CN-121115655-B - Peanut butter production line weighing data processing system and method based on PLC control

CN121115655BCN 121115655 BCN121115655 BCN 121115655BCN-121115655-B

Abstract

The invention relates to the technical field of food industrial processing and intelligent processing, in particular to a peanut butter production line weighing data processing system and method based on PLC control, which comprises a data acquisition module for capturing an original signal by adopting a sliding window mechanism, a dynamic preprocessing module for wavelet transformation denoising and Kalman filtering compensation, the digital twin modeling module builds production process digital mapping, the multi-objective optimization module processes multi-objective optimization by using a non-dominant ordering genetic algorithm to generate a Pareto solution set, the self-adaptive control module adjusts PLC controller parameters through a model predictive control algorithm, and the decision support module adopts a federal learning framework to generate a raw material classification model and a process parameter strategy. The modules form a closed-loop feedback system through the message middleware, so that the raw material loss quantization, the process optimization and the cost control are realized.

Inventors

HAN YONGZENG
PAN SHENGBO
Zhan Bangfan
YANG QINGZHE
LI HAOYU

Assignees

广州杰尔古格食品有限公司
杰尔古格智能科技有限公司

Dates

Publication Date: 20260505
Application Date: 20251029

Claims (10)

1. The peanut butter production line weighing data processing system based on PLC control is characterized by comprising a physical equipment device and a control device, wherein the control device is in real-time communication connection with the physical equipment device through an industrial Ethernet; The physical equipment device comprises a distributed weighing sensor array, an environment temperature and humidity sensor, a PLC (programmable logic controller) cluster, a variable frequency conveying mechanism, intelligent baking equipment, self-adaptive grinding equipment and a man-machine interaction terminal, wherein the distributed weighing sensor array is arranged at a raw material input station, a screening and stoning station, a baking and stoning station and a rough grinding station, and each station is provided with a temperature compensation unit and a vibration suppression module; The control device comprises a data acquisition module, a dynamic preprocessing module, a digital twin modeling module, a multi-objective optimization module, a self-adaptive control module and a decision support module; The data acquisition module captures original signals of the distributed weighing sensor array in real time by adopting a sliding window mechanism, acquires environmental temperature and humidity data and outputs the original data; The dynamic preprocessing module receives the original data, performs wavelet transform denoising and Kalman filtering dynamic compensation on the original data, and outputs a stable weight sequence; The digital twin modeling module receives the stable weight sequence, constructs digital mapping of the production process, adopts a discrete event simulation technology to simulate the material flow process, and outputs simulation data; the multi-objective optimization module receives simulation data, and utilizes a non-dominant ordering genetic algorithm to process multi-objective optimization of minimizing raw material loss, maximizing product consistency and minimizing energy consumption so as to generate a Pareto solution set; The self-adaptive control module receives the Pareto solution set and dynamically adjusts the operation parameters of the PLC controller cluster through a model predictive control algorithm; the decision support module adopts a federal learning framework to aggregate historical production data, generates a raw material grading model and a process parameter recommendation strategy, and reversely transmits the strategy to the self-adaptive control module to correct control parameters; The digital twin modeling module sends simulation data to the multi-objective optimization module, the multi-objective optimization module feeds the Pareto solution set back to the digital twin modeling module, the simulation data and the Pareto solution set are compared in real time through a particle swarm optimization algorithm, an optimal working point is found, and the decision support module periodically sends updated constraint conditions to the self-adaptive control module to form a closed loop system with a feedback loop.
2. The peanut butter production line weighing data processing system based on the PLC control according to claim 1 is characterized in that the data acquisition module comprises a signal conditioning unit and a communication protocol analysis unit, wherein the signal conditioning unit processes millivolt-level signals output by a distributed weighing sensor array by adopting a differential amplification technology, realizes signal digitization through a high-precision analog-to-digital converter and outputs digital signals, the communication protocol analysis unit supports Modbus TCP and PROFINET industrial protocols, reads weight register values of a PLC controller cluster through a polling mechanism, acquires equipment operation status words and outputs analysis data, and the data acquisition module receives the digital signals and analysis data, combines the digital signals and the analysis data into original data and sends the original data into a first-in first-out buffer area for data alignment for eliminating sensor acquisition time difference.
3. The weighing data processing system of the peanut butter production line based on PLC control according to claim 2, wherein the dynamic preprocessing module comprises a signal processing unit and a data quality evaluation unit, wherein the signal processing unit receives original data, performs multi-layer decomposition to remove high-frequency noise by applying a wavelet basis function, outputs a denoised signal, the Kalman filtering unit receives the denoised signal, builds a state space model to perform recursive calculation to compensate a dynamic weighing process, outputs a filtered signal, the data quality evaluation unit receives the filtered signal, calculates variance and kurtosis indexes of the filtered signal, triggers a re-acquisition mechanism when data abnormality is detected, and the dynamic preprocessing module outputs the filtered signal as a stable weight sequence and issues the stable weight sequence to a message bus and simultaneously sends a data updating event to the digital twin modeling module.
4. The PLC control-based peanut butter production line weighing data processing system according to claim 3, wherein the digital twin modeling module comprises a physical entity modeling unit and a process simulation unit, wherein the physical entity modeling unit adopts a three-dimensional engine to construct an equipment model, defines the speed and baking temperature attribute of a conveyor belt and outputs the equipment model, the process simulation unit receives the equipment model and a stable weight sequence, builds a discrete event simulation model, defines peanut particles as an agent and simulates state migration of the agent at a process node, outputs simulation data, and the digital twin modeling module writes the simulation data into a shared memory area and sets a data valid flag bit to trigger the multi-objective optimization module.
5. The PLC control-based peanut butter production line weighing data processing system according to claim 4, wherein the multi-objective optimization module comprises an algorithm library and a constraint management unit, wherein the multi-objective optimization module periodically checks data valid flag bits in a shared memory area, reads simulation data from the shared memory area as an initial population when the flag bits are detected to be valid, the algorithm library realizes non-dominant sorting and congestion degree calculation on the initial population, generates new solutions by using crossover and mutation operators, the constraint management unit defines equipment capacity boundary conditions, guides the solutions which violate the constraint to a feasible domain, and the multi-objective optimization module processes multi-objective optimization with minimized raw material loss, maximized product consistency and minimized energy consumption through a non-dominant sorting genetic algorithm, generates a Pareto solution set, writes the Pareto solution set into an optimization result section of the shared memory area, and simultaneously sends an optimization completion event to the self-adaptive control module.
6. The PLC control-based peanut butter production line weighing data processing system according to claim 5, wherein the self-adaptive control module comprises a prediction model unit and a rolling optimization unit, the self-adaptive control module receives an optimization completion event, reads a Pareto solution set from an optimization result section of a shared memory area, establishes a recursive least square model with forgetting factors based on the Pareto solution set, predicts future output of the system, the rolling optimization unit receives a prediction result, solves a quadratic programming problem with constraint, generates an optimal control sequence and sends the optimal control sequence to the PLC controller cluster, and the self-adaptive control module collects actual control effect data and sends the actual control effect data to the decision support module.
7. The PLC control-based peanut butter production line weighing data processing system is characterized in that the decision support module comprises a federal learning unit and a knowledge graph unit, the decision support module receives actual control effect data sent by the self-adaptive control module, the federal learning unit adopts a transverse federal architecture, a neural network model is trained locally based on the actual control effect data and uploads model parameters to a server for safe aggregation, the knowledge graph unit receives the aggregated model parameters, a process parameter association rule is stored by a graph database, an implicit relation is mined through a graph traversal algorithm, the decision support module generates a raw material classification model and a process parameter recommendation strategy based on the mining result, the strategy is published to a message bus in a JSON format, and the digital twin modeling module subscribes the message bus and receives the strategy to update a simulation model business rule.
8. The PLC-based peanut butter production line weighing data processing system of claim 7, further comprising: The data acquisition module issues an original data event; the dynamic preprocessing module subscribes to an original data event, receives and processes the original data and issues a stable weight sequence event, the digital twin modeling module subscribes to a stable weight sequence event, receives and simulates the stable weight sequence and issues a simulation data event, the multi-objective optimization module subscribes to a simulation data event, receives simulation data and optimizes and issues an optimization completion event, the self-adaptive control module subscribes to an optimization completion event, receives a Pareto solution set and adjusts control parameters, issues a control effect data event, the decision support module subscribes to a control effect data event, receives control effect data and updates a federal learning model, issues a strategy event, the digital twin modeling module subscribes to a strategy event, receives a strategy and updates a simulation model business rule, and the data acquisition module, the dynamic preprocessing module, the digital twin modeling module, the multi-objective optimization module, the self-adaptive control module and the decision support module are decoupled through message middleware, and a message header comprises a sequence number and a timestamp to ensure a data sequence.
9. The peanut butter production line weighing data processing system based on PLC control according to claim 8, wherein six station data of the distributed weighing sensor array in the physical equipment device are cooperatively processed by a control device module, a digital twin modeling module sends simulation data to a multi-objective optimization module as an initial population, the multi-objective optimization module sends a Pareto solution set to an adaptive control module for establishing a prediction model, the adaptive control module sends actual control effect data to a decision support module for training a federal learning model, and the decision support module sends generated strategies to the digital twin modeling module for updating simulation model business rules.
10. The peanut butter production line weighing data processing method based on PLC control, which is applied to the peanut butter production line weighing data processing system based on PLC control as claimed in any one of claims 1 to 9, is characterized by comprising the following steps: the method comprises the steps of 1, capturing original signals of a distributed weighing sensor array in real time by a data acquisition module through a sliding window mechanism, acquiring environmental temperature and humidity data, and outputting the original data; Step2, receiving original data through a dynamic preprocessing module, carrying out wavelet transformation denoising and Kalman filtering dynamic compensation on the original data, and outputting a stable weight sequence; Step 3, receiving a stable weight sequence through a digital twin modeling module, constructing digital mapping of a production process, simulating a material flow process by adopting a discrete event simulation technology, and outputting simulation data; Step 4, receiving simulation data through a multi-objective optimization module, and processing multi-objective optimization with minimized raw material loss, maximized product consistency and minimized energy consumption by using a non-dominant sorting genetic algorithm to generate a Pareto solution set; step 5, receiving a Pareto solution set through an adaptive control module, and dynamically adjusting the operation parameters of the PLC controller cluster through a model predictive control algorithm; Step 6, aggregating historical production data by adopting a federal learning framework through a decision support module, generating a raw material grading model and a process parameter recommendation strategy, and reversely transmitting the strategy to a self-adaptive control module to correct control parameters; And 7, sending simulation data to a multi-objective optimization module through a digital twin modeling module, feeding the Pareto solution set back to the digital twin modeling module through the multi-objective optimization module, comparing the simulation data with the Pareto solution set in real time through a particle swarm optimization algorithm, and searching an optimal working point, and periodically sending updated constraint conditions to an adaptive control module by a decision support module to form a closed loop system with a feedback loop.

Description

Peanut butter production line weighing data processing system and method based on PLC control Technical Field The invention relates to the technical field of food industrial processing and intelligent processing, in particular to a weighing data processing system and method of a peanut butter production line based on PLC control. Background The existing peanut butter production line based on PLC control adopts a programmable logic controller as a system core, monitors the state of raw materials and the operation parameters of equipment through an integrated sensor, processes input signals according to preset program logic, outputs instructions to drive an executing mechanism such as a conveying belt, a heater and a stirring device, and accordingly automatically completes procedures such as cleaning, baking, grinding and mixing of peanuts, the coordination and reliability of the production flow are improved by the aid of the centralized control mode, and the production line can adapt to process changes due to flexible programming characteristics of the PLC, so that efficient and continuous operation is achieved. The existing peanut butter production line based on PLC control has the technical pain that the weight change of raw materials in key processes such as screening, baking, paste grinding and the like cannot be continuously recorded because the production line is not integrated with a multi-node real-time automatic weighing monitoring device. For example, the weight loss of the baked peanuts caused by the evaporation of water cannot be automatically compared with the initial weight, so that the influence of the water content difference of different batches of raw materials on the loss is difficult to quantify, and the peeling rate cannot be accurately calculated according to the variety difference of the peanuts due to the lack of real-time data of the weight occupancy of the skin before and after rough grinding in the peeling process of the peanuts. The breaking of the data chain causes the lack of a demonstration foundation for process optimization, so that decisions such as baking temperature adjustment, formula proportioning and the like depend on empirical estimation, and further cause the problems of cost accounting deviation and product flavor stability. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a peanut butter production line weighing data processing system and method based on PLC control, and solves the technical problems that raw material loss cannot be quantified, process optimization lacks data support and cost control precision is insufficient due to the lack of multi-node real-time automatic weighing monitoring in the peanut butter production process. In order to solve the technical problems, the invention comprises the following specific contents: The weighing data processing system of the peanut butter production line based on PLC control is characterized by comprising a physical equipment device and a control device, wherein the control device is in real-time communication connection with the physical equipment device through an industrial Ethernet; The physical equipment device comprises a distributed weighing sensor array, an environment temperature and humidity sensor, a PLC (programmable logic controller) cluster, a variable frequency conveying mechanism, intelligent baking equipment, self-adaptive grinding equipment and a man-machine interaction terminal, wherein the distributed weighing sensor array is arranged at a raw material input station, a screening and stoning station, a baking and stoning station and a rough grinding station, and each station is provided with a temperature compensation unit and a vibration suppression module; The control device comprises a data acquisition module, a dynamic preprocessing module, a digital twin modeling module, a multi-objective optimization module, a self-adaptive control module and a decision support module; The data acquisition module captures original signals of the distributed weighing sensor array in real time by adopting a sliding window mechanism, acquires environmental temperature and humidity data and outputs the original data; The dynamic preprocessing module receives the original data, performs wavelet transform denoising and Kalman filtering dynamic compensation on the original data, and outputs a stable weight sequence; The digital twin modeling module receives the stable weight sequence, constructs digital mapping of the production process, adopts a discrete event simulation technology to simulate the material flow process, and outputs simulation data; the multi-objective optimization module receives simulation data, and utilizes a non-dominant ordering genetic algorithm to process multi-objective optimization of minimizing raw material loss, maximizing product consistency and minimizing energy consumption so as to generate a Pareto solution set; The self-adaptive control module r