CN-121390954-B - Flour production line safety intelligent supervision system based on cloud platform

CN121390954BCN 121390954 BCN121390954 BCN 121390954BCN-121390954-B

Abstract

The invention relates to the technical field of flour production line safety supervision, in particular to a cloud platform-based flour production line safety intelligent supervision system which comprises a data acquisition module, a risk assessment module, a decision module and an execution and repair module, wherein the system calculates instant physical risks by acquiring production line sensor data and an HMI operation log, the core is to quantify trust bare words of operators based on the HMI operation log and predict bypass probability of safety interlocking according to the trust bare words, when the instant physical risks exceed a preset threshold value, the system executes final actions according to whether the trust bare words exceed the threshold value, and the invention brings the trust state of the operators into assessment, so that the problem of systematic failure caused by neglecting human factors and being easy to bypass in the traditional system is solved.

Inventors

BAO LIFENG
ZHAO JINHUI
Bao Yanchao
DONG SHAOXIONG

Assignees

石家庄市藁城区立峰面业有限公司

Dates

Publication Date: 20260512
Application Date: 20251031

Claims (4)

1. Flour production line safety intelligent supervision system based on cloud platform, characterized by comprising: the data acquisition module is used for acquiring the multidimensional sensor data vector of the production line sensor in real time and acquiring the HMI operation log of the human-machine interface; The risk assessment module is used for determining the probability of the instantaneous ignition source based on the multidimensional sensor data vector and calculating the instantaneous physical risk by combining the preset probability of the exceeding of the concentration of the combustible; The risk assessment module is also used for quantifying trust bare words of operators based on the HMI operation log and determining bypass probability of safety interlocking according to the trust bare words; The decision module is used for determining a final execution action according to the comparison result of the instantaneous physical risk and a preset physical risk threshold value and the trust red word and a preset trust red word threshold value; the execution and repair module is used for executing a final execution action, synchronously pushing the interpretable warning information and feeding back and correcting the trust bare character when the final execution action is a preset toughness repair action; The multidimensional sensor data vector comprises environment relative humidity, material flow rate, electrostatic sensor reading and key bearing vibration; The risk assessment module quantifies operator confidence bare words, including: Monitoring the HMI operation log to identify an operator to continuously override AI alarms triggered by the transient ignition source probability and to mark as a trust destruction event; Based on the trust destruction event, applying a preset exponential decay and pulse accumulation model to calculate trust red words; The decision module determines a final execution action, including: when the instantaneous physical risk is greater than a preset physical risk threshold and the trust red is not greater than a preset trust red threshold, determining a preset default optimal action as a final execution action; When the instantaneous physical risk is greater than a preset physical risk threshold and the trust red is greater than the preset trust red threshold, calculating the total explosion probability of the default optimal action through a preset toughness utility function, wherein the total explosion probability is determined based on the expected bypass probability with extremely high expected value, and the expected bypass probability is equal to the bypass probability; The decision module is further configured to calculate a total explosion probability for the ductile repair action, wherein the total explosion probability is determined based on an expected bypass probability that is expected to be very low; The decision module is also used for comparing the total explosion probability of the default optimal action with the total explosion probability of the ductile repair action, and selecting an action with a lower total explosion probability as a final execution action; The risk assessment module determines an instantaneous ignition source probability, comprising: Nonlinear mapping is carried out on the multidimensional sensor data vector by utilizing a neural network function through a pre-trained federal graph neural network so as to output instantaneous ignition source probability; The pre-trained model weight adopted by the neural network function is obtained by performing federal learning training on historical sensor data and simulated failure events on a cloud platform; The risk assessment module calculates an instantaneous physical risk, comprising: According to the event concurrency probability principle of classical failure tree analysis, carrying out product operation on the instantaneous ignition source probability and the preset combustible concentration exceeding probability; the preset probability of exceeding the concentration of combustible substances is a high-value constant, and the high-risk steady state that the dust concentration of the production line is maintained in an explosion lower limit area is represented; The risk assessment module determines a bypass probability of the safety interlock, comprising: And mapping the trust red characters into bypass probabilities through a preset standard logic Stiff function.
2. The cloud platform based flour production line safety intelligent supervision system according to claim 1, wherein the calculation of the total explosion probability is defined as the probability addition of two mutually exclusive cases, the cases including the probability that an action is performed but still a physical explosion occurs, and the probability that an action is bypassed and the bypass results in an explosion; wherein the conditional probability of the bypass leading to an explosion is preset to 1.0.
3. The cloud platform based flour production line safety intelligent supervision system according to claim 1, wherein the toughness restoration action is as a predefined physical action combination, comprising: Sending an instruction to a ventilation and dust removal system to enable the power to be instantaneously raised to a preset peak value; And sending a command to the production line control system to reduce the speed of the grinding and screening unit by a preset percentage.
4. The cloud platform based flour production line safety intelligent supervision system according to claim 1, wherein the execution and repair module feedback corrects the trust red word, comprising: Monitoring a human-computer interface to identify the behavior of an operator who does not execute bypass operation in a preset decision window, and calibrating the behavior as a trust repair event; And feeding the trust repairing event back to the risk assessment module, and actively reducing the accumulated value of the trust red word by introducing preset negative weight.

Description

Flour production line safety intelligent supervision system based on cloud platform Technical Field The invention relates to the technical field of flour production line safety supervision, in particular to a cloud platform-based flour production line safety intelligent supervision system. Background In high-risk man-machine interaction environments such as flour production lines, dust explosion is a continuous serious threat, a traditional AI safety supervision system tends to operate as a black box and is focused on physical risk assessment, but generally lacks human factor consideration on the trust state of an operator, when the AI system gives out an alarm or performs actions, the operator can select bypass safety interlocking, namely the personnel fail, due to distrust, understanding or alarm fatigue, the bypass behavior causes the safety system to face the risk of systematic failure, so that the problem of supervision failure caused by lack of human factor consideration and easiness in bypass of the operator of the traditional black box AI safety system is solved, and the problem of supervision failure caused by lack of human factor consideration and easiness in bypass of the operator is solved in the field. Disclosure of Invention In order to solve the technical problems, the invention provides a cloud platform-based flour production line safety intelligent supervision system, which comprises the following steps: the data acquisition module is used for acquiring the multidimensional sensor data vector of the production line sensor in real time and acquiring the HMI operation log of the human-machine interface; The risk assessment module is used for determining the probability of the instantaneous ignition source based on the multidimensional sensor data vector and calculating the instantaneous physical risk by combining the preset probability of the exceeding of the concentration of the combustible; The risk assessment module is also used for quantifying trust bare words of operators based on the HMI operation log and determining bypass probability of safety interlocking according to the trust bare words; The decision module is used for determining a final execution action according to the comparison result of the instantaneous physical risk and a preset physical risk threshold value and the trust red word and a preset trust red word threshold value; and the execution and repair module is used for executing the final execution action, and synchronously pushing the interpretability warning information and feeding back and correcting the trust bare letter when the final execution action is the preset toughness repair action. Preferably, the multidimensional sensor data vector comprises ambient relative humidity, material flow rate, electrostatic sensor readings and key bearing vibration. Preferably, the risk assessment module determines an instantaneous ignition source probability, including: Nonlinear mapping is carried out on the multidimensional sensor data vector by utilizing a neural network function through a pre-trained federal graph neural network so as to output instantaneous ignition source probability; the pre-trained model weight adopted by the neural network function is obtained by federal learning training of historical sensor data and simulated failure events on a cloud platform. Preferably, the risk assessment module calculates an instantaneous physical risk, including: According to the event concurrency probability principle of classical failure tree analysis, carrying out product operation on the instantaneous ignition source probability and the preset combustible concentration exceeding probability; the preset probability of exceeding the concentration of combustible substances is a high-value constant, and the high-risk steady state that the dust concentration of the production line is maintained in an explosion lower limit area is represented. Preferably, the risk assessment module quantifies the trust bare word of the operator, including: Monitoring the HMI operation log to identify an operator to continuously override AI alarms triggered by the transient ignition source probability and to mark as a trust destruction event; based on the trust disruption event, a preset exponential decay and pulse accumulation model is applied to calculate trust red. Preferably, the risk assessment module determines a bypass probability of the safety interlock, comprising: And mapping the trust red characters into bypass probabilities through a preset standard logic Stiff function. Preferably, the determining module determines the final execution action, including: when the instantaneous physical risk is greater than a preset physical risk threshold and the trust red is not greater than a preset trust red threshold, determining a preset default optimal action as a final execution action; When the instantaneous physical risk is greater than a preset physical risk threshold and the trust red is grea