85results about How to "Reduce inconsistency" patented technology

The invention discloses an intelligent navigation system and method for a ship in the technical field of ship navigation, and the system comprises a sensing module which is used for collecting the surrounding environment information and the state information of the ship in real time; the data processing and fusing module is used for preprocessing, calibrating and fusing the surrounding environment information and the self state information of the sensing module; the route optimization module is used for formulating a navigation plan and a collision avoidance strategy; the navigation control module is used for autonomously controlling the ship according to the output of the ship route optimization module; and the communication monitoring module monitors the navigation state of the ship and the change of the surrounding environment through real-time communication between the ship and a shore-based control center, other ships and a maritime affair management department. According to the method, data of different sensing units are associated, and the associated data are fused, so that mutual verification and complementation can be realized, and the accuracy and reliability of target identification are improved.

The invention discloses a lifelong target re-identification method based on descriptive text prompt prototype compensation, relates to the field of computer vision, and aims to display and solve the problem of feature drift caused by model increment update by introducing a prompt prototype compensation module. The module generates a multi-modal vision and text prototype for each target category through a vision language pre-training model, and maps old task features to a new task feature space through a cross attention mechanism so as to realize fine-grained semantic drift compensation. In addition, a boundary perception prototype correction module is designed to solve the problem of boundary confusion caused by updating of old prototypes. The module dynamically samples false features in compensation distribution to correct old prototypes with fuzzy boundaries, focuses on categories easy to confuse to maintain clear decision boundaries, optimizes cross-task and intra-task correction loss to reduce category confusion, effectively alleviates the problem of disastrous forgetting, and improves the adaptability and generalization ability of the model.

The invention discloses a short-term power load prediction method based on deep learning, and the method comprises the steps: S1, collecting historical power load data and corresponding weather and holiday and festival information, S2, carrying out the preprocessing of the historical data based on the historical power load data and the change of a power utilization behavior mode of a user, reducing the data inconsistency, and carrying out the prediction of a short-term power load. The method comprises the following steps: S1, processing historical power load data, weather information and holiday and festival information, S3, constructing a deep learning model which inputs processed historical power load data, weather information and holiday and festival information, S4, training the constructed deep learning model by using a training data set and optimizing model parameters, and S5, carrying out short-term power load prediction by using the trained deep learning model and generating a prediction result. According to the short-term power load prediction method based on deep learning, the problem of inconsistency of historical data caused by frequent change of power utilization habits of users in a changeable user behavior mode in the prior art is solved.

The invention discloses a construction method and system for a membrane material database, and relates to the technical field of database construction.The construction method comprises the following steps that in the operation process of a database management system, multi-dimensional parameter information generated by the system is collected in real time, collected original data is cleaned and preprocessed, different types of data are normalized, and a database management result is obtained; and the consistency and the validity of the data are ensured. According to the method, the multi-dimensional parameters of the database management system are collected in real time, cleaning and anomaly analysis are carried out, the system intelligently evaluates the operation state, the operation anomaly coefficient is generated, and early warning and fault prediction are achieved. Based on the comparison between the abnormal coefficient and the multi-level threshold, the system divides the risk into low, middle and high levels, and takes corresponding measures. According to the refined management mechanism, the stability of the database is effectively improved, data loss, inconsistency and crash are prevented, and the accuracy and efficiency of fault processing are improved.

The invention provides an unmanned ship target tracking decision control method based on deep reinforcement learning. The method comprises the following steps: establishing an unmanned ship motion model to obtain a current motion state of an unmanned ship; establishing a target tracking task reward function; constructing a deep reinforcement learning network model based on a soft actor-commentator algorithm, and performing initialization; acquiring a current position and a predicted position of the target ship; obtaining an optimal control strategy of the unmanned ship and an updated motion state of the unmanned ship by using a deep reinforcement learning network model; randomly selecting a control strategy to obtain an updated motion state of the target ship, storing the updated motion state of the unmanned ship and the updated motion state of the target ship as training samples in an experience pool, and repeating the above process when the target ship is not tracked; and when the target ship is tracked, randomly sampling from the experience pool to update the parameters of the deep reinforcement learning network model. The intelligent decision-making capability and tracking performance of the unmanned ship are improved, and the application development of the unmanned ship in the civil and military fields is promoted.

The invention belongs to the field of big data, and relates to a data processing method which comprises the following steps: acquiring initial data; sending the initial data to a pre-constructed real-time queue system, wherein the real-time queue system is used for caching and scheduling the received data; according to a preset priority strategy and a preset scheduling strategy, performing real-time processing on the initial data in the real-time queue system to obtain candidate data; analyzing the candidate data by adopting a preset graph data model to obtain graph structure data corresponding to the initial data; and importing the graph structure data into a graph database. The invention further provides a data processing device, equipment and a medium. In addition, the invention further relates to a block chain technology, and the initial data, the candidate data and the graph structure data can be stored in a block chain. According to the method, the real-time performance, accuracy and efficiency of importing the data into the graph database can be improved.

The invention relates to the technical field of image data processing, in particular to a license plate recognition method and system in a strong backlight environment, and the method comprises the steps: carrying out the graying of a collected image, and obtaining a gray backlight image; extracting a grey level histogram of each image, determining a maximum value point, evaluating a peak value attribute through a grey level difference and a pixel quantity of the maximum value point, and further determining a grey level separation point; performing threshold segmentation on the image, extracting foreground and background regions, calculating a reference value index, and screening out high-quality foreground and background images through threshold judgment; sorting the high-quality images to obtain a retention index, and performing threshold judgment again to determine foreground and background images which are preferentially retained; and the license plates are arranged and combined to form a spliced image, the backlight degree and the edge features of the spliced image are evaluated, finally, the adaptation degree is obtained through sorting, license plate character codes are extracted, a complete license plate number is formed, and license plate recognition under strong backlight is achieved. The invention aims to improve the accuracy of license plate recognition in a backlight environment.

The present invention relates to the technical field of sleep diagnosis, and in particular to a sleep staging method based on deep learning, the method comprising: collecting a number of target physiological signals, preprocessing each target physiological signal to obtain a number of standard physiological signals; synchronously aligning and integrating each standard physiological signal in turn to obtain a standard sleep data set; determining the actual correlation between each physiological signal to be integrated in the standard sleep data set, and determining an actual signal combination strategy according to the actual correlation; generating an actual sleep data set according to the actual signal combination strategy, and extracting features from the actual sleep data set to generate actual feature parameters; inputting the actual feature parameters into a deep learning model to determine the actual sleep state; analyzing the actual sleep state, and performing corresponding sleep intervention according to the analysis results. The present invention realizes low-cost non-intrusive sleep monitoring, thereby improving people's physical health.

The invention relates to the technical field of sleep electroencephalogram analysis, in particular to a sleep electroencephalogram analysis method and device, which more comprehensively assesses the sleep state and improves the reliability and credibility of the result through double verification of frequency domain analysis and image analysis. The method comprises the following steps: acquiring sleep electroencephalogram time domain information of a to-be-analyzed user; the sleep electroencephalogram time domain information at least comprises one physiological signal including an electrooculogram, an electroencephalogram, an electrocardiogram and a mandibular electromyogram; inputting the sleep electroencephalogram time domain information into an electroencephalogram time-frequency domain conversion model to obtain a sleep electroencephalogram frequency domain feature set; performing classified statistics on the sleep electroencephalogram frequency domain feature set by using a preset sleep activity feature reference to obtain a sleep state staging frequency domain analysis result of the to-be-analyzed user; the sleep states comprise sleep states in different time periods; and inputting the sleep electroencephalogram time domain information of the user to be analyzed into the electroencephalogram activity image recognition model.

The invention is suitable for the technical field of data processing, and particularly relates to a sensor data time synchronization error compensation method and device, and the method comprises the steps: obtaining the multi-sensor original data of a current frame, and extracting the multi-sensor feature data; obtaining an error estimation vector based on the error estimation network; performing error compensation on the multi-sensor original data to obtain multi-sensor compensation data; constructing a factor graph, and performing optimization solution on the factor graph to obtain a residual set; calculating a comprehensive quality score according to the residual set; according to the comprehensive quality score and the comprehensive quality scores of the first M frames, determining that a triggering condition is met, and triggering fine adjustment; and based on the multi-sensor original data, the multi-sensor feature data and the comprehensive quality score in the time window, performing fine adjustment on the model parameters of the error estimation network to obtain the error estimation network after fine adjustment. According to the method, more accurate compensation of time sequence errors and data distortion among multiple sensors is realized, and continuous self-adaptive optimization is realized.

The invention discloses an automatic news event extraction and combination system based on chapter-level text analysis, which is characterized in that text noise is cleaned and paragraphs are divided through a preprocessing module, and strong event candidates are screened by combining multi-modal conjoint analysis of picture titles and paragraph texts and utilizing semantic similarity calculation and named entity recognition technologies; a multi-task learning framework based on a Transform architecture is adopted to cooperatively extract elements such as time, places, figures, events, causes and passes, and the comprehensiveness and accuracy of event element extraction are improved; and the event similarity is calculated through a Bi-Encoder model, and the events are combined and sorted in combination with timestamps, so that an event chain with clear logic is generated. The system breaks through the limitation of existing sentence-level or paragraph-level analysis, remarkably improves the comprehensiveness, consistency and coherence of news event identification through chapter-level global context, data multi-mode combination and intelligent event chain construction, and has wide application value in the fields of news monitoring, public opinion analysis and the like.

The invention discloses a rapid assembly welding construction method for boiler steam blowpipes. The method comprises the steps that small-pipe-diameter and large-pipe-diameter welding auxiliary structures are manufactured, pipelines are hoisted to the installation position for butt joint, the corresponding welding auxiliary structures are selected according to the pipe diameters so that pipe openings of the pipelines can be aligned and preliminarily welded and positioned, and final butt joint welding is completed after the auxiliary structures are dismantled. An auxiliary mounting and welding structure of the automatic screwing device is further manufactured, and the device enables a screwing driving mechanism to be in butt joint with a connecting bolt, a nut or an adjusting bolt by adjusting the positions of all parts and drives the screwing driving mechanism to rotate. According to the assembly welding device, the technical effects of quickly and accurately completing assembly welding of the boiler steam blowpipes and improving the construction efficiency and quality are achieved.

The invention provides an intelligent electronic contract auditing and risk early warning method based on AI, and belongs to the field of contract auditing, and the method comprises the following steps: building a standard database for contract auditing, carrying out the pre-classification of contract types, selecting one contract from contracts of the same type as a standard template, and carrying out the risk early warning of the contracts; the standard templates are imported into a standard database to serve as templates, and each input standard template is classified and marked; a general auditing process and an independent auditing process are set for each type of contract, the general auditing process is used for comparing and verifying each preset item in the standard template with the contract signed by adopting the standard template, and target marking is carried out on distinguishing points in the general auditing process; wherein in the independent auditing process, terms added in a standard template are subjected to cross auditing, similar contracts of the same kind are selected through classification marking, and the whole contract is traversed to select similar terms for comparison auditing.

The utility model discloses an aluminum sheet polishing device, which belongs to the technical field of aluminum sheet processing and comprises a polishing device, supporting legs are connected below the polishing device, a control panel is connected to the front side of the polishing device, a conveying belt is arranged in the polishing device, a connecting rod is connected to one side of the polishing device, and the other side of the polishing device is connected with the control panel. And an aluminum sheet box is arranged in the connecting rod, one side of the polishing equipment is connected with a pushing device, and one side of the aluminum sheet box is connected with a cleaning device. According to the aluminum sheet polishing device, the supporting rod, the motor, the half gear, the tooth block and the push rod are arranged, the half gear drives the tooth block and the push rod to move, the push rod drives the push plate to move, the push plate drives an aluminum sheet to move through the sliding groove, then the aluminum sheet is pushed onto the conveying belt, and pushing of the aluminum sheet is completed at the moment; the uncertainty possibly caused by manual placement is eliminated, and the risks of potential inconsistency and quality fluctuation caused by manual operation are reduced.

The invention is suitable for the field of internet order management, and discloses an order management method, system and device based on a delay queue and a medium, the method comprises the following steps: generating a delay message through a merchant platform, storing the delay message in the delay queue, automatically monitoring a payment period, and constructing an event address book; interface addresses and security credentials of a plurality of platforms are integrated, and rapid retrieval and automatic configuration of information are realized; when an order needs to be processed, the system captures necessary information from an address book in time, automatically notifies a merchant and calls a payment platform to confirm a payment state, and once payment is confirmed, a delivery process is triggered, and a logistics platform is quickly connected to generate a waybill; according to the scheme, the configuration and maintenance process is simplified, the response speed and maintainability of the system are remarkably improved, and a more efficient and flexible solution is provided for Internet order management.

The invention provides a data processing method and device, electronic equipment and a storage medium, which are applied to a front-end service in a micro-service architecture, and the method comprises the following steps: receiving response data returned by a back-end service in response to a user service request; determining whether the response data accords with a preset format or not, wherein the preset format predefines field types included in the response data and data types corresponding to the field types; if not, inputting the response data into a preset interface adapter chain, and sequentially detecting a matching relationship between a preset processing format corresponding to each interface adapter and a data format corresponding to the response data until a target interface adapter matched with the data format is determined; and converting the data format into a preset format through the target interface adapter. The data format returned by the back-end service can be unified to improve the development efficiency, reduce the code redundancy, ensure the consistency and accuracy of the front end during data integration, and reduce the occurrence of data inconsistency and errors.

The invention relates to a storage battery dynamic balance management method and device, a storage medium and electronic equipment. The balance efficiency and the balance precision of storage battery dynamic balance management can be improved. The method comprises: determining a first difference value between a working parameter of each single battery in a storage battery pack and an average working parameter of the storage battery pack, the average working parameter of the storage battery pack being an average value of the working parameters of each single battery in the storage battery; under the condition that the absolute value of the first difference value is greater than a first threshold value, performing dynamic balance management on the storage battery pack based on a constant current control strategy; and under the condition that the absolute value of the first difference value is greater than a second threshold value and less than the first threshold value, carrying out dynamic balance management on the storage battery pack based on a constant voltage control strategy.

The invention provides an MPPT type solar charging and discharging light source integrated street lamp system, and relates to the technical field of solar street lamps. The method comprises the following specific implementation steps: firstly, acquiring actual electric energy through a solar energy conversion module, and acquiring battery parameters through a battery data acquisition module; secondly, the battery charging control module determines a charging mode based on the calibration parameters of different battery types, and dynamically adjusts charging current; then, the light source demand analysis module adjusts the actual illuminance of the street lamps according to the distance between the adjacent street lamps and meteorological data; then, the battery discharge control module adjusts the discharge current through the actual power demand of the light source and the discharge current nonlinear mapping; and finally, the battery state evaluation module predicts the health state of the battery through a battery state evaluation model and feeds back the health state to the battery charging control module and the battery discharging control module, so that the charging and discharging efficiency of different battery types is effectively improved, the service life of the battery is prolonged, and the stability and consistency of the street lamp system are further improved.

The utility model is suitable for the technical field of contrast agent preparation, and provides an automatic contrast agent preparation and injection device capable of diluting in a fixed proportion, which comprises a contrast agent storage tank, a normal saline storage tank, a preparation tank and a liquid injection pipe, the contrast agent storage tank and the normal saline storage tank are connected with the preparation tank through a first liquid supply pipeline and a second liquid supply pipeline correspondingly, peristaltic pumps and proportional valves are installed on the first liquid supply pipeline and the second liquid supply pipeline correspondingly, and a stirring mechanism is installed in the contrast agent storage tank; the bottom of the contrast agent storage tank is connected with the liquid injection pipe through a connecting pipe, a pushing and injecting mechanism is installed in the liquid injection pipe, contrast agent dilution in proportion can be achieved, the pushing and injecting speed and the pushing and injecting amount can be accurately controlled, manual intervention is not needed in the process, personal errors and inconsistency are reduced, the image quality and reliability are improved, and the working efficiency is improved. And a stable ultrasonic contrast effect is ensured, and medical professionals can be helped to more accurately diagnose the internal structure and lesion of a patient.

The invention discloses a sketch pedestrian re-identification method and device based on cross-modal reliable subjective features, and relates to the technical field of character sketch image processing, and the method mainly comprises the steps: obtaining a training set, a verification set and a test set according to character RGB and a sketch image; constructing a dual-path feature enhancement network by using the residual network, and training by using the training set to obtain a trained dual-path feature enhancement network; and according to the test set, utilizing the trained dual-path feature enhancement network to obtain a character sketch retrieval result. According to the sketch pedestrian re-identification method and device based on the cross-modal reliable subjective features, the accuracy and robustness of a cross-modal retrieval task can be improved.

The invention relates to the technical field of biological tissue culture, in particular to a biological tissue culture device which comprises a tissue culture frame, a plurality of tissue culture bottles are placed on the lower surface of the inner side of the tissue culture frame, each tissue culture bottle is composed of a bottle body and a bottle cap, a moving plate is slidably connected to the upper surface of the inner side of the tissue culture frame, and a servo motor is fixedly connected to the lower surface of the moving plate. The output end of the servo motor is fixedly connected with a lead screw, the outer surface of the lead screw is in threaded connection with a lifting plate, the upper surface of the lifting plate is fixedly connected with a driving motor, and the output end of the driving motor is fixedly connected to the upper surface of one rotating piece. When the driving motor rotates, the rotating piece can drive the bottle cap clamp to rotate the caps of the multiple tissue culture bottles at the same time, the time for manually unscrewing the caps one by one is greatly saved, it can be ensured that the rotating speed and time parameters of all the tissue culture bottles are kept consistent, it is ensured that all the caps are evenly unscrewed, and the working efficiency is improved. And the standardization degree of the experiment can be effectively improved.

The utility model relates to a wire fixed-segment peeling and cutting mechanism, which comprises a wire feeding assembly, the wire feeding assembly comprises a wire feeding moving pair, a wire guide seat arranged on the wire feeding moving pair and a wire clamping seat arranged at the tail end of the stroke of the wire feeding moving pair, and the wire feeding assembly is used for pulling a wire raw material out of a raw material wheel; the wire stripping fixed section assembly comprises a first wire stripping fixed section moving pair, two groups of wire stripping parts arranged on the first wire stripping fixed section moving pair at intervals, a cutting part arranged beside the wire stripping parts, and a second wire stripping fixed section moving pair connected with the cutting part, and is used for stripping and cutting off the wire; the wire transfer assembly comprises a wire transfer moving pair and a wire clamp arranged on the wire transfer moving pair and is used for providing wire stripping pulling force when the wire is pulled out from the two groups of wire stripping parts and transferring the wire subjected to wire stripping and section fixing to a to-be-welded area, wire stripping and welding feeding operation can be taken into consideration at the same time, dependence on additional carrying equipment is reduced, and the welding efficiency is improved. The production process is simplified.

The application discloses a register configuration combination verification information generation method, a verification method and a device, and relates to the technical field of chip verification. First, the bit width and the value range of a domain segment are obtained by analyzing a register description file, and then the value range is divided according to the bit width, and verification points are automatically obtained, so that the generation efficiency and consistency of the verification points are improved. Secondly, the value range of a random variable corresponding to the domain segment is determined according to one or more verification points, and the diversity of the test configuration is increased. Thirdly, each verification point is mapped into a corresponding coverage bin of a coverage point, so that each coverage bin corresponds to a corresponding verification point, and whether the corresponding verification point is covered can be determined according to the hit condition of the coverage bin in the subsequent process. Finally, the test constraint condition and the coverage rate model are both generated according to the same set of verification points, so that the generation efficiency, completeness and accuracy of the verification information are improved. When the register configuration combination verification is performed, the efficiency, completeness and accuracy of the register configuration combination verification are ensured.

The present application relates to the field of data processing technology, and discloses a data processing method, device, equipment and storage medium for a SaaS system. The method comprises: processing multiple data streams respectively through multiple data processing units in the SaaS system to obtain a first processing result for each data stream; performing parameter correction on the first processing results respectively to obtain multiple second processing results; performing data integration processing on the multiple second processing results to obtain target data integration information; performing data compression and view reconstruction on the target data integration information to obtain an initial data view model; performing feature analysis on the initial data view model to obtain a data view feature set; performing view optimization on the data view feature set and the initial data view model to obtain the target data view model of the SaaS system. The present application improves the accuracy of data processing in the SaaS system.

The utility model belongs to the technical field of glass processing equipment, and particularly relates to a glass cutting device for sound insulation glass production, an inclined conveying belt is matched with a guide assembly, so that glass can be automatically cut off and separated after being cut by a glass cutting machine body, the processing time of the cutting device is shortened, and the production efficiency is improved. The labor intensity of workers is reduced, the production efficiency is improved, auxiliary rollers can effectively prevent the right side of the glass from tilting when the glass is broken by the guide assembly, the reliability of equipment is improved, the glass can be automatically skimmed and separated in the vertical direction of the glass through a top plate and a top strip which are installed in an up-and-down sliding mode, and the production efficiency is improved. Dependence of the device on operators can be reduced, errors and inconsistency caused by human factors are reduced, manual intervention is reduced, operation safety is enhanced, the production process is more stable and controllable, and the reliability of the breaking mechanism is improved.

This application relates to the field of two-dimensional semiconductor material preparation technology, specifically to a method for preparing monolayer molybdenum disulfide by dual-sulfur source chemical vapor deposition, comprising the following steps: Step 1, providing a chemical vapor deposition reaction chamber having an upstream region, a growth region, and a downstream region; Step 2, arranging a molybdenum-containing precursor in the growth region, and arranging a substrate above or opposite to the molybdenum-containing precursor; Step 3, arranging a first sulfur-containing precursor in the upstream region and a second sulfur-containing precursor in the downstream region; Step 4, introducing a carrier gas into the reaction chamber and heating it to cause the first sulfur-containing precursor, the second sulfur-containing precursor, and the molybdenum-containing precursor to volatilize and undergo a chemical vapor deposition reaction on the substrate surface to form a monolayer molybdenum disulfide. This invention employs a dual-sulfur source strategy, achieving large-size, uniform, and low-defect preparation of monolayer MoS2, facilitating widespread application.

The utility model discloses a stamping device. The stamping device comprises a base, a stamping driving mechanism, a stamping and punching mechanism and a lower die stamping mechanism. The lower die stamping mechanism comprises a die body and a stamping flange, and the stamping flange is matched with the stamping and punching mechanism in a jacking mode so as to be used for punching one side wall of the workpiece. A workpiece receding groove is formed in the die body, the stamping flange is inserted into the workpiece receding groove, and a stamping hole is further formed in the die body. According to the air bag punching device, after the punching flange is sleeved with the air bag workpiece, the punching driving mechanism drives the punching and punching mechanism to be inserted into the punching hole in the punching flange, so that the punching and punching mechanism is matched with the upper side wall of the punching flange in a jacking mode, and efficient and accurate punching of the upper side wall of the air bag workpiece is achieved; according to the air bag side wall punching device, the automation degree and the machining precision of air bag workpiece side wall punching machining are improved, the production efficiency is improved, the inconsistency of manual operation is reduced, the size precision of an air bag is guaranteed, and the problem of air leakage of a product is solved.

The invention relates to a programmable optical computing chip, comprising: an input layer comprising an on-chip light source module, a wavelength division multiplexer and an optical modulator group; the input end of the linear layer is connected with the output end of the input layer so as to perform matrix-vector multiplication operation on the plurality of input light splitting signals input by the input layer to obtain a plurality of output light signals; the output layer is connected with the output end of the linear layer, and the output layer detects the optical signal intensity of a plurality of output optical signals and determines a matrix-vector multiplication operation result according to the optical signal intensity of each output optical signal; and the control module is respectively connected with the input layer, the linear layer and the output layer. The method at least can realize large-scale parallel data operation, has polarization insensitivity, can process a plurality of signals in different polarization states at the same time, and improves data throughput and calculation efficiency.

The invention belongs to the technical field of disinfection and nursing, and provides a medical staff double-hand disinfection and nursing device which comprises a shell, a displacement assembly is arranged in the shell, the displacement assembly is connected with a rotating assembly, and the rotating assembly is connected with a cleaning assembly; the cleaning assembly comprises a roller, a water spraying ring is mounted in the roller, a nozzle is mounted on the water spraying ring, and a flexible brush is arranged on the inner wall of the roller; a disinfectant is sprayed to the hand of a medical worker in an annular mode through the water spraying ring and the spray head, omission is avoided, the hand of the medical worker is brushed and cleaned through the flexible brush, the disinfection process is carried out through a machine and only needs auxiliary cooperation of the medical worker, disinfection nursing is not tedious any more, the disinfection process is carried out through the machine, and the disinfection efficiency is improved. The roller is driven to move through the displacement assembly, the roller is driven to rotate through the rotating assembly, then the spray head and the flexible brush can move and rotate, and the disinfection and cleaning effects are improved.

The utility model discloses a container corner fitting cutting, jacking and feeding device, which belongs to the technical field of container corner fittings and comprises a feeding support beam, a horizontal pushing cylinder, a lifting pushing cylinder, a lifting platform, a sinking groove and a bottom plate. The feeding supporting beam is a double-straight-beam body which is horizontally arranged and bilaterally symmetrical, a gap is formed between the two beams, and the upper end face of the feeding supporting beam is a straight supporting end face. The horizontal pushing cylinder is located on the side of the end of the feeding supporting beam and provided with a horizontal telescopic cylinder rod extending in the axial direction of the beam. The sinking groove is formed below the end of the supporting beam and is a space sinking to the bottom face. The lifting material pushing cylinder is installed in the sinking groove and provided with a vertical telescopic cylinder rod which vertically stretches upwards. The lifting platform is installed on the vertical telescopic cylinder rod and is parallel to the gap between the feeding supporting beams. And the bottom plate covers the sinking groove. When the vertical telescopic cylinder rod retracts, the lifting platform and the bottom plate form a plane; when the cylinder rod extends out, the lifting platform is in butt joint with the end of the feeding supporting beam to form a plane. According to the device, the working safety is improved.