CN-121979058-A - Intelligent early warning control system for mixing uniformity of food compound additive

CN121979058ACN 121979058 ACN121979058 ACN 121979058ACN-121979058-A

Abstract

The invention belongs to the technical field of intelligent early warning control, and discloses an intelligent early warning control system for mixing uniformity of a food compound additive; the system comprises a dynamic early warning model updating and learning module, a multistage intelligent early warning generation traceability analysis module, a model prediction feedforward instruction generation module and a self-adaptive feedback optimization coordination module, wherein the deviation of a mixed state prediction report and a state synchronization report is monitored, the dynamic early warning model is updated to obtain a new dynamic early warning model, the comprehensive state synchronization report, the mixed state prediction report and the new dynamic early warning model are analyzed to obtain a comprehensive early warning diagnosis report, an optimization target is set for calculation based on the mixed process prediction report to obtain a feedforward control instruction report, and a final execution instruction report is obtained.

Inventors

ZHAO XINZHEN
GE XINGXING
ZHONG XUANKAI
ZHOU HAO
KOU DANDAN
CHEN XUETING
ZHANG LING
HAN XINYU
WANG WENYU
LIU XIAOHAN

Assignees

山东素康食品科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260202

Claims (10)

1. The intelligent early warning control system for the mixing uniformity of the food compound additive is characterized by comprising a dynamic early warning model updating and learning module, a multistage intelligent early warning generation traceability analysis module, a model prediction feedforward instruction generation module and an adaptive feedback optimization coordination module, wherein: the dynamic early warning model updating learning module is used for monitoring deviation between the mixed state prediction report and the state synchronization report, and updating the dynamic early warning model to obtain a new dynamic early warning model; The multistage intelligent early warning generation traceability analysis module is used for analyzing the comprehensive state synchronous report, the mixed state prediction report and the new dynamic early warning model to obtain a comprehensive early warning diagnosis report; The model prediction feedforward instruction generation module is used for predicting a report based on a mixed process, setting an optimization target for calculation, and obtaining a feedforward control instruction report; the self-adaptive feedback optimization coordination module is used for analyzing based on the feedforward control instruction report, the comprehensive early warning diagnosis report and the state synchronization report to obtain a final execution instruction report.
2. The intelligent pre-warning control system for mixing uniformity of a food compound additive according to claim 1, wherein the system further comprises a multi-source data collaborative acquisition preprocessing module, a state characteristic digital twin synchronization module and a digital twin mixing process simulation prediction module, wherein: The multi-source data collaborative acquisition preprocessing module is used for acquiring an original data stream in real time based on a sensor network and preprocessing the data to obtain a real-time standardized data set; the state feature digital twin synchronization module is used for processing based on the real-time standardized data set to obtain a fusion state feature vector, and performing data synchronization to obtain a state synchronization report; the digital twin hybrid process simulation prediction module is used for predicting the evolution of the hybrid state in a preset future time period based on the state synchronous report and running the hybrid process simulation model to obtain a hybrid state prediction report.
3. The intelligent early warning control system for mixing uniformity of a food compound additive according to claim 2, wherein the steps of collecting an original data stream in real time based on a sensor network and preprocessing the data comprise: s1.1, collecting data streams from different sensors in a sensor network in real time to obtain an original data set, wherein the data streams comprise a collection time stamp, a sensor ID and original readings; S1.2, maintaining a global clock by the system, and aligning acquisition time stamps of other data streams in the original data set based on the time stamp of the first acquired data stream in the original data set as a reference to obtain an aligned data set; S1.3, performing threshold checksum logic verification based on original readings in the aligned data set, and marking data items with invalid verification results as missing to obtain a cleaning data set; s1.4, converting original readings in the cleaning data set into engineering units according to corresponding physical meanings to obtain a real-time standardized data set; And S1.5, outputting the real-time standardized data set to a state characteristic digital twin synchronization module.
4. The intelligent early warning control system for mixing uniformity of a food compound additive according to claim 3, wherein the step of processing based on a real-time standardized data set to obtain a fusion state feature vector and performing data synchronization comprises: s2.1, processing based on a real-time standardized data set to obtain a primary characteristic report; s2.2, fusing the primary characteristics in the primary characteristic report by using a weighted fusion method to obtain a fusion state characteristic vector; S2.3, packaging the fusion state feature vector and the acquisition time stamp to obtain a state synchronization report; S2.4, outputting the state synchronization report to a digital twin hybrid process simulation prediction module.
5. The intelligent pre-warning control system for mixing uniformity of a food compounding additive according to claim 4, wherein the step of predicting the evolution of the mixing state within a preset future time period by running a mixing process simulation model based on a state synchronization report comprises: s3.1, based on a state synchronization report, calling a hybrid process simulation model according to a database, and calibrating an internal simulation clock of the hybrid process simulation model to an acquisition time stamp; inputting the fusion state feature vector in the state synchronization report into a hybrid process simulation model, and taking the fusion state feature vector as an initial state of the hybrid process simulation model; S3.2, retrieving production formula parameters of the current batch and current setting parameters of equipment based on a database to obtain an actual parameter report; s3.3, running a mixed process simulation model based on the fusion state feature vector and the actual parameter report, and outputting to obtain a mixed state prediction report; and S3.4, outputting the mixed state prediction report to a dynamic early warning model updating learning module.
6. The intelligent pre-warning control system for mixing uniformity of a food compound additive according to claim 4, wherein the steps of monitoring deviation of a mixed state prediction report from a state synchronization report and updating a dynamic pre-warning model comprise: s4.1, based on a state synchronization report, whenever the state feature digital twin synchronization module generates a new state feature digital twin synchronization module, invoking a fusion state feature vector corresponding to the new state synchronization report time in the mixed state prediction report, and performing deviation calculation to obtain a prediction deviation; S4.2, continuously monitoring a predicted deviation sequence formed by predicted deviations, comparing according to a deviation threshold, generating a model updating signal when N continuous predicted deviations are larger than the deviation threshold, and entering into a step S4.3; S4.3, updating the dynamic early warning model based on the model updating signal to obtain a new dynamic early warning model; And S4.4, replacing the original dynamic early-warning model with the new dynamic early-warning model.
7. The intelligent pre-warning control system for mixing uniformity of a food compounding additive according to claim 4, wherein the steps of analyzing the integrated status synchronization report, the mixed status prediction report and the new dynamic pre-warning model comprise: S5.1, inputting the fusion state feature vector in the state synchronization report into a new dynamic early warning model, and outputting to obtain a risk score; S5.2, inputting all fusion state feature vectors in the mixed state prediction report into a new dynamic early warning model respectively, outputting to obtain a predicted risk score sequence, and taking the largest risk score in the predicted risk score sequence as a predicted risk score; S5.3, based on the risk score and the predicted risk score, making a decision according to the early warning threshold intervals (Q1, Q2) to obtain a risk early warning report; S5.4, based on the risk early warning report, performing traceability analysis when a serious alarm report or an early warning alarm report exists in the risk early warning report to obtain a traceability analysis report; s5.5, packaging a risk early warning report and a traceability analysis report to obtain a comprehensive early warning diagnosis report; And S5.6, outputting the comprehensive early warning diagnosis report to the adaptive feedback optimization coordination module.
8. The intelligent pre-warning control system for mixing uniformity of a food compounding additive according to claim 5, wherein the step of calculating based on a mixing process prediction report and setting up an optimization target comprises: S6.1, setting an optimization target based on a database, and loading constraint conditions; S6.2, solving control sequences of H moments in the future by using a gradient descent method and taking the state synchronous report of the step S2.4 as an event starting point, and sequencing according to the order of objective function values from small to large to obtain a control optimization sequence; s6.3, taking the first control quantity in the control optimization sequence as a feedforward control instruction to obtain a feedforward control instruction report; And S6.4, outputting a feedforward control instruction report to the adaptive feedback optimization coordination module.
9. The intelligent pre-warning control system for mixing uniformity of a food compounding additive according to claim 7, wherein the step of analyzing based on a feedforward control instruction report, a comprehensive pre-warning diagnostic report, and a status synchronization report comprises: S7.1, reading a comprehensive early warning diagnosis report, and generating a feedback compensation instruction when a risk early warning report in the comprehensive early warning diagnosis report is a serious warning report or a prompt report and the content in a traceable analysis report is an adjustment parameter; S7.2, comparing the feedback compensation instruction with the feedforward control instruction report, and directly fusing when the feedback compensation instruction and the feedforward control instruction report have no conflict, so as to obtain a final execution instruction report; When the feedback compensation instruction and the feedforward control instruction report have conflict, the final execution instruction report is obtained after instruction adjustment; and S7.3, outputting a final execution instruction report to the process control system for execution.
10. An intelligent early warning control method for the mixing uniformity of a food compound additive, which is realized by the intelligent early warning control system for the mixing uniformity of the food compound additive according to any one of claims 1 to 9, is characterized by comprising the following working steps: S1, acquiring an original data stream in real time based on a sensor network, and preprocessing data to obtain a real-time standardized data set; S2, processing based on a real-time standardized data set to obtain a fusion state feature vector, and performing data synchronization to obtain a state synchronization report; s3, based on the state synchronization report, running a mixed process simulation model to predict the evolution of the mixed state in a preset future time period, and obtaining a mixed state prediction report; s4, monitoring deviation between the mixed state prediction report and the state synchronization report, and updating the dynamic early warning model to obtain a new dynamic early warning model; s5, analyzing the comprehensive state synchronous report, the mixed state prediction report and the new dynamic early warning model to obtain a comprehensive early warning diagnosis report; S6, based on the mixed process prediction report, setting an optimization target for calculation to obtain a feedforward control instruction report; And S7, analyzing based on the feedforward control instruction report, the comprehensive early warning diagnosis report and the state synchronization report to obtain a final execution instruction report.

Description

Intelligent early warning control system for mixing uniformity of food compound additive Technical Field The invention relates to the technical field of intelligent early warning control, in particular to an intelligent early warning control system for mixing uniformity of a food compound additive. Background The intelligent early warning control system for the mixing uniformity of the food compound additive is an advanced process control system integrating the functions of real-time monitoring, intelligent analysis, early warning and active intervention, and mainly aims to ensure that a plurality of food additives reach a highly uniform dispersion state in mixing equipment, so that the consistency of the final food product in terms of safety, stability and quality is ensured. However, the traditional intelligent early warning control system for mixing uniformity of the food compound additive mostly has the following defects in the using process, firstly, the traditional system mostly depends on single sensor equipment or final detection, and lacks the capability of acquiring defects of mixing uniformity in the mixing process in real time, so that the situation that the additive is unevenly mixed can be found after products are processed, enterprise benefits and food safety are affected, secondly, the traditional system early warning is based on a fixed threshold value or a static statistical model, the mixing process is easily influenced by a plurality of factors such as environment, raw materials and equipment, the false alarm rate and the false alarm rate of the traditional system are high, the early warning reliability of the traditional system is poor, thirdly, the traditional system control strategy mostly adopts simple PID mechanism feedback based on deviation and is mutually independent with the early warning system, therefore, the control mode of the traditional system is mostly monitored to alarm and then to manual control processing, the time cost of coping is greatly increased, and quality problems can be easily met through manual operation after the quality problems have occurred, so that the traditional system early warning is difficult to solve the problems of poor quality control performance, and poor quality control performance of the traditional system is difficult to effectively solve the problems of the traditional intelligent early warning system. In view of the above, the invention provides an intelligent early warning control system for mixing uniformity of a food compound additive to solve the problems. Disclosure of Invention In order to overcome the above drawbacks of the prior art, the present invention provides the following technical solutions, including: the multi-source data collaborative acquisition preprocessing module is used for acquiring an original data stream in real time based on a sensor network and preprocessing the data to obtain a real-time standardized data set; Further, the steps of collecting the original data stream in real time based on the sensor network and preprocessing the data comprise the following steps: s1.1, collecting data streams from different sensors in a sensor network in real time to obtain an original data set, wherein the data streams comprise a collection time stamp, a sensor ID and original readings; S1.2, maintaining a global clock by the system, and aligning acquisition time stamps of other data streams in the original data set based on the time stamp of the first acquired data stream in the original data set as a reference to obtain an aligned data set; S1.3, performing threshold checksum logic verification based on original readings in the aligned data set, and marking data items with invalid verification results as missing to obtain a cleaning data set; s1.4, converting original readings in the cleaning data set into engineering units according to corresponding physical meanings to obtain a real-time standardized data set; S1.5, outputting the real-time standardized data set to a state characteristic digital twin synchronization module; the state feature digital twin synchronization module is used for processing based on the real-time standardized data set to obtain a fusion state feature vector, and performing data synchronization to obtain a state synchronization report; Further, S2.1, processing based on the real-time standardized data set to obtain a primary characteristic report; s2.2, fusing the primary characteristics in the primary characteristic report by using a weighted fusion method to obtain a fusion state characteristic vector; S2.3, packaging the fusion state feature vector and the acquisition time stamp to obtain a state synchronization report; s2.4, outputting a state synchronization report to a digital twin hybrid process simulation prediction module; The digital twin hybrid process simulation prediction module is used for predicting the evolution of the hybrid state in a preset future time period based on the state sy