Search

CN-122022602-A - Valve fitting online quality regulation and control method and system

CN122022602ACN 122022602 ACN122022602 ACN 122022602ACN-122022602-A

Abstract

The invention relates to the technical field of intelligent manufacturing and industrial automatic detection, in particular to a valve fitting online quality control method and system. The invention monitors the quality data flow of the production line in real time, identifies the deviation trend, generates the defect consciousness pulse, and deduces the defect burst time by combining with the time sequence prediction model, and reversely maps the process path. The system adopts an intelligent compensation mechanism of fuzzy logic and historical case weighting to dynamically fine-tune processing parameters, and introduces a cooperative compensation and fusing rollback mechanism to ensure safe and effective regulation and control. And finally, storing the successful regulation cases to a knowledge base. The advanced prejudgment and real-time accurate regulation and control of quality defects in the valve fitting production process are realized, the rejection rate is effectively reduced, and the intelligent level of the production line and the quality stability of products are improved.

Inventors

  • HE XINYU
  • JIANG HE
  • ZHANG HAN
  • WANG XIAODONG
  • LAI JUNLING

Assignees

  • 温州阿泰科机械科技有限公司

Dates

Publication Date
20260512
Application Date
20260414

Claims (10)

  1. 1. An on-line quality control method for valve fittings, comprising: When any detection station on the valve fitting automatic production line recognizes that the quality data flow has a deviation trend, defect consciousness pulse containing fluctuation characteristics and equipment states is automatically generated; The production management system deduces the time point of the burst of the defect-aware pulse into the real defect in the future work piece period based on the time sequence prediction model after receiving the defect-aware pulse, reversely maps the working procedure path to be passed by the defect-aware pulse by taking the time point of the real defect as an end point; Monitoring the mass data flow of the workpiece after compensation, if the mass data flow still has a deviation trend, judging that the local compensation is insufficient; If the continuous two times of verification fail, judging that the current cooperative compensation fails, triggering a fusing rollback mechanism, and re-executing the cooperative compensation until the verification passes; When the quality data flow is restored to a steady-state interval and lasts for at least three work piece periods, parameters, paths and collaborative strategies of the current regulation and control process are automatically saved as self-adaptive regulation and control cases, and meanwhile, a processing report of the current quality event is pushed to a production management system.
  2. 2. The valve fitting on-line quality control method according to claim 1, wherein the defect-aware pulse generating step includes: Disposing multi-source heterogeneous detection nodes and collecting standardized data, and maintaining fixed length through edge calculation modules of all detection stations Fitting a time sequence of data in the window by using a linear regression algorithm or performing deviation trend identification on the mass data flow by using an accumulation and control graph algorithm; If the deviation trend is judged to exist, triggering a pulse generation mechanism, extracting the amplitude, the direction, the fluctuation frequency and the change acceleration of fluctuation as fluctuation feature vectors, and acquiring a feed speed multiplying power, a spindle rotating speed set value and the residual percentage of the service life of a cutter as equipment state snapshots through an industrial Ethernet; and (3) mapping the deviation degree normalization into a confidence degree based on the degree that the deviation point exceeds the control limit, and packaging the pulse generation time, the fluctuation feature vector, the equipment state snapshot and the confidence degree into a defect awareness pulse with a four-element structure.
  3. 3. The valve fitting on-line quality control method according to claim 1, wherein the step of confirming the point in time of the true defect includes: The production management system retrieves historical period data similar to the current defect consciousness pulse characteristics from a historical database, inputs a deviation sequence in a sliding window into a pre-trained long-short-period memory network or a gating circulation unit deep learning model, and outputs a deviation prediction sequence of a future workpiece period; Traversing the deviation prediction sequence, finding out a first index meeting the deviation value greater than or equal to the rejection threshold, and determining the cycle number corresponding to the current time plus the index as a predicted burst time point of the real defect; If the deviation prediction sequence is not out of limit in the prediction range and the prediction curve shows a regression trend, the predicted explosion time point of the real defect is set to be a null value and the depth regulation is suspended.
  4. 4. The valve fitting on-line quality control method according to claim 3, wherein the compensation instruction generation step includes: inquiring real-time process schedule and material flow tracking information in a manufacturing execution system, determining a unique workpiece identification code at a processing position at a real defect burst time point, and tracing a digital process list according to the unique workpiece identification code to determine a process node corresponding to the real defect burst time point; Starting from the current detection station node in the process topological graph, reversely searching paths of process nodes corresponding to the actual defect burst time point, and determining all process node sets on the reverse paths as node sets to be compensated; Inputting the current fluctuation feature vector into a fuzzy logic controller, obtaining a preliminary compensation quantity by inquiring a predefined fuzzy rule base and resolving fuzzy, and simultaneously calculating the similarity of the current fluctuation feature vector and the best matching case in a historical successful regulation case base to obtain a historical reference compensation quantity; Dynamically adjusting the fuzzy weight and the historical weight according to the stability of the working condition, and calculating the final compensation quantity by using the following formula: Wherein, the method comprises the steps of, For real-time membership values calculated based on fuzzy rule bases, Is the current feature And historical case library Similarity of the best matching cases; And Respectively fuzzy weight and historical weight, satisfies 。
  5. 5. The method for online quality control of valve fittings according to claim 1, wherein the judging and generating of the cooperative compensation comprises: establishing a post-compensation monitoring window, extracting key process quality characteristic data of the workpiece subjected to compensation operation to form a post-compensation data stream, and comparing the post-compensation data stream with a data stream before compensation is executed; If the mean value of the post-compensation data stream does not return to the vicinity of the target center line or the fluctuation amplitude is not obviously reduced, judging that the quality data stream still has a deviation trend; Traversing a node set to be compensated, and collecting real-time state information of each node, wherein the real-time state information comprises the residual time of a processing beat, the equipment load rate and the adjustable range of process parameters; according to a preset progressive priority regulation protocol, comprehensively considering the influence of the working procedure on final defects, the flexibility of parameter adjustment and the real-time load of the current node, calculating the regulation priority of each node, and sequentially generating cooperative compensation instructions of all nodes according to the order of the regulation priorities from high to low.
  6. 6. The valve fitting online quality tuning method of claim 5, wherein performing collaborative compensation further comprises performing parameter collaborative corrections and process rebalancing according to a progressive priority tuning protocol: When the processing parameters of a certain process are adjusted, the influence on the processing allowance of the subsequent process is synchronously evaluated, if the cutting amount is increased in the certain process, the reserved allowance of the subsequent finish processing process is automatically reduced, the balance of the total process allowance is kept, and overload of the certain process due to overlarge allowance or blank of the subsequent process due to insufficient allowance are avoided; And minimizing the energy consumption or the processing time of the path where the node set to be compensated is positioned on the premise of meeting the final quality constraint by a global process parameter optimization algorithm.
  7. 7. The method for online quality control of valve fittings according to claim 1, wherein the parameter convergence trend and defect pulse decay rate synchronous verification method is as follows: running monitoring threads in parallel, drawing a time response curve of key control parameters in real time, calculating the change rate of the response curve, and judging that the parameters show a convergence trend if the absolute value of the change rate gradually decreases and approaches 0 and the curve gradually approaches a target value; Continuously generating temporary defect consciousness pulse according to the real-time data, extracting the current confidence coefficient, and calculating the defect pulse attenuation rate by using the following formula: Wherein, the method comprises the steps of, Representing the decay rate of the defect pulse, In order to regulate the confidence of the pulse at the beginning, The confidence level is the confidence level of the current moment; if the defect pulse attenuation rate is continuously greater than 0 and the value is gradually increased along with the time, the defect pulse is rapidly attenuated and the regulation direction is correct.
  8. 8. The method for online quality control of valve fittings according to claim 7, wherein the fusing back-off mechanism is triggered by: setting a verification period, and if the parameter divergence or defect pulse attenuation rate is lower than a lower threshold in a certain evaluation, marking that one verification fails; maintaining a continuous failure counter, and if the continuous two verification results are failed, judging that the current collaborative compensation strategy is not suitable for the current working condition and judging that the collaborative compensation is invalid; Immediately stopping the output of all the cooperative compensation instructions currently being executed, issuing an emergency back-off instruction to related equipment, and forcibly recovering all the modified processing parameters to original values before regulation or default safety values of the system; And adjusting the fuzzy weight and the historical weight coefficient based on the experience of the failure, or reselecting different cooperative compensation strategy combinations, and re-executing the regulation and control flow from the generation of the compensation instruction until the verification is passed.
  9. 9. The valve fitting on-line quality control method of claim 1, wherein the adaptive control case generating step comprises: Constructing a feature index of a case, wherein the feature index comprises an original defect-aware pulse feature vector, a deviation type and a related process node sequence; Recording the adopted regulation and control decision, wherein the regulation and control decision comprises the distribution value of fuzzy weight and historical weight, the specific compensation quantity of each procedure, the priority sequence of cooperative compensation and a process rebalancing strategy; recording regulation effect evaluation indexes, wherein the regulation effect evaluation indexes comprise regulation response time, overshoot and steady state errors, storing the regulation response time, overshoot and steady state errors in a historical successful regulation case library, and establishing a multidimensional index.
  10. 10. A valve fitting on-line quality control system for performing the method of any of claims 1-9, the system comprising: the defect consciousness pulse generation module is used for automatically generating defect consciousness pulses containing fluctuation characteristics and equipment states when any detection station on the valve fitting automatic production line recognizes that the quality data flow has a deviation trend; the prediction and compensation instruction generation module is used for deducing the time point of the burst of the defect-aware pulse into the real defect in the future work piece period based on the time sequence prediction model after receiving the defect-aware pulse, reversely mapping a process path to be passed by the defect-aware pulse by taking the time point of the real defect as an end point; The cooperative compensation module is used for monitoring the mass data flow of the workpiece after compensation is executed, and judging that the local compensation is insufficient if the mass data flow still has a deviation trend; The safety control and fusing module is used for synchronously verifying the parameter convergence trend and the defect pulse attenuation rate in real time in the cooperative compensation process, judging that the current cooperative compensation fails if the two continuous verifications fail, triggering a fusing rollback mechanism, and re-executing the cooperative compensation until the verification passes; and the self-adaptive learning and feedback module is used for automatically storing parameters, paths and collaborative strategies of the current regulation and control process as self-adaptive regulation and control cases when the quality data flow is restored to a steady-state interval and lasts for at least three workpiece periods, and simultaneously pushing a processing report of the current quality event to the production management system.

Description

Valve fitting online quality regulation and control method and system Technical Field The invention relates to the technical field of intelligent manufacturing and industrial automatic detection, in particular to an online quality control method and system for valve fittings, which are suitable for real-time quality monitoring, defect prediction and active compensation control in an automatic production line of the valve fittings. Background The valve fittings are widely applied to key industrial fields of petrochemical industry, electric power, water supply and drainage and the like, and the manufacturing precision and reliability of the valve fittings are directly related to the safety and stability of the whole fluid control system. With the development of industry 4.0 and intelligent manufacturing, the production of valve fittings has gradually moved from traditional stand-alone operations to highly automated, in-line production. However, because the valve fittings have complex structures and various processing procedures, and involve multiple links such as casting, machining, surface treatment and the like, quality influence factors in the production process have high coupling property and nonlinearity, so that real-time monitoring and accurate regulation and control of the product quality become a problem to be solved in the industry. The existing valve fitting production quality control technology mainly depends on post spot check or a single-station threshold alarm mechanism. This passive quality control mode is often subject to hysteresis, and when a real defect is detected, the batch may already produce a lot of rejects, resulting in serious economic losses. In addition, for the detected quality deviation, the traditional adjusting method often depends on manual experience or a fixed rule base, and is difficult to cope with complex and variable interference factors of a production site. The lack of early prediction and dynamic compensation of the trend of quality fluctuations results in an inability to effectively intervene before the defect occurs, and the lack of systematic multi-process coordinated regulation strategies when single point compensation is ineffective. In addition, when the existing production management system processes quality data streams, correlation between time sequence characteristics and equipment states in the data streams is often ignored, and early signals of defect awareness are difficult to accurately identify. In the face of complex process parameter perturbations, the lack of adaptive adjustment mechanisms based on fuzzy matching of historical success cases with real-time data in the prior art results in insufficiently accurate generation of compensation parameters and may even exacerbate process instability due to erroneous adjustments. Meanwhile, the existing system lacks an effective fault-tolerant and rollback mechanism, and once the execution of an adjustment instruction fails, damage cannot be stopped in time, so that the continuity of a production line and the yield of a final product are affected. Disclosure of Invention In order to overcome the problems of lag in quality detection, lack of predictability and insufficient dynamic response capability to complex production interference in the prior art, the invention provides an online quality control method and system for valve fittings, which automatically generate defect conscious pulse comprising fluctuation characteristics and equipment states, deduce real defect burst time by using a time sequence prediction model and reversely map a process path, by combining the fuzzy-historical weighted compensation principle, the cooperative compensation and the fusing rollback mechanism, the advanced prejudgment and real-time accurate regulation and control of quality defects in the production process of valve fittings are realized, the rejection rate is effectively reduced, and the intelligent level of the production line and the quality stability of products are improved. The technical scheme of the application specifically comprises the following steps: According to an aspect of the present application, there is provided a valve fitting on-line quality control method, including: When any detection station on the valve fitting automatic production line recognizes that the quality data flow has a deviation trend, defect consciousness pulse containing fluctuation characteristics and equipment states is automatically generated; The production management system deduces the time point of the burst of the defect-aware pulse into the real defect in the future work piece period based on the time sequence prediction model after receiving the defect-aware pulse, reversely maps the working procedure path to be passed by the defect-aware pulse by taking the time point of the real defect as an end point; Monitoring the mass data flow of the workpiece after compensation, if the mass data flow still has a deviation trend, judgin