sentence is automatically corrected, so the correction efficiency and correction speed of the identification error are improved.

The invention discloses a bridge finite element model modifying method, which comprises the following steps that: 1, an entity finite element model of a whole bridge is built by universal finite element calculation software ANSYS; and 2, the entity finite element model of the whole bridge is subjected to primary modification by a uniform design method. The bridge finite element model modifying method has the following advantages that the uniform design method is used as a test optimization design method with obvious superiority, and the application of the uniform design method to the bridge structure finite element model modification is feasible; on the basis of the uniform design method, a more correct result can be found in a very short time by the uniform design method in the efficiency aspect, and the bridge large-scale finite element model modification efficiency is improved; a response surface method is used as a test design and mathematical statistics combined mathematical analysis method, and the application of the response surface method to the finite element model modification of a large-scale complicated bridge structure is feasible; and higher modification efficiency and higher modification precision can be ensured at the same time.

An optical proximity correction method includes the following steps: acquiring a current correction mask plate figure subjected to at least one time of optical proximity correction, and performing optical simulation on the current correction mask plate figure to acquire a simulation figure; comparing the simulation figure and a target figure to obtain an error of multiple border positions; when the error of the border positions is greater than a preset value and the distance from a to-be-corrected edge of the current correction mask plate figure to an adjacent mask plate figure is greater than the minimum design size of a mask plate, correcting the to-be-corrected edge of the current correction mask plate figure; when the error of the border positions is greater than a preset value and the distance from a to-be-corrected edge of the current correction mask plate figure to an adjacent mask plate figure is equal to the minimum design size of the mask plate, moving outwards an adjacent edge of the to-be-corrected edge of the current correction mask plate figure; and repeating the procedures until the error of the border positions is within the preset value. The method disclosed by the embodiment of the invention improves the efficiency of optical proximity correction.

The invention relates to a device and a method for judging subjective questions in a test paper based on a natural language, the method including: processing a text of an answer of an examinee by clauses and dividing the text of the answer of the examinee into a plurality of clauses; carrying out word segmentation and part-of-speech tagging on the clauses after the clauses in the clause processingmodule; conducting keyword extraction of part-of-speech tagged words in part-of-speech tagging module; calling a syntactic analysis module to analyze the clause after the clause processing module, and analyzing the syntactic structure information of the clause; an obtaining module matching the keywords of the examinees' answers with the scoring keywords provided by the reference answers, and obtaining the similarity between the keywords of the examinees' answers and the scoring keywords; match the text clauses of the candidates' answers with the clauses provided by the reference answers, andobtaining the sentence similarity between the text clauses of the candidates' answers and the clauses provided by the reference answers. The invention realizes the machine correction of the subjectivequestion, lightens the burden of the marking teacher, improves the marking efficiency, and guarantees the fairness and justice of the scores of the examinees.

The present invention relates to a pipe-jacking tunnelling machine, in particular to an active-type automatic deviation rectification device of the pipe-jacking tunnelling machine. The device can realize automatic, highly efficient, safe and reliable deviation rectification according to deviations of proceeding direction of the pipe-jacking tunnelling machine with respect to a pipeline designed axis. Deviation rectification output force of a hydraulic cylinder is weak; the active-type automatic deviation rectification device of the pipe-jacking tunnelling machine is capable of being used for curve pipe-jacking tunneling; technical basis for automation pipe-jacking tunnelling of the pipe-jacking tunnelling machine is provided; and problems of unsuccessful deviation rectification and low deviation rectification efficiency caused by misjudgments of operators or operating instruction input errors which are caused by influences of factors such as ever-changing construction conditions, extremely high labor intensity and insufficiency in operating experience or negligence of operation personnel are avoided; and therefore, pipe-jacking tunnelling efficiency is raised and quality of pipe-jacking construction is improved.

The invention relates to the technical field of power dispatching data network comprehensive operation and maintenance management systems, in particular to a power dispatching data network comprehensive operation and maintenance management platform which comprises a monitoring data acquisition module used for acquiring operation data and operation states of all devices of a power dispatching datanetwork; a data analysis module which is used for analyzing and counting the operation data, generating a data analysis result, and meanwhile, carrying out exception analysis on the data to obtain exceptional data; a data display module which is used for converting the data analysis result into a visual display format; and an information pushing module which comprises a conventional information pushing module and an alarm information pushing module, and the alarm information pushing module is used for generating alarm information according to the abnormal data and performing alarm pushing to corresponding personnel according to the alarm information. The power dispatching data network comprehensive operation and maintenance management platform provided by the invention can reduce the manpower consumption, improve the operation and maintenance efficiency, and is beneficial for ensuring the safe and stable operation of a power system.

The invention discloses an automatic deviation rectifying system for a shield machine and an operating method of the automatic deviation rectifying system. The automatic deviation rectifying system comprises an automatic laser position and posture measuring system, a data storage device, an industrial personal computer, a controller and a driving mechanism, wherein the automatic laser position and posture measuring system is used for monitoring the position and the posture of the shield machine in real time, the data storage device is used for storing correct and real-time position and posture data of the shield machine, the industrial personal computer is used for comparing the correct and real-time position and posture data of the shield machine and transmitting deviation rectifying control signals, the controller is used for controlling the driving mechanism to start, and the driving mechanism is used for rectifying deviation of the shield machine. The automatic deviation rectifying system for the shield machine and the operating method of the automatic deviation rectifying system can be used for automatically rectifying the running track of the shield machine, and has the advantages of high precision, high efficiency and the like. Besides, manual intervention in deviation rectifying is decreased, and constructional safety is improved.

The present invention provides a defect detecting method for electroluminescent display device. A dark spot defect caused by short-circuiting in an electroluminescence element is detected based on an emission brightness or a current flowing through the electroluminescence element when an element driving transistor which controls a drive current supplied to the electroluminescence element is operated in its linear operating region and the electroluminescence element is set to an emission level. Prior to the dark spot defect inspection, the dark spot defect is screened by applying a reverse bias voltage between an anode and a cathode of the electroluminescence element. With this process, disappearance of the dark spot defect at a later time resulting in an impossibility of laser repairing or the like is prevented, and inspection and repairing efficiencies are improved. A dim spot defect caused by a characteristic variation of the element driving transistor is detected based on a current flowing through the electroluminescence element or an emission brightness when the element driving transistor is operated in its saturation operating region and the electroluminescence element is set to an emission level.

The invention relates to a stochastic model modification method based on uncertainty of a stochastic response surface estimated parameter, which comprises the following steps: 1, representing an uncertain parameter of a structure system as a function of a standard stochastic variable and representing a response of the structure system as a polynomial chaos expansion which uses the standard stochastic variable as an independent variable and is based on a Hermite polynomial, and solving an undetermined coefficient of the polynomial chaos expansion so as to establish a stochastic response surface model of the response of the structure system and calculate a statistical characteristic value of the response of the structure system; 2, utilizing an error function of a statistical characteristic value of the stochastic response surface model and the statistical characteristic value of the actually measured response to establish a target function required by stochastic model modification; 3, utilizing the target function to construct optimization inverse problems and modifying a parameter statistical characteristic value step by step; and 4, on the basis of the parameter statistical characteristic value obtained by stochastic modification, utilizing a stochastic response surface to calculate and obtain the statistical characteristic value of the response of the structure system. The method ensures modification accuracy when improving modification efficiency.

The invention relates to a T-type material straightener, which comprises a stander and a transmission mechanism arranged on the stander, and is characterized by also comprising a wave bend straightening mechanism in front of the transmission mechanism and sidewise bend straightening mechanisms distributed on two sides of the wave bend straightening mechanism, wherein the wave bend straightening mechanism comprises an upper main pressure wheel and a lower main pressure wheel which are vertical and opposite; the sidewise bend straightening mechanisms comprise three side pressure wheel sets which are vertical and opposite, and a first side pressure wheel set, the wave bend straightening mechanism, a second side pressure wheel set and a third side pressure wheel set are arranged in turn along the feed direction; and pressing roll pressing planes on the three side pressure wheel sets are on the same horizontal plane as upper and lower pressing roll pressing planes of the upper and lower main pressure wheels. The T-type material straightener has compact structure and reasonable design, compared with manual flame straightening, the T-type material straightener has higher straightening efficiency and precision, and compared with common three-roll straightening, the T-type material straightener has the capability of straightening sidewise bending and angular deformation of T-type materials.

The invention discloses a finite element model correction method based on a positive substructure. The finite element model correction method includes the steps: firstly, dividing an integral structure finite element model into independent substructure models; building an integral structure characteristic equation according to the substructure models to obtain an integral structure characteristic solution; building an integral structure characteristic sensitivity equation according to the integral structure characteristic equation to obtain an integral structure characteristic solution sensitivity matrix; building a target function according to residual errors of the integral structure characteristic solution and a measured integral structure test mode; adjusting structural parameters of the substructure and minimizing the target function to obtain optimal structural parameters; adjusting structural parameters of the finite element model and correcting the finite element model; recognizing structural damage according to change of result parameters of the finite element model. When a structure is locally damaged and a local area needs to be corrected, large structure finite element model correction efficiency and precision are effectively improved only by iteratively analyzing the local substructure models without repeatedly analyzing the integral structure model.

The invention discloses a grinding wheel corrector, provides the grinding wheel corrector which has the advantages of simple structure, time-saving and labor-saving operation and use, high positioning accuracy of a correcting tool, high correction efficiency and high correction accuracy, and solves the technical problems of complex structure, time-saving and labor-saving operation and use, low grinding wheel correction efficiency, poor quality and the like of the grinding wheel corrector existing in the prior art. The grinding wheel corrector comprises a bracket and a tool rest which can horizontally slide along a cross beam on the bracket, wherein the tool rest is provided with the correcting tool; a longitudinal dovetail slide rail lug boss and a longitudinal dovetail slide rail groove which are matched with each other are arranged between the cross beam of the bracket and the tool rest; and the longitudinal dovetail slide rail lug boss is slidably connected in the longitudinal dovetail slide rail groove through a longitudinal lead screw and can drive the tool rest to horizontally slide along the longitudinal direction of the bracket.

The invention relates to a CMOS temperature sensor which comprises a bipolar transistor, an over-sampling analog-digital conversion module, a correction module and at least two unit current sources. The output end of each unit current source is coupled to the collector electrode of the bipolar transistor through a switching module. The collector electrode and the base electrode of the bipolar transistor are connected with a reference ground and then connected to the analog-digital conversion module; the analog-digital conversion module is connected with a reference voltage; the switching module switches all the unit current sources to the bipolar transistor; the analog-digital conversion module outputs a first conversion result; then the unit current sources are switched to the bipolar transistor within a collecting cycle of the analog-digital conversion module in sequence, only one unit current source can be connected each time, and the analog-digital conversion module outputs a second conversion result; the correction module combines the first conversion result and the second conversion result to obtain a final correction value. The CMOS temperature sensor can improve correcting efficiency.

The invention relates to random model updating method based on an interval response surface model. The method is characterized by including the steps of firstly, building a second-order polynomial response surface model without cross terms according to experiment design and regression analysis; secondly, using a square completing method to convert a polynomial response surface expression into perfect square; thirdly, substituting interval parameters into the response surface expression to allow the definite response surface model to be changed into the interval response surface model; fourthly, performing interval calculation on the interval response surface model to obtain predicted structural response intervals, and combining the predicted structural response intervals with actual response intervals to build a target function; fifthly, building a optimization inversion problem to identify interval distribution of parameters. By the method, the expansion problem of interval calculation is avoided, fast calculation of structural response intervals is considered, finite element analyzing calculation and sensitivity matrix building during (interval) random model updating are avoided, a large amount of calculation time and cost is saved, and ill-conditioned optimization is avoided as much as possible.

The invention discloses an automatic paper marking method and device, and belongs to the technical field of computer equipment. The method comprises the steps that a target test question image is acquired, the target test question image comprises at least one target answer area, the target answer area comprises an image of answer content of a target question, and the target question is a questioncorresponding to the target answer area; Positioning a target answer area in the target test question image, and obtaining the type of the target test question; According to the type of the target question, identifying the image of the answer content in the positioned target answer area; And judging whether the answer content is correct or not according to the recognition result, and outputting ajudgment result. According to the technical scheme provided by the invention, the problem of low test question correction efficiency can be solved.

An online correcting device of absolute grating ruler reading head parameters relates to the technical field of grating ruler measuring. The online correcting device comprises a PCI (peripheral component interconnect) communication card and a PC (personal computer). The PCI communication card comprises a PCI interface controller and a second FPGA (field programmable gate array). The PC comprises a PCI interface and a main processor. Command parameters are sent to the PCI interface controller, the second FPGA receives the command parameters and transmits the same to a first FPGA of a grating ruler reading head, and the first FPGA writes data into a ROM (read only memory) in the reading head. During data transmission, state of the PCI communication card is inquired, when the PCI communication card is idle, data and data CRC (cyclic redundancy check) codes are transmitted, so that integrity and accuracy of data transmission is guaranteed. By the correcting device, online correction of grating ruler reading head parameters is achieved, and correction efficiency of grating ruler reading head is improved.

The invention discloses an integral correction device for a long and thin beam structure and a correction method, and belongs to the field of welded structure correction. In order to solve the problem in the prior art that the long and thin beam structure is liable to deform during the welding process, the correction device is formed by assembling a plurality of correction units, and two adjacent correction units are fixedly connected through a connection mechanism, wherein each correction unit structurally comprises an adjusting support, a platform which is arranged on the adjusting support and used for carrying a beam and compaction mechanisms which are arranged on the two sides of the platform and used for compacting the beam. The correction method adopts the correction device to correct the long and thin beam structure with good correction effect. The invention effectively overcomes the difficulty in correction process of the beam due to length, tool and other factors, can integrally handle welding deformation of the beam, makes the correction process more centralized and simple, greatly increases work efficiency, and reduces the labor intensity of workers.

The invention provides a correction method of power authorization of an E-DCH physical uplink channel (E-PUCH). The method comprises the following steps: determining time slots in which a plurality of the E-PUCHs are located in a current transmission time interval (TTI), wherein a plurality of the E-PUCHs are scheduled for UE; selecting a first type of the time slots possessing the single E-PUCH and a second type of the time slots possessing a plurality of E-PUCHs from the time slots; for any time slot in the first type of the time slots, when a power sum of all carrier waves in the time slot is greater than a largest power authorization of the UE, recording the single E-PUCH in the time slot and performing down-regulation to emission power of the single E-PUCH; for any time slot in the second type of the time slots, when the power sum of all the carrier waves in the time slot is greater than the largest emission power of the UE, recording the time slot and determining priorities of a plurality of E-PUCHs in the each time slot which is recorded, and successively performing the down-regulation to the emission power of a plurality of E-PUCHs in the each time slot. By using the method of the invention, an efficiency of power authorization correction can be increased and a communication quality can be guaranteed.

The invention discloses a video timing correction method which comprises the steps of extracting audio data in a video to be corrected, and identifying voice text corresponding to human voice in the audio data; acquiring a text mouth shape waveform corresponding to a speech text from a pre-established text mouth shape database; recording a voice time section for playing the voice of a person, andgenerating a voice mouth waveform according to the voice time section and the text mouth waveform; extracting video data from a video to be corrected, and obtaining a picture mouth shape waveform corresponding to a portrait face in the video data; comparing the similarity of speech waveform and picture waveform, and correcting the timing of speech and picture according to the result of comparison.The invention also discloses a terminal, a computer-readable storage medium. The invention realizes the automatic detection of the consistency of audio and video playing when playing video, and automatically completes the timing correction of audio and video playing when inconsistency is detected, so as to improve the correction efficiency of audio and video synchronous playing.

The invention relates to a plate straightener, which comprises a power mechanism, a stander, a transmission mechanism, and a drive straightening mechanism, wherein the transmission mechanism comprises a first transmission shaft and a second transmission shaft which are arranged on a base by a shaft seat, and the head end of the transmission mechanism is connected with the power mechanism; the drive straightening mechanism comprises a first straightener, a second straightener, a third straightener and a fourth straightener, and the four straighteners have the same structure and comprise a drive part arranged on the base and transmitted by a transmission shaft and a straightening part arranged on a top seat; and the drive parts of the first straightener and the second straightener is arranged on the first transmission shaft, the third straightener and the fourth straightener are arranged on the second transmission shaft, the first straightener, the second straightener, the third straightener and the fourth straightener are in staggered arrangement in turn along the direction from the head end to the tail end of the transmission shafts, and the straightening part and the drive part of the straightener are vertical and opposite. The plate straightener has high automatic degree and good reliability, and can effectively reduce the labor intensity and improve the production efficiency and economic benefits.

The invention provides an optical proximity effect correcting method. A setting mode of a target position of a line segment is not fixed, but is changed along with a graph, and generally, the curve endpoint of an analog graph corresponding to the line segment is used as the reference position of the target position. The curve endpoint can be judged by the following method: if the line segment translates in a direction perpendicular to the line segment, a point, which is tangential to the analog graph, of the line segment is the curve endpoint, and at the moment, the curve endpoint coincides with the target position. For the line segment, EPE at the position, which corresponds to the curve endpoint, of the line segment is maximum generally, image correction is performed according to the EPE, correction frequency can be reduced effectively, insufficient correction degree or excessive correction degree is reduced, an analog result is matched with a target value, and the correction efficiency and the correction precision are improved. The invention further provides an optical proximity effect correcting system.

The invention provides an automatic wrongly written character correcting method in a search engine and a server. The wrongly written character in the text can be corrected more effectively; each character is mapped to a high-order space by deeply learning a model and high-dimensionally vectoring, the relation between the characters is represented through the high-dimension vector, and the context information of the character and the effect thereof in the sentence are used for recognizing whether the character is the wrongly written character. According to the method provided by the invention, a one-to-one corresponding relation between the wrongly written characters and the correct characters does not need to be constructed by consuming a range of costs, but only a proper wrongly written training correction model needs to be constructed to learn the features of the wrongly written character. The wrongly written character in the sentence is recognized and corrected by considering the semantic and grammar of the sentence, the word characteristic and the context information of the word by the technical scheme adopted by the invention, the recognized wrongly written character is not only the homophonous character or characters with the similar form, and according to the technical scheme, the wrongly written words of other types can be further recognized and corrected, and the correction efficiency of the wrongly written character is greatly improved.

The invention discloses a method for straightening the flame of a large-deformation automobile drive axle housing. The method is characterized in that a deformed axle housing is put on a detection platform to measure the axle housing deformation; sand is fully filled in an inner cavity of the deformed axle housing, the deformed axle housing is sealed and is then put on a 'V'-shaped piece fixed bearing, the other parts of the deformed axle housing are kept suspended in the air, the back side of a working surface of the deformed axle housing is put upwards, and loading application force points for assisting the straightening are positioned on two end positions, at which an axle housing spring seat assembly is installed; and heating wires which have the same length and width and are symmetrically distributed are marked on the axle housing along the longitudinal direction sequentially, acetylene flame is used to move from one ends to the other ends of the heating wires to heat, after each heating is completed, water is adopted to cool the axle housing, and flame heating and loading force applying are repeatedly carried out until the axle housing deformation reaches the tolerance allowed range of the axle housing design. By the method, the material shaping reserve consumption can be effectively avoided, and accordingly, the reutilization ratio of the large-deformation drive axle housing is increased.

The invention relates to a lingual bracket. A bracket body consists essentially of a side wing and a bracket seat, wherein the side wing is convexly arranged on an upper surface of the bracket seat, and a lower surface of the bracket seat is provided with anti-skid grains; the lingual bracket is characterized in that: the side wing is arranged by deviating from one end of the bracket seat; a ligature hook is arranged at one side of the side swing positioned at the inner side of the bracket seat, which is toward the middle part; the side wing is arranged in an offset way, thus being capable of vacating an upper space at the other end of the bracket seat, leading the bracket to be capable of being firmly cemented at one end of teeth close to toothcusp, increasing orthodontic force thereof, and also decreasing the influence on chewing as jaw teeth bit the bracket to further reduce the possibility of bracket dropping, and save clinical time. One side wing can be arranged as a replaceable and movable side wing, so that the width of the groove can be adjusted; and the movable side wings with different types can be replaced when in clinical needs, and individualized adjustment can be carried out on the torque angle and protrusion and the like of the teeth.

The invention discloses a license plate inclination correction method and belongs to the field of computer vision and intelligent transportation technology. The license plate inclination correction method is a new inclination correction method proposed for solving the problem that inclination of a license plate may disturb license plate character segmentation and character recognition. According to the method, an inclination angle is determined according to a projection value mean square error of the license plate, projection value information is well utilized, projection values of a license plate image in different directions are calculated by changing the angle theta, a horizontal and vertical correction mode is adopted, and it is not to calculate projection values of an original image when the inclination angle is calculated but to extract parts of the original image to form new small images for operation. Therefore, a large calculation amount can be reduced, efficiency is also greatly improved, and correctness of inclination angle detection is improved to a certain extent.

The invention provides a test paper answering area identifying and test question marking method and system, and belongs to the technical field of mobile Internet. The method comprises that m pieces of answering handwriting information are obtained; according to handwriting interval distances in each piece of answering handwriting information, answering areas are divided successively, and a test question amount set C and an answering area set R are formed; the amount n of test questions and the amount of effective answering areas are determined; information of each effective answering area of each piece of answering handwriting information is identified and counted; answers to test questions are determined; and each effective answering area of each piece of answering handwriting information is marked according to each answer to the test question, and a test question marking result of each piece of answering handwriting information is obtained. The method and system can be used to identify the answering areas automatically and accurately, determine the amount of test questions, mark objective questions of test paper, save time and improve the efficiency.

The test paper correction method provided by the embodiment of the invention comprises the following steps: segmenting question stem image information by utilizing a pre-trained neural network model,and segmenting the question stem image information into a plurality of first fields; identifying question numbers, question types and knowledge points in the plurality of segmented first fields and corresponding position information of the question numbers, the question types and the knowledge points by utilizing a computer identification technology, and generating a unified first segmentation image; segmenting the answer image information by using a neural network model, and segmenting the answer image information into a plurality of second fields; identifying characters, numbers and formulasin the plurality of segmented second fields and corresponding position information of the characters, the numbers and the formulas by utilizing a computer identification technology, and generating aunified second segmented image; and comparing the second segmented image with the corresponding standard answer, and determining a score corresponding to the second segmented image. According to the test paper correction method based on artificial intelligence, the workload of manual reading is reduced, and the test paper correction efficiency is improved.