366results about How to "Eliminate dependencies" patented technology

The invention discloses a mechanical arm operation method based on a multi-mode large visual language model, which comprises the following steps of: firstly, acquiring image data of a mechanical arm operation scene through a visual sensor, and acquiring a user instruction in combination with a voice recognition module or a text input module; and inputting the natural language instruction and the visual scene data into a multi-modal large visual language model for multi-modal task analysis. The model identifies the category, position and attribute of a target object in a scene, and generates a specific operation sequence in combination with an input instruction. And then, according to the operation sequence generated through analysis, the system directly drives the mechanical arm to execute corresponding tasks including specific actions such as grabbing, moving and placing of the target object, and in the task execution process, the mechanical arm sequentially completes the actions according to the operation sequence generated in advance. The problems that in the prior art, a mechanical arm is inaccurate in complex task instruction analysis, poor in operation flexibility and adaptability and poor in real-time response capacity in a dynamic environment are effectively solved.

The invention relates to the technical field of power system automation, and provides a photovoltaic inverter low-voltage ride-through test result automatic analysis method, which comprises the following steps: reading low-voltage ride-through test data of a photovoltaic inverter, firstly judging whether no-load test data is qualified, and further judging whether load test data is offline; if the load test data is not offline, classifying the test data into a three-phase fault, a two-phase fault and a single-phase fault according to fault types, further calculating the voltage drop depth, the drop duration, the reactive current response time, the reactive current duration, the reactive current exit time and the active power recovery rate, and if the calculation result meets the standard requirement, determining that the test data is offline. And if not, automatically storing the test waveform and the calculation result in word, and otherwise, outputting that the test waveform and the calculation result are unqualified. The method and the system do not depend on manual identification, automatically and quickly calculate the low-voltage ride-through test result of the photovoltaic inverter in batches, automatically store the test waveform and the calculation result in the word document, and are simple to implement and high in accuracy.

The invention discloses an electric power system state estimation method based on graph contrast learning. The method comprises the following steps: constructing graph sequence data of electric power system state estimation; constructing an electric power system state estimation model based on graph contrast learning, and putting the graph sequence data into the electric power system state estimation model based on graph contrast learning; inputting the node feature sequence and the masked edge features into a hybrid encoder of a dynamic edge convolutional neural network combining a self-attention mechanism and a residual block and a graph convolutional neural network with a dynamic propagation strategy of a gating mechanism to comprehensively process the features of nodes and edges; comparing an updating mechanism and a sampling symmetry training strategy of model parameters of the learning framework through a graph; and calculating a node voltage phase angle by using the branch phase angle difference, and outputting a state estimation result. According to the method, the internal structure and mode of the power system data can be effectively learned by combining graph contrast learning and a deep neural network technology, and the capability of identifying and filtering abnormal data is improved.

The invention provides a learning-based map-free autonomous navigation obstacle avoidance method, a medium and equipment. The method comprises the following steps: carrying out map-free autonomous navigation obstacle avoidance on the mobile robot by adopting a reusable planning capability library; the multiplexing planning capability library comprises a global planning module and a local planning module; the global planning module is used for guiding a global navigation direction and generating ordered and discrete path points; the local planning module generates real-time actions based on a learned local navigation obstacle avoidance model; and controlling the mobile robot to move according to the action generated by the local planning module so as to sequentially navigate the mobile robot to each path point generated by global planning. According to the method, through a global and local combined planning architecture, an algorithm is modularized, so that the method adapts to different navigation tasks, a large amount of retraining based on a deep reinforcement learning navigation method is avoided, and the universality and interpretability of a model are improved.

The invention provides a secondary network thermal load prediction method based on a CEEMDAN-KPCA-LSTM algorithm. The method comprises the following steps: dividing a complemented data set into a training set and a test set according to a proportion; key factors influencing the dynamic change of the thermal load of the secondary network in the training set data are used as input variables, and the thermal load data are used as output variables; performing local encryption and standardization processing on the thermal load data; building an improved neural network model for predicting the dynamic thermal load of the building; training the improved neural network model, and when the improved neural network model meets a convergence condition, obtaining a trained improved neural network model; performing standardization processing on the test set data, and setting test set input data according to an input variable selected from the test set data after standardization processing of the training set data; and inputting test set input data into the trained improved neural network model to realize prediction of the dynamic thermal load of a certain building in the future 24 hours.

The invention discloses a GNSS interference identification method based on spatial transformation, multi-scale feature fusion and KAN. The method comprises the steps of forming an interference signal data set; obtaining a time-frequency image training set and a time-frequency image test set; constructing a GNSS interference identification classifier composed of a spatial transformation module, a multi-scale feature fusion module and a KAN module; a trained GNSS interference identification classifier is obtained; and obtaining a final GNSS interference identification classifier. And finally, outputting the category of the GNSS interference signal acquired in real time by the GNSS interference identification classifier. According to the invention, the spatial transformation module is used to carry out translation, scaling and other operations on the interference signal time-frequency image, and the number of training samples is increased. And the multi-scale feature fusion module adopts parallel convolution kernels of different scales to extract multi-scale spatial features of the time-frequency graph, so that the richness of feature representation is increased, and the recognition accuracy is further improved. The KAN module uses a KAN to replace a traditional multi-layer perceptron to execute a final classification task, and a more complex nonlinear function can be fitted by using fewer parameters.

The invention discloses a remote sensing image ship segmentation method based on a U-KAN network model. The method comprises the following steps: constructing the U-KAN network model; the U-KAN network model is a hybrid network architecture fusing a U-Net network and a KAN network; wherein the U-Net network comprises an encoder and a decoder which are in jump connection; a plurality of KAN layers are nested in the KAN network; u-KAN network model training is carried out in combination with the remote sensing image data set, and an optimized U-KAN network model is obtained; and inputting a remote sensing image to be processed into the optimized U-KAN network model to obtain a corresponding ship segmentation result. The method not only can effectively capture ship targets with different scales and complexities in the remote sensing image, but also can realize effective fusion between shallow detail features and deep semantic features, thereby overcoming the limitation in processing complex backgrounds and diversified ship forms in the prior art.

The invention belongs to the technical field of nonlinear cooperative control, and particularly relates to a data-driven nonlinear cluster unmanned system security cooperative control method and device. The method comprises the following steps: S100, aiming at a nonlinear cluster unmanned system, designing a distributed state feedback controller based on a feedback gain matrix, and constructing an equivalent closed-loop system based on the controller; s200, acquiring input data and state data of the nonlinear cluster unmanned system through an offline experiment, and realizing non-parametric representation of the equivalent closed-loop system by using the input data and the state data; and S300, state constraint is converted into limitation on an attraction domain of a closed-loop system, a distributed feedback gain matrix is calculated based on parameterized representation of the closed-loop system, then a state feedback controller is obtained, data-driven cooperative control over the nonlinear cluster unmanned system is achieved, and the situation that the state of a single body in the system enters a dangerous area is avoided.

The invention discloses a lightweight vehicle intranet identity authentication and key negotiation method based on a PUF (Physical Unclonable Function). The method comprises a system initialization stage, a PUF registration stage, an identity authentication and key negotiation stage and a block chain synchronization stage. The method is oriented to an automobile domain architecture, direct low-cost identity authentication is carried out on the electronic control unit by utilizing the PUF (namely a physical unclonable function), the necessity of installing safety hardware on the electronic control unit can be eliminated, and secret key leakage is prevented; meanwhile, symmetric cryptography is adopted, a hash function, a pseudo-random function and a Lagrange interpolation formula are combined to balance the speed and safety of the key negotiation process of the electronic control unit, and compared with an existing vehicle intranet identity authentication and key negotiation method, the method has remarkable performance advantages.

According to the encrypted video playing method and device, the storage medium and the computer equipment provided by the invention, in response to a playing request of target video data of a user in a client, an encrypted SM4 symmetric key is acquired from the playing request; determining a user certificate corresponding to the playing request, and decrypting the encrypted SM4 symmetric key by using a private key of the user certificate to obtain a temporary SM4 symmetric key; determining a VKEK plaintext of the target video data, and encrypting the VKEK plaintext by using the temporary SM4 symmetric key to obtain a target VKEK ciphertext; and returning the target VKEK ciphertext to the client, so that the client plays the target video data after decrypting the target VKEK ciphertext by using the temporary SM4 symmetric key, and destroys the temporary SM4 symmetric key. Therefore, the video data playing method and the video data playing device do not depend on USB-Key hardware equipment, so that various clients can play the video data conforming to the C level of the GB35114-2017 protocol specification.

The invention discloses and introduces a linear complexity model architecture for speech recognition in the technical field of speech recognition, an AMLP branch uses an MLP to replace an attention mechanism, an attention average pooling layer is added on the basis of the MLP, uniform distribution of attention weights is ensured while the linear complexity is kept, comprehensive global features are obtained, and the accuracy of speech recognition is improved. According to the convolution branch, an enhanced local feature relationship is captured by using a convolution space gating unit, local features and global features are mixed through a channel interacting information with the AMLP branch, and the local features are supplemented for the global features. Advanced fusion is properly carried out during local feature and global feature extraction processing, so that the problem of real-time mutual influence of the global features and the local features is solved.

The invention discloses a DAG-based cross-fragmentation point-to-point energy transaction optimization method, and belongs to the technical field of cross-fragmentation transaction processing. Comprising the following steps: defining a cross-fragment transaction; describing a dependency relationship among a plurality of sub-transactions based on a directed graph (DAG) structure; constructing a coordination layer based on a DAG account book, and selecting a node with the highest reputation value to serve as a DAG committee member; and carrying out transaction verification, cross-fragment transaction forwarding and verification during forwarding on the nodes serving as DAG committee members. According to the method, the optimistic strategy is adopted to process the cross-fragmentation transaction in parallel, and the cross-fragmentation transaction is decomposed into a plurality of sub-transactions in combination with a DAG structure, so that the atomicity and consistency of the transactions are ensured; in a high-frequency transaction scene, the throughput and the processing rate of the system are remarkably improved, and the transaction delay is effectively reduced; a DAG committee is formed in cross-fragment internal nodes, a DAG account book is maintained, efficient cross-fragment transaction processing is achieved, dependence on a trusted third party is eliminated, and communication and verification cost is remarkably reduced.

The invention belongs to the field of robot motion control, and discloses a dynamic sliding mode motion control method for a mobile manipulator based on dynamic online identification feedforward, which comprises the following steps of: 1, designing a physical consistency linearization coupling dynamic model of the mobile manipulator; 2, designing a self-adaptive Kalman filtering algorithm based on modeling error covariance dynamic updating, and realizing high-precision online identification of the physical consistency kinetic parameters of the mobile mechanical arm; 3, designing a recursive dynamics feedforward algorithm of the mobile mechanical arm; 4, a self-adaptive over-torque dynamic sliding mode control algorithm based on recursive dynamics feedforward is designed, the continuity of the control law is guaranteed, and high-precision finite time convergence of the motion error of the mobile mechanical arm is achieved; and 5, the stability of the controller is achieved through a Lyapunov design method, a paradigm framework of robot dynamics feedforward finite time motion control is formed, and the response speed and precision of robot motion control can be effectively improved.

The invention relates to a voice assistant resource search test method and device, a platform, equipment and a medium, and the method comprises the steps: constructing a request body according to the resource information of any test case in a test case set, and transmitting the request body to a to-be-tested interface of a resource search end to request target resource path information, the test case set is pre-constructed based on a resource data table, and the test cases are literary and style format cases containing resource information; constructing a test message based on the resource information and the target resource path information, and sending the test message to at least one target test client in a notification mode to request to return information; and identifying target resource test information from the resource page screenshot according to the preloaded configuration file and the resource type in the return information, and determining a test result according to the target resource test information and a preset resource identifier in the return information.

The invention discloses a magnetic catheter, an intervention device and a catheter intervention method.The magnetic catheter comprises an inner-layer tube and an outer-layer tube which are arranged in a nested mode to form the inner diameter and the outer diameter of the magnetic catheter, a middle layer is fixedly embedded between the inner-layer tube and the outer-layer tube, and the middle layer is an elastic layer and can be bent in the circumferential direction; the thrust effect can be transmitted and delivered along the axial direction; the magnetic part and the developing part are both arranged at the front end of the magnetic catheter in the advancing direction, the magnetic part has single polarity, and the developing part is used for improving the detectability of the magnetic catheter under rays. The inner-layer tube, the outer-layer tube and the middle-layer structure are combined to form a main body structure of the magnetic catheter, so that the catheter has supporting performance, pushing performance and bending adjusting performance, meanwhile, through the magnetic part and the developing part, the end of the catheter can be clearly observed, a non-contact magnetic field force turning effect is provided for the catheter, and the magnetic catheter can be conveniently used. The compatibility of the magnetic catheter is improved, the operation convenience is improved, and the operation difficulty of the operation is reduced.

The invention is suitable for the technical field of oil-gas exploration, and provides a seismic data convolution cross-correlation intelligent FWI method, which integrates a proposed seismic data convolution cross-correlation FWI target equation into a deep learning framework, and combines an automatic encoder technology to significantly reduce the nonlinearity of the traditional FWI, effectively avoid the cycle jump problem, and improve the accuracy of the FWI. Therefore, the speed model inversion precision of the FWI is improved. Finally, the FWI technology which can still carry out stable and accurate high-resolution speed modeling under the conditions that seismic source wavelets are unknown, the initial speed is unknown and low-frequency data are missing is realized, and the method has important significance on seismic exploration.

The invention provides a component management system and method. The system comprises a plurality of server nodes and a logic controller. The server node is used for generating a component management request of a to-be-controlled component according to the management information corresponding to the to-be-controlled component in the plurality of components, and sending the component management request to the logic controller; the logic controller is respectively connected with each server node and each component, and is used for determining a target register in a plurality of register modules in the logic controller according to component information and management requirements obtained by the component management request, and determining the target register according to the component information and the management requirements as well as the current management state of the to-be-controlled component; and calling the target register to perform management operation on the to-be-controlled component. According to the method, the logic controller centrally processes the management operation requests of all server nodes on the component, the dependence on a single main node is eliminated, decoupling between the nodes is realized, and the timeliness, reliability and flexibility of component information acquisition are improved.

The invention belongs to the technical field of robots, and discloses a quick'eye-on-hand 'hand-eye calibration method for accurate clamping at the tail end of a robot, which is used for independently calibrating two relative relations of rotation and translation of a coordinate system in order to reduce the solving difficulty of a nonlinear optimization problem. The method comprises the following steps: firstly, obtaining a CCS rotation matrix which is the same as an ECS through a three-step rotation calibration process; a translation vector between the CCS and the ECS is then calculated using the coordinate transformation. And finally, the relative relation between the center of a clamping jaw at the tail end of the robot and the binocular camera is calibrated, and real-time, in-place and accurate clamping of visual guidance is achieved. According to the method, high-precision positioning of the tail end of the robot for accurate clamping can be achieved, after calibration is completed, the center of a clamping jaw at the tail end of the robot can be accurately guided to any space target point in real time through a binocular camera, and extra LT guiding is not needed.

The invention discloses a digital pattern reference image library construction method for multi-style fusion building restoration. The method comprises the following steps: constructing a reference image library consisting of a plurality of single-style building component reference images; acquiring a plurality of images of the same building component, performing preprocessing to extract contour information, selecting one of the images with the most complete contour information, calculating the similarity between the selected image and the corresponding reference image in the reference image library, and matching the reference image with the similarity higher than a similarity threshold value; comparing the contour information of the to-be-repaired building component image with the reference image to obtain a defect position; fusing and reconstructing the plurality of to-be-repaired building component images with different defect positions to obtain a fused image; the pictures meeting the requirements are added into a reference picture library through manual approval; and if the defect exceeds the defect threshold, matching the reference images with the same age, the same region and the same main style to perform fusion reconstruction, performing manual approval again, and adding the fused image meeting the requirement into the reference image library.

The invention discloses a locomotive brake detection performance determination method, and the method comprises the steps: employing a vehicle-mounted device on a locomotive to collect dynamic data generated in a brake test process in real time, and the dynamic data comprise air cylinder pressure, pipeline pressure, a brake handle position signal and other operation related parameters; the collected dynamic data are preprocessed, a test standard curve set is constructed in combination with historical test data, and the standard curve set is classified and stored according to the locomotive model and the brake model. According to the invention, a standard test data set is constructed based on a large amount of test data of locomotives and brakes of different models, feature extraction and fitting analysis of dynamic data are realized in combination with a feature sampling algorithm, a least square fitting algorithm and a multiple linear regression model, and through comparison of a fitting curve and a standard test curve, a dynamic test result is obtained. The system automatically judges the normalization and integrity of the operation, and subjective errors and omissions of manual analysis are avoided.

The invention relates to the technical field of data security and privacy protection, in particular to a block chain and edge computing-based medical data sharing and access control platform, which comprises a multi-layer data storage system, a block chain network, an intelligent contract module, a plurality of data consumers, a plurality of data providers and a plurality of hospital servers. According to the invention, the calculation and communication overhead in data transmission of the medical internet of things is reduced, and the sharing efficiency and security requirements of the medical data between a provider and a consumer can be met.

The invention discloses a production method of hexachlorotrisiloxane. Production devices mainly comprise a reaction synthesizer, a condensation system, a rectification system, a final-stage condensation system and a storage system. The method comprises the following steps: reacting raw materials silicon tetrachloride and primary alcohol according to a certain ratio under the catalytic action of a boron Lewis acid catalyst in an environment that the reaction temperature is controlled to be 650-800 DEG C and the reaction pressure is controlled to be 0.6-0.8 MPa to obtain a mixed gas of hexachlorotrisiloxane and a by-product; and the high-purity hexachlorotrisiloxane is obtained after cooling of the condensation system and separation of the rectification system. The invention breaks through the blank of the production process of the hexachlorotrisiloxane in China, thereby eliminating the dependence of the demand of the hexachlorotrisiloxane product in China on foreign import, providing a stable production process, expanding the productivity of the hexachlorotrisiloxane, and realizing the production of the high-purity hexachlorotrisiloxane product. The problems of high-end industries in China are solved, and sustainable development of related industries in China is promoted.

A magnet array holder configured to hold a magnet piece and / or a non-magnet piece to form a magnet array, the magnet array configured to manipulate one or more electron beams in a vacuum electronic device when assembled, the magnet array holder comprising: a set of slots configured to receive the magnet piece and / or the non-magnet piece; a set of pockets for accommodating the magnetic and / or non-magnetic parts; and one or more attachment interfaces configured to couple the magnet array holder to the vacuum electronic device.

The utility model relates to the technical field of cranes, and particularly discloses a portal crane anchoring device which comprises a crane body and a track, a mounting plate is arranged on one side of the crane body, an anchoring component is pivotally connected onto the mounting plate, a connecting plate is arranged on the side surface of the mounting plate, and one end of the connecting plate is pivotally connected onto the mounting plate. The connecting plate is connected with the anchoring assembly through a bolt, a first anchoring block is arranged on the side wall of the track, and a second anchoring block is arranged on the ground on one side of the track. According to the anchoring device, the anchoring assembly is matched with the first anchoring block and / or the second anchoring block in an inserted mode, anchoring of the crane body is achieved, anchoring is stable and reliable, the anchoring assembly can be selected to be in anchoring connection with the first anchoring block and / or the second anchoring block according to different field conditions, flexibility is high, anchoring efficiency is improved, and the anchoring device is convenient to use. A pure mechanical structure is adopted as a whole, the cost is low, dependence on electricity, gas and liquid is eliminated, the failure rate is reduced, and use is stable and reliable.

The invention discloses an MIRA-LFD kit and method for rapidly detecting alternaria kikuchiana, and belongs to the technical field of rapid detection of plant diseases. The primer probe combination for rapidly detecting the alternaria kikuchiana is designed and screened by targeting the Alt a1 gene of the alternaria kikuchiana, and the primer probe combination is further developed into a freeze-dried product which can be stably stored and has good redissolution detection performance; according to the detection scheme, a NaOH solution (0.5 M) can be adopted to coarsely extract DNA at normal temperature, meanwhile, MIRA reaction can be completed by utilizing palm center temperature, and color development can be well performed on the colloidal gold test strip, so that field on-site detection of alternaria kikuchiana is effectively realized; moreover, the detection scheme also has the advantages of simplicity in operation, short detection time, no dependence on instruments and equipment, low detection limit, accuracy, reliability and the like, is convenient to popularize and use in grassroots, and has important significance on early diagnosis and precise prevention and treatment of the pear black spot.

The invention discloses a biasing circuit and a radio frequency power amplifier, and belongs to the technical field of amplifiers. The bias circuit comprises bias control modules which are arranged in one-to-one correspondence with amplifiers at all levels in the radio frequency power amplifier; the bias control module comprises a coupling control unit and a clamping transistor. The coupling control unit is used for filtering the radio-frequency signal of the bias end and then outputting the radio-frequency signal through a second connecting end of the coupling control unit; a parasitic clamping diode is arranged between the control electrode and the first electrode of the clamping transistor, and the parasitic clamping diode is used for chopping a signal of the second connecting end of the coupling control unit, so that the change trend of the direct current component of the voltage of the first electrode of the clamping transistor is the same as the change trend of the envelope amplitude of the radio frequency signal; the voltage of the first electrode of the clamping transistor passes through the coupling control unit and then is provided to the bias end as a bias voltage. According to the embodiment of the invention, the working state of the amplifier is optimized by adjusting the bias voltage in real time.

The present invention discloses an automobile intelligent interaction system for realizing AI interaction, including an intelligent control box, a Beidou chip, a communication chip, a camera, a massager, a sleepiness reminder module, a screen projection module, an emergency rescue module, an AI voice invitation module and a line recommendation module. The emergency rescue module is connected to the intelligent control box, and the intelligent control box analyzes whether it is necessary to actively send rescue information based on external sensors. If emergency rescue is required, the specific coordinates of the vehicle, the owner's contact information and the image are automatically uploaded to the emergency rescue module, and the rescue personnel go to rescue according to the information; the AI ​​voice invitation module is connected to the intelligent control box, and the address book personnel are invited by AI voice; the line recommendation module is connected to the intelligent control box, and the line recommendation module is based on user behavior and resource characteristics. The present invention can realize multiple functions such as emergency rescue, AI voice invitation, intelligent line recommendation, sleepiness reminder, and intelligent screen projection.

A batch process model-free output feedback control method with non-repetitive interference belongs to the technical field of industrial process control, and specifically comprises the following steps: step 1, giving a mathematical description for an optimal tracking control problem of a non-repetitive batch process; 2, designing an optimal control law based on output feedback; 3, a game Q function construction method based on two-dimensional output feedback in the intermittent process; 4, solving a two-dimensional output feedback problem of batch operation by utilizing reinforcement learning; 5, solving an optimal control strategy based on output feedback through iteration to enable the optimal control strategy to be finally converged to an ideal optimal value; according to the method, the problem that a traditional method for processing the complex characteristics of a two-dimensional system is too tedious is solved, through the output feedback control method based on reinforcement learning, the method does not depend on a system model, the influence on the control performance is reduced, and the control effect is greatly improved.

The invention discloses a tyramine signal amplification system based on a rare earth fluorescent material and application of the tyramine signal amplification system in immunodetection. Firstly, the preparation method of the tyrosine functionalized rare earth fluorescent nanoparticles is provided, and the prepared tyrosine functionalized rare earth fluorescent nanoparticles can realize a tyramine signal amplification system without combination of a biotin-avidin system, so that the biotin-avidin combination step of traditional tyramine signal amplification can be simplified; and meanwhile, the detection cost is also reduced. The invention further provides a method for hypersensitive digital fluorescence immunodetection based on tyramine signal amplification, synchronous detection of multiple targets can be realized, fluorescence images of the multiple targets can be synchronously acquired under the same excitation light source, and tedious operation of frequent optical filter switching in multiple digital immunodetection is avoided. Furthermore, the rare earth fluorescent material doped based on different lanthanide ions can be subjected to digital detection through time resolution.

The invention discloses recombinant escherichia coli for efficiently producing gamma-aminobutyric acid as well as a construction method and application of the recombinant escherichia coli, and belongs to the technical field of gene engineering or biology. The recombinant escherichia coli capable of efficiently producing gamma-aminobutyric acid is obtained by heterologous expression of a mutant glutamate decarboxylase gene gadB C11 in E.coli BL21 (DE3) and heterologous expression of a GABA / Glu reverse transport channel gene gadC C41 connected in series with weak RBS and a pyridoxal phosphokinase gene pdxY connected in series with strong RBS under the action of an araBAD promoter. The gamma-aminobutyric acid is inoculated into an LB liquid culture medium containing kanamycin and ampicillin to be cultured until OD600 is equal to 0.6, and IPTG, 1-arabinose and 1-glutamic acid are added for conversion, so that the gamma-aminobutyric acid can be obtained. According to the method, the pH value does not need to be adjusted, the gamma-aminobutyric acid can be synthesized by constructing a reverse transport path and an endogenous PLP refilling path in escherichia coli and directionally modifying a rate-limiting enzyme, and the problems that the GABA production cost is high, the enzyme activity is low, and the optimal pH value of GAD is incompatible with one-step production of GABA are solved.