CN-122022488-A - Safety risk monitoring method and system for pharmaceutical workshop, electronic equipment and medium

CN122022488ACN 122022488 ACN122022488 ACN 122022488ACN-122022488-A

Abstract

The application provides a safety risk monitoring method, a system, electronic equipment and a medium for a pharmaceutical workshop, wherein the method comprises the steps of obtaining multi-source monitoring data of the type of a current operation scene in the pharmaceutical workshop and in a current time period; the method comprises the steps of predicting occurrence probability of each preset accident type in future preset time through a Bayesian network model, inputting the current operation scene type and multi-source monitoring data into a fusion network in a risk monitoring model to obtain multi-mode fusion characteristics, and determining a first safety risk monitoring result at the current moment according to the current operation scene type, the multi-mode fusion characteristics and on-site operation images in the multi-source monitoring data so as to perform safety risk early warning. The application can be fused with the field operation depth of a pharmaceutical workshop, solves the problem of the existing risk management and control deficiency and the manual inefficiency of fire control inspection, and realizes the intelligent field risk inspection.

Inventors

LU XIAOJING
Ma Lvli
ZHANG WEI
XU GUANHUA

Assignees

杭州胡庆余堂药业有限公司

Dates

Publication Date: 20260512
Application Date: 20260213

Claims (10)

1.A method for monitoring safety risk of a pharmaceutical plant, the method comprising: Acquiring multi-source monitoring data of the current operation scene type in a pharmaceutical workshop in a current time period; Predicting occurrence probability of each preset accident type in a future preset time period through a Bayesian network model based on multi-source monitoring data in a current time period, wherein a graph structure in the Bayesian network model is constructed by taking accident causes of the environment and equipment as root nodes, taking abnormal symptoms of the pharmaceutical workshop environment as intermediate nodes and taking preset accident types as leaf nodes; inputting the current operation scene type and the multi-source monitoring data at the current moment into a fusion network in a risk monitoring model, and carrying out weighted fusion on the multi-source monitoring data according to the current operation scene type to obtain a multi-mode fusion characteristic; Determining a first security risk monitoring result at the current moment according to the current operation scene type, the multi-mode fusion characteristic and a field operation image in multi-source monitoring data at the current moment; And taking the occurrence probability of each preset accident type and the first safety risk monitoring result as a second safety risk monitoring result so as to perform safety risk early warning.
2. The method of claim 1, wherein prior to predicting the probability of occurrence of each preset accident type within a future preset time period and weighting the multi-source monitoring data, the method further comprises: Aiming at the multisource monitoring data at each moment, when the environmental humidity in the multisource monitoring data is larger than a preset humidity threshold value, defogging enhancement is carried out on the field operation image in the multisource monitoring data according to the difference value between the environmental humidity and the preset humidity threshold value so as to carry out safety risk monitoring through the multisource monitoring data after defogging enhancement.
3. The method for monitoring the safety risk of a pharmaceutical shop according to claim 2, wherein the defogging enhancement of the field operation image in the multi-source monitoring data according to the difference between the ambient humidity and the preset humidity threshold value comprises: Calculating a contrast compensation coefficient corresponding to a field operation image in the multi-source monitoring data according to a difference value between the environmental humidity in the multi-source monitoring data and the preset humidity threshold value, wherein the contrast compensation coefficient is used for quantifying the influence degree of the environmental humidity on the image definition; and defogging and contrast enhancement operations are carried out on the field operation image in the multi-source monitoring data according to the contrast compensation coefficient.
4. A method for monitoring safety risk of a pharmaceutical plant according to claim 3, wherein calculating a contrast compensation coefficient corresponding to a field operation image in the multi-source monitoring data according to a difference between the environmental humidity in the multi-source monitoring data and the preset humidity threshold value comprises: substituting the difference value between the ambient humidity and the preset humidity threshold value into the following formula to obtain a contrast compensation coefficient corresponding to the field operation image; ; Wherein, the As the contrast compensation coefficient, a reference is made to, For the ambient humidity at time t, For a first gain factor to be preset, For a second gain factor to be preset, In order to preset the humidity threshold value, And the difference value between the ambient humidity and the preset humidity threshold value is obtained.
5. The method for monitoring the safety risk of a pharmaceutical workshop according to claim 1, wherein predicting the occurrence probability of each preset accident type in a future preset time period through a bayesian network model based on multi-source monitoring data in a current time period comprises: inputting multi-source monitoring data in a current time period into an accident condition detection model to obtain an accident condition set, wherein the accident condition set comprises accident causes of the environment and equipment and/or abnormal symptoms of the environment of the pharmaceutical workshop; Inputting the accident occurrence condition set into the Bayesian network model for reasoning to obtain the occurrence probability of each preset accident type in the future preset time length.
6. The method for monitoring the safety risk of a pharmaceutical workshop according to claim 1, wherein the determining the first safety risk monitoring result at the current moment according to the current operation scene type, the multi-mode fusion feature and the field operation image in the multi-source monitoring data at the current moment comprises: Inputting the field operation image into a target detection model to obtain the position information of each target; inputting the position information of each target and the multi-mode fusion characteristic into a security risk detection model corresponding to the current operation scene type to obtain corresponding security risk detection characteristic data; and calculating a first security risk monitoring result at the current moment according to the current operation scene type and the security risk detection characteristic data.
7. The method for monitoring the safety risk of a pharmaceutical workshop according to claim 6, wherein if the current operation scene type is a live fire operation scene type, calculating a first safety risk monitoring result at the current time according to the current operation scene type and the safety risk detection feature data, comprises: Calculating a first security risk monitoring result at the current moment according to the following formula; ; Wherein, the A fire operation risk assessment score in the first safety risk monitoring result; detecting a characteristic diagram containing open flame and splash sparks in the characteristic data for safety risk; In order to be a characteristic diagram of the open flame, A splash spark map; Is a physical and chemical property dangerous coefficient of a medium in a flammable solvent pipeline of a pharmaceutical workshop; detecting fire source coordinates in the characteristic data for safety risk; detecting the first in the feature data for security risk Coordinates of the flammable solvent lines; detecting whether a guardian is on duty or not as a binary variable in the characteristic data for safety risk; All flammable solvent lines in the characteristic data are collected for safety risk detection.
8. A system for monitoring safety risk in a pharmaceutical plant, the system comprising: The acquisition module is used for acquiring the current operation scene type in the pharmaceutical workshop and multi-source monitoring data in the current time period; The system comprises a prediction module, a Bayesian network model, a leaf node and a leaf node, wherein the prediction module is used for predicting occurrence probability of each preset accident type in a future preset time period through the Bayesian network model based on multi-source monitoring data in a current time period, and a graph structure in the Bayesian network model is constructed by taking accident causes of an environment and equipment as a root node, abnormal symptoms of a pharmaceutical workshop environment as an intermediate node and preset accident types as the leaf node; The input module is used for inputting the current operation scene type and the multi-source monitoring data at the current moment into a fusion network in a risk monitoring model so as to perform weighted fusion on the multi-source monitoring data according to the current operation scene type and obtain multi-mode fusion characteristics; the determining module is used for determining a first safety risk monitoring result at the current moment according to the current operation scene type, the multi-mode fusion characteristic and a field operation image in multi-source monitoring data at the current moment; And the risk early warning module is used for taking the occurrence probability of each preset accident type and the first safety risk monitoring result as a second safety risk monitoring result so as to perform safety risk early warning.
9. An electronic device comprising a processor, a storage medium and a bus, the storage medium storing machine-readable instructions executable by the processor, the processor in communication with the storage medium via the bus when the electronic device is in operation, the processor executing the machine-readable instructions to perform the steps of the method of monitoring safety risk in a pharmaceutical shop as claimed in any one of claims 1 to 7.
10. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a computer program which, when executed by a processor, performs the steps of the safety risk monitoring method of a pharmaceutical workshop according to any one of claims 1 to 7.

Description

Safety risk monitoring method and system for pharmaceutical workshop, electronic equipment and medium Technical Field The invention relates to the field of pharmaceutical factories, in particular to a safety risk monitoring method, a system, electronic equipment and a medium for a pharmaceutical workshop. Background The safety production risk management of the pharmaceutical factory is a key link for guaranteeing personnel safety, product quality and environmental protection, and has important significance for stable operation of enterprises. Currently, most pharmaceutical enterprises establish a basic safety production risk management system, and apply risk management tools such as HAZOP analysis and PHA analysis to carry out risk assessment work, and the whole management mode is still mainly based on the traditional manual mode, so that the intelligent and automatic levels are low. In the actual operation process, the existing risk management system lacks on-site risk check, the phenomenon of irregular operation is prominent, the fire control check depends on manual development, the flow is complicated, the efficiency is low, the problem feedback is delayed, various risk analysis results are mostly recorded on paper, the on-site operation depth is not fused, the risk management and control measures are difficult to execute in a landing mode, and great hidden danger is brought to the safety production of pharmaceutical factories. Disclosure of Invention In view of the above, the application aims to provide a safety risk monitoring method, a system, electronic equipment and a medium for a pharmaceutical workshop, which can be fused with the field operation depth of the pharmaceutical workshop, solve the problems of existing risk management and control deficiency, manual inefficiency of fire protection inspection and difficulty in landing measures, and realize intelligent field risk inspection. In a first aspect, an embodiment of the present application provides a method for monitoring safety risk in a pharmaceutical workshop, The method comprises the following steps: Acquiring multi-source monitoring data of the current operation scene type in a pharmaceutical workshop in a current time period; Predicting occurrence probability of each preset accident type in a future preset time period through a Bayesian network model based on multi-source monitoring data in a current time period, wherein a graph structure in the Bayesian network model is constructed by taking accident causes of the environment and equipment as root nodes, taking abnormal symptoms of the pharmaceutical workshop environment as intermediate nodes and taking preset accident types as leaf nodes; inputting the current operation scene type and the multi-source monitoring data at the current moment into a fusion network in a risk monitoring model, and carrying out weighted fusion on the multi-source monitoring data according to the current operation scene type to obtain a multi-mode fusion characteristic; Determining a first security risk monitoring result at the current moment according to the current operation scene type, the multi-mode fusion characteristic and a field operation image in multi-source monitoring data at the current moment; And taking the occurrence probability of each preset accident type and the first safety risk monitoring result as a second safety risk monitoring result so as to perform safety risk early warning. In a possible implementation manner, before predicting occurrence probability of each preset accident type in a future preset time period and performing weighted fusion on the multi-source monitoring data, the method further comprises: Aiming at the multisource monitoring data at each moment, when the environmental humidity in the multisource monitoring data is larger than a preset humidity threshold value, defogging enhancement is carried out on the field operation image in the multisource monitoring data according to the difference value between the environmental humidity and the preset humidity threshold value so as to carry out safety risk monitoring through the multisource monitoring data after defogging enhancement. In a possible implementation manner, the defogging enhancement on the field operation image in the multi-source monitoring data according to the difference between the ambient humidity and the preset humidity threshold value includes: Calculating a contrast compensation coefficient corresponding to a field operation image in the multi-source monitoring data according to a difference value between the environmental humidity in the multi-source monitoring data and the preset humidity threshold value, wherein the contrast compensation coefficient is used for quantifying the influence degree of the environmental humidity on the image definition; and defogging and contrast enhancement operations are carried out on the field operation image in the multi-source monitoring data according to the contrast co