CN-121785110-B - Method and system for process parameter multi-procedure associated modeling and self-adaptive control

CN121785110BCN 121785110 BCN121785110 BCN 121785110BCN-121785110-B

Abstract

The invention discloses a method and a system for multi-procedure associated modeling and self-adaptive control of process parameters, and relates to the technical field of production process control. The method comprises the steps of collecting process parameter time sequence data, equipment running state time sequence data and quality detection result data associated with production batch identification, writing time stamps and aligning and associating the time stamps to generate an associated data set comprising a three-dimensional associated identification and a hash check code, extracting energy consumption stability characteristic quantity and fault hidden risk characteristic quantity in an evaluation period, calculating equipment health score, inputting the process time sequence characteristic and the equipment health score into a process optimization model, outputting target process parameter combinations under process constraint, evaluating risk grades according to adjustment amplitude and the equipment health score before issuing, controlling issuing modes, and collecting the quality detection result data and updating the process optimization model to form a closed loop after the quality detection result data and the energy consumption index update process optimization model are validated. The invention realizes the safe and self-adaptive optimization of the technological parameters under the perception of the health state of the equipment.

Inventors

CHEN FENG
LI XIAOYOU
XU FANGHUI
WANG HANBO
QI WENFU

Assignees

奥特赛斯(天津)数字科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260305

Claims (9)

1. The method for multi-procedure associated modeling and self-adaptive control of the technological parameters is characterized by comprising the following steps: S1, acquiring multi-source time sequence data of target production equipment in a production process, wherein the multi-source time sequence data comprises process parameter time sequence data, equipment running state time sequence data and quality detection result data related to production batch identification; s2, extracting equipment health characteristic quantity based on the equipment running state time sequence data in a preset evaluation period, and calculating equipment health score based on the equipment health characteristic quantity, wherein the equipment health characteristic quantity comprises an energy consumption stability characteristic quantity and a fault hiding risk characteristic quantity; S3, constructing the process parameter time sequence data into multi-process time sequence features according to process marks, generating multi-process associated input features based on parameter transfer relations among different processes, taking the multi-process time sequence features, the multi-process associated input features and the equipment health scores together as inputs of a process optimization model, and outputting target process parameter combinations under preset process constraint conditions; the step of outputting the target process parameter combination comprises the following steps: Based on the process parameter time sequence data, segmenting a process parameter sampling sequence of each process according to process identification, extracting multi-working-procedure time sequence features reflecting the change trend, fluctuation amplitude and change rate of the process parameters according to a preset time window, constructing the time sequence features and the equipment health score by combining the features to form combined input features for representing the current running state and the process adaptation degree of the equipment, inputting the combined input features into a process optimization model and outputting the target process parameter combination under the common limitation of process parameter value range constraint, equipment running capacity constraint, quality stability constraint and equipment cooperative constraint, wherein the equipment cooperative constraint is determined by constructing a scheduling adaptation degree function based on the equipment health score, equipment replacement time, equipment cooperative punishment coefficient and qualified deviation punishment coefficient, searching by adopting a genetic algorithm with the maximized scheduling adaptation degree function as a target to obtain an optimal production path, and determining the equipment cooperative constraint according to the optimal production path, wherein the scheduling adaptation degree function meets the following formula: ; Wherein: A function value for scheduling fitness; the number of the replacement links is the number; Is the first Equipment changing time of each changing link; Is extremely small normal quantity; is a synergistic and quality comprehensive item; a penalty coefficient is coordinated for the device; Punishment coefficients for the qualified deviation; and (3) with Is a preset weight coefficient and meets the following requirements ; S4, before issuing the target process parameter combination, evaluating and obtaining a risk level of process parameter adjustment based on the adjustment amplitude of the target process parameter combination relative to the currently effective process parameter and the equipment health score associated with the production lot identification, and controlling the issuing mode of the target process parameter combination according to the risk level; S5, after the target process parameter combination is effective, corresponding quality detection result data and energy consumption index data are collected, and the process optimization model is updated based on the quality detection result data and the energy consumption index data, so that a process parameter self-adaptive optimization closed loop is formed.
2. The method of claim 1, wherein the step of collecting multi-source time series data of the target production equipment during the production process comprises: Taking the same production batch period or continuous production period as an acquisition period, and acquiring process parameter time sequence data, equipment running state time sequence data and quality detection result data associated with production batch identification; Writing uniform time stamps into the process parameter time sequence data, the equipment running state time sequence data and the quality detection result data, and executing alignment processing on the multi-source time sequence data based on the uniform time stamps; Generating a three-dimensional association identifier based on the unified timestamp, the material identifier and the equipment number, and binding the three-dimensional association identifier with the aligned multi-source time sequence data to form an association data set; Calculating a hash check code for key binding fields at least comprising the three-dimensional association identifier, the unified timestamp, the equipment number and the production batch identifier in the association data set, and storing the hash check code in association with a corresponding association data record to realize data tracing and tamper-proof check.
3. The method of process parameter multi-process correlated modeling and adaptive control of claim 2, wherein said quality inspection result data is generated by a test management module comprising a centralized process test and a time control test; the centralized processing test comprises the steps of executing pre-analysis processing on a test file associated with a production batch identifier, and generating a quality index entry based on a pre-analysis result, wherein the quality index entry forms at least one part of the quality detection result data; The time control test comprises triggering the start and the end of the test according to a preset time control condition, recording breakpoint information in the test process, and continuously executing a corresponding test flow based on the breakpoint information after the test is interrupted to generate a quality index item.
4. The method of process parameter multi-process correlated modeling and adaptive control of claim 1, wherein calculating an equipment health score based on the equipment health feature comprises: respectively constructing an equipment energy consumption stability characteristic quantity and an equipment fault hiding risk characteristic quantity based on equipment running state time sequence data; Performing scale unification processing on the equipment energy consumption stability characteristic quantity and the equipment fault hidden risk characteristic quantity to obtain a health evaluation characteristic quantity with comparability, wherein the scale unification processing comprises normalization processing or standardization processing; Comprehensively evaluating the health evaluation feature quantity based on a preset evaluation fusion rule to generate an equipment health score; the evaluation fusion rule comprises the steps of performing normalization processing on the energy consumption change coefficient, the risk hiding coefficient and the temperature related index and the vibration related index extracted from the equipment operation state time sequence data, inputting the normalized temperature related index and the vibration related index into a fuzzy rule base, and performing anti-blurring processing on an output result of the fuzzy rule base to obtain the equipment health score representing the equipment health state.
5. The method of process parameter multi-process related modeling and adaptive control of claim 4, wherein the energy consumption stability characteristic comprises an energy consumption coefficient of variation The energy consumption change coefficient Calculated as follows: ; Wherein: The ratio of standard deviation to mean value is calculated based on the equipment energy consumption sampling sequence of the target production equipment in a preset evaluation period; Is a natural logarithmic function; the fault-concealment risk feature comprises a risk-concealment coefficient The risk concealment coefficient Calculated as follows: ; Wherein: the method comprises the steps that as the time difference between the current time and the equipment failure high-frequency center time, the equipment failure high-frequency center time is obtained by adopting a DBSCAN density clustering method based on an equipment failure occurrence time sequence; the method comprises the steps of presetting similar fault times in a historical time window; is natural constant and In hours, 24 represents a 24 hour time reference.
6. The method of claim 1, wherein the evaluating obtains a risk level of process parameter adjustment and controlling the issuing method of the target process parameter combination according to the risk level comprises executing issuing control logic by a rule engine, and determining a corresponding issuing method based on an adjustment amplitude of the target process parameter combination relative to a currently effective process parameter, wherein the adjustment amplitude is an absolute value or a relative change proportion of each process parameter in the target process parameter combination relative to a change amount of the currently effective process parameter; When the adjustment amplitude is not more than a first threshold value, executing automatic issuing, when the adjustment amplitude is more than the first threshold value and not more than a second threshold value, executing manual confirmation and then issuing, and when the adjustment amplitude is more than the second threshold value, executing approval and then issuing; When the equipment health score is lower than a preset health threshold value or the risk hiding coefficient is higher than a preset risk threshold value, the upper limit value of at least one process parameter in the target process parameter combination is adjusted downwards according to a preset proportion; when the early warning event state which is not closed exists, the issuing mode is controlled to be issued after manual confirmation; The early warning event processing comprises the steps of constructing an alarm three-dimensional identifier based on equipment numbers, alarm types and parameter names, executing alarm duplication elimination in a preset sliding window, executing gradient upgrading of alarm levels based on alarm unprocessed time length and a preset upgrading threshold value, and merging low-level alarms corresponding to the same equipment numbers into summarized alarms based on a preset merging time window, wherein alarm rules support online updating and version rollback.
7. The method of process parameter multi-process modeling and adaptive control of claim 1, wherein updating the process optimization model comprises: based on the production batch identification, binding quality detection result data, energy consumption index data and joint input features, and generating a training sample for model training, wherein the joint input features are feature vectors obtained by joint construction of process parameter time sequence features and equipment health scores; when the number of the newly added training samples reaches a preset number threshold or reaches a preset model updating time period, executing incremental training of the process optimization model and updating model parameters; and generating corresponding version identifiers for model versions before and after model updating, and storing the version identifiers and the corresponding production batch identifiers in a correlated manner.
8. The method for multi-process related modeling and self-adaptive control of technological parameters according to claim 1, wherein the method further comprises a quality closed loop control step and a packaging release verification step, wherein the quality closed loop control step comprises the steps of locking circulation authorities of detected bad products, generating and issuing corresponding maintenance tasks; checking the filling test completion state and the bad processing state of the product to be packaged in a product packaging link based on the production batch identification and the associated quality detection result data; checking the consistency of the product model identification, the production batch identification and the outer box identification in the product boxing step; When the boxing quantity reaches a preset value, generating an outer box tracing total code associated with the product serial number set in the box, wherein the outer box tracing total code is a coded mark generated based on the product serial number set, and writing the outer box tracing total code into an outer box label.
9. A system for process parameter multi-procedure associated modeling and adaptive control, which is applied to the method for process parameter multi-procedure associated modeling and adaptive control according to any one of claims 1 to 7, and is characterized by comprising a multi-source data acquisition module, a multi-source data processing module and a multi-source data processing module, wherein the multi-source data acquisition module is used for acquiring multi-source time sequence data of target production equipment in the production process, and the multi-source time sequence data comprises process parameter time sequence data, equipment running state time sequence data and quality detection result data associated with production batch identification; The alignment association and tamper-proof module is used for executing alignment and association processing on the multi-source time sequence data to generate an association data set containing a three-dimensional association identifier and a corresponding hash check code; The equipment health scoring module is used for extracting equipment health characteristic quantity based on equipment running state time sequence data in a preset evaluation period and calculating equipment health scoring based on the equipment health characteristic quantity; The process parameter optimization module is used for constructing process parameter time sequence data into multi-process time sequence characteristics according to process marks, generating multi-process associated input characteristics based on parameter transfer relations among different processes, taking the multi-process time sequence characteristics, the multi-process associated input characteristics and equipment health scores as inputs of a process optimization model, and outputting target process parameter combinations under the condition that preset process constraint conditions are met; The issuing risk control module is used for evaluating the risk level of process parameter adjustment based on the adjustment amplitude of the target process parameter combination relative to the current effective process parameter and the corresponding equipment health score before issuing the target process parameter combination, and controlling the issuing mode of the target process parameter combination according to the risk level; and the model updating module is used for acquiring corresponding quality detection result data and energy consumption index data after the target process parameter combination is effective, and updating the process optimization model based on the data.

Description

Method and system for process parameter multi-procedure associated modeling and self-adaptive control Technical Field The invention relates to the technical field of production process control, in particular to a method and a system for multi-procedure associated modeling and self-adaptive control of process parameters. Background In the industrial production process, information such as process parameters, equipment operation states, quality detection results and the like are usually generated in a time sequence manner and stored in different production and detection systems in a scattered manner. The process parameter optimization and process control based on the data are common, but the following defects still exist in practical application. On the one hand, the multi-source data has complex sources and more circulation links, and is easy to generate the conditions of inconsistent time, unclear association relation, incomplete record and the like, so that the process tracing and reproduction for production batches are difficult, and the credibility and verifiability of the optimization decision basis are affected. On the other hand, the existing technology optimizes single targets such as multi-side quality or energy consumption, or considers the equipment state insufficiently, so that the health condition and potential risk of the equipment are difficult to synchronously reflect in the optimizing process. Under the condition of equipment performance fluctuation or risk accumulation, if the parameter adjustment lacks constraint on equipment state, unstable process, quality fluctuation or abnormal energy consumption are easily caused, and stable operation of production is affected. Process parameter adjustment is generally required to meet constraints such as safety, equipment carrying capacity, and quality stability. In the aspect of risk assessment and execution control of parameter adjustment, the prior art often lacks a consistent and controllable treatment mechanism, and is difficult to adopt a differentiated execution mode according to the risk degree, so that the problems of bias of adjustment strategies, false issuing or response lag and the like are caused. Meanwhile, equipment aging, material difference and working condition change exist in long-term operation of the production process, and the suitability of the optimization strategy may be reduced with time. If an effective result feedback and continuous updating mechanism is lacking, the optimization effect is easy to drift and is difficult to stably play a role for a long time. Disclosure of Invention The invention aims to provide a method and a system for multi-procedure associated modeling and self-adaptive control of process parameters, which enable multi-source data to be reliably associated and reliably traced in an industrial production process, realize optimized output and controllable execution of the process parameters considering equipment states, and form a continuous improvement mechanism by combining production results, thereby improving the reliability of process optimization and the stability of the production process. In order to achieve the above object, the present invention is realized by the following technical scheme: In one aspect, the invention provides a method for multi-process associated modeling and adaptive control of process parameters, comprising the following steps: S1, acquiring multi-source time sequence data of target production equipment in a production process, wherein the multi-source time sequence data comprises process parameter time sequence data, equipment running state time sequence data and quality detection result data related to production batch identification; s2, extracting equipment health characteristic quantity based on the equipment running state time sequence data in a preset evaluation period, and calculating equipment health score based on the equipment health characteristic quantity, wherein the equipment health characteristic quantity comprises an energy consumption stability characteristic quantity and a fault hiding risk characteristic quantity; S3, constructing the process parameter time sequence data into multi-process time sequence features according to process marks, generating multi-process associated input features based on parameter transfer relations among different processes, taking the multi-process time sequence features, the multi-process associated input features and the equipment health scores together as inputs of a process optimization model, and outputting target process parameter combinations under preset process constraint conditions; S4, before issuing the target process parameter combination, evaluating and obtaining a risk level of process parameter adjustment based on the adjustment amplitude of the target process parameter combination relative to the currently effective process parameter and the equipment health score associated with the production lot identifi