A method and device for fusing multiple round verification data of glass liquid thermometer with double-type standard device

Abstract

The application discloses a kind of glass liquid thermometer multi-round verification data fusion processing method and device of adaptation double type standard device.The method comprises: obtaining the effective multi-round data set after preprocessing;According to the effective multi-round data set and the preset global parameter configuration table, generate the standard device sequence after correction and single round fusion standard temperature sequence;According to the effective multi-round data set after preprocessing, generate the correction multi-round reading matrix of all to-be-verified thermometer;According to single round fusion standard temperature sequence and to-be-verified thermometer correction reading matrix, generate the final error value of each to-be-verified thermometer after fusion.The application defines double type standard device, for each set temperature verification point, system control rotating column multiple rotation alignment, camera multi-round lifting shooting, collect multi-round image recognition original reading;Make full use of the magnitude advantage of two kinds of standard devices, suppress multi-round shooting random error, improve batch glass liquid thermometer verification data stability and measurement accuracy.

Description

Technical Field

[0001] This application relates to the field of metrology and testing technology, specifically to a method and apparatus for fusion processing of multi-round verification data of a glass liquid thermometer adapted to dual-type standards. Background Technology

[0002] Existing automatic verification systems only support single standard thermometers of a single type as references. In the measurement field, it is often necessary to use two types of standard instruments with different accuracy levels, temperature ranges, and response characteristics (such as a standard mercury thermometer and a standard platinum resistance thermometer) for joint verification. The original readings of the two types of standard instruments have systematic deviations. There is no standardized fusion algorithm, and only manual selection is required for comparison. The data utilization rate is low and the traceability reliability needs to be improved.

[0003] Multiple rounds of repeated verification are required at the same temperature verification point to eliminate random errors in image recognition: camera 19 lifting and positioning, rotating column self-rotation positioning, LED supplementary light brightness fluctuation, and slight disturbances in the temperature field of the temperature measuring medium will cause discrete random errors in the readings of the thermometer to be verified 7 and the standard thermometer in each round of image recognition. Direct judgment based on single-round data is prone to false detection and missed detection.

[0004] Traditional data processing only performs simple mean calculations, without distinguishing the traceability weights of the two types of standards, the temperature field response hysteresis error, and the scale recognition system error. It also fails to combine the characteristics of this system, such as batch clamping of the water droplet annular array, rotating batch shooting, and adjustable supplementary lighting imaging, to perform error layering correction. As a result, the reliability of the verification results after fusion is only slightly improved.

[0005] The simultaneous verification of multiple thermometers to be verified in a circular arrangement generates a massive amount of raw image readings through multiple rounds of shooting. However, the system lacks a hierarchical data cleaning, weight allocation, and fusion correction process adapted to the hardware working logic of this automatic verification system, resulting in a missing data closed loop across the entire automated verification process.

[0006] Existing traditional processing methods lack methods to correct optical errors caused by adjacent glass refraction and reflection due to the annular water droplet array, and cannot rely on data from multiple batches of historical calibration points to compensate for the inherent liquid column thermal hysteresis material error of the glass liquid thermometer. This invention addresses an additional technological gap in the prior art. Summary of the Invention

[0007] The purpose of this invention is to provide a method for fusion processing of multi-round calibration data of a glass liquid thermometer adapted to dual-type standards, which at least solves one of the above-mentioned technical problems.

[0008] One aspect of the present invention provides a method for fusing and processing multi-round verification data of a glass liquid thermometer adapted to dual-type standards, the method comprising:

[0009] Step 1: Obtain the preset global parameter configuration table;

[0010] Step 2: Obtain the original multi-round dataset, which includes the original multi-round data output by each thermometer to be calibrated during the multi-round testing on the automatic calibration system for glass liquid thermometers;

[0011] Step 3: Preprocess the original multi-round dataset to obtain a preprocessed valid multi-round dataset;

[0012] Step 4: Generate the corrected standard sequence and the single-round fused standard temperature sequence based on the preprocessed valid multi-round dataset and the preset global parameter configuration table;

[0013] Step 5: Generate the corrected multi-round reading matrix of all thermometers to be calibrated based on the preprocessed valid multi-round dataset;

[0014] Step 6: Generate the final error value of each thermometer after fusion based on the single-round fusion standard temperature sequence and the correction reading matrix of the thermometer under test.

[0015] Optionally, the preset global parameter configuration table includes basic configuration parameters for the verification task, original parameters for the traceability certificate of the dual-type standard instrument, fixed coefficients for the system hardware error database, thresholds for determining equipment hardware anomalies, and thresholds for determining metrological procedures.

[0016] Optionally, step 3 includes:

[0017] Data cleaning and preprocessing are performed in two levels and in layers, based on the complete original multi-round dataset and the hardware anomaly judgment threshold in the preset global parameter configuration table:

[0018] The first level eliminates invalid whole-wheel data that exceeds the hardware positioning deviation threshold of the rotating column and camera;

[0019] The second level uses the 3σ statistical criterion to remove outlier abnormal readings from the standard instrument readings and the original readings of each thermometer to be calibrated within the remaining qualified rounds.

[0020] After two-stage cleaning, valid collection rounds without hardware failures or random outlier interference from readings are selected and summarized to generate a preprocessed valid multi-round dataset.

[0021] Optionally, step 4 includes:

[0022] By iterating through all valid rounds, double-layer error correction is performed on the standard mercury thermometer round by round, thereby obtaining the fully corrected reading of the standard mercury thermometer for each valid round.

[0023] By traversing all valid rounds, the high-precision thermometer is subjected to double-layer error correction round by round, thereby obtaining the fully corrected reading of the high-precision thermometer in each valid round.

[0024] A single-round fusion standard temperature is generated based on the fully corrected readings of the standard mercury thermometer and the fully corrected readings of the high-precision thermometer in each valid round. The fully corrected readings of the standard mercury thermometer in each round constitute the corrected first standard sequence; the fully corrected readings of the high-precision thermometer in each round constitute the corrected second standard sequence; and the single-round fusion standard temperatures in each round constitute the single-round fusion standard temperature sequence.

[0025] Optionally, step 5 includes:

[0026] The clamping sway compensation fixed value is obtained by reading the type identifier of the tensioning component selected for this verification from the global parameter configuration table.

[0027] Calculate the overall offset correction for each effective round by iterating through the remaining rounds.

[0028] For each round, perform the following processing:

[0029] Calculate the corrected reading for each thermometer to be calibrated in this round, thereby obtaining the corrected reading for a single round;

[0030] Generate a matrix of corrected multi-round readings for all thermometers to be calibrated based on the corrected single-round readings of each thermometer.

[0031] Optionally, step 6 includes:

[0032] The stability weight of each valid round is calculated for each valid round, thereby obtaining the weight sequence of all valid rounds;

[0033] The weighted average of the single-round fusion standard temperature sequence is obtained by using the weight sequence of all effective rounds to obtain the unique standard reference temperature of this constant temperature calibration point.

[0034] Calculations are performed on each thermometer to be tested to obtain the final error value of each thermometer after fusion.

[0035] Optionally, the method for fusing multi-round calibration data of the glass liquid thermometer adapted to dual-type standards further includes:

[0036] Store the final readings of the same thermometer under test after fusion at all historical temperature calibration points and the corresponding single-round fusion standard temperature sequence, and construct a piecewise linear compensation model based on the thermal hysteresis characteristics of the liquid column of the glass liquid thermometer.

[0037] For the final fused reading obtained from the current target calibration point, retrieve the historical fused data from adjacent high and low temperature calibration points of the thermometer and substitute it into the piecewise linear compensation model to solve for the thermal hysteresis compensation amount. Then, use the thermal hysteresis compensation amount to correct the current final fused reading, thereby obtaining the corrected final calibration reading.

[0038] Optionally, after obtaining the corrected final verification value after thermal hysteresis compensation, the multi-round verification data fusion processing method for the glass liquid thermometer adapted to dual-type standards further includes:

[0039] Retrieve the pre-stored parameters in the global parameter configuration table, such as the annular arrangement spacing of the water droplets on the mounting plate and the single rotation division angle of the rotating column.

[0040] Obtain a pre-constructed adjacent optical cross-interference coefficient library, which includes the thermometer to be tested and the adjacent optical cross-interference coefficient corresponding to each thermometer to be tested.

[0041] Read the ring installation position number corresponding to each thermometer to be tested in the multi-round reading matrix after correction of all thermometers to be tested, and match the adjacent optical cross-interference coefficient corresponding to the number.

[0042] Using the adjacent optical cross-interference coefficient obtained by matching, crosstalk cancellation calculation is performed on the corrected final verification value after thermal hysteresis compensation, and the final verification value is output to eliminate the refraction and reflection imaging deviation of the annular adjacent glass.

[0043] Optionally, the crosstalk cancellation operation performed after the adjacent optical crosstalk coefficient matching satisfies the following calculation formula:

[0044] ;

[0045] in, The final calibration reading of the i-th thermometer to be calibrated after eliminating optical crosstalk;

[0046] The final calibration reading of the i-th thermometer to be calibrated is corrected after thermal hysteresis compensation.

[0047] Let i be the set of adjacent thermometer numbers on the annular array of water droplets on the mounting plate for the i-th thermometer.

[0048] Let be the adjacent optical cross-interference coefficient between the i-th thermometer and the p-th adjacent thermometer;

[0049] The corrected final calibration value is obtained after thermal hysteresis compensation processing for the adjacent thermometer to be calibrated, numbered i.

[0050] This application also provides a data fusion processing device for multi-round verification of a glass liquid thermometer adapted to dual-type standards, the data fusion processing device for multi-round verification of a glass liquid thermometer adapted to dual-type standards includes:

[0051] A global parameter configuration table acquisition module is used to acquire a preset global parameter configuration table.

[0052] The original multi-round dataset acquisition module is used to acquire the original multi-round dataset, which includes the original multi-round data output by each thermometer to be tested in the automatic calibration system for glass liquid thermometers.

[0053] The preprocessing module is used to preprocess the original multi-round dataset to obtain a preprocessed valid multi-round dataset.

[0054] The sequence acquisition module is used to generate a corrected standard sequence and a single-round fused standard temperature sequence based on the preprocessed valid multi-round dataset and a preset global parameter configuration table.

[0055] A matrix acquisition module is used to generate a corrected multi-round reading matrix of all thermometers to be calibrated based on a preprocessed valid multi-round dataset.

[0056] The error value acquisition module is used to generate the final error value of each thermometer after fusion based on the single-round fusion standard temperature sequence and the correction reading matrix of the thermometer under test.

[0057] This application defines two types of standards: a first type (a standard mercury thermometer, installed in the water droplet hole of the mounting plate and simultaneously immersed in the temperature measuring medium along with the rotating column, with images taken and readings taken simultaneously), and a second type (an external high-precision digital standard thermometer, which collects the real standard temperature of the constant temperature bath medium in real time as an independent measurement standard). For each set temperature calibration point, the system controls the rotating column to rotate and align multiple times, and the camera to rise and fall multiple times to take pictures, collecting multiple rounds of images to identify the original readings. By fully utilizing the measurement advantages of the two types of standards, random errors from multiple rounds of shooting are suppressed, improving the stability and measurement accuracy of batch glass liquid thermometer calibration data. Attached Figure Description

[0058] Figure 1 This is a flowchart illustrating a method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards, according to an embodiment of this application.

[0059] Figure 2 yes Figure 1 This is a schematic diagram of the structure of an automatic thermometer calibration system according to the present invention;

[0060] Figure 3 This is a top view of the thermometer of the present invention when it is clamped and fixed by the mounting plate and the toothed plate;

[0061] Figure 4 This is a top view of the installation disk;

[0062] Figure 5 It is a top view of a thermometer held by a pair of toothed discs;

[0063] Figure 6 It is a top view of a single toothed disc;

[0064] Figure 7 It is one type of structure for tensioning components;

[0065] Figure 8 This is a schematic diagram of the tension adjustment structure of one type of tensioning component;

[0066] Figure 9 It is another structural form of a pair of toothed discs with a tensioning element;

[0067] Figure 10 yes Figure 9 The structure shown is a top view after one of the teeth of the upper gear disk has been removed.

[0068] Figure 11 yes Figure 9 Enlarged view of point A in the middle;

[0069] Figure 12 yes Figure 10 Enlarged view of point B in the middle;

[0070] Figure 13 yes Figure 2 A diagram showing the situation after the tensioning components and storage wheels have been removed.

[0071] Figure 14 yes Figure 11 A schematic diagram of the wheel axle after axial sectioning.

[0072] Component labeling: Rotating column 1, Fill light 2, Ring lamp holder 3, Mounting plate 4, Water drop hole 401, Gear plate 5, Upper gear plate 501, Guide strip hole 50101, Lower gear plate 502, Guide strip groove 50202, Spring 6a, Pressure band 6b, Thermometer to be calibrated 7, Socket tube 8, Slide rod 9, Rubber rod 10, Locking screw 11, Storage wheel 12, Wheel axle 13, Slotted groove 1301, Cylindrical gear 14, Locking nut 15, Guide gear 16, Rack 17, Slider 18, Camera 19, Cabinet 20, Threaded sleeve 21, Lead screw 22, Viewing window 23. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be described in more detail below with reference to the accompanying drawings. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are some, but not all, embodiments of this application. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0074] The method for fusing multi-round verification data of glass liquid thermometers adapted to dual-type standards in this application is applicable to... Figures 2 to 14 The automatic thermometer calibration system shown is separately applied for (application number 2026107528711). Since some steps in the method of this application involve specific corresponding structures, this application provides a simplified description of the automatic thermometer calibration system, the specific structure of which is as follows:

[0075] Please see Figures 2-5This embodiment provides an automatic thermometer calibration system. The system mainly consists of a constant temperature bath containing a temperature measuring medium and a mounting frame for batch loading thermometers. The mounting frame can move vertically and extend into the constant temperature bath, ensuring that the measuring end of the mounted thermometer is completely immersed in the temperature measuring medium for temperature calibration. The mounting frame has a vertically arranged rotating column 1 that can rotate autonomously around its central axis. Multiple rings of supplementary lights 2 are arranged axially on the outer side of the rotating column 1 to provide supplementary lighting for photographing the thermometer scale. Near the bottom of the rotating column 1, a mounting plate 4 and multiple pairs of toothed discs 5 are fixedly mounted from top to bottom, for example, two pairs of toothed discs 5. Water droplet-shaped holes 401 are arranged in a ring array on the surface of the mounting plate 4 to accommodate thermometers of different sizes. In this design, V-shaped notches are provided on the edge of the toothed discs 5, directly opposite the water droplet holes 401. The water droplet holes 401 and the clamping openings are coaxially connected, allowing the thermometers 7 to be calibrated to be placed vertically through them. Annular tensioning elements are arranged between each pair of toothed discs 5, and these tensioning elements are taut and surround the outside of the rotating column 1. Using their own clamping force, they limit and fix the thermometers inserted into the water droplet holes 401 and the clamping openings, ensuring that these thermometers to be calibrated are arranged in a ring array around the rotating column 1, preventing the thermometers from shifting or shaking during the calibration process. During actual assembly, a standard thermometer with compliant accuracy is manually inserted into any one of the water droplet holes 401 on the mounting plate 4 as a reference. The remaining water droplet holes 401 are then evenly inserted into the thermometers 7 to be calibrated, achieving simultaneous clamping of multiple thermometers in a single batch. The system is equipped with a camera 19 capable of vertical reciprocating movement and a matching image recognition system. During the calibration and photography process, the camera 19 is raised and lowered to the designated shooting height according to a preset program. Simultaneously, the rotating column 1 rotates at a set angle, rotating the thermometer to be photographed to the position directly opposite the lens of the camera 19. The camera 19 captures images of the thermometer scale, liquid column, and other positions at fixed points. The captured images are transmitted to the image recognition system, which automatically compares and identifies the readings and scale positions of the thermometer 7 to be calibrated with those of the standard thermometer to determine whether the thermometer calibration is qualified, thus completing the automated calibration operation.

[0076] The supporting system of this application sets differentiated acquisition timing for two types of standards: the standard mercury thermometer is placed in the ring-shaped No. 0 standard water droplet hole of the mounting plate and performs image capture synchronously with each rotation of the column; the sensor (temperature probe) of the high-precision thermometer is immersed in the temperature measuring medium and does not move with the rotating column, continuously outputting temperature values ​​throughout the entire acquisition cycle. The acquisition positions and acquisition frequencies of the two types of standards are different, corresponding to the double-layer and single-layer differentiated error correction logic in step 4.

[0077] As one of the specific implementation details, the supplementary light 2 in this embodiment uses LED lights as the light source. All the LED beads of a single supplementary light 2 are uniformly and sealed inside an integrated lamp cover. The lamp cover is made of light-transmitting protective material, which can isolate moisture and dust around the constant temperature bath, prevent the LED beads from getting damp and accumulating dust, thus avoiding light emission failure and improving the service life of the supplementary light 2. All the encapsulated lamp covers are arranged in a circular array and fixedly mounted on the bottom of the circular lamp holder 3. The lamp covers are illuminating diagonally downwards, and the light can evenly cover the mounting plate 4, the toothed plate 5, and the thermometer clamped and fixed below, eliminating light and shadow on the surface of the thermometer, ensuring that the scale image captured by the subsequent camera 19 is clear and without dark areas, and adapting to the verification shooting operation under different lighting conditions.

[0078] In this embodiment, the top of the annular lamp holder 3, which supports all the lamp covers, is fixedly connected to a slide (not shown in the figure). The slide adopts a sliding sleeve assembly method and is vertically installed on the outside of the rotating column 1. The inner wall of the slide is in close contact with the outer wall of the rotating column 1 and can slide freely along the length of the rotating column 1. A locking bolt is installed horizontally through the side wall of the slide. The operator can manually slide the slide to adjust the vertical height of the annular lamp holder 3 and the bottom LED light source supplement light 2 according to the length of the thermometer and the shooting point. After adjusting to a suitable lighting position, the locking bolt is tightened so that the end of the bolt presses tightly against the side wall of the rotating column 1. The relative fixation of the slide and the rotating column 1 is achieved by friction, locking the lighting height of the supplement light 2, and adapting to the supplementary lighting shooting needs of thermometers of different specifications and lengths.

[0079] In specific manufacturing, the first embodiment of the aforementioned tensioning component, such as... Figure 2 , Figure 3 , Figure 5 Alternatively, the tensioning element can be a circular spring 6a, which is wrapped around the paired toothed discs 5 and kept under constant tension. The spring 6a has elastic expansion and contraction characteristics. When thermometers of different outer diameters are inserted into the water droplet hole 401 and the clamping opening, the spring 6a adapts to the outer wall of the thermometer, shrinking and tightening around the thermometer rod. Its own elastic clamping force presses each thermometer into the water droplet hole 401 and the corresponding clamping opening, thus fixing the thermometer. This structure requires no manual adjustment, is compatible with thermometers of various sizes, has a simple structure, and strong adaptability. Furthermore, the flexible clamping method of the spring 6a will not cause compression or damage to the thermometer, protecting the integrity of the test sample.

[0080] As a specific embodiment, such as Figures 7-8As shown, the connection structure of the annular spring 6a is defined, and the tightening of spring 6a is specifically explained: the two free ends of spring 6a are respectively fixed to the arc-shaped socket tube 8 and the arc-shaped sliding rod 9. A hollow sliding cavity is reserved inside the socket tube 8. The arc-shaped sliding rod 9 is slidably inserted into the cavity of the socket tube 8. The outer wall of the sliding rod 9 slides against the inner wall of the socket tube 8, which can change the overall inner diameter of spring 6a. Tightening spring 6a allows thermometers of different diameters to be clamped onto the mounting plate 4 and the toothed plate 5. In addition, a locking screw 11 is vertically inserted through the side wall of the socket tube 8. When the operator adjusts the depth of the sliding rod 9 inserted into the socket tube 8 according to the diameter of the thermometer, and adjusts spring 6a to the clamping state against the outer wall of the thermometer, the locking screw 11 is tightened, so that the end of the screw presses and fixes the sliding rod 9, locking the opening and closing diameter of spring 6a. This prevents spring 6a from loosening and the tube diameter from increasing, which would cause the thermometer to slip and shift during the calibration process, further improving the stability of clamping and fixing.

[0081] To better pre-tighten spring 6a when adjusting the installation temperature, such as... Figure 7 As shown, a curved rubber rod 10 can be axially fixed inside the cavity of the arc-shaped socket tube 8. The rubber rod 10 is made of wear-resistant and non-slip rubber. At the same time, a blind hole is opened on the axial end face of the arc-shaped slide rod 9, and the body of the rubber rod 10 is frictionally inserted into the blind hole. During the sliding adjustment of the slide rod 9, the outer wall of the rubber rod 10 and the inner wall of the blind hole generate sliding friction. On the one hand, this can increase the damping of the fit between the slide rod 9 and the socket tube 8, and prevent the slide rod 9 from sliding and shifting on its own when there is no external force for adjustment. So that when it is necessary to adjust the tightness, the slide rod 9 can be directly sleeved on the rubber rod 10, and the friction can be used to maintain a temporary relatively stable connection, which facilitates the subsequent tightening of the locking screw 11 to achieve the final tightening and positioning of the spring 6a.

[0082] In addition to the structure described above, this embodiment also proposes a second implementation structure for the tensioning member: such as... Figures 9-12The tensioning element is replaced with a rectangular strip-shaped clamping band 6b. Since each pair of toothed discs 5 is divided into an upper toothed disc 501 and a lower toothed disc 502 arranged parallel to each other, two symmetrically arranged receiving wheels 12 are rotatably mounted in the gap between each pair of toothed discs 5, for example, between the upper and lower teeth of a pair of toothed discs 5. The two ends of the strip-shaped clamping band 6b are respectively wrapped around and stored on the outside of the wheel body of the two receiving wheels 12, so that the clamping band 6b tightens the thermometer around the rest of the ring. Specifically, the top ends of the axles 13 of the two receiving wheels 12 extend vertically out of the upper surface of the upper toothed disc 501, and the protruding ends are coaxially fixedly mounted with cylindrical gears 14, and the two cylindrical gears 14 mesh with each other for transmission. When the operator manually rotates one of the cylindrical gears 14, the two meshing cylindrical gears 14 rotate synchronously in opposite directions, driving the two storage wheels 12 below to rotate synchronously in opposite directions. The rotating storage wheels 12 synchronously wind up the strip-shaped clamping band 6b, shortening the clamping band 6b around the inner diameter, thereby clamping the thermometer inserted between the water droplet hole 401 and the clamping opening, achieving rigid and controllable clamping, which is suitable for high-precision calibration operation scenarios that require high-strength fixation and avoid slight shaking.

[0083] More specifically, such as Figure 14 As shown, the axle 13 has a threaded section machined on the outer side of the top of the upper gear 501, and a locking nut 15 is fitted onto the threaded section. When the operator rotates the cylindrical gear 14 to drive the receiving wheel 12 to complete the winding of the clamping band 6b, and after the thermometer is tightened, the locking nut 15 is tightened, so that the nut is pressed tightly against the upper surface of the upper gear 501. The rotation angle between the axle 13 and the cylindrical gear 14 is locked by the compressive force, preventing the gear from rotating on its own and the clamping band 6b from loosening during the calibration process. At the same time, a slot 1301 is opened at the top of the axle 13. The size of the slot 1301 is compatible with a standard flathead screwdriver. The operator can insert the flathead screwdriver into the slot 1301 and manually apply torque to rotate the axle 13. The tightening and loosening of the clamping band 6b can be adjusted without the need for additional special tooling, making the operation convenient and labor-saving.

[0084] In the above design, the synchronous counter-rotation of the storage wheels 12 further tightens the length of the clamping band 6b surrounding the thermometer, which is equivalent to reducing the circumference of the clamping band 6b around the thermometer. To better achieve the ring-shaped clamping effect, the actual radius of the clamping band 6b should also be reduced. Therefore, the two storage wheels 12 in the above embodiment are preferably made movable. As the clamping band 6b is further stored, the storage wheels 12 move closer to the center of the rotating column 1 by a certain distance to better form a clamping effect. Specifically: as Figure 11 A pair of guide strip holes 50101 are radially drilled through any one of the teeth of the upper gear 501, and simultaneously... Figures 12-13A pair of guide grooves 50202 are formed on the upper surface of the lower gear disk 502, with the two guide grooves 50202 vertically opposite to the upper guide hole 50101. Furthermore, a rack 17 is fixedly installed along its length inside each guide groove 50202, with the teeth of the rack 17 facing inwards. A guide gear 16 is coaxially fixed to the bottom end of the axle 13 corresponding to the receiving wheel 12. The guide gear 16 is recessed and positioned inside the guide groove 50202 of the lower gear disk 502, and its teeth mesh with the rack 17 inside the groove. To ensure stable linear sliding, the corresponding shaft segment of the axle 13 between the receiving wheel 12 and the cylindrical gear 14 is, for example... Figure 14 The slider 18, which is rotatably mounted inside a rectangular block, is essentially a bearing. Its shape matches the guide bar hole 50101, and it slides within the hole, allowing it to slide linearly along the extension direction of the hole. During actual adjustment, when the cylindrical gear 14 is manually rotated, the bottom guide gear 16 rolls along the surface of the rack 17, simultaneously driving the slider 18 to make linear displacement along the guide bar hole 50101 and guide bar groove 50202. This achieves gear rotation to retract the clamping band 6b while adaptively bringing the collecting wheel 12 closer to the center of the clamping band 6b, adapting to the clamping distance adjustment of thermometers of different sizes, improving the clamping adaptability of the clamping band 6b, and ensuring that the clamping band 6b always adheres to the outer wall of the thermometer.

[0085] Finally, regarding the vertical drive mounting structure of camera 19, such as Figure 2 The camera 19 can be assembled entirely inside a vertical cabinet 20. The cabinet 20 is vertically positioned outside the mounting frame and the constant temperature bath, without occupying space in the calibration work center. Inside the cabinet 20, a lead screw 22 is vertically mounted. The bottom end of the lead screw 22 is connected to a stepper motor, which provides power to drive the lead screw 22 to rotate forward and backward based on gear transmission. A threaded sleeve 21 is threaded onto the lead screw 22. One side of the threaded sleeve 21 is attached to the inner wall of the cabinet 20 to form a sliding limit structure, preventing the sleeve from rotating synchronously with the lead screw 22 and allowing it to only move vertically and linearly along the lead screw 22. A camera 19 is fixedly installed on the extended end of the threaded sleeve 21. A viewing window 23 is opened on the side wall of the cabinet 20 facing the shooting direction of the camera 19. The inside of the viewing window 23 is sealed with a transparent glass panel, which can isolate the water vapor and temperature airflow emitted from the constant temperature bath, protect the camera 19 inside the cabinet from environmental interference, and ensure that the camera 19 can clearly shoot the scale of the thermometer on the outside through the glass panel. The camera 19 is raised and lowered precisely by a stepper motor, so as to realize fixed-point shooting of different scale segments of the thermometer, meet the shooting requirements of multi-segment scale verification, and can perform verification under different immersion depths such as full immersion and partial immersion.

[0086] like Figure 1The method for fusing and processing multi-round calibration data of glass liquid thermometers adapted to dual-type standards, as shown, includes:

[0087] Step 1: Obtain the preset global parameter configuration table;

[0088] Step 2: Obtain the original multi-round dataset, which includes the original multi-round data output by each thermometer to be calibrated during the multi-round testing on the automatic calibration system for glass liquid thermometers;

[0089] Step 3: Preprocess the original multi-round dataset to obtain a preprocessed valid multi-round dataset;

[0090] Step 4: Generate the corrected standard sequence and the single-round fused standard temperature sequence based on the preprocessed valid multi-round dataset and the preset global parameter configuration table;

[0091] Step 5: Generate the corrected multi-round reading matrix of all thermometers to be calibrated based on the preprocessed valid multi-round dataset;

[0092] Step 6: Generate the final error value of each thermometer after fusion based on the single-round fusion standard temperature sequence and the correction reading matrix of the thermometer under test.

[0093] In this embodiment, the preset global parameter configuration table is obtained in the following way:

[0094] The system pre-enters, calculates, and stores all basic task parameters, dual-type standard traceability parameters, system hardware error correction parameters, equipment anomaly judgment thresholds, and metrological qualification judgment thresholds required for this batch verification of glass liquid thermometers. It also completes the automatic normalization calculation of the dual-standard traceability weights and integrates them to generate a global parameter configuration table that is uniquely bound to the verification task ID. After the data fusion processing flow is started, the system directly retrieves and reads this preset global parameter configuration table, providing a unified and fixed calculation standard and correction coefficient for the entire process of subsequent raw data acquisition, data preprocessing, standard value correction, thermometer reading compensation, multi-round data weighted fusion, and indication error judgment.

[0095] Specifically, obtain the following information:

[0096] 1. Basic configuration parameters for the verification task

[0097] 1) Nominal temperature of the target constant temperature calibration point ;

[0098] 2) The total number of sampling rounds N is preset for single-inspection fixed-point sampling;

[0099] 3) Total quantity of 7 thermometers to be calibrated in batches: I;

[0100] 4) This test uses the following tensioner type identification: distinguish between spring (spring 6a flexible clamping) and belt (compression belt 6b rigid clamping).

[0101] 2. Original parameters of dual-type standard traceability certificate

[0102] 1) First standard Expanded uncertainty Fixed source tracing correction value Δ ;

[0103] 2) Second standard Expanded uncertainty Fixed source tracing correction value Δ .

[0104] 3. Fixed coefficients in the system hardware error database

[0105] 1) Image offset correction coefficient matrix for fill light 2 h represents the locking height of the ring lamp holder 3 slide;

[0106] 2) Camera 19-degree scale offset correction coefficient matrix : This represents the positioning height of the lead screw 22 threaded sleeve 21;

[0107] 3) Clamping wobbling inherent compensation value: spring 6a corresponding compensation Compensation corresponding to the 6b compression band belt.

[0108] 4. Equipment hardware anomaly detection threshold

[0109] 1) Maximum permissible deviation threshold for the rotation angle of rotating column 1 ;

[0110] 2) Maximum permissible deviation threshold for camera 19 lead screw lifting and positioning .

[0111] 5. Measurement procedure judgment threshold

[0112] Maximum permissible indication error limit for glass liquid thermometers of corresponding accuracy class .

[0113] The above parameters are calculated, integrated, and encapsulated to generate a global parameter configuration table, as detailed below:

[0114] Automatic calculation and normalization verification of source weights for dual standard instruments

[0115] Based on the inverse proportional allocation of traceability weights according to the expanded uncertainties of the two types of standard instruments, the calculation formula is as follows:

[0116] ;

[0117] Verify constraints + =1; If there is a slight floating-point deviation in the calculated sum, normalization correction is performed to ensure that the weighted sum is strictly equal to 1.

[0118] in: for The fusion weight of the glass standard thermometer for External digital standard fusion weights.

[0119] Establish task identifier binding relationship:

[0120] Generate a unique task ID for this verification, and combine the task ID with all the entered raw parameters and calculated values. Establish one-to-one binding relationships to distinguish between different batches and different inspection points for inspection tasks, thereby achieving data traceability.

[0121] All parameters are encapsulated in layers to build a structured global parameter configuration table.

[0122] The global parameter configuration table is stored uniformly in 5 major parameter modules:

[0123] Basic Task Modules: N, I, tensioner type markings;

[0124] Dual Standard Tracing Module: ;

[0125] Hardware error correction module: matrix, matrix, ;

[0126] Equipment anomaly threshold module: , ;

[0127] Measurement judgment module: Maximum allowable error limit .

[0128] In this embodiment, the original multi-round dataset in step 2 is obtained through the following steps:

[0129] Equipment pre-positioning and constant temperature waiting:

[0130] The control mounting bracket is lowered as a whole into a constant temperature bath containing the temperature measuring medium, so that the first standard instrument... All thermometers to be calibrated have their measuring ends completely immersed in the measuring medium.

[0131] The system controls the temperature control module of the constant temperature bath to stabilize at the preset target calibration point in the global parameter configuration table. Let it stand at a constant temperature until the temperature field stabilizes;

[0132] Based on the preset height h in the global parameter configuration table, lock the position of the fill light 2 ring light holder 3 slide to keep the fill light height constant.

[0133] Initialize the round count variable k=1 and start the cyclic data collection process.

[0134] Single-round k-automatic synchronous data acquisition loop:

[0135] Rotational alignment control: Drive the rotating column 1 to rotate sequentially at preset angles, thereby sequentially positioning the first standard device. The first thermometer to be tested, 7, is aligned with the shooting area of ​​the camera lens, 19.

[0136] Camera lifting and positioning: The stepper motor drives the lead screw 22 to drive the threaded sleeve 21, raising and lowering the camera 19 to the preset shooting height l. The camera completes the scale image shooting through the viewing window 23 with glass panel of the cabinet 20.

[0137] Image reading output: The image recognition system analyzes the captured image and outputs the original readings for this round:

[0138] Round k Image recognition of raw readings from glass standard thermometers;

[0139] : The original reading of the image recognition of the i-th thermometer to be tested in the k-th round (i=1,2,…,I);

[0140] Synchronous acquisition of medium standard temperature: Real-time reading of the output of the high-precision thermometer to obtain the original reading of the medium standard for this round. ;

[0141] Hardware positioning deviation acquisition: The system acquires the difference between the actual movement position of the current wheel mechanism and the target position, and records the rotation angle deviation of the rotating column. Camera lifting height deviation ;

[0142] Single-round data encapsulation: Integrating and encapsulating all collected information in this round into a single-round raw dataset. :

[0143] ;

[0144] Cycle determination and cycle termination:

[0145] Round count increments: k = k + 1;

[0146] Judgment conditions: If k≤N, proceed to the next round and execute the next round of complete data collection; if k>N, terminate the multi-round data collection loop.

[0147] After all loops terminate, retrieve all round-specific datasets from the cache partition, aggregate and sort them to form a complete and ordered original multi-round dataset. Lock the dataset storage partition.

[0148] Complete original multi-round dataset { , ,…, The dataset contains N sets of ordered single-round data acquisition data. Each set of single-round data synchronously carries the readings of the standard instrument in this round, the original readings of all thermometers to be calibrated, the fixed height of the supplementary light, the shooting height of the camera, and the original information of the hardware positioning deviation between the rotating column and the camera.

[0149] Step 3 includes:

[0150] Data cleaning and preprocessing are performed in two levels and in layers, based on the complete original multi-round dataset and the hardware anomaly judgment threshold in the preset global parameter configuration table:

[0151] The first level eliminates invalid whole-wheel data that exceeds the hardware positioning deviation threshold of the rotating column and camera;

[0152] Specifically, it iterates through each single-round dataset in the original multi-round dataset. The validity of the entire round is determined based on the hardware deviation threshold:

[0153] Judgment rule: If satisfied or This indicates that the alignment deviation of the rotating column 1 in this round exceeds the standard, or the positioning deviation of the camera 19 driven by the lead screw 22 exceeds the standard. There is a systematic offset in the image capture of this round, and all data of the entire round is invalid and directly discarded.

[0154] Only when and At that time, retain the dataset from this round to form a subset of hardware-qualified rounds. ;

[0155] The total number of rounds within the qualified hardware subset is counted and denoted as n′; if n′=0, the system directly issues a data acquisition anomaly alarm and terminates the data fusion process.

[0156] The second level uses the 3σ statistical criterion to remove outlier abnormal readings from the standard instrument readings and the original readings of each thermometer to be calibrated within the remaining qualified rounds.

[0157] Specifically, for the hardware qualified subset In all rounds, respectively, for the first standard instrument Reading sequence, second standard Reading sequence, each thermometer to be calibrated The reading sequences were independently subjected to 3σ outlier removal, using the first standard. Complete calculation logic using a reading sequence as an example:

[0158] The 3σ outlier removal process proceeds as follows: First, the reading sequences of the first type of standard mercury thermometers are processed, then the reading sequences of the second type of high-precision thermometers are processed, and finally, the reading sequences of each thermometer to be calibrated are processed one by one. If any set of readings in any round meets the outlier determination criteria, then the entire single-round dataset Dk of that round is removed. Different reading sequences are judged independently but share the same set of effective round removal rules to ensure that the round data of the standard instrument and the tested sample are retained or removed synchronously in the same round.

[0159] Extract all from the subset Read the data and calculate the sample mean. Sample standard deviation :

[0160] ;

[0161] Outlier removal criteria:

[0162] ;

[0163] Rounds of data that satisfy the above formula are marked as outliers and all data from that round are directly removed.

[0164] Repeat the above logic for calculating the mean and standard deviation and identifying outliers, sequentially applying it to the second standard. All reading sequences, the first thermometer to be calibrated The reading sequence completes outlier screening;

[0165] After two-stage cleaning, valid collection rounds without hardware failures or random outlier interference from readings are selected and summarized to generate a preprocessed valid multi-round dataset.

[0166] After all sequences have been filtered, the remaining collection rounds without hardware errors or outliers are summarized, the total number of remaining valid rounds m is recorded, and a set of valid round numbers is generated. .

[0167] Step 4 includes:

[0168] By iterating through all valid rounds, double-layer error correction is performed on the standard mercury thermometer round by round, thereby obtaining the fully corrected reading of the standard mercury thermometer for each valid round.

[0169] Specifically, for each valid round Perform the following calculations:

[0170] Retrieve the image correction coefficients corresponding to the current hardware:

[0171] ;

[0172] in, Based on the height of this round of supplementary lighting Obtained from the hardware error database. The altitude is determined by the camera in this round. Obtained from the hardware error database; This represents the total offset correction amount for this round of image recognition.

[0173] The dual-layer error correction for the standard mercury thermometer described in this application is defined as follows: the first layer is the correction for the light and shadow shift error caused by changes in the height of the supplementary lighting, with a corresponding coefficient... The second layer corrects for the scale imaging offset error caused by the camera's vertical movement, with corresponding coefficients. The two layers of error are independently calibrated and superimposed for calculation, which is called double-layer error correction. The high-precision thermometer has no image imaging process and does not have double-layer imaging error. It only performs single-layer traceability fixed correction.

[0174] Eliminating image recognition offset errors yields the first standard image after correction. reading:

[0175] ;in, The first standard after image correction reading; This represents the total offset correction amount for this round of image recognition; For the k-th round within the valid dataset Raw readings;

[0176] Superimposed first standard By tracing the source and fixing the correction value, the fully corrected indication is obtained:

[0177] ;

[0178] in, for Source-based fixed correction values; The first standard after image correction reading; After the kth round of correction Indicated value;

[0179] By iterating through all valid rounds, the high-precision thermometer is corrected for errors round by round, thereby obtaining the fully corrected reading of the high-precision thermometer for each valid round.

[0180] Specifically, This is a high-precision temperature measuring instrument, with no image capture or light and shadow offset errors, only superimposed fixed correction values ​​from the traceability certificate.

[0181] ;

[0182] in, for Source-based fixed correction values; For the k-th round within the valid dataset Original readings; This is the reading of the high-precision thermometer after the k-th round of complete correction;

[0183] A single-round fusion standard temperature is generated based on the fully corrected readings of the standard mercury thermometer and the fully corrected readings of the high-precision thermometer in each valid round. The fully corrected readings of the standard mercury thermometer in each round constitute the corrected first standard sequence; the fully corrected readings of the high-precision thermometer in each round constitute the corrected second standard sequence; and the single-round fusion standard temperatures in each round constitute the single-round fusion standard temperature sequence.

[0184] Preset source tracing weights using a global parameter configuration table. , The two types of corrected standard instrument readings are weighted and fused to generate a globally unified standard temperature for this round:

[0185] ;

[0186] in, The standard reference temperature for the medium in the kth round of this calibration point is used. All thermometers 7 to be calibrated in this round share the same standard temperature.

[0187] The following data were compiled and generated: complete correction sequences for two types of standards and single-round fusion standard temperature sequences.

[0188] After traversing all valid rounds and completing the calculations, summarize all rounds in ascending order of their round numbers. Generate the corrected first standard sequence: ;

[0189] Summarize all in the same round order. Generate the corrected second standard sequence: ;

[0190] Summarize all in round order Generate a single-round fusion standard temperature sequence: .

[0191] In this embodiment, step 5 includes:

[0192] The clamping sway compensation fixed value is obtained by reading the type identifier of the tensioning component selected for this verification from the global parameter configuration table.

[0193] Read the type identifier of the tensioning component used in this verification from the global parameter configuration table, and determine the clamping sway compensation fixed value:

[0194] If the tensioning element is labeled "spring" (6a flexible clamping spring), then... ;

[0195] If the tensioning element is labeled as belt (compression band 6b rigid clamp), then .

[0196] Calculate the overall offset correction for each effective round by iterating through the remaining rounds.

[0197] The total offset correction coefficient for this round of image recognition is obtained by superposition:

[0198] Specifically as follows:

[0199] For each valid data collection round implement:

[0200] Based on the height of the fill lights stored in the valid dataset of this round. Retrieve the corresponding image offset coefficient from the hardware error database. ;

[0201] Based on the camera positioning height stored in the valid dataset of this round Retrieve the corresponding scale offset coefficient from the hardware error database. ;

[0202] The total offset correction coefficient for this round of image recognition is obtained by superposition:

[0203] ;

[0204] This is a combined offset correction for light and shadow and camera positioning, shared by all thermometers to be calibrated in this round.

[0205] For each round, perform the following processing:

[0206] Calculate the corrected reading for each thermometer to be calibrated in this round, thereby obtaining the corrected reading for a single round;

[0207] Specifically, for each thermometer to be calibrated in this round Read the raw reading Layered elimination of image offset error and clamping wobbling error yields the corrected single-wheel reading. The calculation formula is as follows:

[0208] ;

[0209] in, To eliminate the image recognition system deviation caused by the shadows of supplementary lighting and the camera's lifting and positioning; The slight wobbling of the thermometer caused by either the spring 6a or the clamping band 6b introduces a reading offset.

[0210] Generate a matrix of corrected multi-round readings for all thermometers to be calibrated based on the corrected single-round readings of each thermometer.

[0211] Specifically, with thermometer number i as the matrix row and effective round number k as the matrix column, all... Fill the two-dimensional matrix in order;

[0212] The matrix has a row dimension of I rows, corresponding to I batches of thermometers to be calibrated (7); and a column dimension of m columns, corresponding to m groups of valid data collection rounds.

[0213] Step 6 includes:

[0214] The stability weight of each valid round is calculated for each valid round, thereby obtaining the weight sequence of all valid rounds;

[0215] Specifically, for each valid round Calculate sequentially:

[0216] The hardware total normalized deviation calculation combines the deviations of the rotating cylinder and the camera by normalizing and superimposing them according to their respective thresholds:

[0217] ;

[0218] The hardware total normalization deviation in the k-th round is [value]. The smaller the hardware positioning deviation, the better. The smaller the value, the higher the reliability of the data collected in this round; The rotation angle deviation of the k-th rotating column; The deviation of the camera's lifting height in the kth round; The threshold value for the rotation angle deviation of the rotating cylinder; The threshold for camera height deviation;

[0219] The rotation column angle deviation and the camera lifting deviation have different dimensions and cannot be directly added. This solution performs dimensionless normalization processing on their respective hardware thresholds, converts the two types of deviations into dimensionless values ​​in the range of 0 to 1, and then superimposes them to eliminate the influence of unit differences and ensure a unified standard for comprehensive evaluation of hardware deviations.

[0220] Inverse proportional mapping generates round-stabilized weights; the smaller the deviation, the higher the weight.

[0221] ;

[0222] Summarize the weights of all valid rounds to generate a weight sequence. .

[0223] The weighted average of the single-round fusion standard temperature sequence is obtained by using the weight sequence of all effective rounds to obtain the unique standard reference temperature of this constant temperature calibration point.

[0224] Specifically, the single-round fusion standard temperature sequence is weighted and averaged using the stable weights of each round to obtain the unique standard reference temperature for this temperature calibration point:

[0225] ;

[0226] The weighted average standard reference temperature for all rounds; The stability weight for the k-th round of data collection; The standard temperature for single-round fusion in the kth round;

[0227] Calculations are performed on each thermometer to be tested to obtain the final reading of each thermometer after fusion.

[0228] Specifically, each thermometer to be tested was examined. Read the corrected readings for all valid cycles of the thermometer. Combined with round-based stable weights Weighted fusion:

[0229] ;

[0230] in, The final reading of the i-th thermometer to be tested after fusion is the final test reading of the thermometer after eliminating hardware random positioning disturbance, image system error, clamping shaking error and discrete error of multiple rounds of random reading; The value of a single round after correction for the i-th thermometer to be calibrated in the k-th round; The stability weight for the k-th round of data collection.

[0231] In this embodiment, the multi-round calibration data fusion processing method for the glass liquid thermometer adapted to dual-type standards of this application further includes:

[0232] Indication error calculation and automatic pass / fail determination:

[0233] Calculation of error in single thermometer reading:

[0234] ;

[0235] Measurement qualification judgment rules ( (Taken from the global parameter configuration table)

[0236] like The i-th thermometer to be tested has passed the test at this testing point;

[0237] like The i-th thermometer to be tested failed the test at this testing point.

[0238] This method sets up a three-level metrological judgment mark to automatically distinguish the risk level of samples:

[0239] 1. Level 1 Qualified: ≤0.5 No re-inspection is required;

[0240] 2. Suspicious at the edge: 0.5 < ≤ The system automatically generates a re-inspection checklist and prompts the operator to retest;

[0241] 3. Unqualified: > Directly mark as unqualified;

[0242] Level 3 markers are saved synchronously with the traceability data package and displayed in sections when exporting the verification report.

[0243] The method for fusing and processing multi-round calibration data of the glass liquid thermometer adapted to dual-type standards further includes:

[0244] Store the final readings of the same thermometer under test after fusion at all historical temperature calibration points and the corresponding single-round fusion standard temperature sequence, and construct a piecewise linear compensation model based on the thermal hysteresis characteristics of the liquid column of the glass liquid thermometer.

[0245] Specifically, after completing the calculations in steps 1-6 above for each temperature calibration point, the system automatically binds the current calibration task ID, thermometer number i, and the nominal temperature of the current calibration point. Simultaneously archive and store two types of historical data to a dedicated historical database:

[0246] (1) All single-round fusion standard temperature sequences at this verification point ;

[0247] (2) The final reading of the thermometer after fusion at this calibration point The database is indexed by "thermometer number - calibration point temperature" to enable rapid retrieval of full temperature gradient data for a single thermometer, providing data support for coordinated retrieval of data from multiple calibration points.

[0248] For the final fused reading obtained from the current target calibration point, retrieve the historical fused data from adjacent high and low temperature calibration points of the thermometer and substitute it into the piecewise linear compensation model to solve for the thermal hysteresis compensation amount. Then, use the thermal hysteresis compensation amount to correct the current final fused reading, thereby obtaining the corrected final calibration reading.

[0249] Specifically, for the i-th thermometer to be calibrated and the current target calibration point Perform the search:

[0250] Filter the historical database to find all historical temperature calibration points for this thermometer that have been calibrated;

[0251] Filtering rules:

[0252] Adjacent low temperature testing points The value closest to and less than among all historical test points. Temperature points, retrieve corresponding historical fusion values ;

[0253] Adjacent high temperature testing points The value closest to and greater than among all historical checkpoints. Temperature points, retrieve corresponding historical fusion values ;

[0254] Boundary handling: If only low-temperature historical points exist and no high-temperature historical points exist, only the low-temperature segment single-piece model is used; if only high-temperature historical points exist and no low-temperature historical points exist, only the high-temperature segment single-piece model is used; if there is no historical data, skip thermal hysteresis compensation and directly use the existing model. As the corrected indication.

[0255] A piecewise linear compensation model is constructed based on the thermal hysteresis characteristics of liquid column:

[0256] Based on the physical laws of thermal hysteresis in glass liquid thermometers, the core expression of the piecewise linear compensation model is as follows:

[0257] ;

[0258] This is the compensation amount for the thermal hysteresis of the liquid column; For thermometer i at the high-temperature calibration point The final value after historical integration; For thermometer i at the low temperature calibration point The final value after historical integration; The adjacent high-temperature calibration point is higher than the current calibration point; The adjacent low-temperature calibration point is lower than the current calibration point; The output of step 6 is the final reading of the thermometer after fusion at the current target calibration point; The current weighted average standard reference temperature is used as the benchmark temperature for the calibration points.

[0259] The global parameter configuration table contains pre-stored low-temperature boundary temperatures. High temperature boundary temperature The entire temperature measurement range is divided into a low-temperature segment T< medium temperature range ≤T≤ High temperature range T> Three segments, each with an independent linear correction coefficient stored in the hardware error database, are calculated. Based on the current inspection point The calculation result of the coefficient correction corresponding to the interval matching;

[0260] Boundary handling rules: When only low-temperature historical calibration points exist, only the low-temperature single-segment calculation model is used; when only high-temperature historical calibration points exist, only the high-temperature single-segment model is used; when there is no historical calibration data... =0, use directly As a thermal hysteresis correction value, the program will not interrupt the operation.

[0261] Retrieve the temperature segment threshold from the global parameter configuration table, based on... It belongs to the low temperature / medium temperature / high temperature range, and switches the corresponding linear slope correction coefficient to adapt to the difference in liquid column hysteresis in different temperature ranges.

[0262] Substitute the values ​​into the model to solve for the thermal hysteresis compensation and correct the current fused reading:

[0263] Substitute the retrieved data into the piecewise linear compensation model to calculate the thermal hysteresis compensation amount. ;

[0264] The final fused value output in this step is corrected for hysteresis bias by offsetting the hysteresis, and the output is the corrected final verification value after thermal hysteresis compensation:

[0265] ;

[0266] Branch-by-branch cycle: ..., I sequentially perform retrieval, modeling, and compensation calculation on all thermometers to be calibrated, generating a set of thermal hysteresis correction values ​​for each thermometer.

[0267] The final calibration readings of all thermometers were corrected after thermal hysteresis compensation. Stored in a dedicated read-only cache, fully preserving the corresponding thermometer number, current calibration point, parameters of adjacent high and low temperature historical calibration points, and thermal hysteresis compensation. .

[0268] After obtaining the corrected final verification value after thermal hysteresis compensation, the multi-round verification data fusion processing method for the glass liquid thermometer adapted to dual-type standards further includes:

[0269] Retrieve the pre-stored parameters in the global parameter configuration table, such as the annular arrangement spacing of the water droplets on the mounting plate and the single rotation division angle of the rotating column.

[0270] Specifically, the global parameter configuration table generated in step 1 is read, and the fixed hardware parameters are extracted: the spacing parameter of the annular arrangement of the water droplet holes on the mounting plate and the single rotation division angle parameter of the rotating column; these two types of parameters are used as the basic modeling conditions for constructing and verifying the adjacent optical cross-interference coefficient library.

[0271] The system calls upon the locally pre-built adjacent optical cross-interference coefficient library, which is pre-calibrated based on the annular water droplet hole arrangement and rotating shooting optical path characteristics of this device.

[0272] In this embodiment, the method for constructing the adjacent optical cross-interference coefficient library is as follows:

[0273] All calibration procedures use standardized test conditions: the temperature bath medium is kept constant at 20°C, and the supplemental lighting is locked at the standard height. Camera locked at standard shooting height The rotating column has a fixed standard graduation angle, and the environment remains unchanged throughout the process, eliminating external interference with the calibration coefficient.

[0274] Calibration fixture setup: Insert the same standard glass liquid thermometer into each water droplet hole of the mounting plate sequentially, keeping only one thermometer as the measured object at a time, leaving the other water droplet holes empty. Control the rotating column and camera to perform the standard shooting procedure and read the standard reading under conditions of no adjacent interference. ;

[0275] A2 Single Adjacent Calibration: Fix the position of the thermometer being measured, insert a glass liquid thermometer of the same specification into the adjacent water droplet hole on one side of the thermometer, leaving the other holes empty. Repeat multiple rounds of shooting under the same supplementary lighting height, camera height, and rotation angle to identify the current value affected by interference. Calculate the one-sided interference coefficient:

[0276] ;

[0277] Bilateral adjacent calibration: Simultaneously insert the same specification thermometer into the left and right adjacent water droplets of the thermometer being measured, repeatedly take pictures and read the bilateral interference readings, verify the linear relationship of superimposed interference, and correct the coefficient value.

[0278] If the difference in coefficients obtained from two calibrations (single-adjacent and double-adjacent) exceeds 0.002, it is determined that there is nonlinear interference in the optical path, and the double-adjacent calibration value is used as the final value for storage. If the difference is ≤0.002, the one-sided calibration coefficient is used directly.

[0279] Full position traversal: Traverse all water droplet positions in the ring array of the mounting plate, and assign each position number its left and right neighbor pairing interference coefficients. ;

[0280] Using the thermometer ring installation position number as the primary key, and the adjacent position number and corresponding optical cross-interference coefficient as auxiliary fields, the data is structured and stored in a local read-only database to form a fixed adjacent optical cross-interference coefficient library. This library is stored along with the global parameter configuration table for direct access in subsequent calibration processes. When the supplementary light, camera, or mounting plate clamping structure is disassembled or replaced, the complete calibration process is re-executed to update the coefficient library.

[0281] The coefficient database storage rules are as follows: The primary index is the ring-shaped installation position number of the thermometer. Each position number is associated with the numbers of its two adjacent thermometers, as well as the corresponding optical refraction and reflection interference coefficients for each pair. .

[0282] Read the ring installation position number corresponding to each thermometer to be tested in the multi-round reading matrix after correction of all thermometers to be tested, and match the adjacent optical cross-interference coefficient corresponding to the number.

[0283] Read the corrected multi-round reading matrix of all thermometers to be calibrated output in step 5, and extract the ring installation position number of each thermometer in the matrix.

[0284] Using the installation location number of a single thermometer as the search key, an index match is performed in the adjacent optical cross-interference coefficient database; the search output is as follows:

[0285] The set of neighboring thermometer numbers corresponding to this thermometer and the cross-interference coefficient corresponding to each pair of adjacent pairs. ;

[0286] Iterate through all I thermometers to be calibrated, complete the matching of adjacent numbers and interference coefficients of all thermometers, and temporarily cache them.

[0287] Using the adjacent optical cross-interference coefficient obtained by matching, crosstalk cancellation calculation is performed on the corrected final verification value after thermal hysteresis compensation, and the final verification value is output to eliminate the refraction and reflection imaging deviation of the annular adjacent glass.

[0288] Specifically, the operational logic is based on the following calculation formula:

[0289] ;

[0290] in, The final calibration reading of the i-th thermometer to be calibrated after eliminating optical crosstalk;

[0291] The final calibration reading of the i-th thermometer to be calibrated is corrected after thermal hysteresis compensation.

[0292] Let i be the set of adjacent thermometer numbers on the annular array of water droplets on the mounting plate for the i-th thermometer.

[0293] Let be the adjacent optical cross-interference coefficient between the i-th thermometer and the p-th adjacent thermometer;

[0294] The corrected final calibration value is obtained after thermal hysteresis compensation processing for the adjacent thermometer to be calibrated with the serial number p.

[0295] Iterative calculation for each thermometer: For each thermometer numbered i=1,2,…,I, retrieve the matching... and corresponding adjacent Substitute into the formula to solve ;

[0296] Boundary handling: If the thermometer is located at the beginning or end of the circular array and has only one adjacent position on one side, then interference cancellation is performed only on that one adjacent position; for single-branch calibration positions without adjacent positions, the summation term value is 0. = No compensation is required;

[0297] After the calculation is completed, each thermometer is numbered and its final calibration value is determined after eliminating the refraction and reflection deviation of the adjacent annular glass. Each item is bound individually, and an ordered result set is generated and stored in a read-only storage partition.

[0298] The final set of calibration readings for all thermometers to be calibrated after eliminating optical cross-interference. Each data entry in the set contains: thermometer number, ring installation location number, adjacent thermometer number, corresponding cross-interference coefficient, reading before thermal hysteresis compensation, and final reading after crosstalk cancellation. It fully retains the traceability information of the entire compensation chain and can be directly used for reading error calculation and measurement qualification judgment.

[0299] This application has the following advantages:

[0300] It is compatible with both standard mercury thermometers (visual acquisition) and high-precision thermometers (electric direct reading). It automatically assigns traceability weights based on their respective expanded uncertainties, eliminating the need for manual selection of a single standard for comparison. For glass standard thermometers, it uses superimposed lighting and double-layer image offset correction from cameras, while digital standard instruments only perform traceability corrections. This differentiated error processing fully leverages the advantages of both types of standards, resulting in significantly higher reliability of standard temperature measurement compared to single-standard verification schemes. Each round independently generates a fused standard temperature, and then a unified standard reference for the verification point is obtained through multiple rounds of hardware stability weighted averaging, eliminating standard drift caused by random disturbances in single-round acquisition.

[0301] Two-level hierarchical data preprocessing is used to screen out invalid data in advance. The first level is based on the rotation column and camera positioning threshold to remove whole-wheel data with out-of-tolerance mechanism offset. The second level uses the 3σ criterion to remove outliers in the readings, eliminating hardware failures and random jump data from the source, avoiding abnormal data from participating in the fusion calculation, preventing result distortion, and greatly reducing the probability of false detection and false negative detection.

[0302] Five-layer progressive error compensation: Specifically, the clamping sway compensation distinguishes between two types of ring clamping structures, namely the flexible spring 6a and the rigid clamping band 6b, and matches them with exclusive sway compensation values ​​to eliminate the slight offset reading deviation of the thermometer clamping.

[0303] Image offset compensation is based on the height of the fill light and the height of the camera to retrieve the corresponding offset coefficients, and uniformly cancels out the errors in lighting, shadow, and vertical alignment imaging.

[0304] Multi-round weighted fusion noise reduction calculates the acquisition weight based on the rotation and camera normalization deviation of each round. The smaller the deviation, the higher the weight, thus weakening the interference of unstable rounds on the final reading.

[0305] Multi-point thermal hysteresis linkage compensation utilizes historical high and low temperature calibration data to construct a piecewise linear model, specifically to offset the inherent thermal hysteresis material error of the liquid column temperature rise and fall of the glass liquid thermometer, breaking through the industry pain point that single-point calibration cannot correct hysteresis.

[0306] The ring-shaped adjacent optical crosstalk cancellation addresses the unique optical path interference caused by the ring array of water droplets and rotating imaging in this device. It uses an offline calibrated adjacent interference coefficient library to perform linear crosstalk cancellation, thus solving the unique imaging error caused by mutual interference from the refraction and reflection of multiple thermometers.

[0307] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards, characterized in that, The method for fusing and processing multi-round calibration data of the glass liquid thermometer adapted to dual-type standards includes: Step 1: Obtain the preset global parameter configuration table; Step 2: Obtain the original multi-round dataset, which includes the original multi-round data output by each thermometer to be calibrated during the multi-round testing on the automatic calibration system for glass liquid thermometers; Step 3: Preprocess the original multi-round dataset to obtain a preprocessed valid multi-round dataset; Step 4: Generate the corrected standard sequence and the single-round fused standard temperature sequence based on the preprocessed valid multi-round dataset and the preset global parameter configuration table; Step 5: Generate the corrected multi-round reading matrix of all thermometers to be calibrated based on the preprocessed valid multi-round dataset; Step 6: Generate the final error value of each thermometer after fusion based on the single-round fusion standard temperature sequence and the correction reading matrix of the thermometer under test.

2. The method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards as described in claim 1, characterized in that, The preset global parameter configuration table includes basic configuration parameters for the verification task, original parameters for the traceability certificate of the dual-type standard instrument, fixed coefficients for the system hardware error database, thresholds for determining equipment hardware anomalies, and thresholds for determining metrological procedures.

3. The method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards as described in claim 2, characterized in that, Step 3 includes: Data cleaning and preprocessing are performed in two levels and in layers, using the complete original multi-round dataset and the hardware anomaly detection threshold in the preset global parameter configuration table: The first level eliminates invalid whole-wheel data that exceeds the hardware positioning deviation threshold of the rotating column and camera; The second level uses the 3σ statistical criterion to remove outlier abnormal readings from the standard instrument readings and the original readings of each thermometer to be calibrated within the remaining qualified rounds. After two-stage cleaning, valid collection rounds without hardware failures or random outlier interference from readings are selected and summarized to generate a preprocessed valid multi-round dataset.

4. The method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards as described in claim 3, characterized in that, Step 4 includes: By iterating through all valid rounds, double-layer error correction is performed on the standard mercury thermometer round by round, thereby obtaining the fully corrected reading of the standard mercury thermometer for each valid round. By iterating through all valid rounds, the high-precision thermometer is corrected for errors round by round, thereby obtaining the fully corrected reading of the high-precision thermometer for each valid round. A single-round fusion standard temperature is generated based on the fully corrected readings of the standard mercury thermometer or the fully corrected readings of the high-precision thermometer in each valid round. The fully corrected readings of the standard mercury thermometer in each round constitute the corrected first standard sequence; the fully corrected readings of the high-precision thermometer in each round constitute the corrected second standard sequence; and the single-round fusion standard temperatures in each round constitute the single-round fusion standard temperature sequence.

5. The method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards as described in claim 4, characterized in that, Step 5 includes: The clamping sway compensation fixed value is obtained by reading the type identifier of the tensioning component selected for this verification from the global parameter configuration table. Calculate the overall offset correction for each effective round by iterating through the remaining rounds. For each round, perform the following processing: The corrected reading is calculated for each thermometer to be calibrated in this round, thereby obtaining the corrected reading for a single round; Generate a matrix of corrected multi-round readings for all thermometers to be calibrated based on each corrected single-round reading.

6. The method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards as described in claim 5, characterized in that, Step 6 includes: The stability weight of each valid round is calculated for each valid round, thereby obtaining the weight sequence of all valid rounds; The weighted average of the single-round fusion standard temperature sequence is obtained by using the weight sequence of all effective rounds to obtain the unique standard reference temperature of this constant temperature calibration point. Calculations are performed on each thermometer to be tested to obtain the final error value of each thermometer after fusion.

7. The method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards as described in claim 6, characterized in that, The method for fusing and processing multi-round calibration data of the glass liquid thermometer adapted to dual-type standards further includes: Store the final readings of the same thermometer under test after fusion at all historical temperature calibration points and the corresponding single-round fusion standard temperature sequence, and construct a piecewise linear compensation model based on the thermal hysteresis characteristics of the liquid column of the glass liquid thermometer. For the final fused reading obtained from the current target calibration point, retrieve the historical fused data from adjacent high and low temperature calibration points of the thermometer and substitute it into the piecewise linear compensation model to solve for the thermal hysteresis compensation amount. Then, use the thermal hysteresis compensation amount to correct the current final fused reading, thereby obtaining the corrected final calibration reading.

8. The method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards as described in claim 7, characterized in that, After obtaining the corrected final verification value after thermal hysteresis compensation, the multi-round verification data fusion processing method for the glass liquid thermometer adapted to dual-type standards further includes: Retrieve the pre-stored parameters in the global parameter configuration table, such as the annular arrangement spacing of the water droplets on the mounting plate and the single rotation division angle of the rotating column. Obtain a pre-constructed adjacent optical cross-interference coefficient library, which includes the thermometer to be tested and the adjacent optical cross-interference coefficient corresponding to each thermometer to be tested. Read the ring installation position number corresponding to each thermometer to be tested in the multi-round reading matrix after correction of all thermometers to be tested, and match the adjacent optical cross-interference coefficient corresponding to the number. Using the adjacent optical cross-interference coefficient obtained by matching, crosstalk cancellation calculation is performed on the corrected final verification value after thermal hysteresis compensation, and the final verification value is output to eliminate the refraction and reflection imaging deviation of the annular adjacent glass.

9. The method for fusing and processing multi-round calibration data of a glass liquid thermometer adapted to dual-type standards as described in claim 8, characterized in that, The crosstalk cancellation operation performed after the adjacent optical crosstalk coefficient matching satisfies the following calculation formula: ； in, The final calibration reading of the i-th thermometer to be calibrated after eliminating optical crosstalk; The final calibration reading of the i-th thermometer to be calibrated is corrected after thermal hysteresis compensation. Let i be the set of adjacent thermometer numbers on the annular array of water droplets on the mounting plate for the i-th thermometer. Let be the adjacent optical cross-interference coefficient between the i-th thermometer and the p-th adjacent thermometer; The corrected final calibration value is obtained after thermal hysteresis compensation processing for the adjacent thermometer to be calibrated with the serial number p.

10. A data fusion processing device for multi-round calibration of a glass liquid thermometer adapted to dual-type standards, characterized in that, The multi-round calibration data fusion processing device for the glass liquid thermometer with dual-type standard adapter includes: A global parameter configuration table acquisition module is used to acquire a preset global parameter configuration table. The original multi-round dataset acquisition module is used to acquire the original multi-round dataset, which includes the original multi-round data output by each thermometer to be tested in the automatic calibration system for glass liquid thermometers. The preprocessing module is used to preprocess the original multi-round dataset to obtain a preprocessed valid multi-round dataset. The sequence acquisition module is used to generate a corrected standard sequence and a single-round fused standard temperature sequence based on the preprocessed valid multi-round dataset and a preset global parameter configuration table. A matrix acquisition module is used to generate a corrected multi-round reading matrix of all thermometers to be calibrated based on a preprocessed valid multi-round dataset. The error value acquisition module is used to generate the final error value of each thermometer after fusion based on the single-round fusion standard temperature sequence and the correction reading matrix of the thermometer under test.

Images