An on-line roller bed type slab continuous weighing device and method

Abstract

The application discloses an online roller type continuous casting billet weighing device and method, and belongs to the technical field of metallurgical continuous casting. The continuous casting billet weighing device comprises a conveying roller system, a weighing sensor system, a visual identification system and a central control system, the weighing sensor system is installed below the weighing roller, is used for weighing the casting billet conveyed onto the weighing roller, the visual identification system is used for acquiring the moving track of the casting billet in the conveying process through image acquisition and processing, so that it is judged whether the weighing start or stop signal needs to be sent, and the visual identification system and the weighing sensor system are connected with the central control system. The application realizes the online continuous weighing of the casting billet, does not need to stop the production line, significantly improves the production efficiency and the weighing precision, realizes the full-automatic operation, and reduces the labor cost.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of metallurgical continuous casting, and particularly relates to an online roller type continuous casting billet weighing device and method. BACKGROUND

[0002] In the process of steel continuous casting production, especially in the production line of producing continuous casting billets such as rectangular billets, special-shaped billets and square billets, the cut continuous casting billets need to be weighed to ensure product quality and production management. The traditional weighing method usually needs to stop the production line and place the object to be weighed on a static weighing device for measurement.

[0003] This intermittent weighing process has many technical defects: the traditional roller will produce mechanical vibration and jolt during the movement of the continuous casting billet, and these dynamic disturbances will seriously affect the measurement accuracy of the weighing sensor, resulting in large fluctuations in the weighing data and the inability to obtain accurate weight values, so the continuous casting billet must be stopped to be accurately weighed; since the weighing operation needs to interrupt the production process, it not only seriously affects the continuity of the production line and the overall production efficiency, but also takes a long time in the weighing process, resulting in slow production rhythm; during the stoppage weighing process, auxiliary equipment such as heating furnaces and heat preservation devices need to be maintained in a running state, causing a large amount of energy waste; frequent start and stop of the production line also causes the equipment to wear out, further reducing the overall production efficiency; at the same time, this method relies on manual weighing time judgment, equipment operation and data recording, which not only increases the labor cost, but also introduces measurement errors due to the randomness of human operation and environmental interference factors, making it difficult to ensure the stability and consistency of weighing precision, and far from meeting the technical needs of modern continuous production lines for high efficiency, high precision and automatic weighing.

[0004] In addition, the traditional stoppage weighing method also has the following efficiency loss: the weighing operation time is 7-35 seconds, the restart time is 1-2 seconds, and the cumulative time loss is 8-37 seconds / time; considering the high production value characteristics of the continuous casting production line, the economic loss of each stoppage weighing can reach hundreds of yuan, which seriously affects the economic benefits of enterprises. At present, the weighing process of most continuous casting production in steel plants relies on manual tracking and adjustment, which has low work efficiency, large labor input, long adjustment time and low precision, and far from meeting the needs of production.

[0005] Although some existing online weighing technologies can solve the above problems to some extent, they still have technical defects such as difficulty in accurately controlling the weighing start and stop time, low weighing precision and low system integration. SUMMARY

[0006] The application provides an online roller type continuous casting blank weighing device and method, so that the problems of low weighing precision, influence on production continuity and high labor cost of a traditional stopping and weighing mode can be solved, the online continuous and automatic weighing of the casting blank can be realized by adopting the technical scheme of the application, the weighing precision of the casting blank can be effectively ensured, and the problems of low precision of existing online weighing technologies can be solved.

[0007] To achieve the above object, the application provides the following scheme.

[0008] The application provides an online roller type continuous casting blank weighing device, which comprises the following.

[0009] A conveying roller system is used for conveying the casting blank, and comprises multiple flow conveying roller tracks which are arranged in parallel and side by side, each flow conveying roller track comprising a feeding roller track, a weighing roller track and a discharging roller track which are sequentially arranged along the conveying direction of the casting blank.

[0010] A weighing sensor system is installed below the weighing roller track, and is used for weighing the casting blank conveyed onto the weighing roller track, and comprises multiple independent weighing units which are arranged at intervals along the conveying direction of the casting blank, each weighing unit comprising two weighing sensors, i.e. a first sensor and a second sensor, which are located below the same weighing roller and close to the outer sides of both ends of the weighing roller.

[0011] A visual recognition system is used for obtaining the moving track of the casting blank in the conveying process through image acquisition and processing, so as to determine whether a weighing start or stop signal needs to be sent.

[0012] A central control system is connected with the visual recognition system and the weighing sensor system, and is used for receiving the visual recognition signal to control the start and stop of the weighing process, and receiving and processing the weighing signal collected by the weighing sensor system to obtain the weighing result of the casting blank.

[0013] According to any of the technical schemes of the first aspect of the application, the feeding roller track, the weighing roller track and the discharging roller track in each flow conveying roller track are respectively composed of multiple feeding rollers, weighing rollers and discharging rollers which are arranged at intervals along the conveying direction, the rotating direction of each roller corresponds to the conveying direction of the casting blank, and each roller is driven by a motor.

[0014] According to any of the technical schemes of the first aspect of the application, each flow conveying roller track is installed and supported by a support frame, the support frame comprises a support seat which is arranged below each weighing roller and extends along the length direction of the weighing roller, and two roller track beams which are arranged in parallel and at intervals and extend along the conveying direction of the casting blank and are installed on the top of each support seat, and the feeding roller, the weighing roller and the discharging roller on the flow conveying roller track are respectively and intervally rotatably installed between the two roller track beams.

[0015] According to any of the technical solutions of the first aspect of the present application, the load cells are all installed below the support frame through the mounting seats, and the top of each load cell is in spherical contact with the support frame through a spherical contact pad.

[0016] According to any of the technical solutions of the first aspect of the present application, the mounting seat is processed with a mounting groove for accommodating the load cell, and the load cell is installed on the bottom plate of the mounting groove through a support screw, and the mounting seat is fixedly installed on the concrete foundation together with the entire load device through the support screw.

[0017] According to any of the technical solutions of the first aspect of the present application, the load cell system comprises a three-stage adjusting mechanism for adjusting the installation height of the load cell, wherein:

[0018] The first-stage coarse adjusting mechanism: four support screws are symmetrically arranged below each load cell to connect the mounting seat, thereby used for coarse adjustment;

[0019] The second-stage medium adjusting mechanism: equipped with precision shims with different thicknesses for medium adjustment;

[0020] The third-stage fine adjusting mechanism: the third-stage fine adjusting mechanism adopts a wedge adjusting mechanism for fine adjustment;

[0021] The wedge adjusting mechanism comprises a fixed wedge block and a movable wedge block which are matched with each other through wedge surfaces, the fixed wedge block is fixedly installed at the bottom of the load cell, and the movable wedge block is connected with the mounting seat through a support screw.

[0022] According to any of the technical solutions of the first aspect of the present application, the load cell is connected with the central control system through a signal box, and the signal box is used for signal conditioning and A / D conversion of the data collected by the load cell, and then transmitted to the central control system.

[0023] The visual recognition system comprises a first camera and a second camera which are respectively installed at the feeding end and the discharging end of the weighing roller way, wherein the first camera is used for image acquisition and processing of the casting blank at the feeding end of the weighing roller way to determine whether the casting blank completely enters the weighing area; and the second camera is used for image acquisition and processing of the casting blank at the discharging end of the weighing roller way to determine whether the casting blank starts to leave the weighing area.

[0024] The second aspect of the present application further provides an online roller way type continuous casting blank weighing method, which adopts any of the online roller way type continuous casting blank weighing devices of the first aspect of the present application, and comprises the following steps:

[0025] The casting blank is conveyed through the conveying roller way, and the state of the casting blank on the conveying roller way is monitored in real time through the visual recognition system to identify the color feature and position information of the casting blank.

[0026] When the visual recognition system detects that the tail end of the billet has completely entered the weighing area, it sends a weighing start signal to the central control system. The central control system then controls the weighing sensor system to start data acquisition, and the weighing sensor system transmits the acquired weight data to the central control system in real time.

[0027] When the visual recognition system detects that the billet head begins to leave the weighing area, it sends a weighing end signal to the central control system, and the weighing sensor system automatically stops weighing.

[0028] After receiving the weighing data sent by the weighing sensor system, the central control system processes the data in real time and calculates the final weighing result.

[0029] According to any of the technical solutions described in the second aspect of the present invention, after receiving the weighing data sent by the weighing sensor system, the central control system processes the data in real time and calculates the final weighing result, specifically including:

[0030] Data preprocessing: The repetitive data is preprocessed, including filtering.

[0031] Prediction and dynamic adjustment: Based on the casting speed, position and historical data of the billet, the system makes predictions and estimates the upcoming position and weight change trend of the billet;

[0032] Data fusion and weighted processing: Data from different weighing sensors are fused using a weighted averaging technique;

[0033] Output final weight data: The weighing data after multiple layers of processing will be sent to the central control system for display and processing, updating the weight information of the billet in real time, and outputting the final weight data.

[0034] According to any of the technical solutions described in the second aspect of the present invention, in the prediction and dynamic adjustment step, the window length and weight value of the data processing are automatically adjusted based on the speed and weight of the billet, wherein the window length calculation takes into account the billet passing speed and the lateral spacing between adjacent roller conveyor sensors, and the final window length is: L window = L base × k v × k stability ;in:

[0035] Base window length L base :L base = ceil(S / (v×f s S is the sensor spacing between adjacent rollers, v is the billet casting speed, and f is the casting speed. s The sampling frequency;

[0036] velocity correlation coefficient k v :kv = 1+μ×(v-v ref ) / v ref , μ is an adjustment factor, v ref is a reference speed, i.e. a theoretical value;

[0037] The stability coefficient k stability : k stability = 1 + 0.2 × σ v / v mean , σ v is a speed standard deviation, and v mean is an average speed.

[0038] According to any of the technical solutions of the second aspect of the application, the data fusion and weighting processing step specifically comprises:

[0039] (1) The data of two weighing sensors corresponding to the same weighing roller is internally fused, and the weight coefficients are assigned considering the sensor positions and the center of gravity of the casting blank:

[0040] w front = (L rear / L total );

[0041] w rear = (L front / L total );

[0042] wherein, w front , w rear are the fusion weight coefficients of the first sensor (4001) and the second sensor (4002) corresponding to the same weighing roller, respectively; L rear , L front are the longitudinal distances of the center of gravity of the casting blank from the second sensor (4002) and the first sensor (4001), respectively, and L total is the longitudinal spacing between the two weighing sensors, then the fusion weight W i of the two weighing sensors corresponding to the ith weighing roller is: w front = the first sensor reading × w rear + the second sensor reading ×

[0043] (2) The sensor data between adjacent weighing rollers is fused using a recursive fusion strategy, i.e. the weight corresponding to each weighing roller is dynamically adjusted based on the length of the casting blank covered w i0 : w i0Coverage i Coverage j Coverage i is the coverage ratio of the casting blank on the weighing roller i, and Σ(Coverage j ) is the sum of the coverage ratios of the casting blank on all the weighing rollers;

[0044] The weight W0 of the casting blank = Σ(W i × w i0 )。

[0045] According to any of the technical solutions of the second aspect of the application, the filtering processing in the data preprocessing comprises three-level filtering processing: hardware filtering, digital pre-filtering, and adaptive Kalman filtering, wherein:

[0046] The adaptive Kalman filtering adopts a state equation x(k+1) = Fx(k) + Gu(k) + ω(k) and an observation equation z(k) = Hx(k) + v(k), for a multi-sensor system, the state vector x(k) of which contains weight estimation values and change rates of each sensor; F, G, and H are respectively system matrix, control matrix, and observation matrix, z(k) is an observation vector, ω(k) and v(k) are respectively process noise and observation noise; u(k) represents control input or external influence of the system at time k, which can be an adjustment signal of the sensor or other external force; the process noise covariance Q is set as a diagonal matrix diag([σ ω² ,σ v² ]); r 2 ;

[0047] According to any of the technical solutions of the second aspect of the application, when fusing the sensor data between adjacent weighing rollers, the weight w i0 of each weighing roller is adjusted based on the weight of the casting blank, and the maximum weight w i = basic weight w i0 ×k w ;

[0048] wherein k w is a weight-related coefficient, k w =1+β×(W-W ref ) / W ref , β is an adjustment coefficient (0.05-0.15); W ref is a reference weight.

[0049] Compared with the prior art, the present application can achieve the following beneficial effects:

[0050] (1) The present application provides an online roller type continuous casting slab weighing device, which can accurately control the start and stop time of weighing by recognizing the timing of the complete entry or exit of the casting slab into the weighing area through a visual recognition system, thereby reducing the weighing error caused by improper timing control, ensuring the realization of online continuous weighing of the casting slab without stopping the production line, and significantly improving the production efficiency; and the reliability and fault tolerance of the system are further improved by adopting a distributed weighing sensor design.

[0051] (2) The present application optimizes the mounting structure of the conveying roller and the weighing sensor, especially through the spherical contact structure between the top of the sensor and the weighing roller support frame, and the three-stage adjusting mechanism between the bottom of the sensor and the mounting seat, thereby not only reducing the influence of mechanical vibration and jolt on the weighing accuracy to realize the accuracy of dynamic weighing, but also allowing rough, medium and fine adjustment of the height of the weighing sensor as needed.

[0052] (3) The present application further adopts a multi-sensor data fusion technology to weight and fuse the data of multiple weighing sensors, wherein two weighing sensors are correspondingly arranged below the same weighing roller, and the weight coefficients of the two sensors are determined based on the distance between the two sensors and the center of gravity of the casting slab; the sensor data between adjacent weighing rollers adopts a recursive fusion strategy, i.e., the weight corresponding to each weighing roller is dynamically adjusted based on the coverage length of the casting slab, thereby facilitating the improvement of the detection accuracy of the weight of the casting slab.

[0053] (4) The present application further adopts a Kalman filter algorithm to perform real-time filtering on the weight signal, thereby effectively suppressing vibration and noise interference, and facilitating the further improvement of the detection accuracy of the weight of the casting slab, and the data processing delay is less than 10 ms, which can meet the real-time control requirements.

[0054] In summary, the present application can realize fully automated operation, reduce manual intervention, reduce labor costs, significantly improve weighing accuracy and production efficiency, and keep the casting slab continuously moving during the entire weighing process without affecting the production rhythm; at the same time, the present application can be applied to the weighing of casting slabs of different specifications and temperatures, has good versatility, and can provide strong support for the digital transformation of the steel industry. BRIEF DESCRIPTION OF DRAWINGS

[0055] In the drawings, the dimensions and proportions do not represent the actual dimensions and proportions of the product. The drawings are merely illustrative and certain non-essential elements or features are omitted for clarity.

[0056] Figure 1 is a flowchart of the online roller type continuous casting slab weighing method of the embodiments of the present application;

[0057] Figure 2 The whole structure schematic diagram of the online roller type slab continuous weighing device of the embodiment of the present application;

[0058] Figure 3 The longitudinal installation structure schematic diagram of the weighing roller of the embodiment of the present application;

[0059] Figure 4 The transverse installation structure schematic diagram of the weighing sensor of the embodiment of the present application;

[0060] Figure 5 The longitudinal installation structure schematic diagram of the weighing sensor of the embodiment of the present application;

[0061] Figure 6 The installation structure schematic diagram of the weighing sensor and the wedge adjusting mechanism of the embodiment of the present application;

[0062] Figure 7 The state schematic diagram of the wedge adjusting mechanism when not weighing in the embodiment of the present application;

[0063] Figure 8 The state schematic diagram of the wedge adjusting mechanism when weighing in the embodiment of the present application.

[0064] Wherein: 101, roller beam; 102, support seat; 103, concrete foundation; 2, feeding roller; 3, weighing roller; 4, weighing sensor; 4001, first sensor; 4002, second sensor; 401, spherical contact pad; 402, mounting seat; 403, support screw; 404, wedge adjusting mechanism; 4041, fixed wedge block; 4042, movable wedge block; 405, gasket; 5, first camera; 6, second camera; 7, signal box; 8, central control system; 9, weighing roller; 10, discharging roller; 11, slab; 12, motor. DETAILED DESCRIPTION

[0065] For further understanding of the present application, the present application will be described in detail below in combination with the drawings and embodiments. However, it should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present disclosure. In the following detailed description, many specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments can be implemented without these specific details. In addition, in the following description, the description of well-known structures and techniques is omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0066] The structures, proportions, sizes, etc. shown in the drawings of the present disclosure are only intended to facilitate the understanding of the present disclosure, and are not intended to limit the conditions under which the present disclosure can be implemented. Therefore, any modification, change in proportion relationship or adjustment in size that does not affect the effects and purposes of the present disclosure should still fall within the scope of the technical content disclosed by the present disclosure.

[0067] In the present application, the terms "upper", "lower", "left", "right", "front", "back", "top", "bottom", "inner", "outer", "middle", "vertical", "horizontal", "lateral", "longitudinal", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. These terms are mainly used to better describe the present application and its embodiments, and are not intended to limit the devices, elements or components indicated to have a specific orientation, or to be constructed and operated in a specific orientation. In addition, in addition to indicating the orientation or positional relationship, the above-mentioned terms can also be used to indicate other meanings, for example, the term "upper" can also be used to indicate a certain dependent relationship or connection relationship in some cases. For those skilled in the art, the specific meaning of these terms in the present application can be understood according to the specific circumstances.

[0068] Meanwhile, the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so as to describe the embodiments of the present application described herein.

[0069] In addition, the terms "include", "contain" and the like in the present application indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components. The terms "mount", "set", "provided with", "connected" and the like should be interpreted broadly. For example, it can be a fixed connection, a detachable connection, or a monolithic structure; it can be a mechanical connection, or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or internal communication between two devices, elements or components. For those skilled in the art, the specific meaning of the above-mentioned terms in the present application can be understood according to the specific circumstances.

[0070] It should also be understood that certain features of the disclosure, which are, for clarity, described in the context of separate embodiments, can also be provided in combination in a single embodiment. That is, unless explicitly stated otherwise, each separate embodiment is considered an embodiment independent of the other embodiments, and combinations of features of a single embodiment can be provided in combination with any other embodiment or in combination with one another. Conversely, various features of the disclosure, which are, for brevity, described in the context of a single embodiment, can also be provided separately or in any sub-combination. Finally, while a particular feature can be described in the context of a series of steps or a method, that feature can also be provided or implemented in the context of a separate embodiment or in combination with any other embodiment.

[0071] As shown in Figure 2 , Figure 3 The online roller type slab continuous weighing device provided by the embodiment of the application comprises a conveying roller system, a weighing sensor system, a visual recognition system and a central control system 8, wherein:

[0072] The conveying roller system is used for conveying slabs and comprises multiple flow conveying roller tracks arranged in parallel, each flow conveying roller track comprising a feeding roller track 2, a weighing roller track 3 and a discharging roller track 10 arranged in sequence along the conveying direction (transverse direction) of the slabs;

[0073] The weighing sensor system is installed below the weighing roller track 3 and used for weighing the slabs conveyed onto the weighing roller track 3;

[0074] The visual recognition system is used for obtaining the moving track of the slabs in the conveying process through image acquisition and processing, so as to determine whether a weighing start or stop signal needs to be sent;

[0075] The visual recognition system and the weighing sensor system are connected with the central control system 8, and the central control system 8 is used for receiving the weighing signal and the visual recognition signal and controlling the start and stop of the weighing process.

[0076] In the embodiment of the application, the feeding roller track 2, the weighing roller track 3 and the discharging roller track 10 of each flow are respectively composed of multiple feeding rollers, weighing rollers 9 and discharging rollers (collectively referred to as conveying rollers) distributed at intervals along the conveying direction, the rotating direction of each conveying roller corresponds to the conveying direction of the slabs, and each conveying roller is driven by a motor 12, and the rotating speed is adjustable, so as to ensure the stable conveying of the slabs 11.

[0077] The number of each conveying roller and the spacing between adjacent conveying rollers can be determined according to the length of the billet 11 to be conveyed. For example, the number of conveying rollers is 12 to 16, and the spacing between adjacent conveying rollers is 1 to 1.5 m. Furthermore, in this embodiment of the invention, each conveying roller is preferably made of 45# steel with a surface hardness of HRC45 to 50, a diameter of 200 mm, and a length that can be customized according to the width of the billet.

[0078] In some embodiments, conveying rollers distributed along the same flow path are supported by a support frame, which includes a support base 102 and roller beams 101. The support base 102 includes multiple spaced support seats located in the area where the weighing roller conveyor 3 is located, corresponding to each weighing roller 9 (each located below the weighing roller and extending along the length direction of the weighing roller, i.e., extending longitudinally), and is fixedly installed on a concrete foundation 103. The top of the support base 102 is provided with two parallel, spaced roller beams 101 extending along the conveying direction of the billet 11. Both ends of each conveying roller are rotatably mounted between the two roller beams 101, and one end of the feed roller, weighing roller 9, and discharge roller are coaxially fixed to an output shaft and driven to rotate by a motor 12. Specifically, in this embodiment of the invention, the motor 12 is a variable frequency speed control motor with a speed range of 0~100 rpm.

[0079] In some embodiments, the weighing sensor system includes a plurality of independent weighing units spaced apart along the billet conveying direction, which detect weight changes on corresponding weighing rollers 9. In a preferred embodiment, each weighing unit includes two weighing sensors 4 located below the same weighing roller 9: a first sensor 4001 (or a front sensor, for ease of expression)... Figure 2 The first weighing sensor 4002 (or rear sensor) and the second weighing sensor 4002 are symmetrically located below and near the outer sides of the same weighing roller to ensure uniform force distribution on the roller conveyor. All the weighing sensors 4 constitute a distributed weighing detection system. More preferably, a laser tracker is used for three-dimensional coordinate measurement to establish a spatial position database of the roller conveyor and sensors, facilitating subsequent data processing.

[0080] Specifically, the height difference between the two sensors corresponding to the same weighing roller 9 is controlled within 0.05mm, and their signals are differentially amplified and fused to effectively eliminate the influence of off-center loading and improve measurement accuracy. Furthermore, the sensor spacing S between adjacent rollers is controlled at 1~2m, and the specific spacing S is determined based on the minimum length Lmin of the billet: S = Lmin / n, where n is the minimum number of sensors (≥3) that the billet must cover in the transverse direction.

[0081] Specifically, the length of the support base 102 is greater than the distance between the two roller beams 101, that is, the two ends of the support base 102 extend a certain length to the outside of the two roller beams 101, and the first sensor 4001 and the second sensor 4002 are installed below the support base 102 and close to the two ends of the support base 102, respectively, and the longitudinal horizontal distance between the corresponding end (front end or tail end) of the support base 102 is 200-300 mm. The weighing sensor 4 of the embodiment of the application adopts a high-precision strain weighing sensor, the range of which is 0-7.5 tons, the accuracy level is C3 level, the non-linearity is ≤0.02% F.S., and the repeatability is ≤0.01% F.S.

[0082] As shown in Figure 4 some embodiments, each weighing sensor 4 installation position is provided with a mounting seat 402 corresponding to the support base 102 (fixedly connected with the support base 102), and a mounting groove for placing the weighing sensor 4 is processed on the mounting seat 402, and the weighing sensor 4 is fixedly installed on the mounting groove bottom plate through a support screw 403, and the top thereof is in spherical surface contact with the support base 102 through a spherical contact pad 401 to ensure vertical transmission of the load, and the maximum allowed angle deviation is ±3°.

[0083] Specifically, the top of the weighing sensor 4 is processed with an arc-shaped groove matched with the spherical contact pad 401, and the spherical contact pad 401 is directly movably supported in the arc-shaped groove, and a rectangular mounting groove is correspondingly processed on the mounting seat 402, and the weighing sensor 4 is installed in the rectangular mounting groove, so that on the one hand the weighing sensor 4 can be effectively protected, and on the other hand the spherical contact pad 401 can be limited to prevent the spherical contact pad 401 from sliding off the sensor.

[0084] Further, the weighing sensor 4 and the mounting seat 402 are fixedly installed on the concrete foundation 103 through the support screw 403, that is, the weighing sensor 4 and the support base 102 are fixedly supported on the concrete foundation 103 through the support screw 403. The concrete foundation 103 is preferably made of C30 concrete pouring, and the rigidity is greater than 3-5 times the rigidity of the sensor, and each sensor is preferably fixed by four screws, and the pre-tightening force is controlled according to the 200-400 N·m torque, and further preferably equipped with three anti-loose measures of double-nut anti-loose, spring washer and thread locking agent.

[0085] As shown in Figure 5As shown, in some further preferred embodiments, the weighing sensor system includes a three-stage adjustment mechanism: (1) a primary coarse adjustment mechanism: four support screws 403 are provided under each sensor for coarse adjustment; (2) a secondary intermediate adjustment mechanism: precision shims 405 of different specifications (e.g., 0.1mm, 0.2mm, 0.5mm) are provided for intermediate adjustment; and (3) a tertiary fine adjustment mechanism: i.e., a wedge adjustment mechanism 404. The three-stage adjustment mechanism ensures the accuracy of the sensor and protects it from continuous high-precision operation, thus extending its service life.

[0086] Specifically, such as Figure 6 As shown, the wedge adjustment mechanism 404 of this embodiment includes a fixed wedge 4041 and a movable wedge 4042. The fixed wedge 4041 is fixedly installed on the bottom of the weighing sensor 4, and the movable wedge 4042 is connected to the mounting base 402 and the concrete foundation 103 through a support screw 403. The top surface of the movable wedge 4042 and the fixed wedge 4041 are machined into wedge-shaped surfaces that cooperate with each other. More preferably, the inclination angle of the wedge-shaped surface is 3°±0.1°.

[0087] Combination Figure 7 , Figure 8 As shown, when no weighing is performed, the sensor is positioned at H1. When a billet is being weighed, the movable wedge 4042 moves forward under the influence of the force in the direction of motion. At this time, the bolt rotates, causing the sensor height to rise, resulting in higher accuracy. The maximum rotation angle of the bolt is 360°, at which point the sensor height is at its maximum. H2 = H1 + h, where h is the height the sensor rises when the bolt rotates 360°. H is preferably 0.02 mm. When the billet leaves the weighing roller, the movable wedge 4042 is subjected to a reaction force and returns to the unweighed state. The height adjustment calculation formula is: Where P is the screw pitch, specifically 1.25 mm in this embodiment; α is the wedge angle, i.e., the inclination angle of the wedge surface; The adjustment accuracy of the wedge-shaped adjustment mechanism 404 can reach 0.02mm, which is the screw rotation angle.

[0088] As a further preferred embodiment, the weighing sensor 4 is connected to the central control system 8 through a signal box 7. The signal box 7 is used to perform signal conditioning and A / D conversion on the data collected by the weighing sensor 4 and transmit it to the central control system 8.

[0089] Specifically, the signal box 7 is internally provided with a high-precision signal conditioning circuit and a 24-bit A / D conversion module, which can amplify, filter and digitize the analog signal of the weighing sensor. As a preferred solution, the signal box 7 is internally provided with an anti-vibration filtering algorithm (existing algorithm, not described here), which can effectively filter out mechanical vibration and bump interference generated during the operation of the roller bed, ensuring the stability and accuracy of the weighing data. Further preferably, the signal box 7 adopts an industrial-grade protection design with a protection level of IP65, which can work stably in harsh environments. Multiple signal boxes 7 can be set up as needed, and the multiple signal boxes 7 form a distributed signal processing system.

[0090] In some embodiments, the visual recognition system includes a first camera 5 and a second camera 6 respectively installed at the inlet and outlet ends of the weighing roller bed 3 for real-time monitoring of the position of the casting blank on the weighing roller bed. The first camera 5 is used to collect and process images of the casting blank at the inlet end of the weighing roller bed 3 to determine whether the casting blank has completely entered the weighing area, thereby sending a weighing start signal to the central control system 8. The second camera 6 is used to collect and process images of the casting blank at the outlet end of the weighing roller bed 3 to determine whether the casting blank has started to leave the weighing area, thereby sending a weighing stop signal to the central control system 8.

[0091] Specifically, the first camera 5 and the second camera 6 in the embodiments of the present application both adopt an industrial-grade CCD camera with a resolution of 3088x2064 and a frame rate of 30fps, and are equipped with a high-temperature filter with a working temperature range of -20℃ to +70℃. As a preferred installation method, the first camera 5 is installed 3 meters above the inlet end of the weighing roller bed 3 at a downward angle of 15°~30°, preferably 15°; the second camera 6 is installed 3 meters above the outlet end of the weighing roller bed 3 at a downward angle of 15°~30°, preferably 15°. The cameras are directly connected to the central control system 8 through Ethernet to transmit real-time image data.

[0092] As a further optimization scheme, the first camera 5 and the second camera 6 have color recognition function (color recognition algorithm, which can directly use existing technology, not described here), which can recognize casting blanks of specific colors, such as red high-temperature casting blanks. The camera is internally provided with an image processing chip that supports real-time image processing with a processing delay of less than 50ms; and the lens adopts a zoom design with a focal length range of 8~50mm, which can adjust the field of view range according to the site conditions.

[0093] In some embodiments, the central control system 8 adopts an industrial control computer, which controls the operation and stop of the load cell system according to the identification result of the visual recognition system, and processes the data collected by the load cell system in real time. Further, the central control system 8 integrates an image processing module, a data acquisition module, a communication module and a human-computer interface module. The central control system 8 is connected with the signal box 7 through a Profibus-DP bus with a communication rate of 12 Mbps, and is connected with the camera through a gigabit Ethernet with a communication rate of 1000 Mbps.

[0094] Signal transmission architecture:

[0095] Analog signal of the load cell 4 → signal box 7 (signal conditioning, A / D conversion) → central control system 8 (data processing, control decision);

[0096] Digital image signal of the first camera 5 and the second camera 6 → central control system 8 (image processing, position recognition);

[0097] The advantages of this design architecture are as follows: the signal box 7 undertakes the functions of signal conditioning and preliminary digital processing, effectively improving the signal quality and reducing the noise interference in the transmission process; the camera is directly connected with the central control system 8, reducing the transmission delay of image data and improving the real-time performance of visual recognition; the distributed signal processing architecture improves the reliability and maintainability of the system.

[0098] The embodiment also provides an online roller type continuous casting slab weighing method, which combines the continuous weighing device and the roller type continuous casting slab weighing method. Figure 1 As shown in the figure, the method adopts any of the continuous weighing devices, and specifically includes the following steps:

[0099] S1, system initialization stage;

[0100] Start the visual recognition system and the load cell system, and establish a communication connection with the central control system 8.

[0101] Specifically, the first camera 5 and the second camera 6 start to work, and the image acquisition and processing system is initialized; the load cell 4 is calibrated to zero to ensure the measurement accuracy; the signal box 7 is self-checked to verify that the signal conditioning circuit and the A / D conversion module are working normally; and the central control system 8 establishes a data communication link with each subsystem and sets data acquisition parameters.

[0102] S2, casting slab detection and tracking stage;

[0103] The casting slab is conveyed through the conveying roller, and the state of the casting slab on the conveying roller is monitored in real time through the visual recognition system to identify the color feature and position information of the casting slab.

[0104] The rotating speed of the weighing roller on the weighing roller way 3 is set according to the production rhythm requirement, usually 20-60 rpm, and the casting blank 11 keeps stable movement on the roller way, and the movement speed is preferably 3.5 m / min.

[0105] The first camera 5 and the second camera 6 in the visual recognition system collect the position information of the casting blank, and calculate the position coordinates of the casting blank in real time through an image processing algorithm (including but not limited to using an edge detection algorithm, a contour detection algorithm, etc.). Further, the image processing includes but is not limited to image enhancement, edge detection, contour extraction, feature matching, etc.

[0106] Specifically, the first camera 5 adopts a color recognition algorithm, and identifies the contour features of the high-temperature red casting blank 11 through infrared imaging and visible light imaging functions; then the system establishes a casting blank movement trajectory model, and calculates the movement speed and position coordinates of the casting blank in real time.

[0107] S3, weighing start determination stage

[0108] When the first camera 5 located at the feeding end detects that the tail end of the casting blank 11 completely falls on the weighing roller way 3, the tail end boundary of the casting blank 11 is determined through an image processing algorithm, and after confirming that the tail end completely enters the weighing area, the visual recognition system sends a weighing start signal to the central control system 8.

[0109] As a preferred embodiment, the determination standard for the tail end of the casting blank 11 completely entering the weighing area is that the tail end boundary of the casting blank 11 exceeds the feeding end boundary line of the weighing roller way 3 by at least 50 mm, so as to ensure that the casting blank 11 completely separates from the influence of the feeding roller way 2.

[0110] As a further optimization scheme of the embodiment of the present application, the first camera 5 located at the feeding end identifies the red high-temperature casting blank through color recognition function, so as to ensure that the tail end of the casting blank completely enters the weighing area. Specifically, the color recognition adopts HSV color space, and sets the red threshold range: hue H value 160-180, saturation S value 100-255, and brightness V value 100-255.

[0111] S4, data acquisition start stage

[0112] After the central control system 8 receives the weighing start signal, the weighing sensor system starts data acquisition immediately. The casting blank 11 continues to move normally along the weighing roller way 3, so that its movement speed remains within the drawing speed range. At the same time, the time stamp at the weighing start moment is recorded, which provides a time reference for subsequent data processing.

[0113] S5, continuous weighing data acquisition stage

[0114] The weighing sensor system continuously collects weight data, which is transmitted to the central control system 8 for real-time processing and storage of weighing data after signal conditioning and A / D conversion by the signal box 7. As a further optimization scheme, the sampling frequency of the weighing sensor system is 100 Hz to ensure the continuity and accuracy of data acquisition; the data acquisition accuracy in this embodiment is 0.1 kg, and the weighing range is 0-20 tons.

[0115] The processing of the weight data collected by the signal box 7 includes amplification, filtering and digitization of the original weighing signal, especially using a band-pass filter to filter out low-frequency vibration (1-5 Hz) and high-frequency jolt interference (above 50 Hz) generated by the roller operation, so as to retain the effective weighing signal frequency band (5-50 Hz).

[0116] Specifically, the data of the plurality of weighing sensors 4 is fused by the signal box 7, and the Kalman filtering algorithm is used to reduce the influence of local jolting on the overall weighing result, and the total weight is calculated. The central control system 8 monitors the stability of the data in real time, and removes abnormal data points through sliding window algorithm and outlier detection algorithm; the processed weight data is stored in time sequence.

[0117] The length of the weighing roller 3 in this embodiment is calculated as follows:

[0118] The roller length L and the key parameters of the weighing process have the following relationship:

[0119] L = L cast +v × t + L0;

[0120] Wherein:

[0121] L: total length of the weighing roller (m), which refers to the total length of the weighing roller 9 along the direction of the casting blank conveying in the same flow;

[0122] L cast : length of the casting blank (m);

[0123] v: casting blank pulling speed (m / s);

[0124] t: total weighing time (s);

[0125] L0: safety margin length (m);

[0126] The detailed calculation process of the safety margin length L0 is as follows:

[0127] L0= L a + L b + L c

[0128] Wherein:

[0129] La = n x d (n is the number of buffer roller, d is the roller spacing);

[0130] L b = v x t b (t b is the system response time, usually 0.1-0.2s);

[0131] L c = 0.1 x L cast (safety factor, usually 10% of the length of the casting billet);

[0132] Specific to the present embodiment:

[0133] (1) L a Calculation:

[0134] Roller spacing d = 1.0m; the number of buffer roller n = 1; then L a = 1 x 1.0 = 1.0m;

[0135] (2) L b Calculation:

[0136] System response time t b = 0.15s, casting speed v = 3.5m / min = 0.058m / s; then L b = 0.058 x 0.15 = 0.009m = 0.01m;

[0137] (3) L c Calculation:

[0138] Casting length L cast = 12.0m, then L c = 0.1 x 12.0 = 1.2m;

[0139] Therefore: L0= 1.0 + 0.01 + 1.2 = 2.21m.

[0140] Basic calculation principle: from the end of the casting billet completely into the weighing roller, to the head of the casting billet contact the end of the weighing roller, the length of the roller is L. Considering the system response time, the requirement of roller buffer and the measurement accuracy, the design of safety margin L0 must consider mechanical structure, control accuracy and process safety and other factors.

[0141] Actual design parameters: casting speed v: 3.5 m / min, casting length L castThe weighing time is ≤ 7s; the safety margin L0 is 2.21, therefore the roller conveyor length L = 12.0 + 0.058 × 7 + 2.21 = 14.61 m. This embodiment uses L = 15m, which facilitates standardized design and is applicable to most working conditions.

[0142] S6. Weighing Completion Judgment Stage

[0143] When the second camera 6 located at the discharge end detects that the head of the billet 11 contacts the end of the weighing roller conveyor 3, the system automatically stops weighing. Specifically, a position detection algorithm or a contour detection algorithm can be used for detection. In this embodiment of the invention, the head of the billet 11 is preferably identified accurately using a contour recognition algorithm to ensure the accuracy of the timing when weighing ends.

[0144] The specific criteria for determining the end of weighing are as follows: when the boundary of the billet head reaches the boundary line of the weighing roller conveyor discharge end, the vision recognition system sends a weighing end signal to the central control system 8 to ensure that the weighing process covers the entire length of the billet.

[0145] S7. Data Processing and Result Output Stage

[0146] After synchronous acquisition, all weighing sensors 4 transmit analog signals to signal box 7. Signal box 7 conditions and converts the signals to A / D, then transmits them to central control system 8 via PROFIBUS-DP bus. The first camera 5 and the second camera 6 transmit image data directly to central control system 8 via Ethernet. Central control system 8 processes the data in real time, including signal filtering, data fusion, anomaly detection, and result calculation, and calculates the final weighing result (including billet weight, weighing time, weighing accuracy, etc.). This result can be displayed through a human-machine interface, stored in a database, and further transmitted to the upper-level system via network.

[0147] In some embodiments, it is further preferred to introduce a dynamic measurement window (i.e., a sliding window) and a weighted fusion algorithm when processing the collected weight data: the sliding window technique is used to balance casting speed and accuracy; as the billet moves, the data acquisition frequency and processing time need to be precisely controlled, and by setting the dynamic window length, it can be ensured that the data used in each calculation does not exceed the necessary window range. Simultaneously, the readings from different sensors are weighted and averaged based on the billet's position, speed, and sensor accuracy, thereby improving the accuracy of the measurement results. For example, sensors closer to the weighing area are given higher weights than those farther away.

[0148] Specifically, in this embodiment of the invention, the central control system 8 processes the collected weight data, including the following processing steps:

[0149] S71, Data Preprocessing

[0150] The data preprocessing comprises filtering processing, specifically comprising: adopting Kalman filtering to denoise the original data and eliminate environmental interference.

[0151] Further, in the embodiment of the present application, the data preprocessing preferably comprises three-stage filtering processing, and the specific processing process is as follows:

[0152] (1) First stage: hardware filtering

[0153] Analog low-pass filter, cut-off frequency is 50Hz;

[0154] Differential amplification circuit, common-mode rejection ratio > 120dB.

[0155] (2) Second stage: digital pre-filtering

[0156] Eliminate high-frequency random noise, adopt moving average filtering: , N is the window length, preferably the window length is 5;

[0157] Remove impulse interference, adopt median filtering, sort the data window with a length of 5, and take the median value x(n): .

[0158] (3) Third stage: adaptive Kalman filtering

[0159] The Kalman filter adopts state equation x(k+1) = Fx(k) + Gu(k) + ω(k) and observation equation z(k) = Hx(k) + v(k), for a multi-sensor system, the state vector contains the weight estimation value and the change rate of each sensor; the process noise covariance Q is set as a diagonal matrix diag([σ ω² , σ v² ])。

[0160] State space model:

[0161] State equation: x(k+1) = Fx(k) + Gu(k) + ω(k);

[0162] Observation equation z(k) = Hx(k) + v(k);

[0163] State vector: x(k) = [W(k), dW / dt(k)] T ;

[0164] Wherein, W(k) is the weight of the casting blank, dW / dt(k) is the weight change rate; ω(k) and v(k) are process noise and observation noise respectively;

[0165] System matrix: ;

[0166] Control matrix: ;

[0167] Observation matrix: ;

[0168] Noise covariance adaptive adjustment:

[0169] Process noise covariance: Q = diag[σ ω 2 , σ v 2 ];

[0170] Observation noise covariance: R = σ r 2 ;

[0171] Dynamic adjustment strategy:

[0172] - High speed drawing (v > 4.5 m / min): σ ω 2 = 0.01, σ v 2 = 0.001, σ r 2 = 0.05;

[0173] - Low speed drawing (v ≤ 4.5 m / min): σ ω 2 = 0.1, σ v 2 = 0.01, σ r 2 = 0.1;

[0174] Innovation sequence test:

[0175] Innovation: ;

[0176] Innovation refers to the difference between the actual observation and the predicted observation in Kalman filtering, which represents the deviation of the system estimate from the observation value, and is used to measure the inconsistency between the current prediction and the observation, so as to update the Kalman gain to correct the estimate of the system state; wherein represents the observation error at time k .

[0177] Innovation covariance: ; P represents the state estimate error covariance matrix, which represents the uncertainty of the state estimate, represents the one-step prediction error covariance.

[0178] Abnormality detection: Time-triggered exception handling.

[0179] Adaptive parameter adjustment algorithm optimizes Q and R parameters through grid search. In Q range 10 -4 ~10 0 , R range 10 -3 ~10 1 The optimal parameter combination is selected based on the mean square error of the training data. Especially for the roller double sensor configuration, the correlation analysis between the sensors is added to optimize the covariance matrix structure.

[0180] S72, prediction and dynamic adjustment

[0181] Based on the casting speed, position and historical data, the system will make predictions to estimate the position and weight change trend of the casting. According to the casting speed and weight, the window length and weight value of data processing are automatically adjusted. The faster moving casting is calculated by a shorter processing window, while the slower casting uses a longer window.

[0182] The window length calculation considers the casting passing speed and the lateral distance between adjacent roller sensors. The basic window length L base = ceil(S / (v×f s )), where S is the sensor spacing between adjacent rollers (1-2m), v is the casting speed, and f s is the sampling frequency.

[0183] In a typical configuration, the window length is 64-256 sampling points; further optimization is v>4.5m / min, the window length is set to 64-128 sampling points; and when v ≤ 4.5m / min, the window length is set to 128-256 sampling points. Since the sensor spacing between adjacent rollers is 1-2m, the casting will produce a time delay when passing through each sensor, so the algorithm uses timestamp alignment technology to spatially register each sensor data according to the casting position, and realizes synchronous fusion processing.

[0184] Speed correlation coefficient: k v = 1+μ×(v-v ref ) / v ref ;

[0185] Where μ is the adjustment coefficient (0.1-0.3), v ref is the reference speed, i.e. the theoretical value;

[0186] Final window length: L window = L base × k v × k stability ;

[0187] Stability coefficient: k stability = 1 + 0.2 x s v v mean ;

[0188] Where: s v is the standard deviation of speed, v mean is the average speed.

[0189] S73, Data fusion and weighted processing

[0190] The data from different sensors are fused through weighted average technology to ensure that the contribution of each sensor in the calculation matches its position and accuracy. At the same time, by comparing the sensor data with the position information provided by the vision system in real time, to ensure the accuracy of the sensor data. If the data is inconsistent, the system will make adaptive correction.

[0191] First, the two sensor data corresponding to the same weighing roller are internally fused, and the weight coefficient distribution considers the sensor position and the center of gravity of the casting blank:

[0192] w front = (L rear / L total );

[0193] w rear = (L front / L total );

[0194] Where, w front , w rear are the fusion weight coefficients of the first sensor 4001 and the second sensor 4002 corresponding to the same weighing roller, respectively; L rear , L front are the longitudinal distances from the center of gravity of the casting blank to the second sensor 4002 and the first sensor 4001, respectively, and L total is the longitudinal distance between the two sensors, that is, the weight of the first and second sensors is dynamically adjusted according to the position of the center of gravity of the casting blank, then the fusion weight W i of the two weighing sensors corresponding to the i-th weighing roller is: w front = first sensor reading x w rear .

[0195] Example: Assuming that the center of gravity of the casting blank is 600 mm away from the front sensor and 400 mm away from the rear sensor, then:

[0196] w front = 400 / (600+400) = 0.4;

[0197] w rear = 600 / (600+400) = 0.6;

[0198] Then the fusion weight = front sensor reading x 0.4 + rear sensor reading x 0.6.

[0199] Then, the sensor data between adjacent weighing rollers adopts a recursive fusion strategy, that is, dynamically adjusting the weight corresponding to each weighing roller based on the coverage length of the casting blank: w i0 = Coverage i / Σ(Coverage j ), Coverage i is the coverage ratio of the casting blank on the weighing roller i, and Σ(Coverage j ) is the sum of the coverage ratios of the casting blank on all weighing rollers; then the weight of the casting blank W0=Σ(W i x w i0 ).

[0200] For example:

[0201] (1) Time T1: the casting blank just enters the weighing area;

[0202] - Weighing roller 1 covers 30%, its weight = 0.3 / 0.3 = 1.0;

[0203] - Weighing rollers 2-4 are not covered, and their weights = 0;

[0204] (2) Time T2: the casting blank continues to advance;

[0205] - Weighing roller 1 covers 100%, its weight = 1.0 / 2.5 = 0.4;

[0206] - Weighing roller 2 covers 100%, its weight = 1.0 / 2.5 = 0.4;

[0207] - Weighing roller 3 covers 50%, its weight = 0.5 / 2.5 = 0.2;

[0208] - Weighing roller 4 is not covered, and its weight = 0.

[0209] As a further preferred embodiment, the weight corresponding to each weighing roller is adjusted based on the weight of the casting blank w i0That is, the basic weight) is further optimized and adjusted, for heavy slabs (>= 10 tons), the weight of the middle sensor is increased, and the weight of the edge sensor is reduced; and for light slabs (<10 tons), the weight of each sensor is relatively balanced.

[0210] Specifically, the weight-related coefficient k in the embodiment of the present application is: w =1+β×(W-W ref ) / W ref ;

[0211] Wherein, beta is the adjustment coefficient (0.05~0.15); W ref is the reference weight, that is, the slab weight W0predicted by the basic weight in the foregoing.

[0212] Then: the maximum weight w i = basic weight w i0 x k w ;

[0213] The corrected slab weight is: W= Σ(W i x w i ).

[0214] S74, output the final weight data

[0215] The weighing data after multi-layer processing will be sent to the central control system for display and processing, real-time updating the weight information of the slab, and outputting the final weight data.

[0216] To sum up, the present application can accurately control the weighing time according to the real-time feedback of the visual recognition system, realize the online continuous weighing of the slab, and achieve the purpose of accurate weighing and improving production efficiency. Among them, the first camera 5 is responsible for identifying the weighing start time, and adopts an edge detection algorithm to identify the tail end of the slab; the second camera 6 is responsible for identifying the weighing end time, and adopts a contour recognition algorithm to identify the head end of the slab. The detection accuracy of the two cameras is better than ±5mm, so as to ensure the accuracy of the weighing data.

[0217] Compared with the prior art, the present application can achieve the following technical effects:

[0218] Weighing accuracy: the weighing accuracy reaches ±0.5%, which meets the requirements of industrial production. Compared with the traditional static weighing, the accuracy can be improved by 20%.

[0219] Production efficiency: continuous non-stop weighing can be realized, the production efficiency is improved by 35%, and the annual production capacity is increased by about 15%.

[0220] Automation degree: full-automatic operation, reducing the number of operators by 60% and reducing labor costs.

[0221] Adaptability: can be applied to different specifications of casting blank, cross-sectional size 150x150mm to 300x300mm, length 6-12 meters.

Claims

1. An online roller conveyor type continuous weighing device for cast billets, characterized in that, include: The conveying roller system is used to convey the billet (11), and includes parallel multi-flow conveying rollers. Each flow conveying roller includes a feed roller (2), a weighing roller (3) and a discharge roller (10) arranged sequentially along the billet conveying direction. The weighing sensor system is installed below the weighing roller (3) and is used to weigh the billet (11) conveyed to the weighing roller (3). It includes several independent weighing units distributed at intervals along the billet conveying direction. Each weighing unit includes two weighing sensors (4) located below the same weighing roller (9) and close to the outer sides of its two ends, namely the first sensor (4001) and the second sensor (4002). A visual recognition system is used to acquire the movement trajectory of the billet during the conveying process through image acquisition and processing, so as to determine whether it is necessary to send a weighing start or stop signal; The central control system (8) is connected to the visual recognition system and the weighing sensor system respectively. It is used to receive visual recognition signals to control the start and stop of the weighing process, and to receive and process the weighing signals collected by the weighing sensor system to obtain the billet weighing results. Each flow conveyor roller is installed and supported by a support frame, which includes a support base (102) distributed below each weighing roller (9) and extending along the length of the weighing roller (9), and two parallel roller beams (101) installed on the top of each support base (102) and extending along the conveying direction of the billet (11). The feed roller, weighing roller (9) and discharge roller on the flow conveyor roller are respectively rotatably installed between the two roller beams (101) at intervals. The weighing sensors (4) are all mounted on the bottom of the support frame via mounting bases (402), and their tops are in contact with the spherical surface of the support frame via spherical contact pads (401). The mounting base (402) is machined with a mounting groove for mounting and accommodating the weighing sensor (4). The weighing sensor (4) is mounted on the bottom plate of the mounting groove by a support screw (403), and the mounting base (402) together with the entire weighing device is fixedly installed on the concrete foundation (103) by the support screw (403).

2. The online roller conveyor type continuous weighing device for cast billets according to claim 1, characterized in that, The weighing sensor system includes a three-stage adjustment mechanism, which adjusts the installation height of the weighing sensor (4). Primary coarse adjustment mechanism: Four support screws (403) are symmetrically arranged below each load cell and connected to the mounting base (402) for coarse adjustment; Secondary adjustment mechanism: Equipped with precision gaskets (405) of different thicknesses for adjustment; Three-stage fine-tuning mechanism: This three-stage fine-tuning mechanism uses a wedge-shaped adjustment mechanism (404) for fine-tuning; The wedge adjustment mechanism (404) includes a fixed wedge (4041) and a movable wedge (4042) that cooperate with each other through wedge surfaces. The fixed wedge (4041) is fixedly installed on the bottom of the weighing sensor (4), and the movable wedge (4042) is connected to the mounting base (402) through a support screw (403).

3. The online roller conveyor type continuous weighing device for cast billets according to claim 1 or 2, characterized in that, The weighing sensor (4) is connected to the central control system (8) through the signal box (7). The signal box (7) is used to perform signal conditioning and A / D conversion on the data collected by the weighing sensor (4) and transmit it to the central control system (8). The visual recognition system includes a first camera (5) and a second camera (6) installed at the feeding end and the discharging end of the weighing roller conveyor (3), respectively. The first camera (5) is used to collect and process images of the billet at the feeding end of the weighing roller conveyor (3) to determine whether the billet has completely entered the weighing area. The second camera (6) is used to collect and process images of the billet at the discharging end of the weighing roller conveyor (3) to determine whether the billet has started to leave the weighing area.

4. A method for continuous weighing of online roller conveyor billets, characterized in that, The online roller conveyor type continuous weighing device for cast billets according to any one of claims 1-3 includes the following steps: The billet (11) is conveyed by the conveyor rollers, and the status of the billet (11) on the conveyor rollers is monitored in real time by the vision recognition system to identify the color characteristics and position information of the billet; When the visual recognition system detects that the tail end of the billet (11) has completely entered the weighing area, it sends a weighing start signal to the central control system (8). The central control system (8) controls the weighing sensor system to start data acquisition, and the weighing sensor system transmits the collected weight data to the central control system (8) in real time. When the visual recognition system detects that the head of the billet (11) begins to leave the weighing area, it sends a weighing end signal to the central control system (8), and the weighing sensor system automatically stops weighing. After receiving the weighing data sent by the weighing sensor system, the central control system (8) processes the data in real time and calculates the final weighing result.

5. The online roller conveyor continuous weighing method for cast billets according to claim 4, characterized in that, After receiving the weighing data sent by the weighing sensor system, the central control system (8) processes the data in real time and calculates the final weighing result, specifically including: Data preprocessing: The repetitive data is preprocessed, including filtering. Prediction and dynamic adjustment: Based on the casting speed, position and historical data of the billet, the system makes predictions and estimates the upcoming position and weight change trend of the billet; Data fusion and weighted processing: Data from different weighing sensors are fused using a weighted averaging technique; Output final weight data: The weighing data after multiple layers of processing will be sent to the central control system for display and processing, updating the weight information of the billet in real time, and outputting the final weight data.

6. The online roller conveyor continuous weighing method for cast billets according to claim 5, characterized in that, In the prediction and dynamic adjustment step, the window length and weight value of the data processing are automatically adjusted based on the billet speed and weight. The window length calculation takes into account the billet's passing speed and the lateral spacing between adjacent roller conveyor sensors. The final window length is: L window = L base × k v × k stability ; in: Base window length L base :L base = ceil(S / (v×f s S is the sensor spacing between adjacent rollers, v is the billet casting speed, and f is the casting speed. s The sampling frequency; velocity correlation coefficient k v :k v = 1+μ×(vv ref ) / v ref μ is the adjustment coefficient, v ref This is the reference speed, i.e., the theoretical value; Stability coefficient k stability :k stability = 1 + 0.2 × σ v / v mean , σ v v is the standard deviation of velocity. mean This represents the average speed.

7. The online roller conveyor continuous weighing method for cast billets according to claim 5, characterized in that, The data fusion and weighting process specifically includes: (1) Internally fuse the data of two weighing sensors corresponding to the same weighing roller, and the weighting coefficients are allocated considering the sensor position and the center of gravity of the billet: w front = (L rear / L total ); w rear = (L front / L total ); in, w front , w rear These are the fusion weighting coefficients for the first sensor (4001) and the second sensor (4002) corresponding to the same weighing roller, respectively; L rear L front L represents the longitudinal distance L from the center of gravity of the billet to the second sensor (4002) and the first sensor (4001), respectively. total Let W be the longitudinal distance between the two weighing sensors, then the fused weight W of the two weighing sensors corresponding to the i-th weighing roller is... i = First sensor reading × w front +Second sensor reading × w rear ; (2) The sensor data between adjacent weighing rollers adopts a recursive fusion strategy, that is, the weights of each weighing roller are dynamically adjusted based on the billet coverage length. w i0 : w i0 = Coverage i / Σ(Coverage j Coverage i Σ(Coverage) represents the coverage ratio of the cast billet on weighing roller i. j () represents the sum of the coverage ratios of the billet on all weighing rollers; Then the weight of the cast billet W0 = Σ(W i × w i0 ).

8. The online roller conveyor continuous weighing method for cast billets according to claim 7, characterized in that, The filtering process in the data preprocessing includes three levels of filtering: hardware filtering, digital pre-filtering, and adaptive Kalman filtering, wherein: The adaptive Kalman filter uses the state equation x(k+1) = Fx(k) + Gu(k) + ω(k) and the observation equation z(k) = Hx(k) + v(k). For a multi-sensor system, its state vector x(k) contains the weight estimate and rate of change of each sensor; F, G and H are the system matrix, control matrix and observation matrix, respectively; z(k) is the observation vector; ω(k) and v(k) are the process noise and observation noise, respectively; u(k) represents the control input or external influence of the system at time k.

Images