A cloud platform-based production quality inspection method and system

Abstract

This invention discloses a production quality inspection method and system based on a cloud platform, relating to the fields of intelligent manufacturing and quality control technology. The method includes: the cloud platform acquiring the spectral and temperature data of drill pipe raw materials, generating a process fingerprint database and risk classification labels; the production system executing differentiated process allocation according to the labels, collecting coaxiality data for each process and uploading it to the cloud platform; the cloud platform constructing a coaxiality drift map, generating compensation instructions to adjust annealing temperature control parameters; after production is completed, conducting pre-assembly inspection of drill pipes and drill bits, determining passability through conical pressure distribution; tracing and coupling non-conforming products, calculating the process contribution rate, and updating process rules. This invention achieves closed-loop quality control throughout the entire process from raw materials to finished products.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing and quality control technology, and in particular to a production quality inspection method and system based on a cloud platform. Background Technology

[0002] In the oil drilling industry, drill pipe, as a key consumable material for oil and gas extraction, directly determines the overall mechanical properties and downhole safety of the drill string assembly through the machining precision of its threaded connections. With the development of deep wells, ultra-deep wells, and unconventional oil and gas resources, stringent requirements have been placed on the sealing reliability, torsional resistance, and fatigue life of drill pipe. However, in existing drill pipe manufacturing lines, quality control models still commonly suffer from problems such as "emphasizing results over process," "severe data silos," and "rigid processes."

[0003] Existing quality inspection methods largely rely on offline sampling and final destructive testing of finished products, lacking the ability to trace the source of fluctuations in raw material smelting composition and changes in temperature field during hot rolling. Because steel from different furnace batches exhibits differences in chemical composition and temperature stability during hot forming, this material property dispersion directly leads to uneven stress release and deformation drift in subsequent machining processes. Traditional fixed process routes cannot adaptively adjust to this dispersion. Furthermore, in the precision thread machining stage, the cumulative effect of coaxiality errors across multiple processes is often overlooked, and existing quality control methods struggle to dynamically compensate for micron-level dimensional and positional tolerances. When the drill rod and drill bit are in a tapered fit, minute contact unevenness can cause stress concentration under high torque conditions, leading to sticking or leakage failure.

[0004] Therefore, there is an urgent need for a production quality inspection method and system that can connect the entire chain of data from raw materials to process to finished products and use cloud-based intelligent decision-making to achieve real-time closed-loop optimization of process parameters, in order to solve the technical pain points of poor quality stability, high scrap rate and difficulty in traceability in the existing manufacturing model. Summary of the Invention

[0005] This invention proposes a production quality inspection method based on a cloud platform, comprising: S1. The cloud platform obtains the chemical composition spectral data and hot rolling forming temperature sequence of the drill pipe raw material furnace batch number; retrieves the preset reference content, calculates the deviation between the actual content and the reference content of each element, and counts the fluctuation amplitude value of the temperature sequence; associates and stores the furnace batch number, spectral data, temperature sequence, deviation and fluctuation amplitude value as a raw material process fingerprint database; generates risk classification labels based on the deviation and fluctuation amplitude value, and sends them to the production quality inspection system. S2. The production quality inspection system performs differentiated process path allocation based on the graded labels, allocating high-risk raw materials to precision process paths that include online annealing and dynamic compensation processes, and allocating medium- and low-risk raw materials to conventional process paths; it also collects coaxiality deviation data at each process node of the drill pipe thread section and uploads it to the cloud platform. S3. The cloud platform constructs a multi-process coaxiality drift map based on the received coaxiality deviation data and extracts the feature value of the cumulative coaxiality drift. Based on the difference between the feature value of the cumulative coaxiality drift and the target value, a process compensation instruction is generated and sent to the production quality inspection system. The temperature control parameters of the online annealing dynamic compensation process are adjusted according to the process compensation instruction. S4. The production quality inspection system completes production according to the adjusted temperature control parameters, performs assembly pre-testing on the finished drill pipe and matching drill bit, collects the cone surface contact pressure distribution data through the ring array thin film pressure sensor, performs frequency domain analysis on the pressure distribution data, extracts the ratio of the fundamental component amplitude to the DC component amplitude, uses this ratio as the cone surface contact uniformity index, and compares it with the qualified threshold to determine whether the finished product is qualified. S5 and the cloud platform trace the source of non-conforming finished products by coupling the furnace batch number, coaxiality cumulative drift, and conical surface contact uniformity index, and calculate their contribution rate to the conical surface contact unevenness defect; update the corresponding furnace batch number process allocation rules according to the contribution rate and issue them for subsequent process allocation of similar furnace batch numbers; continuously accumulate data to iteratively optimize model parameters and form a closed-loop dynamic control.

[0006] The above-described cloud-based production quality inspection method includes the following steps: the cloud platform acquires the chemical composition spectral data and hot rolling temperature sequence of the drill pipe raw material furnace batch number; retrieves a preset reference content, calculates the deviation between the actual content and the reference content of each element, and statistically analyzes the fluctuation range of the temperature sequence; associates and stores the furnace batch number, spectral data, temperature sequence, deviation, and fluctuation range as a raw material process fingerprint database; and generates a risk classification label based on the deviation and fluctuation range, which is then sent to the production quality inspection system. The deviation between the actual content and the reference content of each element is compared with the preset tolerance range of the deviation of each element. If the deviation of any key element exceeds the key element deviation threshold, it is initially marked as a high-risk candidate. The fluctuation amplitude of the temperature sequence is statistically analyzed. When the fluctuation amplitude exceeds the preset upper limit threshold for temperature stability, the risk level is adjusted upward based on the initial labeling. Conversely, when the fluctuation amplitude is lower than the preset lower limit threshold for temperature stability, the risk level is adjusted downward. Based on the combined assessment results of the deviation and fluctuation amplitude values, a final risk classification label is generated, where the high-risk classification label is associated with the more stringent online annealing dynamic compensation initiation conditions in subsequent processes.

[0007] The production quality inspection method based on a cloud platform, as described above, involves the production quality inspection system performing differentiated process path allocation according to graded labels. High-risk raw materials are allocated to precision process paths that include online annealing and dynamic compensation processes, while medium- and low-risk raw materials are allocated to conventional process paths. Coaxiality deviation data is collected at each process node of the drill pipe thread section and uploaded to the cloud platform, including the following sub-steps: After the rough turning process of the drill pipe thread section is completed, the first radial runout between the thread section reference surface and the tool path is collected by the first laser displacement sensor as the initial coaxiality deviation. After the semi-finishing process is completed, the second radial runout between the thread pitch diameter and the spindle rotation center line is collected by the second laser displacement sensor to generate the intermediate coaxiality deviation. After the threading process and the subsequent online annealing dynamic compensation process are completed, the tooth profile half-angle deviation and pitch cumulative deviation of the finished thread are collected by the third laser displacement sensor. The first radial runout, the second radial runout, the tooth profile half-angle deviation and the pitch cumulative deviation are combined according to the process sequence to form a multi-dimensional coaxiality deviation data sequence.

[0008] The production quality inspection method based on a cloud platform, as described above, involves the cloud platform constructing a multi-process coaxiality drift map based on received coaxiality deviation data and extracting the feature value of the cumulative coaxiality drift. Based on the difference between the feature value of the cumulative coaxiality drift and the target value, a process compensation instruction is generated and sent to the production quality inspection system. The temperature control parameters of the online annealing dynamic compensation process are adjusted according to the process compensation instruction. This method includes the following sub-steps: With the process sequence number as the horizontal axis and the root mean square value of the radial runout in the coaxiality deviation data collected at each process node as the vertical axis, a curve showing the change of coaxiality deviation as the process progresses is plotted to form the multi-process coaxiality drift map. Piecewise linear regression was performed on the change curve to calculate the intra-segment drift rate of the roughing section, semi-finishing section and finishing section respectively, and the cumulative deviation value of the final process was used as the characteristic value of the coaxiality cumulative drift. The cumulative coaxiality drift characteristic value is compared with the preset process capability upper limit value, and the difference between the two is calculated. If the difference is positive and exceeds the allowable residual deviation threshold, a process compensation instruction is generated. The process compensation instruction includes the adjustment amount of the temperature control parameters of the online annealing dynamic compensation process.

[0009] The cloud-based production quality inspection method described above includes the following sub-steps: The production quality inspection system completes production according to adjusted temperature control parameters, performs assembly pre-testing on the finished drill pipe and matching drill bit, collects conical surface contact pressure distribution data using a ring array thin-film pressure sensor, performs frequency domain analysis on the pressure distribution data, extracts the ratio of the fundamental component amplitude to the DC component amplitude, uses this ratio as the conical surface contact uniformity index, and compares it with a pass threshold to determine whether the finished product is qualified. The finished drill pipe to be tested and the standard matching drill bit are installed on the assembly pre-testing platform. The drill pipe and the drill bit are driven to perform conical surface mating with a preset rated torque. During the mating process, the axial clamping displacement is recorded by the servo motor until the target torque value is reached. In the conical mating area of ​​the assembly pre-testing platform, multiple discrete pressure values ​​are collected between the drill rod conical surface and the drill bit conical hole in the axial fitting state by thin film pressure sensors arranged in a ring array. Each discrete pressure value corresponds to the contact pressure at a circumferential angle position. Frequency domain analysis is performed on the multiple discrete pressure values ​​to extract the ratio of the fundamental component amplitude to the DC component amplitude. This ratio is used as the cone surface contact uniformity index, which is used to characterize the uniformity of the pressure distribution.

[0010] The production quality inspection method based on a cloud platform, as described above, involves the cloud platform tracing and coupling the furnace batch number, cumulative coaxiality drift, and conical surface contact uniformity index associated with non-conforming finished products to calculate their contribution rate to the conical surface contact unevenness defect; updating and issuing the corresponding furnace batch number process allocation rules based on the contribution rate for subsequent process allocation of similar furnace batch numbers; and continuously accumulating data to iteratively optimize model parameters, forming a closed-loop dynamic control, including the following sub-steps: The system retrieves the cumulative coaxiality drift characteristic values ​​of all drill rods with the same furnace batch number as the current unqualified finished products from the historical database of the cloud platform, forming a set of coaxiality characteristic values. At the same time, it retrieves the cone surface contact uniformity index corresponding to these drill rods, forming a set of uniformity indices. Partial least squares regression analysis was performed on the set of coaxiality cumulative drift characteristic values ​​and the set of conical surface contact uniformity index to extract the loading coefficient of the first principal component. The square value of the loading coefficient was used as the global contribution rate of the raw material of this batch to the conical surface contact non-uniformity defect. Furthermore, the coaxiality deviation data of each process node is used as the independent variable, and the conical surface contact uniformity index is used as the dependent variable. The stepwise regression algorithm is used to screen out the process nodes with statistical significance, calculate the standardized regression coefficient of each screened process, and normalize each standardized regression coefficient as the process transfer contribution rate of each process.

[0011] The production quality inspection method based on a cloud platform, as described above, further uses the coaxiality deviation data of each process node as the independent variable and the conical surface contact uniformity index as the dependent variable. A stepwise regression algorithm is used to screen out process nodes with statistical significance. The standardized regression coefficient of each screened process is calculated, and the normalized standardized regression coefficients are used as the process transfer contribution rate of each process. This method includes the following sub-steps: The global contribution rate is compared with a preset raw material sensitivity threshold. If the global contribution rate exceeds the raw material sensitivity threshold, the initial risk classification label of the raw material of that batch number will be forcibly increased by one level in the next allocation. Based on the process transfer contribution rate of each process, the two key processes with the highest contribution rates are identified, and supplementary detection instructions are generated for these two key processes. The supplementary detection instructions require that when the raw materials of the same batch are put back into production, all process parameters of the two key processes be recorded and the sampling frequency be increased to twice the normal sampling frequency. The upgraded risk classification labels and supplementary testing instructions are integrated to form updated process path allocation rules, which are then stored back in the raw material process fingerprint database as records associated with the batch number of the furnace. At the same time, the updated rules are sent to the rule engine of the production quality inspection system in the form of structured data.

[0012] This invention also proposes a cloud platform-based production quality inspection system, comprising: Cloud platform data processing module: The cloud platform acquires the chemical composition spectral data and hot rolling forming temperature sequence of the drill pipe raw material furnace batch number; retrieves the preset reference content, calculates the deviation between the actual content and the reference content of each element, and counts the fluctuation range value of the temperature sequence; associates and stores the furnace batch number, spectral data, temperature sequence, deviation and fluctuation range value as a raw material process fingerprint database; generates risk classification labels based on the deviation and fluctuation range value, and sends them to the production quality inspection system; Production quality inspection execution module: The production quality inspection system performs differentiated process path allocation based on the grade label, allocating high-risk raw materials to precision process paths that include online annealing and dynamic compensation processes, and allocating medium- and low-risk raw materials to conventional process paths; it also collects coaxiality deviation data at each process node of the drill pipe thread section and uploads it to the cloud platform; The cloud platform collaborative control module: The cloud platform constructs a multi-process coaxiality drift map based on the received coaxiality deviation data and extracts the feature value of the cumulative coaxiality drift. Based on the difference between the feature value of the cumulative coaxiality drift and the target value, it generates a process compensation instruction and sends it to the production quality inspection system. Based on the process compensation instruction, it adjusts the temperature control parameters of the online annealing dynamic compensation process. Assembly pre-test module: The production quality inspection system completes production according to the adjusted temperature control parameters, performs assembly pre-test on the finished drill pipe and matching drill bit, collects the cone surface contact pressure distribution data through the ring array thin film pressure sensor, performs frequency domain analysis on the pressure distribution data, extracts the ratio of the fundamental component amplitude to the DC component amplitude, uses this ratio as the cone surface contact uniformity index, and compares it with the qualified threshold to determine whether the finished product is qualified. Intelligent decision-making and feedback module: The cloud platform traces and couples the furnace batch number, coaxiality cumulative drift, and conical surface contact uniformity index associated with non-conforming finished products to calculate their contribution rate to the conical surface contact unevenness defect; updates the corresponding furnace batch number process allocation rules based on the contribution rate and issues them for subsequent process allocation of similar furnace batch numbers; continuously accumulates data to iteratively optimize model parameters and form a closed-loop dynamic control.

[0013] The beneficial effects achieved by this invention are as follows: By constructing a raw material process fingerprint database and risk classification labels, differentiated process path allocation is achieved, reducing quality risks from the source. Multi-process coaxiality drift maps are used to quantitatively characterize the deviation transmission process, generating precise process compensation instructions to proactively intervene in process drift. The introduction of ring array thin-film pressure sensors and frequency domain analysis transforms conical surface contact uniformity into a quantitative index, improving final inspection accuracy. A complete traceability feedback chain is established from non-conforming finished products to raw materials and key processes, dynamically updating process allocation rules to form closed-loop control, significantly improving the stability and consistency of drill pipe and drill bit fit quality. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0015] Figure 1 This is a flowchart of a cloud-based production quality inspection method provided in an embodiment of this application.

[0016] Figure 2 This is a schematic diagram of a cloud-based production quality inspection system provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1

[0019] As Figure 1 shown, a production quality inspection method based on a cloud platform in an embodiment of the present application includes: Step S1: The cloud platform obtains the chemical composition spectral data and the hot rolling forming temperature sequence of the drill pipe raw material furnace batch number; retrieves the preset reference content, calculates the deviation between the actual content of each element and the reference content, and statistically calculates the fluctuation amplitude value of the temperature sequence; associates and stores the furnace batch number, spectral data, temperature sequence, deviation, and fluctuation amplitude value as a raw material process fingerprint database; generates a risk classification label based on the deviation and fluctuation amplitude value, and issues it to the production quality inspection system; Specifically, compare the deviation between the actual content of each element and the reference content with the preset single-element deviation tolerance interval respectively. If the deviation of any key element exceeds the threshold, it is initially marked as a high-risk candidate; statistically calculate the fluctuation amplitude value of the temperature sequence. When it exceeds the upper threshold of temperature stability, the risk is increased on the basis of the initially judged level, and when it is lower than the lower threshold of temperature stability, it is decreased; comprehensively evaluate the results of the deviation and fluctuation amplitude, and output the final high, medium, and low risk classification labels through the risk level mapping table. The high-risk classification label corresponds to more stringent online annealing dynamic compensation start conditions in subsequent processes, specifically including the following sub-steps: Step S11: Compare the deviation between the actual content of each element and the reference content with the preset single-element deviation tolerance interval respectively. If the deviation of any key element exceeds the key element deviation threshold, it is initially marked as a high-risk candidate;

[0018] Example 1

[0019] As Figure 1 shown, a production quality inspection method based on a cloud platform in an embodiment of the present application includes: Step S1: The cloud platform obtains the chemical composition spectral data and the hot rolling forming temperature sequence of the drill pipe raw material furnace batch number; retrieves the preset reference content, calculates the deviation between the actual content of each element and the reference content, and statistically calculates the fluctuation amplitude value of the temperature sequence; associates and stores the furnace batch number, spectral data, temperature sequence, deviation, and fluctuation amplitude value as a raw material process fingerprint database; generates a risk classification label based on the deviation and fluctuation amplitude value, and issues it to the production quality inspection system; Specifically, compare the deviation between the actual content of each element and the reference content with the preset single-element deviation tolerance interval respectively. If the deviation of any key element exceeds the threshold, it is initially marked as a high-risk candidate; statistically calculate the fluctuation amplitude value of the temperature sequence. When it exceeds the upper threshold of temperature stability, the risk is increased on the basis of the initially judged level, and when it is lower than the lower threshold of temperature stability, it is decreased; comprehensively evaluate the results of the deviation and fluctuation amplitude, and output the final high, medium, and low risk classification labels through the risk level mapping table. The high-risk classification label corresponds to more stringent online annealing dynamic compensation start conditions in subsequent processes, specifically including the following sub-steps: Step S11: Compare the deviation between the actual content of each element and the reference content with the preset single-element deviation tolerance interval respectively. If the deviation of any key element exceeds the key element deviation threshold, it is initially marked as a high-risk candidate; The cloud platform analyzes the actual mass fraction of each chemical element of the drill pipe raw material from the spectral data, including key alloy elements such as carbon, manganese, chromium, and molybdenum, and at the same time retrieves the standard reference content corresponding to this grade of steel. For each key element, calculate the actual content and the reference content difference to obtain a signed deviation value . The cloud platform internally presets a single deviation tolerance interval for each key element, which consists of a lower tolerance value and an upper tolerance value. Compare the actual deviation with the tolerance interval: if the deviation completely falls within the tolerance interval, it is determined that the element is qualified; if the deviation exceeds the lower or upper limit of the interval, it is determined that the element is deviated. When the deviation of any key element exceeds its corresponding key element deviation threshold, the system immediately initially marks this furnace batch number as a high-risk candidate, and records the name and deviation amount of the over-standard element for subsequent comprehensive risk level assessment.

[0020] Step S12: Statistically calculate the fluctuation amplitude value of the temperature sequence. When the fluctuation amplitude value exceeds the preset upper threshold of temperature stability, increase it on the basis of the initially marked risk level. Conversely, when the fluctuation amplitude value is lower than the preset lower threshold of temperature stability, decrease it; The cloud platform analyzes the actual mass fraction of each chemical element of the drill pipe raw material from the spectral data, including key alloy elements such as carbon, manganese, chromium, and molybdenum, and at the same time retrieves the standard reference content corresponding to this grade of steel. For each key element, calculate the actual content and the reference content difference to obtain a signed deviation value . The cloud platform internally presets a single deviation tolerance interval for each key element, which consists of a lower tolerance value and an upper tolerance value. Compare the actual deviation with the tolerance interval: if the deviation completely falls within the tolerance interval, it is determined that the element is qualified; if the deviation exceeds the lower or upper limit of the interval, it is determined that the element is deviated. When the deviation of any key element exceeds its corresponding key element deviation threshold, the system immediately initially marks this furnace batch number as a high-risk candidate, and records the name and deviation amount of the over-standard element for subsequent comprehensive risk level assessment.

[0020] Step S12: Statistically calculate the fluctuation amplitude value of the temperature sequence. When the fluctuation amplitude value exceeds the preset upper threshold of temperature stability, increase it on the basis of the initially marked risk level. Conversely, when the fluctuation amplitude value is lower than the preset lower threshold of temperature stability, decrease it; The cloud platform acquires temperature sequences collected along the time axis during the hot rolling process. These sequences include multiple sampling points from the initial rolling temperature to the final rolling temperature. The standard deviation of this sequence is used as the fluctuation amplitude value to quantify the stability of temperature control. The system presets two boundary thresholds: an upper limit threshold for temperature stability and a lower limit threshold for temperature stability. If the calculated standard deviation exceeds the upper limit threshold, it indicates severe temperature fluctuations, requiring an upward adjustment of one level from the initially marked risk level. If the standard deviation is below the lower limit threshold, it indicates extremely stable temperature control, allowing for a downward adjustment of one level of risk. If the standard deviation is between the two thresholds, the initial risk level remains unchanged. For furnace batches already marked as high-risk candidates, the risk level remains high after the upward adjustment; for low-risk candidates, the risk level remains low after the downward adjustment, without generating a negative rating.

[0021] Step S13: Based on the evaluation results of the deviation and fluctuation amplitude values, generate the final risk classification label, where the high-risk classification label is associated with the more stringent online annealing dynamic compensation initiation conditions in subsequent processes.

[0022] The basic risk level is determined based on the normalization degree of the key element deviations: if the relative deviations of all key elements are less than or equal to 0.5 times the key element deviation threshold, the basic risk is set to low; if the relative deviation of any key element is between 0.5 and 1.0 times the threshold, the basic risk is set to medium; if the relative deviation of any key element is greater than 1.0 times the threshold, the basic risk is set to high. Then, the adjustment amount generated in step S12 is added to the basic risk level, and the final result is constrained to fall within three levels: 0 (low), 1 (medium), and 2 (high). The cloud platform queries the pre-configured risk level mapping table based on the final risk level index and outputs the corresponding classification label. The high-risk classification label is written into the raw material process fingerprint database and sent to the production quality inspection system, triggering stricter dynamic compensation strategies in subsequent online annealing processes, such as shortening the temperature sampling cycle, increasing the PID adjustment coefficient of heating power, and activating the backup heating zone.

[0023] Step S2: The production quality inspection system performs differentiated process path allocation based on the graded labels, allocating high-risk raw materials to precision process paths that include online annealing dynamic compensation processes, and allocating medium- and low-risk raw materials to conventional process paths; coaxiality deviation data is collected at each process node of the drill pipe thread section and uploaded to the cloud platform; Specifically, after the rough turning process, the first radial runout between the thread section reference surface and the tool path is collected by the first laser displacement sensor as the initial coaxiality deviation. After the semi-finish turning process, the second radial runout between the thread pitch diameter and the spindle rotation center line is collected by the second laser displacement sensor to generate an intermediate coaxiality deviation. After the finish turning threading process and the online annealing dynamic compensation process, the tooth profile half-angle deviation and pitch cumulative deviation of the finished thread are collected by the third laser displacement sensor. The first radial runout, the second radial runout, the tooth profile half-angle deviation, and the pitch cumulative deviation are combined according to the process sequence to form a multi-dimensional coaxiality deviation data sequence, which specifically includes the following sub-steps: Step S21: After the rough turning operation of the drill pipe thread section is completed, the first radial runout between the thread section reference surface and the tool path is collected by the first laser displacement sensor as the initial coaxiality deviation. Immediately after the roughing operation, the first laser displacement sensor is activated. This sensor is fixed to the machine tool post and maintains precise parallelism with the spindle axis. During measurement, the sensor probe scans at equal angular intervals along the circumference of the thread section reference surface, acquiring a sequence of radial offsets relative to the ideal rotation center. Simultaneously, to eliminate the influence of spindle rotation error on the measurement results, the system synchronously acquires data from an eddy current displacement sensor mounted at the spindle end, compensating for the original measurements in real time. The maximum peak-to-valley value of the compensated radial offset sequence is defined as the first radial runout, serving as the initial coaxiality deviation characterizing the machining quality of this operation.

[0024] Step S22: After the semi-finishing process is completed, the second radial runout between the thread pitch diameter and the spindle rotation center line is collected by the second laser displacement sensor to generate the intermediate coaxiality deviation. After the semi-finishing process, the second laser displacement sensor begins operation. Its measurement object is the machined thread pitch diameter cylindrical surface. The measurement strategy employs a two-point or three-point error separation technique to eliminate the interference of workpiece shape errors on runout assessment. Specifically, the sensor samples at multiple points along the helix along the axial direction, acquiring a radial cross-sectional profile perpendicular to the spindle axis at each sampling point. The actual center position of each cross-section is fitted using the least squares method, and the coordinates of all cross-section centers are connected to form the actual axis of the thread pitch diameter. The maximum offset and direction of this actual axis relative to the spindle rotation centerline are obtained. After comprehensive evaluation, a composite index including eccentricity and eccentricity angle is generated, namely the second radial runout, which serves as the intermediate coaxiality deviation.

[0025] Step S23: After the thread cutting process and the subsequent online annealing dynamic compensation process are completed, the tooth profile half angle deviation and pitch cumulative deviation of the finished thread are collected by the third laser displacement sensor. The first radial runout, the second radial runout, the tooth profile half angle deviation and the pitch cumulative deviation are combined according to the process sequence to form a multi-dimensional coaxiality deviation data sequence. To comprehensively evaluate thread quality, the system performs two independent measurement tasks: First, by performing linear regression analysis on the left and right sidewalls of a single complete thread profile, the differences between the actual left and right half-angles and the theoretical half-angles are obtained. The sum of their absolute values ​​is taken as the half-angle deviation of the thread profile, and the average of multiple thread profiles is taken as the final thread profile half-angle deviation. Second, a distance is scanned along the thread axis, and the axial position coordinates of multiple thread crests are recorded. By comparing the point-by-point difference with the theoretical cumulative length of the standard pitch, the cumulative pitch deviation is calculated, reflecting the phase error over a long distance, and is expressed by the following formula:

[0026] in, Indicates the cumulative pitch deviation; Indicates the number of the tooth crest; This represents the summation index variable, from 1 to... , indicating the previous The local pitch error is accumulated one by one at each tooth crest; Indicates the first Measured axial coordinates of each tooth crest; This represents the theoretical standard pitch value; It represents the amplitude of periodic error, used to simulate the periodic pitch fluctuations of machine tool lead screws or transmission systems; The period length represents the periodic error.

[0027] Finally, the system encapsulates and serializes the collected first radial runout, second radial runout, tooth profile half-angle deviation, and cumulative pitch deviation according to the process sequence of "rough turning → semi-finish turning → finish turning + annealing" to form a multi-dimensional coaxiality deviation data sequence.

[0028] Step S3: The cloud platform constructs a multi-process coaxiality drift map based on the received coaxiality deviation data and extracts the feature value of the cumulative coaxiality drift. Based on the difference between the feature value of the cumulative coaxiality drift and the target value, a process compensation instruction is generated and sent to the production quality inspection system. The temperature control parameters of the online annealing dynamic compensation process are adjusted according to the process compensation instruction. Specifically, using the process sequence number as the horizontal axis and the root mean square value of the radial runout in the coaxiality deviation data collected at each process node as the vertical axis, a curve showing the change in coaxiality deviation as the process progresses is plotted to obtain a multi-process coaxiality drift map. Piecewise linear regression is performed on this curve to calculate the intra-segment drift rate for the roughing, semi-finishing, and finishing stages, respectively. The cumulative deviation value of the final process is used as the characteristic value of the cumulative coaxiality drift. The characteristic value of the cumulative coaxiality drift is compared with a preset upper limit value for process capability, and the difference between the two is calculated. If the difference is positive and exceeds the allowable residual deviation threshold, a process compensation instruction is generated. This instruction includes adjustments to the temperature control parameters of the online annealing dynamic compensation process, specifically including the following sub-steps: Step S31: Plot the coaxiality deviation as the process progresses, with the process number as the horizontal axis and the root mean square value of the radial runout in the coaxiality deviation data collected at each process node as the vertical axis, to form the multi-process coaxiality drift map. After receiving the coaxiality deviation data for each process reported by the production quality inspection system, the cloud platform performs time-series alignment and outlier removal preprocessing on the original deviation sequence. For each process node, the recorded radial runout time-series waveform is extracted, and the root mean square value of this waveform over one complete rotation cycle is calculated as the radial runout characteristic of that process, expressed by the following formula:

[0029] in, This represents the radial runout characteristic quantity; This represents the total number of sampling points within a complete rotation cycle. Indicates the sequence number of the sampling point (from 1 to ... ); Indicates the first Each process node, at any time The instantaneous value of the original radial runout obtained from the measurement at the location; Indicates the first Each sampling time The measured instantaneous value of the original radial runout; Indicates the first Hanning window weighted coefficients for each sampling point.

[0030] Then, using the order of processing steps as the x-axis and the calculated root mean square value of radial runout for each step as the y-axis, points are plotted in a two-dimensional coordinate system, and adjacent points are connected using cubic spline interpolation to generate a continuous curve reflecting the evolution of coaxiality deviation as it accumulates with each step. This curve is the multi-step coaxiality drift map.

[0031] Step S32: Perform piecewise linear regression on the change curve to calculate the intra-segment drift rate of the roughing section, semi-finishing section and finishing section respectively, and use the cumulative deviation value of the final process as the characteristic value of the coaxiality cumulative drift. After obtaining the coaxiality drift map, the cloud platform divides the entire curve into three continuous intervals based on the process type label: roughing, semi-finishing, and finishing. For the data points within each interval, a univariate linear regression is performed using the least squares method. The slope of the fitted line is the coaxiality drift rate within that processing interval, used to characterize the sensitivity of that process segment to workpiece axis offset. Simultaneously, the root mean square value of radial runout corresponding to the last process is extracted, and this value is weighted and corrected with the drift rate of the previous process to serve as the cumulative coaxiality drift characteristic value reflecting the cumulative effect of axis offset throughout the entire machining process.

[0032] Step S33: Compare the cumulative coaxiality drift characteristic value with the preset process capability upper limit value, calculate the difference between the two, and if the difference is positive and exceeds the allowable residual deviation threshold, generate a process compensation instruction. The process compensation instruction includes the adjustment amount of the temperature control parameters of the online annealing dynamic compensation process. The cloud platform reads the upper limit of process capability and the allowable residual deviation threshold corresponding to the workpiece material and process standard from the local configuration library. The cumulative coaxiality drift characteristic value obtained in step S32 is subtracted from the upper limit of process capability to obtain the drift excess. If this excess is positive and its absolute value is greater than the preset residual deviation threshold, it is determined that the current process route has generated an unacceptable cumulative eccentricity risk, requiring online compensation to be initiated. Based on the magnitude and trend of the drift excess, and combined with the pre-stored process compensation rule library, the platform calculates the heating power adjustment and holding time adjustment required for the online annealing dynamic compensation process, encapsulates these two parameters into a process compensation instruction, and sends it to the production quality inspection system through the industrial IoT interface to drive the annealing equipment to perform real-time dynamic compensation during subsequent workpiece processing.

[0033] Step S4: The production quality inspection system completes production according to the adjusted temperature control parameters, performs assembly pre-testing on the finished drill pipe and matching drill bit, collects the cone surface contact pressure distribution data through the ring array thin film pressure sensor, performs frequency domain analysis on the pressure distribution data, extracts the ratio of the fundamental component amplitude to the DC component amplitude, uses this ratio as the cone surface contact uniformity index, and compares it with the qualified threshold to determine whether the finished product is qualified. Specifically, the finished drill pipe to be tested and the standard matching drill bit are installed on the assembly pre-testing platform. The conical surface is driven to fit with a preset rated torque, and the axial clamping displacement is recorded by a servo motor until the target torque value is reached. Multiple discrete pressure values ​​are collected in the conical surface mating area using a ring array thin-film pressure sensor, each value corresponding to the contact pressure at a circumferential angle position. Frequency domain analysis is performed on the pressure values ​​to extract the ratio of the fundamental component amplitude to the DC component amplitude. After normalization, this ratio is used as the conical surface contact uniformity index. The closer the index is to 1, the more uniform the pressure distribution. The specific steps include the following: Step S41: Install the finished drill pipe to be tested and the standard matching drill bit on the assembly pre-testing platform, drive the drill pipe and drill bit to perform conical surface mating with the preset rated torque, and record the axial clamping displacement through the servo motor during the mating process until the target torque value is reached; The finished drill pipe to be inspected is vertically hoisted onto the fixed fixture of the assembly pre-testing platform, and the standard matching drill bit is installed on the hydraulically driven moving platform. The servo motor is started, driving the drill pipe to feed towards the drill bit's tapered hole at a constant angular velocity, gradually bringing the drill pipe's tapered surface into contact with the drill bit's tapered hole. When the contact force of the tapered surface reaches the preset initial meshing force threshold, the torque control mode is switched, and the driving torque is gradually increased to the rated value. At the same time, the axial pressing displacement of the servo motor is continuously recorded using a high-precision grating ruler until the displacement stabilizes within the allowable fluctuation range under the rated torque, completing the tapered surface fit.

[0034] Step S42: In the conical mating area of ​​the assembly pre-testing platform, multiple discrete pressure values ​​are collected between the drill rod conical surface and the drill bit conical hole in the axial fitting state by thin film pressure sensors arranged in a ring array. Each discrete pressure value corresponds to the contact pressure at a circumferential angle position. In the conical mating area of ​​the assembly pre-testing platform, a ring array thin-film pressure sensor with a flexible substrate is pre-embedded. This sensor contains N independent sensing units, distributed at equal angular intervals along the circumference. After the drill rod and drill bit complete axial clamping under rated torque, the pressure electrical signal of each sensing unit is synchronously read through a multi-channel data acquisition card. After signal conditioning and analog-to-digital conversion, discrete pressure values ​​at the corresponding circumferential angular positions are obtained, forming the original data sequence of the conical contact pressure distribution.

[0035] Step S43: Perform frequency domain analysis on the multiple discrete pressure values, extract the ratio of the fundamental component amplitude to the DC component amplitude, and use this ratio as the cone surface contact uniformity index. The cone surface contact uniformity index is used to characterize the uniformity of pressure distribution. The collected discrete pressure value sequence is treated as a periodic function on a circular angle. Frequency domain analysis is performed to extract the amplitude of the fundamental component reflecting the degree of pressure distribution non-uniformity and the amplitude of the DC component reflecting the average pressure level. The ratio of the fundamental component amplitude to the DC component amplitude is calculated, and the minimum-maximum normalization method is used to map this ratio to the [0,1] interval, which is defined as the conical surface contact uniformity index. The conical surface contact uniformity index is expressed by the following formula:

[0036] in, This represents the contact uniformity index of the conical surface; after normalization, its value ranges between [0,1]. This indicates the number of pressure sampling points arranged at equal angular intervals along the circumference; This represents the sequence index, with values ​​ranging from 1 to... ; Indicates the first Discrete pressure values ​​measured at the location of each sensing unit; and They represent angles respectively. The cosine and sine functions; It represents a very small positive number to prevent numerical calculation errors caused by a denominator of zero.

[0037] The closer the conical surface contact uniformity index is to 1, the more uniform the pressure distribution along the circumference and the higher the quality of the conical surface fit; conversely, it indicates the presence of off-center loading or local contact defects. The index is compared with a preset acceptable threshold. If the index is not lower than the threshold, the finished product is considered acceptable; otherwise, it is considered unacceptable.

[0038] Step S5: The cloud platform traces and couples the furnace batch number, coaxiality cumulative drift, and conical surface contact uniformity index associated with the non-conforming finished products to calculate their contribution rate to the conical surface contact unevenness defect; updates the corresponding furnace batch number process allocation rules based on the contribution rate and issues them for subsequent process allocation of similar furnace batch numbers; continuously accumulates data to iteratively optimize model parameters and form a closed-loop dynamic control. Specifically, the cumulative coaxiality drift characteristic value and conical surface contact uniformity index of all drill rods from the same batch as the defective finished products in the cloud platform's historical database are retrieved during the processing. Partial least squares regression analysis is performed, and the square of the first principal component load coefficient is extracted as the global contribution rate of the raw material of that batch to the conical surface contact non-uniformity defect. Further, the coaxiality deviation data of each process node is used as the independent variable and the conical surface contact uniformity index is used as the dependent variable for stepwise regression. Process nodes with statistical significance are screened out, and the standardized regression coefficients of each process are calculated and normalized to obtain the process transfer contribution rate. The specific steps include the following: Step S51: Retrieve the cumulative coaxiality drift characteristic values ​​of all drill rods with the same furnace batch number as the current unqualified finished products from the historical database of the cloud platform, forming a coaxiality characteristic value set; at the same time, retrieve the cone surface contact uniformity index corresponding to these drill rods, forming a uniformity index set. After receiving the judgment result of the non-conforming finished product, the cloud platform first extracts the furnace batch number associated with the finished product. Based on the furnace batch number, the platform retrieves the processing records of all drill rods under the same furnace batch number from the historical production database. Each record contains the cumulative coaxiality drift feature value collected at each process node of the drill rod, as well as the finally detected cone surface contact uniformity index. The platform arranges the retrieved cumulative coaxiality drift feature values ​​according to the drill rod number and process sequence to form a two-dimensional dataset; at the same time, it arranges the corresponding cone surface contact uniformity indices according to the same drill rod number to form a one-dimensional index set. For records with missing data or abnormal drift exceeding three times the standard deviation threshold, the platform automatically removes them to prevent interference with subsequent modeling.

[0039] Step S52: Perform partial least squares regression analysis on the set of coaxiality cumulative drift characteristic values ​​and the set of conical surface contact uniformity index to extract the loading coefficient of the first principal component, and take the square value of the loading coefficient as the global contribution rate of the raw material of this batch to the conical surface contact non-uniformity defect. The cloud platform uses the set of coaxiality eigenvalues ​​as the independent variable matrix. The set of evenness indices is used as the dependent variable vector. .right and Standardization was performed to eliminate the influence of dimensions. Then, partial least squares regression iterations were executed to extract the first principal component. Within the loading vector of the first principal component, the indices of independent variables strongly correlated with the raw material attributes of the furnace batch were located, and the squares of their loading coefficients were taken as the global contribution rate.

[0040] Step S53: Further, take the coaxiality deviation data of each process node as the independent variable and the conical surface contact uniformity index as the dependent variable, use the stepwise regression algorithm to screen out the process nodes with statistical significance, calculate the standardized regression coefficient of each screened process, and normalize each standardized regression coefficient as the process transfer contribution rate of each process. Step S531: Compare the global contribution rate with the preset raw material sensitivity threshold. If the global contribution rate exceeds the raw material sensitivity threshold, the initial risk classification label of the raw material of this batch number will be forcibly increased by one level in the next allocation. Step S532: Based on the process transfer contribution rate of each process, identify the two key processes with the highest contribution rates, and generate supplementary detection instructions for these two key processes. The supplementary detection instructions require that when the raw materials of the same batch are put back into production, all process parameters of these two key processes be recorded and the sampling frequency be increased to twice the normal sampling frequency. The platform sorts the normalized process transfer contribution rates from highest to lowest and selects the two processes with the highest contribution rates as critical processes. For each critical process, the platform automatically generates supplementary inspection instructions, which explicitly require that when the raw materials of that batch are put into production again, all processing parameters of that process must be fully recorded, and sampling inspection is not allowed; at the same time, the sampling frequency of the coaxiality deviation monitoring points of that process is increased to twice the normal sampling frequency in order to capture more precise drift trends.

[0041] Step S533: Integrate the upgraded risk classification label and supplementary detection instructions to form an updated process path allocation rule, and store it back in the raw material process fingerprint database in the record item associated with the furnace batch number. At the same time, send the updated rule to the rule engine of the production quality inspection system in the form of structured data.

[0042] The platform structurally encapsulates the upgraded risk classification label generated by S531 and the supplementary detection instructions generated by S532 to form a complete process path allocation rule. This rule is stored back in the raw material process fingerprint database as key-value pairs, overwriting the original allocation rule entries for that batch. Simultaneously, the platform pushes this rule as JSON structured data to the rule engine of the production quality inspection system via a message queue. When the rule engine subsequently receives a production request for raw materials from the same batch, it automatically loads this rule and drives the sampling and recording strategy for the corresponding process.

[0043] Example 2

[0044] like Figure 2 As shown, Embodiment 2 of this application provides a cloud platform-based production quality inspection system, including: Cloud platform data processing module 21: The cloud platform acquires the chemical composition spectral data and hot rolling forming temperature sequence of the drill pipe raw material furnace batch number; retrieves the preset reference content, calculates the deviation between the actual content and the reference content of each element, and statistically analyzes the fluctuation range of the temperature sequence; associates and stores the furnace batch number, spectral data, temperature sequence, deviation, and fluctuation range as a raw material process fingerprint database; generates risk classification labels based on the deviation and fluctuation range, and sends them to the production quality inspection system; includes the following sub-modules: Risk candidate preliminary marking submodule 211: The deviation between the actual content and the reference content of each element is compared with the preset single element deviation tolerance range. If the deviation of any key element exceeds the key element deviation threshold, it is initially marked as a high-risk candidate. Temperature fluctuation risk correction submodule 212: Calculates the fluctuation amplitude value of the temperature sequence. When the fluctuation amplitude value exceeds the preset upper limit threshold of temperature stability, it is adjusted upward based on the initially marked risk level. Conversely, when the fluctuation amplitude value is lower than the preset lower limit threshold of temperature stability, it is adjusted downward. Risk Level Comprehensive Mapping Submodule 213: Based on the evaluation results of the aforementioned deviation and fluctuation amplitude values, generate the final risk level label, wherein the high-risk level label is associated with the more stringent online annealing dynamic compensation activation conditions in subsequent processes; Production Quality Inspection Execution Module 22: The production quality inspection system performs differentiated process path allocation based on graded labels, assigning high-risk raw materials to precision process paths that include online annealing and dynamic compensation processes, and assigning medium- and low-risk raw materials to conventional process paths; it collects coaxiality deviation data at each process node of the drill pipe thread section and uploads it to the cloud platform; it includes the following sub-modules: Rough turning process coaxiality quality inspection submodule 221: After the rough turning process of the drill pipe thread section is completed, the first radial runout between the thread section reference surface and the tool path is collected by the first laser displacement sensor as the initial coaxiality deviation; Semi-finishing process coaxiality quality inspection submodule 222: After the semi-finishing process is completed, the second radial runout between the thread pitch diameter and the spindle rotation center line is collected by the second laser displacement sensor to generate the intermediate coaxiality deviation; Finished thread key parameter quality inspection submodule 223: After the thread finishing process and the subsequent online annealing dynamic compensation process are completed, the tooth profile half angle deviation and pitch cumulative deviation of the finished thread are collected by the third laser displacement sensor. The first radial runout, the second radial runout, the tooth profile half angle deviation and the pitch cumulative deviation are combined according to the process sequence to form a multi-dimensional coaxiality deviation data sequence. Cloud platform collaborative control module 23: The cloud platform constructs a multi-process coaxiality drift map based on the received coaxiality deviation data, extracts the feature value of the cumulative coaxiality drift, generates a process compensation instruction based on the difference between the feature value of the cumulative coaxiality drift and the target value, and sends it to the production quality inspection system. The system adjusts the temperature control parameters of the online annealing dynamic compensation process according to the process compensation instruction. This includes the following sub-modules: Coaxiality drift map construction submodule 231: With the process sequence number as the horizontal axis and the root mean square value of the radial runout in the coaxiality deviation data collected at each process node as the vertical axis, plot the coaxiality deviation change curve as the process progresses, and form the multi-process coaxiality drift map. Cumulative drift feature extraction submodule 232: Perform piecewise linear regression on the change curve, calculate the intra-segment drift rate of the roughing section, semi-finishing section and finishing section respectively, and use the cumulative deviation value of the final process as the coaxiality cumulative drift feature value; Process compensation instruction generation submodule 233: compares the cumulative drift characteristic value of coaxiality with the preset upper limit value of process capability, calculates the difference between the two, and if the difference is positive and exceeds the allowable residual deviation threshold, generates a process compensation instruction. The process compensation instruction includes the adjustment amount of the temperature control parameters of the online annealing dynamic compensation process. Assembly Pre-test Module 24: The production quality inspection system completes production according to the adjusted temperature control parameters, performs assembly pre-test on the finished drill pipe and matching drill bit, collects the contact pressure distribution data of the conical surface through a ring array thin-film pressure sensor, performs frequency domain analysis on the pressure distribution data, extracts the ratio of the fundamental component amplitude to the DC component amplitude, uses this ratio as the conical surface contact uniformity index, and compares it with the qualified threshold to determine whether the finished product is qualified; includes the following sub-modules: Assembly execution and displacement monitoring submodule 241: Install the finished drill rod to be tested and the standard matching drill bit on the assembly pre-testing platform, drive the drill rod and drill bit to perform conical surface engagement with a preset rated torque, and record the axial clamping displacement through the servo motor during the engagement process until the target torque value is reached; Conical surface contact pressure acquisition submodule 242: In the conical surface mating area of ​​the assembly pre-testing platform, multiple discrete pressure values ​​between the drill rod conical surface and the drill bit conical hole in the axial fitting state are acquired by thin film pressure sensors arranged in a ring array. Each discrete pressure value corresponds to the contact pressure at a circumferential angle position. Pressure distribution analysis and index calculation submodule 243: Perform frequency domain analysis on the multiple discrete pressure values, extract the ratio of the fundamental component amplitude to the DC component amplitude, and use this ratio as the cone surface contact uniformity index, which is used to characterize the uniformity of pressure distribution; Intelligent Decision and Feedback Module 25: The cloud platform traces and couples the furnace batch number, cumulative coaxiality drift, and conical surface contact uniformity index associated with non-conforming finished products to calculate their contribution rate to the conical surface contact unevenness defect; updates the corresponding furnace batch number process allocation rules based on the contribution rate and issues them for subsequent process allocation of similar furnace batch numbers; continuously accumulates data to iteratively optimize model parameters, forming a closed-loop dynamic control; including the following sub-modules: Data retrieval and set construction submodule 251: Retrieve the coaxiality cumulative drift characteristic values ​​of all drill rods with the same furnace batch number as the current unqualified finished products from the historical database of the cloud platform, forming a coaxiality characteristic value set; at the same time, retrieve the cone surface contact uniformity index corresponding to these drill rods, forming a uniformity index set. Global contribution rate calculation submodule 252: Perform partial least squares regression analysis on the set of coaxiality cumulative drift characteristic values ​​and the set of conical surface contact uniformity index, extract the loading coefficient of the first principal component, and use the square value of the loading coefficient as the global contribution rate of the raw material of this batch to the conical surface contact non-uniformity defect. Process transfer contribution rate analysis submodule 253: Further, the coaxiality deviation data of each process node is used as the independent variable, the conical surface contact uniformity index is used as the dependent variable, and the stepwise regression algorithm is used to screen out the process nodes with statistical significance. The standardized regression coefficient of each screened process is calculated, and the standardized regression coefficient is normalized and used as the process transfer contribution rate of each process.

[0045] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A cloud platform-based production quality inspection method, characterized in that, include: S1. Obtain the chemical composition spectral data and hot rolling forming temperature sequence of the drill pipe raw material furnace batch number from the cloud platform. Retrieve preset reference content, calculate the deviation between the actual content and the reference content of each element, and statistically analyze the fluctuation range of the temperature sequence; associate and store the furnace batch number, spectral data, temperature sequence, deviation, and fluctuation range as a raw material process fingerprint database; generate risk classification labels based on the deviation and fluctuation range values, and send them to the production quality inspection system; S2. The production quality inspection system performs differentiated process path allocation based on the graded labels, allocating high-risk raw materials to precision process paths that include online annealing and dynamic compensation processes, and allocating medium- and low-risk raw materials to conventional process paths; it also collects coaxiality deviation data at each process node of the drill pipe thread section and uploads it to the cloud platform. S3. The cloud platform constructs a multi-process coaxiality drift map based on the received coaxiality deviation data and extracts the feature value of the cumulative coaxiality drift. Based on the difference between the feature value of the cumulative coaxiality drift and the target value, a process compensation instruction is generated and sent to the production quality inspection system. The temperature control parameters of the online annealing dynamic compensation process are adjusted according to the process compensation instruction. S4. The production quality inspection system completes production according to the adjusted temperature control parameters, performs assembly pre-testing on the finished drill pipe and matching drill bit, collects the cone surface contact pressure distribution data through the ring array thin film pressure sensor, performs frequency domain analysis on the pressure distribution data, extracts the ratio of the fundamental component amplitude to the DC component amplitude, uses this ratio as the cone surface contact uniformity index, and compares it with the qualified threshold to determine whether the finished product is qualified. S5 and the cloud platform trace the source of non-conforming finished products by coupling the furnace batch number, coaxiality cumulative drift, and conical surface contact uniformity index, and calculate their contribution rate to the conical surface contact unevenness defect; update the corresponding furnace batch number process allocation rules according to the contribution rate and issue them for subsequent process allocation of similar furnace batch numbers; continuously accumulate data to iteratively optimize model parameters and form a closed-loop dynamic control.

2. The cloud platform-based production quality inspection method of claim 1, wherein, The cloud platform acquires the chemical composition spectral data and hot rolling temperature sequence of the drill pipe raw material furnace batch number; retrieves the preset reference content, calculates the deviation between the actual content and the reference content of each element, and statistically analyzes the fluctuation range of the temperature sequence; associates and stores the furnace batch number, spectral data, temperature sequence, deviation, and fluctuation range as a raw material process fingerprint database; generates risk classification labels based on the deviation and fluctuation range values, and distributes them to the production quality inspection system, including the following sub-steps: The deviation between the actual content and the reference content of each element is compared with the preset tolerance range of the deviation of each element. If the deviation of any key element exceeds the key element deviation threshold, it is initially marked as a high-risk candidate. The fluctuation amplitude of the temperature sequence is statistically analyzed. When the fluctuation amplitude exceeds the preset upper limit threshold for temperature stability, the risk level is adjusted upward based on the initial labeling. Conversely, when the fluctuation amplitude is lower than the preset lower limit threshold for temperature stability, the risk level is adjusted downward. Based on the combined assessment results of the deviation and fluctuation amplitude values, a final risk classification label is generated, where the high-risk classification label is associated with the more stringent online annealing dynamic compensation activation conditions in subsequent processes. 3.The production quality inspection method based on the cloud platform of claim 1, wherein, The production quality inspection system performs differentiated process path allocation based on graded labels, assigning high-risk raw materials to precision process paths that include online annealing and dynamic compensation processes, and assigning medium- and low-risk raw materials to conventional process paths; coaxiality deviation data is collected at each process node of the drill pipe thread section and uploaded to the cloud platform, including the following sub-steps: After the rough turning process of the drill pipe thread section is completed, the first radial runout between the thread section reference surface and the tool path is collected by the first laser displacement sensor as the initial coaxiality deviation. After the semi-finishing process is completed, the second radial runout between the thread pitch diameter and the spindle rotation center line is collected by the second laser displacement sensor to generate the intermediate coaxiality deviation. After the threading process and the subsequent online annealing dynamic compensation process are completed, the tooth profile half-angle deviation and pitch cumulative deviation of the finished thread are collected by the third laser displacement sensor. The first radial runout, the second radial runout, the tooth profile half-angle deviation and the pitch cumulative deviation are combined according to the process sequence to form a multi-dimensional coaxiality deviation data sequence.

4. The cloud platform-based production quality inspection method of claim 1, wherein, The cloud platform constructs a multi-process coaxiality drift map based on the received coaxiality deviation data and extracts the feature value of the cumulative coaxiality drift. Based on the difference between the feature value of the cumulative coaxiality drift and the target value, a process compensation instruction is generated and sent to the production quality inspection system. The temperature control parameters of the online annealing dynamic compensation process are adjusted according to the process compensation instruction, including the following sub-steps: With the process sequence number as the horizontal axis and the root mean square value of the radial runout in the coaxiality deviation data collected at each process node as the vertical axis, a curve showing the change of coaxiality deviation as the process progresses is plotted to form the multi-process coaxiality drift map. Piecewise linear regression was performed on the change curve to calculate the intra-segment drift rate of the roughing section, semi-finishing section and finishing section respectively, and the cumulative deviation value of the final process was used as the characteristic value of the coaxiality cumulative drift. The cumulative coaxiality drift characteristic value is compared with the preset process capability upper limit value, and the difference between the two is calculated. If the difference is positive and exceeds the allowable residual deviation threshold, a process compensation instruction is generated. The process compensation instruction includes the adjustment amount of the temperature control parameters of the online annealing dynamic compensation process.

5. The cloud platform-based production quality inspection method of claim 1, wherein, The production quality inspection system completes production according to the adjusted temperature control parameters, performs assembly pre-testing on the finished drill pipe and matching drill bit, collects the contact pressure distribution data of the conical surface using a ring array thin-film pressure sensor, performs frequency domain analysis on the pressure distribution data, extracts the ratio of the fundamental component amplitude to the DC component amplitude, uses this ratio as the conical surface contact uniformity index, and compares it with the qualified threshold to determine whether the finished product is qualified. This includes the following sub-steps: The finished drill pipe to be tested and the standard matching drill bit are installed on the assembly pre-testing platform. The drill pipe and the drill bit are driven to perform conical surface mating with a preset rated torque. During the mating process, the axial clamping displacement is recorded by the servo motor until the target torque value is reached. In the conical mating area of ​​the assembly pre-testing platform, multiple discrete pressure values ​​are collected between the drill rod conical surface and the drill bit conical hole in the axial fitting state by thin film pressure sensors arranged in a ring array. Each discrete pressure value corresponds to the contact pressure at a circumferential angle position. Frequency domain analysis is performed on the multiple discrete pressure values ​​to extract the ratio of the fundamental component amplitude to the DC component amplitude. This ratio is used as the cone surface contact uniformity index, which is used to characterize the uniformity of the pressure distribution.

6. The cloud platform-based production quality inspection method of claim 1, wherein, The cloud platform traces and couples the furnace batch number, cumulative coaxiality drift, and conical surface contact uniformity index associated with non-conforming finished products to calculate their contribution rate to the conical surface contact unevenness defect; it updates and distributes the corresponding furnace batch number process allocation rules based on the contribution rate for subsequent process allocation of similar furnace batch numbers; it continuously accumulates data to iteratively optimize model parameters, forming a closed-loop dynamic control, including the following sub-steps: The system retrieves the cumulative coaxiality drift characteristic values ​​of all drill rods with the same furnace batch number as the current unqualified finished products from the historical database of the cloud platform, forming a set of coaxiality characteristic values. At the same time, it retrieves the cone surface contact uniformity index corresponding to these drill rods, forming a set of uniformity indices. Partial least squares regression analysis was performed on the set of coaxiality cumulative drift characteristic values ​​and the set of conical surface contact uniformity index to extract the loading coefficient of the first principal component. The square value of the loading coefficient was used as the global contribution rate of the raw material of this batch to the conical surface contact non-uniformity defect. Furthermore, the coaxiality deviation data of each process node is used as the independent variable, and the conical surface contact uniformity index is used as the dependent variable. The stepwise regression algorithm is used to screen out the process nodes with statistical significance, calculate the standardized regression coefficient of each screened process, and normalize each standardized regression coefficient as the process transfer contribution rate of each process.

7. The cloud platform-based production quality inspection method of claim 6, wherein, Further, using the coaxiality deviation data of each process node as the independent variable and the conical surface contact uniformity index as the dependent variable, a stepwise regression algorithm is used to screen out process nodes with statistical significance. The standardized regression coefficient of each screened process is calculated, and the normalized standardized regression coefficients are used as the process transfer contribution rate of each process. This includes the following sub-steps: The global contribution rate is compared with a preset raw material sensitivity threshold. If the global contribution rate exceeds the raw material sensitivity threshold, the initial risk classification label of the raw material of that batch number will be forcibly increased by one level in the next allocation. Based on the process transfer contribution rate of each process, the two key processes with the highest contribution rates are identified, and supplementary detection instructions are generated for these two key processes. The supplementary detection instructions require that when the raw materials of the same batch are put back into production, all process parameters of the two key processes be recorded and the sampling frequency be increased to twice the normal sampling frequency. The upgraded risk classification labels and supplementary testing instructions are integrated to form updated process path allocation rules, which are then stored back in the raw material process fingerprint database as records associated with the batch number of the furnace. At the same time, the updated rules are sent to the rule engine of the production quality inspection system in the form of structured data.

8. A cloud platform-based production quality inspection system, characterized in that, include: Cloud platform data processing module: The cloud platform acquires the chemical composition spectral data and hot rolling forming temperature sequence of the drill pipe raw material furnace batch number; retrieves the preset reference content, calculates the deviation between the actual content and the reference content of each element, and counts the fluctuation range value of the temperature sequence; associates and stores the furnace batch number, spectral data, temperature sequence, deviation and fluctuation range value as a raw material process fingerprint database; generates risk classification labels based on the deviation and fluctuation range value, and sends them to the production quality inspection system; Production quality inspection execution module: The production quality inspection system performs differentiated process path allocation based on the grade label, allocating high-risk raw materials to precision process paths that include online annealing and dynamic compensation processes, and allocating medium- and low-risk raw materials to conventional process paths; it also collects coaxiality deviation data at each process node of the drill pipe thread section and uploads it to the cloud platform; The cloud platform collaborative control module: The cloud platform constructs a multi-process coaxiality drift map based on the received coaxiality deviation data and extracts the feature value of the cumulative coaxiality drift. Based on the difference between the feature value of the cumulative coaxiality drift and the target value, it generates a process compensation instruction and sends it to the production quality inspection system. Based on the process compensation instruction, it adjusts the temperature control parameters of the online annealing dynamic compensation process. Assembly pre-test module: The production quality inspection system completes production according to the adjusted temperature control parameters, performs assembly pre-test on the finished drill pipe and matching drill bit, collects the cone surface contact pressure distribution data through the ring array thin film pressure sensor, performs frequency domain analysis on the pressure distribution data, extracts the ratio of the fundamental component amplitude to the DC component amplitude, uses this ratio as the cone surface contact uniformity index, and compares it with the qualified threshold to determine whether the finished product is qualified. Intelligent decision-making and feedback module: The cloud platform traces and couples the furnace batch number, coaxiality cumulative drift amount and conical surface contact uniformity index associated with non-conforming finished products, calculates their contribution rate to the conical surface contact unevenness defect; updates the corresponding furnace batch number process allocation rules according to the contribution rate and issues them for subsequent process allocation of similar furnace batch numbers. Continuously accumulate data to iteratively optimize model parameters and form a closed-loop dynamic control.

Images