CN-121980505-A - AI fusion method and system for production data acquisition of precast beam mobile production line

CN121980505ACN 121980505 ACN121980505 ACN 121980505ACN-121980505-A

Abstract

The application provides an AI fusion method and system for production data acquisition of a precast beam mobile production line, and belongs to the technical field of data processing. The method comprises the steps of firstly collecting multi-source heterogeneous data of a production line, performing time-space registration to form a fusion data stream, further extracting space structure features and time sequence features from the fusion data stream, intelligently identifying a current process stage, evaluating confidence values of data sources, dynamically distributing feature fusion weights by combining the two data sources, generating comprehensive quality features based on the weight fusion multi-dimensional features, and finally outputting a judging result through an abnormality judging model. According to the application, the production situation awareness and the data reliability dynamic evaluation are combined, so that the self-adaptive optimization of the data fusion strategy is realized, and the accuracy and reliability of the production anomaly judgment and the overall robustness of the system are effectively improved.

Inventors

YANG XI
LV BAO
XU YUANJI
SHEN ZHONGPING
XIAO WENQIANG
Xing Jiaqing

Assignees

上海有间建筑科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260122

Claims (10)

1. An AI fusion method for production data acquisition of a precast beam mobile production line, which is characterized by being applicable to an AI fusion system for production data acquisition of a precast beam mobile production line, wherein the system comprises an execution terminal, and the method is executed by the execution terminal and comprises the following steps: Acquiring corresponding multi-source heterogeneous data of each procedure of the precast beam on a mobile production line based on a plurality of sensors, and registering the multi-source heterogeneous data in time and space dimensions to form a fusion data stream; extracting spatial structural features and time sequence features related to the production quality of the precast beams from the fusion data stream; identifying a current working procedure stage corresponding to a production line, and acquiring a confidence value corresponding to each data source in the current working procedure stage, wherein the confidence value is used for representing the reliability of data provided by the data sources; Assigning a weight to each of the data sources based on the current process stage and the confidence value; fusing the spatial structure features and the time sequence features based on the weights to generate comprehensive quality features; And inputting the comprehensive quality characteristics into a preset abnormality judgment model, and outputting an abnormality judgment result corresponding to the current production state.
2. The AI fusion method for collecting production data of a precast beam moving production line according to claim 1, wherein the AI fusion method is characterized by identifying a current process stage corresponding to the production line and obtaining a confidence value corresponding to each data source in the current process stage, wherein the confidence value is used for representing reliability of data provided by the data source, and comprises the following steps: acquiring frequency domain characteristics of vibration signals in the fusion data stream, slope characteristics of a temperature change curve and material morphological characteristics in an image; And matching the frequency domain features, the slope features and the standard feature templates of all process stages preset by the material morphological features, and identifying the process stage corresponding to the standard feature template with the highest matching degree as the current process stage.
3. The AI fusion method for collecting production data of a precast beam moving production line according to claim 2, wherein the AI fusion method is characterized by identifying a current process stage corresponding to the production line and obtaining a confidence value corresponding to each data source in the current process stage, wherein the confidence value is used for representing reliability of data provided by the data sources, and comprises the following steps: Acquiring the data loss rate of each data source in the current process stage and the deviation degree of the data value of the data source and the historical reference data; and determining the confidence value based on the data loss rate and the deviation degree, wherein the absolute values of the data loss rate and the deviation degree are inversely related to the confidence value.
4. The AI fusion method of claim 3, wherein assigning weights to each of the data sources based on the current process stage and the confidence values comprises: acquiring a preset basic weight associated with the current process stage; the preset base weights are adjusted based on the confidence value of each of the data sources to generate the weights for feature fusion.
5. The AI fusion method for collecting production data of a precast beam moving production line according to claim 3, wherein obtaining the data deletion rate of each data source in the current process stage and the deviation degree of the data value of the data source from the historical reference data comprises: determining the expected duration of the current working procedure stage and the standard sampling frequency of the data source, and calculating to obtain the theoretical data point quantity; Acquiring the number of valid data points actually received from the corresponding data source in the current process stage; and acquiring the data loss rate based on the theoretical data point number and the effective data point number.
6. The AI fusion method for collecting production data of a precast beam moving production line according to claim 3, wherein obtaining the data deletion rate of each data source in the current process stage and the deviation degree of the data value of the data source from the historical reference data comprises: acquiring data value statistical distribution characteristics of the data source corresponding to the current process stage in a historical normal production state; Acquiring a difference value between a data value of the current data source and a central trend measurement of the statistical distribution characteristic of the historical data value; and normalizing the difference value to obtain the deviation degree representing the current deviation history normal fluctuation range.
7. The AI fusion method for collecting production data of a precast beam moving production line according to claim 1, wherein extracting spatial structure features and time sequence features related to precast beam production quality from the fusion data stream comprises: Obtaining geometrical characteristics used for representing Liang Tibiao surface textures and mold gaps from the image data of the fusion data stream, and taking the geometrical characteristics as the spatial structural characteristics; and extracting characteristics representing signal energy distribution and change trend from the vibration and temperature data sequence of the fusion data stream as the time sequence characteristics.
8. The AI fusion method for collecting production data of a precast beam moving production line according to claim 1, wherein the comprehensive quality features are input into a preset abnormality determination model, and an abnormality determination result corresponding to a current production state is output, and the AI fusion method comprises the following steps: comparing the comprehensive quality characteristics with the normal production state characteristic range corresponding to the current working procedure stage; and if the comprehensive quality characteristics exceed the boundary threshold value of the normal production state characteristic range, judging that production abnormality occurs, and outputting a judging result containing abnormality type and abnormality grade information.
9. An AI fusion system for production data acquisition of a precast beam mobile production line, characterized in that the system comprises an execution terminal, and the system is configured to: Acquiring corresponding multi-source heterogeneous data of each procedure of the precast beam on a mobile production line based on a plurality of sensors, and registering the multi-source heterogeneous data in time and space dimensions to form a fusion data stream; extracting spatial structural features and time sequence features related to the production quality of the precast beams from the fusion data stream; identifying a current working procedure stage corresponding to a production line, and acquiring a confidence value corresponding to each data source in the current working procedure stage, wherein the confidence value is used for representing the reliability of data provided by the data sources; Assigning a weight to each of the data sources based on the current process stage and the confidence value; fusing the spatial structure features and the time sequence features based on the weights to generate comprehensive quality features; And inputting the comprehensive quality characteristics into a preset abnormality judgment model, and outputting an abnormality judgment result corresponding to the current production state.
10. An electronic device, comprising: At least one processor; and a memory communicatively coupled to at least one of the processors; Wherein the memory stores instructions executable by at least one of the processors, the instructions being executable by at least one of the processors to enable the at least one of the processors to perform the method as set forth in any one of claims 1-8.

Description

AI fusion method and system for production data acquisition of precast beam mobile production line Technical Field The application relates to the technical field of data processing, in particular to an AI fusion method for collecting production data of a precast beam mobile production line. Background In the mobile production line of precast beams, the monitoring and control of the production quality is highly dependent on the collection and analysis of multi-dimensional data in the production process. At present, a common production data acquisition mode mainly comprises the steps of deploying various sensors, such as a position encoder, a vibration sensor, a temperature sensor, an industrial camera and the like, at each process node so as to acquire multi-source heterogeneous data reflecting beam body states, equipment operation and environmental parameters. However, prior art schemes often isolate or simply time stamp align data from different physical dimensions, different sampling frequencies, and lack deep spatio-temporal correlation and registration when processing such data, making it difficult to form a consistent data view describing the complete state at the same production time. Secondly, the utilization of data stays at a threshold alarm level of single characteristic monitoring or fixed rules, the inherent association between multi-source data cannot be fully mined, and the reliability difference of different data sources under different production situations cannot be effectively evaluated. Furthermore, the process of the production line is dynamically advanced, and quality concerns at different stages are quite different, but the existing method lacks intelligent recognition capability to the current production stage, and cannot enable the data analysis strategy to be adaptively focused on the current key process parameters. Disclosure of Invention The embodiment of the application provides an AI fusion method and system for collecting production data of a precast beam mobile production line, so as to solve the problems. In order to achieve the above purpose, the application adopts the following technical scheme: In a first aspect, the present application provides an AI fusion method for collecting production data of a precast beam mobile production line, which is applicable to an AI fusion system for collecting production data of a precast beam mobile production line, where the system includes an execution terminal, and the method is executed by the execution terminal, and includes: Acquiring corresponding multi-source heterogeneous data of each procedure of the precast beam on a mobile production line based on a plurality of sensors, and registering the multi-source heterogeneous data in time and space dimensions to form a fusion data stream; Extracting spatial structure characteristics and time sequence characteristics related to the production quality of the precast beams from the fusion data stream; Identifying the current working procedure stage corresponding to the production line, and acquiring a confidence value corresponding to each data source in the current working procedure stage, wherein the confidence value is used for representing the reliability of data provided by the data source; Assigning a weight to each data source based on the current process stage and the confidence value; Fusing the space structure features and the time sequence features based on the weights to generate comprehensive quality features; and inputting the comprehensive quality characteristics into a preset abnormality judgment model, and outputting an abnormality judgment result corresponding to the current production state. With reference to the first aspect, in some embodiments, identifying a current process stage corresponding to the production line, and obtaining a confidence value corresponding to each data source at the current process stage, where the confidence value is used to characterize reliability of data provided by the data source, includes: Acquiring frequency domain characteristics of vibration signals in the fusion data stream, slope characteristics of a temperature change curve and material morphological characteristics in an image; and matching the frequency domain features, the slope features and the standard feature templates of each process stage preset by the material morphological features, and identifying the process stage corresponding to the standard feature template with the highest matching degree as the current process stage. With reference to the first aspect, in some embodiments, identifying a current process stage corresponding to the production line, and obtaining a confidence value corresponding to each data source at the current process stage, where the confidence value is used to characterize reliability of data provided by the data source, includes: Acquiring the data loss rate of each data source in the current process stage and the deviation degree of the data va