Cable length metering system and method based on multi-modal identification and dynamic weighing

Abstract

The invention discloses a cable length metering system and method based on multi-modal identification and dynamic weighing, and relates to the technical field of cable measurement, the system comprises a rotary weighing platform, a multi-modal identification module, a cable sheath thickness detection unit and a data processing unit; the rotary weighing platform is used for dynamically acquiring the reference weight of the bobbin and the total weight of the cable; the multi-modal identification module comprises at least two different physical feature identification sub-modules and is used for acquiring geometric parameters, material attributes and identification information of the bobbin; the cable sheath thickness detection unit measures the thickness of a cable sheath in a non-contact mode; the data processing unit integrates the weighing data, the identification data and the thickness data, and establishes a cable length calculation model containing a dynamic compensation factor. The cable length measuring device has the advantages of improving cable length measuring precision, measuring efficiency and adaptability.

Description

Technical Field

[0001] This application relates to the field of cable measurement technology, and in particular to a cable length measurement system and method based on multimodal recognition and dynamic weighing. Background Technology

[0002] In the cable manufacturing process, length measurement is a crucial quality control step. With the continuous development of electronic equipment and industrial equipment, the demand for various electronic cables, industrial equipment connecting wires, and industrial wire harnesses is increasing daily. Accurate cable length measurement is of paramount importance for ensuring product quality, improving production efficiency, and reducing costs. Precise length measurement ensures that the performance of cables in practical applications meets requirements, reducing malfunctions and safety hazards caused by length errors.

[0003] Currently, the mainstream methods for measuring cable length are traditional weighing and pure visual length measurement. Traditional weighing involves subtracting the weight of the empty spool (tare weight) from the total weight of the spool, and then combining this with the weight per unit length of the cable to calculate the length. Pure visual length measurement uses a camera to capture the cable movement process and count the scale. There are also some single-sensor solutions, such as using only laser ranging or RFID to identify the spool.

[0004] However, these existing measurement methods have significant drawbacks. Traditional weighing methods do not consider the differences in spools, as empty spools of different sizes and materials have different weights. They also ignore the impact of wire sheath thickness and supplier material fluctuations on the weight per unit length. Furthermore, static weighing is easily affected by residual glue on the spools and the increased weight due to moisture absorption by the wire sheath, leading to large errors. Pure visual length measurement methods are only suitable for straight-line laying scenarios and cannot be used for measurement in the presence of spools. They also cannot effectively identify irregularly shaped spools or unmarked spools.

[0005] In summary, existing technologies suffer from low accuracy, low efficiency, and poor adaptability in cable length measurement, and urgently need improvement. Summary of the Invention

[0006] To address the shortcomings of existing technologies and improve the accuracy, efficiency, and adaptability of cable length measurement, this application provides a cable length measurement system and method based on multimodal recognition and dynamic weighing.

[0007] Firstly, the objective of this invention is achieved through the following technical solution: The cable length measurement system based on multimodal recognition and dynamic weighing includes: a rotary weighing platform, a multimodal recognition module, a cable sheath thickness detection unit, and a data processing unit; The rotary weighing platform is used to dynamically collect the reference weight of the spool and the total weight of the cable; The multimodal recognition module includes at least two different physical feature recognition sub-modules for obtaining the geometric parameters, material properties and identification information of the spool; The cable sheath thickness detection unit measures the cable sheath thickness in a non-contact manner. The data processing unit integrates the weighing data, identification data, and thickness data to establish a cable length calculation model that includes a dynamic compensation factor.

[0008] By adopting the above technical solution, since the traditional weighing method usually assumes that the cable density and cross-sectional area are constant, while the actual cable will have significant parameter fluctuations due to differences in suppliers, batches or materials, this application obtains the real geometric parameters of the spool (such as outer diameter, inner diameter and height) through a multimodal recognition module, and combines the actual sheath thickness measured by the cable sheath thickness detection unit. At the same time, the data processing unit uses the identified material properties and identification information to call or predict the corresponding density value, so that the unit length mass calculation is close to the physical reality. Specifically, the rotary weighing platform dynamically collects the baseline weight of the spool and the total weight of the cable, avoiding interference from static weighing caused by residual glue on the spool and moisture absorption by the cable sheath. The multimodal recognition module obtains the geometric parameters, material properties, and identification information of the spool through at least two different physical feature recognition sub-modules, integrating multi-dimensional information to adapt to different spools. The cable sheath thickness detection unit measures the cable sheath thickness in a non-contact manner, avoiding errors that may be caused by contact measurement. The data processing unit integrates various data and establishes a cable length calculation model that includes dynamic compensation factors, enabling high-precision and high-efficiency measurement of cable length. It is also applicable to measurement scenarios of different suppliers, cable quality, and even irregularly shaped spools, demonstrating strong adaptability.

[0009] In a preferred embodiment of this application: the rotary weighing platform includes a load-bearing platform, a shock-absorbing bearing, a drive motor, and a high-sampling-rate piezoelectric thin-film sensor. The drive motor drives the spool to be measured on the weighing platform to rotate at a low speed through the shock-absorbing bearing, so as to achieve empty spool tare weight removal; the sampling frequency of the piezoelectric thin-film sensor is not less than 1 kHz.

[0010] By adopting the above technical solution, unlike static tare which requires machine shutdown and is prone to human error and process interruption, the low-speed rotation and continuous sampling used in this invention can complete "dynamic tare" during the laying process, avoiding machine shutdown. Vibration-damping bearings isolate ground vibrations, preventing external disturbances from coupling to the weighing signal; high-frequency piezoelectric sensors (≥1kHz) capture rapid mass changes, supporting real-time net weight calculation.

[0011] In a preferred embodiment of this application, the multimodal recognition module includes: The visual recognition submodule is configured to acquire the contour image of the spool using an industrial camera and a ring light, and extract the outer diameter, inner diameter, and height of the spool based on an improved YOLOv5 model. The spectral analysis submodule is configured to collect the surface reflectance spectrum of the spool using a near-infrared spectrometer and determine the material type using a support vector machine classifier. The OCR recognition submodule is configured to scan the spool label with a high-speed scanner and recognize the supplier name and model code in combination with a custom font library; The results of the visual recognition submodule and the OCR recognition submodule are mutually verified, and the verification result is output.

[0012] By adopting the above technical solution, the single identification method is easily affected by interference such as occlusion, dirt, and reflection (such as label falling off and logo blurring). The present invention significantly improves the accuracy of extracting the geometric parameters of the spool and the reliability of supplier / model identification through the multimodal identification module. Even in scenarios with damaged labels, uneven lighting, or similar materials, the cable identity and structural parameters can still be accurately obtained.

[0013] In a preferred embodiment of this application, the sheath thickness detection unit includes an ultrasonic array probe, an FPGA real-time processing board, and a servo positioning mechanism. The servo positioning mechanism drives the ultrasonic array probe to move along the axis of the current measuring cable and maintain perpendicular contact with the cable sheath surface. The ultrasonic array probe emits ultrasonic waves that penetrate the sheath and acquire echo signals. The FPGA real-time processing board calculates the sheath thickness based on the echo signals and the time-of-flight method.

[0014] By adopting the above technical solution, the thickness measurement accuracy is highly dependent on the perpendicular incident angle between the probe and the cable sheath surface; tilting will cause sound path error. The servo positioning mechanism automatically tracks the cable axis and maintains normal contact to ensure measurement consistency; thus, it achieves high-precision, real-time, and adaptive alignment of the cable sheath thickness online detection, effectively avoiding thickness misjudgment caused by probe tilting or response lag.

[0015] In a preferred embodiment of this application: the dynamic compensation factor is generated by the data processing unit to suppress vibration noise of the weight signal according to the Kalman filter algorithm, and combined with the environmental parameters collected by the built-in temperature and humidity sensor, and to correct the net weight according to the moisture absorption characteristics of the cable material; the moisture absorption characteristic correction is obtained by calculating the water absorption rate of the cable sheath based on a preset material model.

[0016] By adopting the above technical solution, the weighing signal is subject to two main types of interference during actual cable length calculation: mechanical vibration and ambient humidity. Mechanical vibration generates high-frequency noise, while high ambient humidity affects the material's water absorption and weight gain. Kalman filtering can optimally estimate the true mass trajectory; temperature and humidity correction compensates for the mass increment based on the known moisture absorption rate of materials such as PVC, thereby improving the stability and accuracy of the weighing results in complex industrial environments.

[0017] In a preferred example of this application, the expression for the cable length calculation model includes: Where L is the cable length; The dynamic net weight is calculated based on the reference weight of the spool and the total weight of the cable. For dynamic compensation coefficients; These are conductor density and wire sheath density, respectively. These are the conductor cross-sectional area and the wire sheath cross-sectional area, respectively.

[0018] By adopting the above technical solution, the mass per unit cable length consists of two parts: the conductor and the sheath. Due to... Calculated based on measured outer / inner diameter and thickness. Obtained through material identification or prediction, the entire model is based entirely on measured physical quantities rather than nominal values, which helps to ensure that the calculation results directly reflect the true physical state of the cable and fundamentally eliminates system deviations caused by the use of default parameters.

[0019] In a preferred embodiment of this application, the data processing unit is further configured with a machine learning prediction module, which is based on a random forest regressor and predicts the density value of the conductor or wire sheath according to the supplier ID, wire sheath thickness and material type, with a prediction error of no more than ±0.02 g / cm³, and is used to provide density substitution parameters when the supplier has not been registered.

[0020] By employing the above technical solution, density cannot be retrieved from the database when the supplier is not registered. However, density is statistically correlated with material, thickness, and supplier. The random forest model learns this mapping relationship through historical data, and can still provide high-accuracy density estimates even when prior knowledge is missing.

[0021] In a preferred embodiment, this application further includes an output and warning module, which is configured to display the cable length, spool type, supplier, and thickness detection results in real time, and trigger an audible and visual alarm and record abnormal data when any of the following occurs: The cable length deviation exceeds ±1%, the cable sheath thickness exceeds the design tolerance range, or the supplier information is not registered in the system database.

[0022] By adopting the above technical solutions, non-conforming products such as short lengths, thin sheaths, and unauthorized suppliers can be proactively intercepted, preventing problematic cables from flowing into the next process or storage, thus achieving forward quality control.

[0023] Secondly, the objective of this invention is achieved through the following technical solution: A cable length measurement method based on multimodal recognition and dynamic weighing is applied to the cable length measurement system based on multimodal recognition and dynamic weighing as described above. The method includes: The dynamic net weight of the cable is determined by acquiring the baseline weight of the empty spool and the total weight of the cable and spool during the cable laying process. Obtain the bobbin's geometric parameters, material properties, and identification information synchronously output by at least two different physical feature recognition submodules; Acquire cable sheath thickness data measured by the cable sheath thickness detection unit; Based on the dynamic net weight, bobbin geometry parameters, material properties, identification information, and cable sheath thickness data, a cable length calculation model including a dynamic compensation factor is constructed, and the cable length measurement result is output.

[0024] By adopting the above technical solution, a reproducible, programmable, and highly consistent cable length measurement operation specification is provided, ensuring that the same level of accuracy can be achieved when different batches, different operators, and different production lines are executed.

[0025] Thirdly, the objective of this invention is achieved through the following technical solution: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described cable length measurement method based on multimodal recognition and dynamic weighing.

[0026] In summary, this application includes at least one of the following beneficial technical effects: 1. By dynamically collecting the baseline weight of the spool and the total weight of the cable through a rotary weighing platform, and combining the geometric parameters, material properties and identification information of the spool with the multimodal recognition module, as well as the cable sheath thickness measurement unit to measure the cable sheath thickness, and the data processing unit to establish a cable length calculation model including dynamic compensation factors, high-precision measurement of cable length can be achieved. The actual measured high accuracy is ±0.5%, which solves the problem of low accuracy in existing technologies. 2. The multimodal recognition module can quickly acquire information about the spool, and the rotary weighing platform can quickly complete weight acquisition, which improves the efficiency of cable length measurement to ≤2 seconds / spool, solving the problem of low efficiency in existing technologies; 3. The multimodal recognition module can identify irregularly shaped shafts, damaged labels, or non-standard supplier spools, and the system can integrate multi-dimensional information such as weight, material, and thickness, solving the problem of poor adaptability of existing technologies. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of a cable length measurement system based on multimodal recognition and dynamic weighing in one embodiment of this application; Figure 2 This is a flowchart of a cable length measurement method based on multimodal recognition and dynamic weighing in one embodiment of this application. Detailed Implementation

[0028] The present application will be further described in detail below with reference to the accompanying drawings.

[0029] In one embodiment, such as Figure 1 As shown, this application discloses a cable length measurement system based on multimodal recognition and dynamic weighing. The cable length measurement system includes a rotary weighing platform, a multimodal recognition module, a cable sheath thickness detection unit, a data processing unit, and an output and early warning module. The rotary weighing platform is used to dynamically collect the reference weight of the spool and the total weight of the cable; the multimodal recognition module includes at least two different physical feature recognition sub-modules for acquiring the geometric parameters, material properties, and identification information of the spool; the cable sheath thickness detection unit measures the cable sheath thickness in a non-contact manner; the data processing unit integrates the weighing data, recognition data, and thickness data to establish a cable length calculation model including a dynamic compensation factor. The output and early warning module is configured to display the cable length, spool model, supplier, and thickness detection results in real time, and trigger an audible and visual alarm and record abnormal data when any of the following situations occur: the cable length deviation exceeds ±1%, the cable sheath thickness exceeds the design tolerance range, or the supplier information is not registered in the system database. This embodiment achieves high-precision and high-efficiency measurement of cable length, while supporting supplier traceability and quality anomaly early warning. It comprehensively considers multi-dimensional information such as spool differences, cable sheath thickness, and environmental factors, avoiding interference from a single factor, thereby improving the accuracy and efficiency of cable measurement.

[0030] For example, the dynamic compensation factor K is determined as follows: K=K0×(1+α×ΔT+β×ΔH) / (1+γ×a_v), Where K0 is the system calibration coefficient under standard environment, α and β are the influence coefficients of temperature and humidity on cable quality measurement, ΔT and ΔH are the deviation values ​​of current temperature and humidity relative to standard environment, γ is the vibration acceleration sensitivity coefficient, and a_v is the equivalent vibration amplitude estimated by Kalman filtering of the weighing signal.

[0031] Specifically, the rotary weighing platform includes a load-bearing platform, a vibration-damping bearing, a drive motor, and a high-sampling-rate piezoelectric thin-film sensor. The load-bearing platform surface is textured with anti-slip patterns and is typically flat, made of high-strength metals such as stainless steel to ensure stable support of the spool's weight. The anti-slip pattern increases friction between the spool and the load-bearing platform, preventing slippage during rotation. The vibration-damping bearing isolates ground vibrations; its structure is similar to a regular bearing but offers superior shock absorption. It can be made of special rubber or composite materials to achieve this effect. It is mounted below the load-bearing platform and connected to the drive motor and the platform. The drive motor rotates the spool at a low speed to prevent cable tangling. A stepper motor can be used to stably rotate the spool at a suitable speed, such as 0.5–2 rpm. The piezoelectric thin-film sensor, with a sampling frequency of at least 1 kHz and an accuracy of ±0.01 g, is mounted on the load-bearing platform and collects the spool's weight information in real time. Once the spool is placed on the support platform, the drive motor rotates the empty spool for 5 seconds via the shock-absorbing bearing. The piezoelectric film sensor collects a stable weight as the "empty spool reference weight". During the cable laying process, the total weight is continuously collected, and the empty spool reference weight is subtracted to obtain the dynamic net weight of the cable.

[0032] The multimodal recognition module includes a visual recognition submodule, a spectral analysis submodule, and an OCR recognition submodule. The visual recognition submodule includes an industrial camera and a ring-shaped LED fill light. The industrial camera has a resolution of 5 megapixels and a frame rate of 30fps, capable of clearly capturing the outline image of the spool. Its cuboid shape is strategically positioned to ensure complete image capture of the spool. The ring-shaped LED fill light, positioned around the industrial camera, prevents glare and provides uniform illumination.

[0033] The improved YOLOv5-based spool contour detection model can identify the outer diameter, inner diameter, and height of the spool. Combined with the SIFT feature matching library, it can also recognize supplier logos, even for blurred or damaged labels. The spectral analysis submodule includes a near-infrared spectrometer and a fiber optic probe. The near-infrared spectrometer, with a wavelength range of 800-2500 nm, can acquire the surface reflection spectrum of the spool. The fiber optic probe, with its elongated shape, facilitates contact with the spool surface and is used for contact-based acquisition of the spool material's reflection spectrum. By establishing a material spectral database, such as the characteristic spectral peaks of steel, aluminum, and engineering plastics, a support vector machine classifier can be used to identify the spool material. The OCR recognition submodule includes a document scanner and an OCR recognition algorithm using the Tesseract OCR engine and a custom font library. The document scanner is used to scan labels on the side of the spool and is installed in a suitable position for label scanning. The Tesseract OCR engine and custom font library support the recognition of supplier names and model codes in both Chinese and English, and perform double redundancy verification with the visual logo recognition results to improve recognition accuracy. It should be noted that the appropriate location of each module or device in actual application is based on the experience of experts in the field according to the on-site measurement scenario, and this embodiment does not limit it.

[0034] The sheath thickness detection unit comprises an ultrasonic array probe, an FPGA real-time processing board, and a servo positioning mechanism. The ultrasonic array probe operates at a frequency of 5MHz, has 8×8 array elements with a spacing of 0.5mm, and is capable of emitting ultrasonic waves that penetrate the sheath and receiving echo signals. It resembles a small probe array and is mounted on the servo positioning mechanism. The FPGA real-time processing board has a latency of ≤10ms, enabling rapid processing of the echo signals.

[0035] In this embodiment, the FPGA real-time processing board uses the Xilinx Artix-7 series XC7A35T chip as the core processing unit to perform digital bandpass filtering on the echo signal of each channel (center frequency 5 MHz, bandwidth ±1 MHz). The time-of-flight (TOF) extraction method includes a threshold-zero-crossing joint detection algorithm: first, a dynamic threshold is set to 5 times the root mean square of the baseline noise. When the signal first exceeds the threshold, the time t1 of the first zero-crossing point is recorded; simultaneously, the first zero-crossing point t2 of the bottom surface reflected echo is detected; then the propagation time of the ultrasound in the sheath is: Δt = t2 − t1. The FPGA real-time processing board is installed in a control box and connected to the ultrasonic array probe. The servo positioning mechanism is driven by a servo motor, which can move the ultrasonic array probe along the cable axis to the preset detection point, and through the feedback closed-loop control of the tilt sensor, ensure that the probe end face is in perpendicular contact with the cable sheath surface. The ultrasonic array probe emits ultrasonic waves that penetrate the cable sheath and receives the echo signal. The FPGA real-time processing board can calculate the cable sheath thickness based on the echo signal and the time-of-flight (TOF) method.

[0036] For example, the servo positioning mechanism includes: a high-precision linear module, a servo motor, an encoder, a tilt sensor, and a probe clamping arm. The linear module uses a THK KR20 roller guide rail with a repeatability of ±0.01 mm. The servo motor is equipped with a 20-bit absolute encoder; a miniature MEMS tilt sensor is integrated at the end of the probe clamping arm. After system startup, the data processing unit generates the probe target trajectory based on the spool height H and the current cable placement position obtained by the visual recognition submodule. The servo controller drives the motor to move the probe at a constant speed along the cable axis, while simultaneously reading the tilt sensor feedback in real time. If the tilt deviation exceeds ±0.5°, the controller immediately initiates fine-tuning compensation: the probe pitch angle is adjusted in a closed loop using a piezoelectric ceramic micro-displacement device (stroke ±0.2 mm) to ensure that the ultrasonic emitting surface is always aligned with the cable surface normal. The vertical contact status is verified in real time using the echo signal-to-noise ratio (SNR): if the SNR < 20 dB, it is determined to be a poor contact, triggering repositioning. The spool height H is transmitted to the servo positioning mechanism of the sheath thickness detection unit as a travel boundary parameter for the axial movement of the ultrasonic probe. The probe scanning range is automatically set based on H, for example, from 5 mm from the end face to H−5 mm.

[0037] For example, the probe emits ultrasonic pulses that penetrate the sheath and receives the echo signal from the bottom surface. The formula for calculating the sheath thickness Tz is: Tz = (c × Δt) / 2, where c is the propagation speed of ultrasound in PVC, approximately 2400 m / s, and Δt is the round-trip time. The overall sheath thickness measurement delay is ≤10 ms, meeting the requirements for online detection.

[0038] Specifically, the core function of the data processing unit is to fuse multi-source data and perform length measurement and quality analysis. The data processing unit is controlled by an ARM Cortex-A72 microcontroller and an NPU coprocessor. The dynamic compensation algorithm includes vibration suppression and environmental correction. Vibration suppression removes high-frequency vibration noise from the weighing signal using Kalman filtering, a commonly used filtering algorithm that removes noise through signal prediction and updates. The data processing unit incorporates a temperature and humidity sensor with an accuracy of ±0.5℃ / ±2%RH; it corrects the net weight based on the cable's moisture absorption characteristics, such as a PVC sheath water absorption rate of 0.3% / RH%. For example, for PVC sheaths, the mass increment is calculated based on a water absorption rate of 0.3% / RH%. Net weight after compensation for: ,in This is dynamic net weight data. The dynamic compensation coefficient K defaults to 1.02 and can be fine-tuned through on-site calibration. To accommodate differences in materials from different batches, a preset material model provides on-site calibration functionality.

[0039] The moisture absorption characteristic correction is based on the calculation of the cable sheath water absorption rate using a preset material model. The data processing unit has a built-in material moisture absorption characteristic database, which stores the mass change mapping relationships of common cable sheath materials under different temperature and humidity environments. The preset material model refers to the mathematical relationships or data tables that are pre-established and stored in the data processing unit before system deployment or during the initialization phase, describing the mass change laws of different cable sheath materials under specific environmental temperature and humidity conditions. The preset material model correlates environmental parameters with the unit mass increment (or water absorption rate) caused by material moisture absorption. The system collects environmental parameters in real time through temperature and humidity sensors and compensates for the moisture absorption weight gain based on these parameters. The standard environmental parameters are 23°C and 50%RH.

[0040] The expression for the cable length calculation model is as follows: Where L is the cable length, The dynamic net weight is calculated based on the baseline weight of the spool and the total weight of the cable. For dynamic compensation coefficients, These are conductor density and wire sheath density, respectively. These are the conductor cross-sectional area and the wire sheath cross-sectional area, respectively; where the conductor cross-sectional area... Cross-sectional area of ​​the wire Where D is the outer diameter of the spool, that is, the diameter of the outermost layer of the entire spool after the cable is fully wound, in millimeters (mm), and d is the inner diameter of the spool, that is, the diameter of the hollow core around which the cable is wound.

[0041] The data processing unit is also equipped with a machine learning prediction module, which uses a random forest regressor to predict the density value of the conductor or wire sheath based on the supplier ID, wire sheath thickness and material type. The prediction error does not exceed ±0.02 g / cm³, and it is used to provide density substitute parameters when the supplier has not been registered.

[0042] Specifically, the output and early warning module includes a display interface and an early warning mechanism. The display interface can show the cable length, spool type, supplier, and thickness test results in real time. It can be an LCD screen mounted on the control panel for easy viewing by operators. The early warning mechanism triggers an audible and visual alarm and records abnormal data when any of the following conditions occur: cable length deviation exceeds ±1%, cable sheath thickness exceeds the design tolerance range (e.g., ±0.1mm), or supplier information is not registered in the system database. The audible and visual alarm can be implemented using a buzzer and indicator lights, and the recorded abnormal data can be stored in the system database.

[0043] The implementation principle of the cable length measurement system based on multimodal recognition and dynamic weighing in this embodiment is as follows: First, a rotary weighing platform continuously acquires the empty spool reference weight and the total cable weight during the cable laying process, and calculates the dynamic net weight in real time to avoid operational errors introduced by stopping the machine for tare. Second, the multimodal recognition module synchronously extracts the spool geometry, sheath material type, and supplier identification information to provide accurate input for cross-sectional area calculation and density retrieval. The ultrasonic array probe and servo positioning mechanism perform vertical contact thickness detection on the cable sheath. On this basis, the data processing unit constructs a dynamic compensation model with a dual correction mechanism: on the one hand, Kalman filtering is used to eliminate weighing noise caused by mechanical vibration; on the other hand, the net weight is physically corrected based on a preset material moisture absorption characteristic model and real-time temperature and humidity data. Furthermore, when supplier information is missing, the density can be adaptively estimated based on material and thickness through a machine learning prediction module to ensure measurement continuity. Finally, the corrected net weight, measured cross-sectional area, and calibrated density are substituted into the cable length calculation model to output a high-precision length result.

[0044] In another embodiment, such as Figure 2 As shown, this application also discloses a cable length measurement method based on multimodal recognition and dynamic weighing. This cross-platform financial market data communication method is applied to the cable length measurement system based on multimodal recognition and dynamic weighing described above. The cable length measurement method based on multimodal recognition and dynamic weighing specifically includes the following steps: S1: Obtain the baseline weight of the empty spool and the total weight of the cable and spool dynamically collected during the cable laying process to determine the dynamic net weight of the cable.

[0045] In this embodiment, the spool to be tested is placed on a rotary weighing platform, and the drive motor is started to rotate the spool at a low speed of 0.5–2 rpm. The stable weight is collected for 5 seconds without wire being laid as the reference weight of the empty spool. Then the wire is laid, and the piezoelectric film sensor continuously collects the total weight of the cable and spool at a frequency of ≥1 kHz. The dynamic net weight = total weight of cable and spool - reference weight of empty spool.

[0046] S2: Obtain the spool geometric parameters, material properties, and identification information synchronously output by at least two different physical feature recognition submodules.

[0047] In this embodiment, synchronous triggering occurs during the rotation of the spindle: An industrial camera captures the contour image of the spool, and the outer diameter D, inner diameter d, and height H are extracted using an improved YOLOv5 model; a near-infrared spectrometer collects the surface reflectance spectrum, and an SVM classifier is used to determine the sheath material type, such as PVC or PE; a high-speed scanner scans the side label, and an OCR engine is used to identify the supplier name and model code; the visual logo recognition result is cross-validated with the OCR text, and the final identity and structural parameters are output.

[0048] S3: Obtain the cable sheath thickness data measured by the cable sheath thickness detection unit.

[0049] In this embodiment, the servo positioning mechanism drives the ultrasonic array probe to move along the cable axis to the detection point, and ensures that the probe is in perpendicular contact with the sheath surface through tilt feedback control; the probe emits 5 MHz ultrasonic waves and receives the bottom echo signal; the FPGA real-time processing board calculates the sheath thickness based on the time-of-flight (TOF) method.

[0050] S4: Based on dynamic net weight, bobbin geometry parameters, material properties, identification information and cable sheath thickness data, construct a cable length calculation model that includes dynamic compensation factors, and output the cable length measurement results.

[0051] In this embodiment, the data processing unit first applies a Kalman filter to the dynamic net weight to suppress vibration noise, and generates a dynamic compensation factor based on temperature and humidity sensor data and a preset material moisture absorption model; then, it retrieves or predicts conductor density and sheath density based on material type, and finally substitutes them into the cable length calculation model. The output cable length L is displayed in real time through the display interface.

[0052] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0053] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: S1: Obtain the baseline weight of the empty spool and the total weight of the cable and spool dynamically collected during the cable laying process to determine the dynamic net weight of the cable; S2: Obtain the bobbin geometric parameters, material properties, and identification information synchronously output by at least two different physical feature recognition submodules; S3: Obtain cable sheath thickness data measured by the cable sheath thickness detection unit; S4: Based on dynamic net weight, bobbin geometry parameters, material properties, identification information and cable sheath thickness data, construct a cable length calculation model that includes dynamic compensation factors, and output the cable length measurement results.

[0054] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0055] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0056] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A cable length measurement system based on multimodal recognition and dynamic weighing, characterized in that, include: Rotary weighing platform, multimodal recognition module, sheath thickness detection unit and data processing unit; The rotary weighing platform is used to dynamically collect the reference weight of the spool and the total weight of the cable; The multimodal recognition module includes at least two different physical feature recognition sub-modules for obtaining the geometric parameters, material properties and identification information of the spool; The cable sheath thickness detection unit measures the cable sheath thickness in a non-contact manner. The data processing unit integrates the weighing data, identification data, and thickness data to establish a cable length calculation model that includes a dynamic compensation factor.

2. The cable length measurement system based on multimodal recognition and dynamic weighing according to claim 1, characterized in that, The rotary weighing platform includes a load-bearing platform, a shock-absorbing bearing, a drive motor, and a high-sampling-rate piezoelectric thin-film sensor. The drive motor drives the spool to be measured on the weighing platform to rotate at a low speed through the shock-absorbing bearing, so as to achieve empty spool tare. The sampling frequency of the piezoelectric thin-film sensor is not less than 1 kHz.

3. The cable length measurement system based on multimodal recognition and dynamic weighing according to claim 1, characterized in that, The multimodal recognition module includes: The visual recognition submodule is configured to acquire the contour image of the spool using an industrial camera and a ring light, and extract the outer diameter, inner diameter, and height of the spool based on an improved YOLOv5 model. The spectral analysis submodule is configured to collect the surface reflectance spectrum of the spool using a near-infrared spectrometer and determine the material type using a support vector machine classifier. The OCR recognition submodule is configured to scan the spool label with a high-speed scanner and recognize the supplier name and model code in combination with a custom font library; The results of the visual recognition submodule and the OCR recognition submodule are mutually verified, and the verification result is output.

4. The cable length measurement system based on multimodal recognition and dynamic weighing according to claim 1, characterized in that, The sheath thickness detection unit includes an ultrasonic array probe, an FPGA real-time processing board, and a servo positioning mechanism. The servo positioning mechanism drives the ultrasonic array probe to move along the axis of the current measuring cable and maintain perpendicular contact with the cable sheath surface. The ultrasonic array probe emits ultrasonic waves that penetrate the sheath and acquire echo signals. The FPGA real-time processing board calculates the sheath thickness based on the echo signals and the time-of-flight method.

5. The cable length measurement system based on multimodal recognition and dynamic weighing according to claim 1, characterized in that, The dynamic compensation factor is generated by the data processing unit after suppressing vibration noise of the weight signal according to the Kalman filter algorithm, and after combining the environmental parameters collected by the built-in temperature and humidity sensor and correcting the net weight according to the moisture absorption characteristics of the cable material. The moisture absorption characteristic correction is obtained by calculating the water absorption rate of the sheath based on a preset material model.

6. The cable length measurement system based on multimodal recognition and dynamic weighing according to claim 1, characterized in that, The expression for the cable length calculation model includes: Where L is the cable length; The dynamic net weight is calculated based on the reference weight of the spool and the total weight of the cable. For dynamic compensation coefficients; These are conductor density and wire sheath density, respectively. These are the conductor cross-sectional area and the wire sheath cross-sectional area, respectively.

7. The cable length measurement system based on multimodal recognition and dynamic weighing according to claim 1, characterized in that, The data processing unit is also equipped with a machine learning prediction module. The machine learning prediction module is based on a random forest regressor and predicts the density value of the conductor or wire sheath according to the supplier ID, wire sheath thickness and material type. The prediction error does not exceed ±0.02 g / cm³. It is used to provide density substitution parameters when the supplier has not been registered.

8. The cable length measurement system based on multimodal recognition and dynamic weighing according to claim 6, characterized in that, It also includes an output and early warning module, which is configured to display the cable length, spool type, supplier, and thickness detection results in real time, and trigger an audible and visual alarm and record abnormal data when any of the following situations occur: The cable length deviation exceeds ±1%, the cable sheath thickness exceeds the design tolerance range, or the supplier information is not registered in the system database.

9. A cable length measurement method based on multimodal recognition and dynamic weighing, characterized in that, The method, applied to the cable length measurement system based on multimodal recognition and dynamic weighing as described in any one of claims 1 to 8, comprises: The dynamic net weight of the cable is determined by acquiring the baseline weight of the empty spool and the total weight of the cable and spool during the cable laying process. Obtain the bobbin's geometric parameters, material properties, and identification information synchronously output by at least two different physical feature recognition submodules; Acquire cable sheath thickness data measured by the cable sheath thickness detection unit; Based on the dynamic net weight, bobbin geometry parameters, material properties, identification information, and cable sheath thickness data, a cable length calculation model including a dynamic compensation factor is constructed, and the cable length measurement result is output.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the cable length measurement method based on multimodal recognition and dynamic weighing as described in claim 9.

Images