CN-121981496-A - Manufacturing execution system management method for production data

CN121981496ACN 121981496 ACN121981496 ACN 121981496ACN-121981496-A

Abstract

The invention relates to the field of manufacturing execution systems and intelligent manufacturing, in particular to a manufacturing execution system management method for production data, which comprises a multi-source data acquisition step of acquiring physical sensing state data, manual interaction time data and system scheduling data of a manufacturing execution system and integrating the physical sensing state data, the manual interaction time data and the system scheduling data to generate multi-source heterogeneous production data, a trust quantification evaluation step of obtaining an execution trust entropy based on the distortion degree of the quantitative execution state of the multi-source heterogeneous production data, a cross-validation prediction step of comparing the physical sensing state data with the manual interaction time data to calculate the execution trust entropy and predict the failure risk of the system, and a dynamic scheduling decision step of setting a scheduling degradation adjustment mechanism based on the execution trust entropy and outputting a target scheduling instruction to adjust the running state of the system.

Inventors

OU DAPENG
HU XIAOXIAO

Assignees

苏州安软信息科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260403

Claims (9)

1. The manufacturing execution system management method for production data is characterized by comprising the following steps: Acquiring physical sensing state data, manual interaction time data and system scheduling data of a manufacturing execution system; Constructing a multi-source data acquisition layer, and integrating the physical sensing state data, the manual interaction time data and the system scheduling data to generate multi-source heterogeneous production data; Constructing a trust quantization evaluation layer, quantifying the distortion degree of the execution state of the manufacturing execution system based on the multi-source heterogeneous production data to obtain an execution trust entropy, wherein the trust quantization evaluation layer comprises a cross verification module and a risk prediction module, the cross verification module is used for comparing the physical sensing state data with the manual interaction time data to calculate the execution trust entropy, and the risk prediction module is used for predicting the failure risk of the system based on the execution trust entropy; And constructing a dynamic scheduling decision layer, setting a scheduling degradation adjustment mechanism based on the execution trust entropy, outputting a target scheduling instruction according to the scheduling degradation adjustment mechanism, and adjusting the running state of the manufacturing execution system by using the target scheduling instruction.
2. The method of claim 1, further comprising, prior to said building the multi-source data collection layer: Preprocessing the physical sensing state data, the manual interaction time data and the system scheduling data to obtain preprocessed production data for integration of the multi-source data acquisition layer, wherein the preprocessing comprises time delay compensation, noise filtering and space-time alignment.
3. The manufacturing execution system management method for production data according to claim 1, wherein the cross-validation module is configured to calculate the execution trust entropy based on a logic comparison model in combination with a time sequence analysis model, and specifically comprises: extracting fluctuation characteristics of the physical sensing state data and reporting nodes of the manual interaction time data through the logic comparison model, calculating a time difference value between the fluctuation characteristics and the reporting nodes, and taking the time difference value as a consistency deviation value; fitting the historical consistency deviation data through the time sequence analysis model, calculating the cumulative sum of the consistency deviation values in a preset time window by using a time sequence analysis method, and taking the cumulative sum as the execution trust entropy.
4. The manufacturing execution system management method for production data according to claim 1, wherein the risk prediction module is configured to divide the risk of system failure into risk states of different levels, and evaluate a contribution degree of the execution trust entropy to the risk of system failure, and specifically includes: And evaluating the risk evolution trend of the execution trust entropy by combining a preset equipment operation threshold, a capacity constraint condition and a historical failure record, wherein the evaluation process comprises the steps of performing global failure probability prediction by using a probability prediction model trained based on the historical failure record, and calculating a probability value that the manufacturing execution system reaches a preset failure boundary parameter by combining the current scheduling buffer time.
5. The manufacturing execution system management method for production data according to claim 1, wherein the schedule degradation adjustment mechanism includes: setting a preset safety threshold and a preset dangerous threshold, wherein the preset dangerous threshold is larger than the preset safety threshold; Judging a numerical interval in which the execution trust entropy is located; triggering an optimal scheduling strategy when the execution trust entropy is smaller than the preset safety threshold; Triggering a transition scheduling strategy when the execution trust entropy is greater than or equal to the preset safety threshold and is smaller than the preset dangerous threshold; And triggering a suboptimal degradation scheduling strategy when the execution trust entropy is greater than or equal to the preset dangerous threshold.
6. The method of claim 5, wherein adjusting the operating state of the manufacturing execution system using the target scheduling instruction comprises: and implementing the dynamic scheduling based on the execution trust entropy by combining the dynamic scheduling decision layer with the scheduling degradation adjustment mechanism, wherein the dynamic scheduling based on the execution trust entropy comprises the following steps: Selecting a target scheduling strategy according to the numerical interval in which the execution trust entropy is located, and generating the target scheduling instruction by using the dynamic scheduling decision layer based on the target scheduling strategy; taking a real-time production data stream as dynamic input, dynamically adjusting the target scheduling instruction, when the abnormal production beat is monitored, recalculating the execution trust entropy in real time, and rescheduling a future scheduling path by utilizing the dynamic scheduling decision layer; The target scheduling instructions are automatically updated based on a data synchronization mechanism of the manufacturing execution system.
7. The method of claim 6, wherein generating the target scheduling instructions using the dynamic scheduling decision layer based on the target scheduling policy comprises: if the optimal scheduling strategy is triggered, setting the inter-process buffer time as a minimum preset value, keeping a system assessment mechanism in the manufacturing execution system activated, and generating a first scheduling instruction as the target scheduling instruction; if the suboptimal degradation scheduling strategy is triggered, setting the inter-process buffer time to be larger than the minimum preset value, freezing timeout penalty logic in the system assessment mechanism, and generating a second scheduling instruction for guiding real abnormal data to flow back as the target scheduling instruction; and if the transition scheduling strategy is triggered, dynamically interpolating and calculating target buffer time according to the increase rate of the execution trust entropy, and generating a third scheduling instruction as the target scheduling instruction.
8. The manufacturing execution system management method for production data according to claim 1, wherein the physical sensing state data comprises equipment current fluctuation data collected by workshop internet of things equipment, and the manual interaction time data comprises manual code scanning progress reporting data collected by terminal equipment.
9. The manufacturing execution system management method for production data according to claim 1, further comprising: acquiring real state feedback data of the manufacturing execution system after executing the target scheduling instruction; calculating the descending amplitude of the execution trust entropy based on the real state feedback data; And carrying out self-adaptive updating on parameters in the trust quantization evaluation layer by using the descending amplitude.

Description

Manufacturing execution system management method for production data Technical Field The invention relates to the field of manufacturing execution systems and intelligent manufacturing, in particular to a manufacturing execution system management method for production data. Background In the management of a manufacturing execution system, a scheduling plan based on the comprehensive efficiency of equipment gradually becomes a main stream method for improving the production efficiency of a workshop due to the visual capacity guiding capability of the equipment; however, when the scheduling system supporting the multi-source data is constructed in the above manner, the physical sensing state data and the manual interaction time data need to be directly butted, so that the distortion accumulation between the real execution state and the digital twin state of the bottom fragmented data is covered, the logic contradiction between the action time and the manual reporting time of the physical sensor cannot be accurately identified, the scheduling strategy cannot be adjusted according to the real physical bearing capacity when the sudden interference is faced, the problem that the system issues the logic paralysis risk of an abnormal scheduling instruction due to the serious data disjoint exists, meanwhile, the processing manner lacks a quantitative evaluation mechanism for the distortion degree, the whole line shutdown is easily caused by the tiny data disjoint in the high-load running state of the system, the problem of low vulnerability of a complex manufacturing network exists, and the problem that the instruction failure caused by the static scheduling is existed because the real abnormal data backflow is difficult to be guided to reconstruct the data authenticity is also dependent on the static scheduling rule design. Disclosure of Invention The invention aims to provide a manufacturing execution system management method for production data, which solves the following technical problems: the risk that the system generates logic paralysis due to the fact that abnormal instructions are issued due to serious data disconnection is avoided, real abnormal data backflow can be guided through a scheduling degradation adjustment mechanism, dynamic balance is achieved between pursuit efficiency and maintenance of bottom data authenticity, and vulnerability resistance of a complex manufacturing network is effectively improved. The aim of the invention can be achieved by the following technical scheme: the manufacturing execution system management method facing to the production data comprises the following steps: Acquiring physical sensing state data, manual interaction time data and system scheduling data of a manufacturing execution system; Constructing a multi-source data acquisition layer, and integrating the physical sensing state data, the manual interaction time data and the system scheduling data to generate multi-source heterogeneous production data; Constructing a trust quantization evaluation layer, quantifying the distortion degree of the execution state of the manufacturing execution system based on the multi-source heterogeneous production data to obtain an execution trust entropy, wherein the trust quantization evaluation layer comprises a cross verification module and a risk prediction module, the cross verification module is used for comparing the physical sensing state data with the manual interaction time data to calculate the execution trust entropy, and the risk prediction module is used for predicting the failure risk of the system based on the execution trust entropy; And constructing a dynamic scheduling decision layer, setting a scheduling degradation adjustment mechanism based on the execution trust entropy, outputting a target scheduling instruction according to the scheduling degradation adjustment mechanism, and adjusting the running state of the manufacturing execution system by using the target scheduling instruction. Further, before the constructing the multi-source data acquisition layer, the method further comprises: Preprocessing the physical sensing state data, the manual interaction time data and the system scheduling data to obtain preprocessed production data for integration of the multi-source data acquisition layer, wherein the preprocessing comprises time delay compensation, noise filtering and space-time alignment. Further, the cross-validation module is configured to calculate the execution trust entropy based on a logic comparison model and a timing analysis model, and specifically includes: extracting fluctuation characteristics of the physical sensing state data and reporting nodes of the manual interaction time data through the logic comparison model, calculating a time difference value between the fluctuation characteristics and the reporting nodes, and taking the time difference value as a consistency deviation value; fitting the historical consistency deviation data thr