153results about How to "Ensure controllability" patented technology

This invention provides a resource recovery device system for lithium precipitation mother liquor. Through the coordinated use of a pretreatment unit, conditioning unit, extraction unit, washing unit, back-extraction unit, and raffinate post-treatment unit connected in sequence, it achieves maximum recovery and utilization of lithium ions in the lithium precipitation mother liquor, reduces lithium resource waste, ensures the greenness, stability, and controllability of the extraction process, and results in a high-lithium liquid product with low metal impurity ion content and high lithium extraction rate, while the raffinate meets discharge standards. The resource recovery device system has a simple structure, low operation difficulty, and low manufacturing cost.

The present application relates to the technical field of bamboo fiber composite material, in particular to a purification preparation method of bamboo fiber reinforced natural polymer composite material. The bamboo raw material is crushed and treated by steam explosion to obtain primary bamboo fiber; the primary bamboo fiber is sequentially subjected to organic solvent dewaxing treatment and oxidation system selective delignification treatment to obtain purified bamboo fiber; the purified bamboo fiber is treated in a tannin acid solution to form a polyphenol interface layer on the surface thereof; and a solution containing polyvalent metal ions is added to the treated bamboo fiber to form a coordination cross-linking interface layer on the surface of the bamboo fiber. The present application constructs a stable interface layer between the fiber and the natural polymer matrix. The interface layer can effectively transfer stress, significantly improve the tensile strength, bending strength and shear strength of the composite material, and solve the problem of weak interface and easy delamination of traditional natural composite materials.

The invention relates to the technical field of high-speed aircraft health management, and particularly discloses an interpretable aircraft state sensing and predicting system and method. The method comprises the following steps: constructing and training a lightweight anomaly diagnosis network with an expansion perception layer; aircraft dynamics, kinematics and fault mechanism formulas are used as soft constraints to be embedded into the future state estimation network; an SHAP interpretation framework is used for sorting the importance of the input features, and a network structure is optimized accordingly; and finally, accurate diagnosis of abnormal types and positions and rapid prediction of aircraft states at a plurality of moments in the future are realized. According to the method, by introducing three innovation points of an expansion perception layer, physical information embedding and feature importance sorting, the diagnosis accuracy and the model interpretability are remarkably improved, and the high-performance calculation requirement in a scene with limited calculation power is met.

This invention relates to the field of seismic data processing technology, and discloses a method, apparatus, and device for smoothing seismic data. The method includes: acquiring seismic data to be processed; constructing an objective function based on the seismic data to be processed and the smoothed seismic data; and solving the objective function using an alternating multiplier algorithm to obtain the smoothed seismic data corresponding to the seismic data to be processed. The seismic data smoothing method disclosed in this application can eliminate abrupt changes in the drastically changing parts of the seismic data while maintaining the relatively stable parts, effectively ensuring the controllability of the seismic data smoothing result.

An AI-based iterative language expression training system and its usage method include the following steps: Constructing a language expression training task library, where each task unit contains first descriptive information (such as topic, keywords) and second descriptive information (such as standard examples, structural analysis); arranging the task units according to difficulty or logical order to form a training sequence; the user learns according to an iterative forward training method, starting from the first task unit and attempting to generate or express corresponding language content based on the first descriptive information; the system calls an AI language model (such as BERT, GPT) to evaluate the user-generated content in multiple dimensions (including logical coherence, semantic accuracy, and subjective emotional consistency); if the evaluation fails, the system pushes second descriptive information for the user to learn and reverts to the first task unit to start again; if the evaluation passes, the user proceeds to the next task unit until all forward training is completed; then, the training sequence is logically reversed, and iterative reverse training is performed, starting from the original last task unit, requiring the user to reverse-engineer or parse the first descriptive information based on the second descriptive information, also employing an iterative reinforcement mechanism of evaluation-learning-reverting. This invention significantly improves users' abilities and proficiency in language organization, logical expression, and creative writing through a closed loop of forward and reverse iterative training and AI intelligent evaluation.

This application relates to the field of industrial furnace temperature control technology, and discloses an energy-saving brazing furnace heating process temperature adaptive control method and system. The heating process of the brazing furnace is divided into at least three continuous temperature ranges, including a first temperature range, a second temperature range, and a transition temperature range. A first power control strategy and a first temperature fluctuation tolerance are configured for the first temperature range, and a second power control strategy and a second temperature fluctuation tolerance are configured for the second temperature range. The current temperature range of the brazing furnace is determined based on the current temperature. When it is determined that the brazing furnace is currently in the first temperature range, the first power control strategy is executed to output heating power, ensuring that the characteristic temperature is within the first temperature fluctuation tolerance. When it is determined that the brazing furnace is currently in the second temperature range, the second power control strategy is executed to output heating power, ensuring that the characteristic temperature is within the second temperature fluctuation tolerance. When it is determined that the brazing furnace is currently in the transition temperature range, a smooth switching logic is executed.

The application discloses an electrochemical luminescence measuring cell with a hook hand lifting mechanism and relates to the conveying technical field. The electrochemical luminescence measuring cell with the hook hand lifting mechanism comprises an outer box body, an electrochemical measuring cell mechanism and a reaction cup. Two sealing doors are movably arranged on one side of the outer box body. An adjusting mechanism is arranged at the bottom of the electrochemical measuring cell mechanism. The adjusting mechanism comprises a reversible motor one fixedly arranged on one side of the inner box body, a transmission shaft fixedly arranged at the output end of the reversible motor one, a brake one sleeved outside the transmission shaft, a support frame two and a brake two, and a positioning assembly arranged outside the support frame two. The support frame two is provided with the hook hand lifting mechanism at the top. The hook hand lifting mechanism is matched with the adjusting mechanism to clamp and fix a new reaction cup and then place the reaction cup on the top of the electrochemical measuring cell mechanism. The reaction cup can be automatically replaced, the safety of the experiment of the harmful substances escaping in the reaction process is ensured, the replacement speed of the reaction cup is ensured, and the safety hidden danger of the workers is reduced.

This utility model belongs to the field of sludge filtration technology and discloses a dynamic sludge filtration device, including a housing. A filtration unit and a discharge unit are connected to the outside of the housing via a rotating motor and a discharge motor, respectively. The filtration unit and discharge unit are located inside the housing. A water pumping unit is located at the top of the housing and connected to the water pumping unit via a filtration water pumping pipe. The discharge unit is located at the bottom of the filtration unit and the water pumping unit. This utility model can increase the sludge contact area and filtration area, allowing liquid to penetrate into the filter housing more quickly. Combined with the rotating water pumping component, the liquid is promptly extracted, improving the speed and efficiency of sludge dewatering and shortening the processing cycle. The rotating motor drives a rotating rod, which in turn causes the filter housing to rotate within the sludge. Dewatering is achieved using the principle of free liquid permeation. Compared to traditional methods that rely on strong external force for dewatering, this effectively reduces the moisture content of the filter cake and significantly reduces energy consumption.

The application discloses a kind of welding methods of high-strength steel shell and low-strength steel inner shell penetration pipe of jacketed pressure vessel, core step includes: precision pretreatment before welding: size and defect detection, homogenization modification and preheating are carried out to machining bevel; Sequential optimization welding: the sequence of "first inside then outside" and "sectional skip welding" is combined, and prior to pipe side (high-strength steel side) is carried out backing surfacing welding; Strict welding control: the combined process of tungsten argon arc welding backing and electrode arc welding filling is used, interlayer temperature, heat input and weld size are accurately controlled; Timely and effective post-welding treatment: high-temperature post-heat treatment is carried out immediately after welding, or according to the structure constraint degree, choose whole intermediate annealing.The application effectively decomposes and releases welding stress through systematic process combination, inhibits the generation of crack, and significantly improves the welding quality and reliability of jacketed vessel penetration pipe joint.

The application provides a chip burning method, a computer device and a computer readable storage medium. The chip burning method comprises the following steps: generating a plurality of different temporary keys; combining the plurality of temporary keys into a key sequence and storing the key sequence in a temporary key file; receiving an encrypted burning file from a second terminal; matching the temporary keys in the key sequence of the encrypted burning file with the temporary keys stored in a local key database one by one; when the temporary keys are successfully matched with the temporary keys in the local key database, downloading a decrypted program file to a burner, the program file is used for burning to a target chip, and then deleting the matched temporary key from the local key database. By generating a plurality of different temporary keys and storing the temporary keys in the local database, the number of burners that can be used at the same time is controlled by matching the temporary keys one by one and deleting the used keys immediately after successful matching, and the total burning quantity can be accurately controlled.

The invention provides an automatic operation and maintenance scheduling system and method in a cloud computing environment. The method comprises the following steps: predicting a load demand curve through historical load time sequence data and real-time load indexes of an online service cluster; generating an arbitration decision of resource operation and maintenance scheduling according to the simulated resource occupation contour of the online service cluster constructed by the load demand curve and the resource request queue corresponding to the offline task cluster; issuing a reserved resource permission to the second control ring by the arbitration decision, and injecting a virtual node resource view complementary with the reserved resource permission in time and space into the first control ring; and when the real-time load index triggers the first control ring to perform resource scaling operation, the first control ring executes instance scaling in a resource boundary limited by the virtual node resource view, and the second control ring starts and stops the offline task instance in the corresponding resource block according to the current reserved resource permission. According to the technical scheme provided by the invention, collaborative scheduling of operation and maintenance of the online service cluster and the offline task cluster can be realized under load fluctuation.

The invention relates to the technical field of pressure drop adjustment, in particular to a shelf calandria refrigerant pressure drop self-adaptive adjustment method and system, and the method comprises the steps: determining a plurality of target calandria groups based on a shelf calandria; executing the following operations on each target calandria group in the plurality of target calandria groups: determining a target pressure drop range based on the target calandria group; acquiring an initial pressure monitoring value set group based on the pressure monitoring unit and a target calandria group; acquiring an actual pressure drop based on the pressure difference analysis unit and the initial pressure monitoring value set group; a pre-built refrigerant valve is adjusted based on the flow adjusting unit, the actual pressure drop and the target pressure drop range, an updated valve is obtained, and an updated pressure drop is obtained based on the updated valve; and realizing pressure drop self-adaptive adjustment based on the updated pressure drop. According to the invention, accurate adjustment of the pressure drop of the refrigerant can be realized.

The application relates to a time axis interaction method, device and equipment. After a visual time axis of a target business module is created in an interaction interface, the set time span is acquired, and a current time scale is drawn in combination with a current progress time. When a user performs a dragging operation, the dragging action is listened to, and a pixel offset amount accumulated in the dragging process is acquired. According to the pixel offset amount, a time offset value is calculated in combination with a preset unit pixel time length. Then, the time offset value continuously accumulated along with the dragging is superimposed on the current progress time of the target business module to update the current progress time, and the current time scale of the time axis is redrawn based on the updated current progress time, so that the time scale continuously moves backward along with the accumulation of the pixel offset amount generated by each dragging operation. Finally, the time axis sliding range limit is broken through, and infinite time span browsing is supported without increasing the rendering burden.

The application provides a soft structure and a processing method thereof and a soft robot, and belongs to the technical field of growing robots, and comprises a main air chamber and at least two sub air chambers, the sub air chambers are distributed in a ring shape along the outer peripheral surface of the main air chamber and extend along the axial direction of the main air chamber; a plurality of extension turning parts are arranged in each sub air chamber, the extension turning part comprises a first buckle and a second buckle, and the first buckle and the second buckle are fixedly arranged on the inner wall of the sub air chamber; wherein, when the first buckle and the second buckle are buckled, a wrinkle part is formed in the sub air chamber; when the first buckle and the second buckle are separated, the wrinkle part can be unfolded; when the soft structure is stretched out, the corresponding sub air chamber is inflated, and the first buckle and the second buckle are automatically separated; an air nozzle, the air nozzle is communicated with the corresponding sub air chamber and is used for inflating the sub air chamber; the advantage is that when the soft structure is stretched out from the main air chamber, the corresponding sub air chamber is inflated, the cross-sectional size of the sub air chamber is changed, the first buckle and the second buckle are relatively displaced and separated, and the wrinkle part is unfolded as required.

The invention discloses a TBM double-track tunnel trackless transportation cross-line bypassing cushion layer rapid construction method, and belongs to the technical field of tunnel construction. According to the method, a construction area is divided into a plurality of construction sections in the longitudinal direction of a tunnel, construction and transport vehicle framing and side alternate operation is organized, and parallel cushion layer pouring and trackless transport are achieved; the method comprises the following steps of: performing radar detection and scabbling treatment on the top surface of an inverted arch block during construction preparation; installing a steel plate water stop belt, and penetrating through the ground beam and the drainage blind pipe; a detachable steel pipe is adopted for shaping and constructing a cushion drainage ditch; after a reinforcing mesh is laid, concrete pouring, multi-pass surface finishing, steam and heat preservation curing and temperature and humidity monitoring are conducted; by optimizing the construction organization and technology, the problems that transportation is interrupted, the construction period is long and cracking is prone to occurring in a traditional method are solved, efficient, high-quality and low-interference cushion layer construction is achieved, optimal construction parameters can be scientifically screened by setting a contrast test section, and adaptability is high.

The invention discloses a robot control method based on active disturbance observation and phase coupling control, and the method comprises the steps: collecting the current state quantity of a robot in real time, and constructing a joint space dynamic model; generating state estimation and disturbance estimation in the joint space dynamic model; a phase parameterization method is adopted to map the phase of the multi-degree-of-freedom action to an expected trajectory, and a closed-loop trajectory tracking and disturbance compensation control law is designed in combination with disturbance estimation and a sliding manifold mechanism; according to the method, a position error transformation design continuously-derivable robot auxiliary level function is introduced, the stability is analyzed by adopting a Lyapunov method, and a closed-loop trajectory tracking and disturbance compensation control law and a coupling gain adaptive law are substituted into a time derivative of a candidate Lyapunov function to complete the control of the robot. The control output can be kept smooth and continuous, instruction sudden change or attitude instability is avoided, high action success rate and landing stability can still be maintained when the environment changes, and the landing impact load peak value is effectively reduced.

The utility model discloses a kind of adjustable heating device for composite electric contact material, and heating device includes rack and heating furnace;Partition heating pipe for being used to supply belt material to pass is replaced or alternatively inserted in heating cavity;Partition heating pipe inner cavity is not communicated with heating cavity space, and heating cavity and / or partition heating pipe are separately independently ventilated into protective atmosphere;Composite equipment includes belt material unwinding device, material polishing device, heating device, rolling cooling device, belt material winding device sequentially arranged along the direction of transmission of belt material to be compounded;The application adopts variable protective atmosphere structure of insertable partition heating pipe, flexible construction of heating interval is realized, just replace different types or specifications of partition heating pipe, heating and atmosphere protection adjustment under different materials or process conditions can be satisfied;Not limited to the specific heating source or specific setting of heating furnace, without changing the overall structure of heating furnace, improve the adaptability and flexibility of composite equipment in multi-process scene.

The utility model discloses a toy bubble gun which comprises a gun-shaped main body, a foaming assembly and a driving device, the foaming assembly and the driving device are both installed on the main body, and the foaming assembly and the driving device are in transmission connection through a linkage structure. The linkage structure is further in transmission connection with a transmission mechanism, the transmission mechanism is in transmission connection with a doll, the linkage structure drives the foaming assembly to operate and also drives the transmission mechanism to operate, and the transmission mechanism drives the doll to rotate. When the driving device operates, the linkage structure drives the transmission mechanism to operate, and then the doll is driven to rotate. In addition, the linkage structure is arranged to be a stepped gear, so that the transmission structure is simplified, and the efficiency is improved; the transmission mechanism comprises a bevel gear, a coupling rod, a transmission gear and the like so as to achieve power transmission and rotation of the doll. In conclusion, the design of the toy bubble gun aims at combining the dual functions of bubble emission and doll rotation, so that the interestingness and interactivity of the toy are improved.

This utility model discloses a slag cooling device for a vertical kiln, including a base plate. A fixing plate is fixedly connected to the top of the base plate. The vertical kiln body is fixedly connected to the inner side of the fixing plate. A drive motor is fixedly installed on the outer side of the fixing plate by bolts. A conveyor belt is driven to the output end of the drive motor. An electric telescopic rod is fixedly installed on the top of the fixing plate. A baffle is fixedly connected to the output end of the electric telescopic rod. A distance sensor is fixedly connected to the surface of the baffle. A water tank is fixedly connected to one side of the fixing plate. This utility model uses a water pump to circulate water in the water tank through a circulating water pipe, which can remove some heat and play a preliminary cooling role, reducing the temperature around the vertical kiln. The water pump also delivers water to atomizing nozzles through a drain pipe. The atomizing nozzles atomize the water and spray it onto the slag, increasing the contact area between the water and the slag and accelerating the cooling speed of the slag.

The present application relates to the technical field of data desensitization, and in particular to a differential hierarchical desensitization method and system based on dynamic context perception, comprising receiving a data access request; collecting multi-dimensional context information in real time; calculating a dynamic risk score using a risk scoring model; querying a dynamic strategy knowledge base according to the score and the context, matching and determining a differential desensitization instruction; performing real-time desensitization processing according to the instruction and returning a result; recording audit logs throughout the process, and the system accordingly comprising a context perception engine, a strategy decision engine, a desensitization execution engine and an audit tracing center. The present application quantifies access context into a dynamic risk indicator, drives the desensitization strategy to change from a static rule to an adaptive decision, realizes differential and refined data protection for different access scenarios, and effectively balances data security and business availability.

This disclosure relates to the field of core network equipment operation and maintenance, and proposes a method and device for diagnosing the status of core network elements based on dynamic AI algorithms. The method includes: connecting with a third-party platform data source to obtain real-time status data of core network elements; outputting anomalies in the real-time performance indicators of network elements based on dynamic security thresholds; outputting anomaly information for detected abnormal log data and alarm information; and deriving a final health score for the network element based on anomalies, anomaly information in abnormal log data and alarm information, test results, and complaint information, triggering a hidden danger investigation task. This invention analyzes multi-dimensional data (performance, alarms, logs) using AI algorithms, and ultimately derives the network element status through a comprehensive health assessment algorithm. It can automatically adapt to network changes, possesses high intelligence and automation characteristics, and significantly improves the accuracy and efficiency of fault detection and location.

This invention provides a control method for dual-motor driven tracked vehicles based on a neural network prediction model. By introducing a Long Short-Term Memory (LSTM) neural network framework, it trains a neural network prediction model for the tracked unmanned platform, fully leveraging the LSTM's ability to process sequential data. This allows for the extraction of vehicle dynamics features from past driving data to predict the state at the next moment. This method can comprehensively control lateral displacement deviation, heading angle deviation, longitudinal velocity, and yaw rate, as well as track the desired trajectory position, while also ensuring steering controllability. Furthermore, this invention provides a trajectory tracking control method based on Neural Network Model Predictive Control (NN-MPC), simultaneously considering the vehicle's kinematics and dynamics, effectively overcoming the problem of poor trajectory tracking performance caused by the strong coupling of lateral and longitudinal dynamics in the physical model MPC method.

The utility model discloses a gas state monitoring sensor calibration device, which relates to the technical field of sensor calibration and comprises a calibrator, a standard micro-water sensor, a standard pressure sensor, a gas circuit adjusting system, a humidity calibration module and a pressure calibration module. The calibrator is provided with a sealing cavity, a first port and a second port, wherein the first port and the second port are communicated with the sealing cavity. According to the technical scheme of the utility model, the humidity calibration module and the pressure calibration module are adopted, so that the dual calibration function of gas humidity and pressure parameters can be realized, the problem of single function of the existing calibration device is solved, multi-parameter integrated calibration is realized, the calibration efficiency is improved, and the application range is widened; by adopting the modular design and switching the four pipelines through the on-off control assembly, a user can select different humidity calibration modules or pressure calibration modules according to needs, the operation is convenient, and the automation level and the operation consistency of the system are improved.

The utility model discloses a brewing funnel and a coffee machine. The brewing funnel comprises a funnel body and a filter screen assembly connected in the funnel body in a floatable mode. The part, arranged in the funnel body, of the filter screen assembly is matched with the inner wall of the funnel body to form a containing cavity used for containing coffee powder. The end, away from the containing cavity, of the filter screen assembly partially extends out of the funnel body, and after the part, extending out of the funnel body, of the filter screen assembly is subjected to acting force facing the containing cavity, the part can apply acting force to coffee powder so that the coffee powder can be separated from the containing cavity. Through the design of the floatable filter screen assembly, coffee powder can be actively pushed to be separated from the containing cavity, and the problem that the coffee powder in a traditional funnel adheres to a filter screen is solved; after use, the residual coffee powder can be easily removed only by pressing the filter screen assembly, the residual powder does not need to be manually dug, and the cleaning time and workload are greatly reduced; design is ingenious, the structure is relatively simple, and manufacturing and maintenance are easy; a user can clean the coffee powder only through simple pressing, and operation is simple and visual.

This application discloses a knowledge base management method, system, and medium based on data mining technology, mainly relating to the field of knowledge base management technology, to address the problems of difficulty in balancing privacy protection and data utility in existing solutions, and the risk of privacy data leakage during computation. The method includes: migrating encrypted numerical data from a privacy knowledge database to a visual database; obtaining the specific data and formulas involved in the computation through the corresponding display interface of the visual database; when encrypted numerical data exists in the specific data, inputting the encrypted numerical data and calculation formulas into a pre-trained preset neural learning algorithm to obtain first result data, and inputting unencrypted numerical data into a conventional formula to obtain second result data; and inputting the first result data and second result data into a merging result formula to obtain the final calculation result.

The invention discloses an automatically-adjusted workpiece clamping system and a control method, and belongs to the technical field of machining, clamping and positioning. The system comprises a clamp base, a lineation detection unit, an image acquisition unit, a hydraulic adjustment unit and a control unit. The hydraulic adjusting unit comprises hydraulic cylinder sets in the X direction, the Y direction and the Z direction, the two hydraulic cylinders in the Z direction are communicated through a hydraulic communicating device to form linkage, the two hydraulic cylinders in the Z direction and the other two hydraulic cylinders jointly form a three-point supporting structure, and the over-positioning problem of four-point supporting is solved. The control unit carries out processing and feature extraction on the image, calculates the six-degree-of-freedom deviation of the workpiece, calculates the adjustment amount of each hydraulic cylinder according to the deviation and controls the hydraulic cylinders to act. The system realizes closed-loop automatic adjustment of the position of the workpiece by cyclically executing the detection and adjustment process until the six-degree-of-freedom deviation is within the allowable range. Automation and high precision of large workpiece clamping and positioning are achieved, and the device is suitable for precise positioning of large heavy workpieces such as train chassis.

This invention relates to an event-triggered model predictive obstacle avoidance control method and system, belonging to the field of navigation control technology for unmanned vehicles, automobiles, and robots. The method addresses constrained discrete-time systems by establishing a predictive model of the controlled object and constructing a model predictive control problem in the rolling time domain, encompassing state constraints, input constraints, terminal constraints, and obstacle avoidance constraints, thereby achieving attitude adjustment, trajectory tracking, and safe obstacle avoidance. This invention represents the controlled object's occupied region and the obstacle region as a unified convex set, constructing obstacle avoidance constraints based on minimum distance optimization and its dual transformation, avoiding direct handling of non-smooth distance determination and improving optimization efficiency. Control updates employ an event-triggered mechanism, determining whether to resolve the optimization problem based on the deviation between the actual and predicted states; otherwise, the existing control sequence continues execution. Furthermore, the prediction time domain is adaptively adjusted according to the operational phase, reducing the optimization frequency and computational scale while ensuring obstacle avoidance safety and control performance, thus improving the real-time performance and stability of navigation control under complex constraints.

The utility model relates to a dispensing tool for carton forming processing, which comprises a main frame body, a rolling dispensing assembly is rotatably mounted on the main frame body, and the rolling dispensing assembly is connected with a pressurized liquid supply assembly mounted on the main frame body through a liquid supply supporting pipe. The rolling dispensing assembly comprises a rotating disc, supporting rib rods, a mounting ring plate and a telescopic dispensing mechanism, the rotating disc is rotatably connected with the main frame body, and the rotating disc is coaxially connected with the mounting ring plate through the supporting rib rods which are annularly arranged at intervals; and a plurality of telescopic dispensing mechanisms which radially coincide with the mounting ring plate are inserted into the mounting ring plate in a rotational symmetry manner. According to the utility model, the dispensing spacing distance can be changed according to requirements, so that the dispensing quality and the dispensing efficiency are improved, and meanwhile, the pressure-contact glue discharging can be realized, so that the overflow and splashing of glue liquid are reduced.

The embodiment of the invention discloses a data processing system of a dense array and a device thereof. The data processing system of the dense array comprises an equipment unit which comprises an array and a mobile communication device, and the array is composed of seismometers of different models; the acquisition unit is used for providing a gateway for data synchronization and information transmission of the mobile communication equipment and providing a format conversion tool for ensuring that data is transmitted according to a unified format; the data unit is used for providing unified data access, data management and index management for multi-source heterogeneous data, and providing data whole-process management and access such as data acquisition, management, archiving storage, query access and the like; the software function unit is used for carrying out standardization processing on the data in the data unit and carrying out task management on array layout in the equipment unit; and the application unit provides a plurality of platform products according to the functions of the data unit and the software function unit.

The application discloses a kind of multi-modal big model sequence generation-oriented credible decoding method, including obtaining original image and text input;Local focus and global noise processing are carried out to original image respectively;Original image, local visual input and global noise visual input are respectively input to multi-modal big language model with text input, and original output, local focus output and global noise output distribution are obtained;Based on the difference between original output, local focus output and global noise output distribution, the output distribution after calibration is calculated;High-risk hallucination word is dynamically identified in combination with modal dependency evaluation criteria, and the output probability distribution of high-risk hallucination word is calibrated using the output distribution after calibration;Multi-modal big model sequence generation-oriented credible decoding method can effectively alleviate the semantic fragmentation and object hallucination problem caused by statistical bias and language prior, significantly improve the output reliability of multi-modal big language model in sequence generation task.