CN-121998617-A - Intelligent operation and maintenance and assessment method, system, storage medium and equipment for silk manufacturing equipment

CN121998617ACN 121998617 ACN121998617 ACN 121998617ACN-121998617-A

Abstract

The invention discloses an intelligent operation and maintenance method, an intelligent operation and maintenance system, an intelligent operation and maintenance storage medium and intelligent operation and maintenance equipment for silk manufacturing equipment, and belongs to the technical field of intelligent operation and maintenance of industrial equipment; the method comprises the steps of preprocessing multi-source heterogeneous data, respectively executing the following two evaluation paths to obtain a numerical health score and a semantic health score based on the preprocessed multi-source heterogeneous data, calculating uncertainty measurement between the numerical health score and the semantic health score, fusing the numerical health score and the semantic health score based on the uncertainty measurement to obtain a fused health score, calculating comprehensive efficiency of the silk manufacturing equipment through a nonlinear comprehensive efficiency evaluation function based on the fused health score and combining a safety index and a failure index, and automatically executing corresponding operation and maintenance decision according to the risk level of the comprehensive efficiency. The invention realizes accurate, self-adaptive and interpretable intelligent operation and maintenance and evaluation of the state of the yarn manufacturing equipment.

Inventors

ZHANG LUDAN
ZHANG XIAOLING

Assignees

罡正(海南)科技有限责任公司

Dates

Publication Date: 20260508
Application Date: 20260121

Claims (10)

1. The intelligent operation and assessment method for the silk manufacturing equipment is characterized by comprising the following steps of: S1, acquiring multi-source heterogeneous data related to a wire manufacturing device, wherein the multi-source heterogeneous data at least comprises structured data acquired by a sensor in real time and unstructured text data related to operation and maintenance; s2, preprocessing the multi-source heterogeneous data; s3, based on the preprocessed multi-source heterogeneous data, executing the following two evaluation paths: a numerical evaluation path, which is to calculate the preprocessed structured data based on a traditional numerical algorithm to obtain a numerical health score; A semantic evaluation path, which is to perform semantic understanding and reasoning on the unstructured text data and the structured data characteristics based on a multi-agent cooperative architecture and a search enhancement generation technology to obtain semantic health scores; S4, calculating an uncertainty measure between the numerical health score and the semantic health score, and fusing the numerical health score and the semantic health score based on the uncertainty measure to obtain a fused health score; s5, calculating the comprehensive efficiency of the silk manufacturing equipment through a nonlinear comprehensive efficiency evaluation function based on the fusion health score and combining the safety index and the failure index; S6, according to the risk level of the comprehensive efficiency, automatically executing a corresponding operation and maintenance decision.
2. The intelligent operation and maintenance and assessment method for a filament manufacturing device according to claim 1, wherein the preprocessing comprises: performing outlier rejection and standardization processing on the structured data; And converting the unstructured text data into a triplet knowledge graph, and converting the text paragraph into a vector by using a text embedding model.
3. The intelligent operation and maintenance and assessment method of a filament manufacturing apparatus according to claim 1, wherein the numerical health score is calculated by the formula: Wherein, the A numerical health score is represented and is used to determine, Indicating the overall deviation of the yarn making device.
4. The intelligent operation and maintenance and evaluation method of a filament manufacturing device according to claim 1, wherein the numerical evaluation path further comprises: Clustering historical data by using a K-Means algorithm to form a normal working condition cluster and an abnormal working condition cluster; and calculating Euclidean distance from the real-time data point to the center of each working condition cluster, and judging potential failure if the Euclidean distance exceeds a threshold value and the average sliding trend is obviously negative.
5. The method for intelligent operation and maintenance and assessment of a filament manufacturing device according to claim 1, wherein said calculating an uncertainty measure between said numerical health score and said semantic health score comprises: the degree of divergence between the numeric health score and the semantic health score is measured using cross entropy.
6. The intelligent operation and maintenance and assessment method for a filament manufacturing device according to claim 5, wherein the fusion health score is calculated by the following formula: Wherein the said The weight of the numeric health score is represented, Representing a semantic health score of the subject, , Representing the weight-adjusting parameter of the vehicle, Representing the uncertainty measure.
7. The intelligent operation and maintenance and assessment method for a filament manufacturing device according to claim 1, wherein the overall performance is calculated by the following formula: Wherein H represents a health index, S represents a safety index, Indicating the failure index (f) and, The coefficient of steepness is represented by the value of the steepness, , , All represent weight coefficients.
8. An intelligent operation and assessment system for a wire making device, comprising: The data perception layer is used for acquiring multi-source heterogeneous data related to the wire manufacturing equipment, and the multi-source heterogeneous data at least comprises structured data acquired by a sensor in real time and unstructured text data related to operation and maintenance; The algorithm processing layer is used for preprocessing the multi-source heterogeneous data and executing the following two evaluation paths based on the preprocessed multi-source heterogeneous data: a numerical evaluation path, which is to calculate the preprocessed structured data based on a traditional numerical algorithm to obtain a numerical health score; A semantic evaluation path, which is to perform semantic understanding and reasoning on the unstructured text data and the structured data characteristics based on a multi-agent cooperative architecture and a search enhancement generation technology to obtain semantic health scores; the agent cooperation layer is used for calculating the uncertainty measure between the numerical health score and the semantic health score, and fusing the numerical health score and the semantic health score based on the uncertainty measure to obtain a fused health score; And the fusion decision layer is used for calculating the comprehensive efficiency of the wire manufacturing equipment through a nonlinear comprehensive efficiency evaluation function based on the fusion health score and combining the safety index and the failure index, and automatically executing corresponding operation and maintenance decisions according to the risk level of the comprehensive efficiency.
9. A computer readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the intelligent operation and maintenance and assessment method of a wire-making apparatus according to any one of claims 1 to 7.
10. An electronic device comprising a memory and a processor, the memory having stored thereon computer instructions executable on the processor, wherein the processor executes the intelligent operation and assessment method of the thread-making device of any one of claims 1-7 when the processor executes the computer instructions.

Description

Intelligent operation and maintenance and assessment method, system, storage medium and equipment for silk manufacturing equipment Technical Field The invention relates to the technical field of intelligent operation and maintenance of industrial equipment, in particular to an intelligent operation and maintenance and evaluation method, system, storage medium and equipment of a wire manufacturing device. Background In the tobacco shred production process, the running states of core process equipment such as a shredding machine, a cut tobacco drier, a moisture regain machine and the like directly determine the physical characteristics of tobacco shreds and the sensory quality of the final cigarettes. At present, the equipment operation and maintenance of a wire making workshop mainly depends on the technology of the industrial Internet of things (IIoT), time series data are collected through sensors such as vibration, temperature, current and the like, and the monitoring is performed by a monitoring and data acquisition System (SCADA) or an equipment management system (EAM). The mainstream technical solutions of the operation and maintenance of the existing equipment are generally classified into two types, namely a traditional algorithm based on a physical model, such as frequency domain analysis by using Statistical Process Control (SPC), fast Fourier Transform (FFT), or a fixed health evaluation system constructed based on Analytic Hierarchy Process (AHP), and a shallow machine learning based on data driving, such as classification and regression prediction of historical fault data by using Random Forest (Random Forest), XGBoost or K-Means clustering algorithm. The techniques have a degree of maturity in processing structured numerical data, and are widely applied to threshold alarm and simple trend prediction of equipment. Although the prior art realizes the digital monitoring of the equipment, the following technical bottlenecks which are difficult to overcome still exist when dealing with the complex failure mechanism and the high dynamic production environment of the wire making equipment: 1. Numerical value-semantic heterogeneous data cannot be effectively fused, and knowledge utilization rate is low: The prior art mainly deals with structured numerical data of sensors, while a large amount of equipment operation and maintenance knowledge (such as equipment maintenance manuals, fault case libraries, expert experience records, team shift logs) exists in unstructured text form. The traditional algorithm cannot understand text semantics, so that sensor data and maintenance knowledge are in a split state during fault diagnosis. For example, when a sensor captures an abnormal waveform, the system cannot automatically correlate with the corresponding description of "occasional vibration caused by improper operation" in the service manual, and still needs to manually review the document, which results in low diagnosis efficiency and easy misjudgment. 2. The traditional evaluation model has static stiffness of weight and lacks environmental adaptability: Once the existing health evaluation system (such as AHP or weighted average of fixed weights) is set, the index weight (such as vibration 30% and temperature 20%) is always kept unchanged. However, the operating conditions of the shredded tobacco processing equipment are dynamically changed (e.g., the processing parameters of shredded tobacco of different brands are different, or the equipment is in different stages of the aging cycle). In certain failure modes (e.g., sudden overheating), the importance of the temperature indicator should be increased instantaneously, and a model with fixed weight cannot respond dynamically to this, resulting in that the composite score is "diluted" by other normal indicators when critical failure features occur, thereby masking the real risk. 3. Lacking interpretability and reasoning capabilities, it is difficult to implement closed-loop decisions: Existing fault diagnosis models based on deep learning or ensemble learning (e.g., XGBoost) are generally regarded as "black boxes", and only output fault classification probabilities (e.g., bearing fault probabilities 85%), but cannot give a logical inference chain (i.e., why a fault is determined). In addition, the existing system cannot automatically generate a specific maintenance strategy which accords with the current technological situation according to the diagnosis result. The operation and maintenance personnel face the alarm information and still need to make a maintenance scheme according to personal experience, so that the treatment effect of the same fault is different from person to person, and the standardized intelligent closed loop from 'finding the problem' to 'solving the problem' cannot be realized. 4. It is difficult for a single algorithm to compromise real-time and complex logic reasoning: While traditional numerical algorithms are fast in computation but