CN-122022467-A - Method for identifying and processing large risk early warning model of pharmaceutical workshop

CN122022467ACN 122022467 ACN122022467 ACN 122022467ACN-122022467-A

Abstract

The invention belongs to the field of risk early warning, and relates to a method for identifying and processing a large risk early warning model of a pharmaceutical workshop, which comprises the following steps of constructing a risk data system and carrying out fusion training on risk data and safety production standards; and constructing a risk processing large model according to historical production data and a company management system, and carrying out multi-mode generation and processing strategies of data distribution and early warning.

Inventors

CHEN YU
WANG YANGCHAO
ZHANG SHUNDE

Assignees

浙江海翔药业股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. A method for identifying and processing a large risk early warning model of a pharmaceutical workshop is characterized by comprising the following steps of constructing a risk data system and carrying out fusion training on risk data and safety production standards; Acquiring real-time production data, and detecting a production risk index in real time through the real-time production data; And constructing a risk processing large model according to the historical production data and a company management system, and carrying out multi-mode generation and processing strategies of data distribution and early warning.
2. The method for identifying and processing the large risk early warning model of the pharmaceutical workshop according to claim 1, wherein the steps of constructing a risk data system and performing fusion training on risk data and safety production standards comprise the following steps: Wherein the risk sources comprise material risk, equipment risk, environment risk and operation risk; in the construction of the risk data system, the data comprise a safety production specification, historical production data and corresponding data related to risk content, and the fusion training comprises the steps of fusing the risk data with the safety production specification to form a structured risk knowledge graph; And carrying out knowledge extraction on the risk knowledge graph through a BERT-based entity identification model, carrying out knowledge fusion based on an entity link technology, carrying out knowledge reasoning based on a graph neural network, and finally realizing knowledge application.
3. The method for identifying and processing the large risk early warning model of the pharmaceutical workshop according to claim 1, wherein the method is characterized by acquiring real-time production data and detecting the production risk index in real time through the real-time production data, and comprises the following steps of acquiring the real-time production data, namely acquiring an Internet of things system and acquiring special data, wherein the special data comprises chemical raw materials, production equipment and production environment; Based on real-time production data, carrying out real-time risk prediction through a large model, and calculating a real-time risk index comprising chemical treatment risks, equipment abnormality risks and environment exceeding risks by combining the risk weights and the severity degrees; And distributing multi-mode early warning data according to the risk types, wherein the multi-mode early warning data comprises a text report, an image report, an audio report, a short video report and an animation report.
4. The method for identifying and processing the risk early warning large model of the pharmaceutical workshop according to claim 3, wherein the acquisition of the real-time production data comprises the acquisition of an internet of things system and the acquisition of special data, wherein for the management of chemical raw materials, the related information of each batch of raw materials is input through chemical electronic codes, a material management system is established, a dynamic material batch and production batch relation library is formed, and the production speed of each batch of materials is controlled through uploading and checking the production and in-library states of each batch of materials in real time.
5. The method for identifying and processing the risk early warning large model of the pharmaceutical workshop according to claim 3, wherein the acquisition of the real-time production data comprises acquisition of an internet of things system and acquisition of special data, wherein various production parameters are monitored in real time for monitoring and maintaining production equipment, during the production process, abnormal production problems of the production equipment are found through an intelligent risk early warning model and manual inspection, a maintenance scheme is provided in a targeted manner, and auxiliary decisions for production equipment management are provided.
6. The method for identifying and processing the risk early warning large model of the pharmaceutical workshop according to claim 3, wherein the acquisition of the real-time production data comprises the acquisition of an internet of things system and the acquisition of special data, wherein the monitoring and the continuous improvement of the production environment are realized, the safety condition of the production environment is monitored in real time, the production environment index report is generated, the continuous improvement is performed, an environment management system is established, the production environment data of different areas, different places and different time periods are recorded, and the real-time monitoring, the management and the analysis of the production environment data are realized.
7. The method for identifying and processing the risk early warning big model of the pharmaceutical workshop according to claim 1 is characterized in that the method is used for constructing a risk processing big model and carrying out multi-mode generation and processing strategies of data distribution and early warning, and comprises the following steps of realizing risk identification through the risk processing big model, carrying out risk assessment through the risk processing big model, carrying out early warning generation and release through a risk manager, and carrying out risk elimination and modification through the risk processing strategies.
8. The method for identifying and processing the risk early warning large model of the pharmaceutical workshop according to claim 7, wherein the risk identification realized by the risk processing large model comprises data processing and identification, and the data processing comprises real-time data filtering, conversion and cleaning to form standardized data; The identification comprises the steps of carrying out identification analysis on data through a big data identification algorithm, screening abnormal data, carrying out supervised learning through a deep neural network model based on a transducer by combining historical data and a knowledge graph, extracting risk characteristics through characteristic engineering, and training a risk reasoning model base.
9. The method for identifying and processing the risk early warning big model of the pharmaceutical workshop according to claim 7 is characterized in that risk assessment is carried out through the risk processing big model, the risk assessment comprises calculation based on real-time risk indexes RI and risk classification division according to the real-time risk indexes RI, the risk assessment flow specifically comprises the steps of calculating the real-time risk indexes RI through the big model, outputting the real-time risk indexes RI in real time, carrying out preliminary assessment judgment through comparison with a preset threshold value, carrying out further audit confirmation through manual work, and generating early warning information according to a final assessment result.
10. The method for identifying and processing the large risk early warning model of the pharmaceutical workshop according to claim 7 is characterized in that the generation and release of the early warning by a risk manager comprises multi-mode early warning including text, image, audio, short video and animation, hierarchical processing of the early warning, different levels of processing schemes according to the level degree of the early warning, and release by means of system information, short messages, safety broadcasting and manual notification to form an emergency response closed loop.

Description

Method for identifying and processing large risk early warning model of pharmaceutical workshop Technical Field The invention belongs to the field of risk early warning, and particularly relates to a method for identifying and processing a large risk early warning model of a pharmaceutical workshop. Background The production safety is a basic stone for enterprise development, the safety production risk management is a foundation for guaranteeing the harmonious development and continuous operation of enterprises, is basic work for constructing the harmonious labor relation of the enterprises and maintaining the image of the enterprises, is an important guarantee for promoting the development of the enterprises and keeping the basic industry, is a key for realizing the identification, early warning, evaluation and coping of the safety risk timely and accurately, is a system engineering and dynamic engineering for the joint participation of all members of the enterprises, and is a system engineering and a dynamic engineering for ensuring the safety production risk management work task and requirements of all levels, departments and posts of the enterprises, and realizes the transition from passive response to active prevention through continuous improvement, thereby gradually establishing the enterprise safety production risk management system. The prior art discloses a method and a system for intelligently identifying risks in chemical safety production based on a knowledge graph, wherein the method comprises the steps of knowledge graph construction and updating, knowledge reasoning and retrieval and risk identification. Aiming at the chemical industry field, multi-source chemical safety knowledge such as chemical technology, equipment operation and experimental data, laws and regulations, operation rules, process data files, emergency plans and the like is formed into a knowledge base of a directed graph, knowledge patterns of risks and potential safety hazards in the chemical technology and equipment operation process are realized, multi-source heterogeneous knowledge fusion is carried out, knowledge reasoning is carried out based on a graph matching and graph neural network method, data such as operation parameters and the like acquired in real time and the data in the safety knowledge base are subjected to reasoning analysis through the knowledge patterns, and risk identification and analysis based on the graph matching and the graph neural network are realized, so that the accuracy of risk identification is improved. However, the prior art only focuses on the dangerous degree of chemical substances, ignores production equipment, production environment and human factors, does not carry out multi-dimensional early warning on the chemical substances, the environment and the equipment, does not carry out grading treatment on risks of early warning, cannot take different measures through different risk grades, does not carry out systematic description on how to eliminate the risks, does not describe how to improve the removal and solving efficiency of the risks, and is inconvenient for a risk manager to judge and manage. Therefore, a person skilled in the art provides a method for identifying and processing a large risk early warning model of a pharmaceutical workshop, so as to solve the problems in the prior art. Disclosure of Invention The invention provides a method for identifying and processing a large risk early warning model of a pharmaceutical workshop. A method for identifying and processing a large risk early warning model of a pharmaceutical workshop comprises the following steps of constructing a risk data system and carrying out fusion training on risk data and safety production standards; Acquiring real-time production data, and detecting a production risk index in real time through the real-time production data; And constructing a risk processing large model according to the historical production data and a company management system, and carrying out multi-mode generation and processing strategies of data distribution and early warning. Preferably, the construction of the risk data system and the fusion training of the risk data and the safety production specification comprise the following contents: Wherein the risk sources comprise material risk, equipment risk, environment risk and operation risk; in the construction of the risk data system, the data comprise a safety production specification, historical production data and corresponding data related to risk content, and the fusion training comprises the steps of fusing the risk data with the safety production specification to form a structured risk knowledge graph; And carrying out knowledge extraction on the risk knowledge graph through a BERT-based entity identification model, carrying out knowledge fusion based on an entity link technology, carrying out knowledge reasoning based on a graph neural network, and finally realizing knowledge