CN-121326021-B - Intelligent temperature control system and method for circulation heating of frying oil

CN121326021BCN 121326021 BCN121326021 BCN 121326021BCN-121326021-B

Abstract

The invention relates to the technical field of food industrial processing and intelligent manufacturing, in particular to an intelligent temperature control system and method for circulating heating of frying oil, comprising the steps of collecting oil temperature simulation signals and pipeline oil pressure changes by a sensing acquisition module to generate a multidimensional data set; the prediction decision module generates a preliminary control instruction; the self-adaptive learning module constructs a digital twin model, analyzes the thermodynamic behavior of the simulated oil flow through finite elements, and carries out global parameter optimization by adopting a genetic algorithm. The invention solves the problem of large fluctuation of the oil temperature caused by the on-off type temperature control strategy, realizes accurate and stable control of the oil temperature in continuous industrial production, and improves the consistency of the fried food quality and the food safety.

Inventors

Xie Chenpeng
LI HAOYU
GONG HONGYANG
YANG QINGZHE
WANG HAOZHE
LI JINLONG

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 intelligent temperature control system for circularly heating frying oil is characterized by comprising a physical equipment device and a control device; The physical equipment device comprises a fryer, an oil storage tank, an oil outlet pipeline and an oil return pipeline which are connected with the fryer and the oil storage pipeline, a circulating pump, an oil filter, a heating boiler, a proportional burner, a temperature sensor array, a pressure sensor, an electric flow control valve, a frequency converter and an electric executing mechanism; The control device establishes real-time communication connection with the physical equipment device through an industrial bus, and comprises a sensing acquisition module, a prediction decision module, a collaborative optimization module and a self-adaptive learning module which are sequentially used for data interaction; The sensing acquisition module acquires an oil temperature analog signal through the temperature sensor array, monitors the oil pressure change of a pipeline through the pressure sensor, converts the acquired analog signal into a digital signal, and generates a multidimensional data set through filtering and feature extraction; The prediction decision module receives the multi-dimensional data set, performs dimension reduction processing on the multi-dimensional data set to extract key characteristic parameters, performs rolling optimization calculation based on a model prediction control framework and the key characteristic parameters, predicts an oil temperature change track, and generates a preliminary control instruction by adopting a fuzzy PID composite control algorithm; The collaborative optimization module receives the preliminary control command, decomposes the preliminary control command into a power control quantity aiming at the proportional burner, a rotating speed control quantity aiming at the circulating pump and an opening control quantity aiming at the electric flow control valve, adjusts the heating power of the proportional burner by adopting a pulse width modulation and pulse frequency modulation mixing technology, adjusts the motor rotating speed of the circulating pump by a vector control algorithm and adjusts the opening of the electric flow control valve based on a fuzzy control rule, introduces a multi-objective particle swarm optimization algorithm, takes temperature stability, energy consumption minimization and oil protection as optimization targets, calculates the optimal weights of the power control quantity, the rotating speed control quantity and the opening control quantity, and sends a control signal carrying the optimal weights to the physical equipment device by the electric actuator; The self-adaptive learning module receives the multidimensional dataset from the sensing acquisition module, builds a digital twin model of a system physical entity, analyzes the thermodynamic behavior of the simulated oil flow through finite elements, and online updates parameters of the digital twin model by adopting a system identification technology, the self-adaptive learning module compares the multidimensional dataset with a prediction result of the digital twin model in real time, detects system abnormality by using a state estimation algorithm, builds a control experience library based on a long-short-term memory neural network, and carries out global parameter optimizing through a genetic algorithm, and the self-adaptive learning module feeds back an optimization result to the prediction decision module for correcting a prediction model in a model prediction control framework and feeds back a learning result to the collaborative optimization module for updating a control strategy.
2. The intelligent temperature control system for hydronic heating of frying oil of claim 1, wherein the sensing acquisition module comprises: the temperature sensor array synchronously collects oil temperature analog signals at millisecond sampling frequency, and the pressure sensor monitors the oil pressure change of a pipeline to generate pressure analog signals; The sensing acquisition module takes the oil temperature analog signal and the pressure analog signal as acquired analog signals, and inputs the acquired analog signals into a high-precision A/D conversion circuit to be converted into digital signals; the sensing acquisition module executes a Kalman filtering algorithm on the digital signal through a digital signal processor to inhibit random noise, and performs multi-scale decomposition on the filtered digital signal by adopting a wavelet transformation technology to extract temperature mutation characteristics and long-term trends; In the signal acquisition and processing process, the sensing acquisition module integrates a self-adaptive sampling mechanism, namely dynamically adjusting the sampling frequency of the temperature sensor array according to the temperature change rate, automatically increasing the sampling frequency when the temperature change rate exceeds a preset threshold value, and reducing the sampling frequency in a temperature stabilization stage; The sensing acquisition module generates a multi-dimensional dataset based on the processed signals.
3. The intelligent temperature control system of claim 1, wherein the predictive decision module is configured to: Receiving the multi-dimensional data set, performing dimension reduction on the multi-dimensional data set by using a principal component analysis method, calculating eigenvalues and eigenvectors of a covariance matrix, and reserving principal components with highest contribution rate to extract key eigenvalues; Based on a model predictive control framework, taking the key characteristic parameters as input, performing rolling optimization calculation by using a system transfer function model, and predicting an oil temperature change track in a future time domain; on the basis of the rolling optimization calculation, a fuzzy PID composite control algorithm is adopted to generate a preliminary control instruction, wherein the fuzzy PID composite control algorithm adopts a Mamdani fuzzy reasoning system, the input variable is temperature deviation and deviation change rate, the output variable is PID parameter adjustment quantity, and the proportional coefficient, the integral time and the differential time are dynamically adjusted through a fuzzy rule base; Meanwhile, a deep reinforcement learning agent is integrated, a state space, an action space and a reward function are defined by adopting a Q learning algorithm, an optimal control strategy is learned by iteratively updating a Q value table, and the output of the deep reinforcement learning agent is used for optimizing the parameter adjustment of the fuzzy PID composite control algorithm.
4. The intelligent temperature control system of claim 1, wherein the co-optimization module is configured to: receiving the preliminary control command, and decomposing the preliminary control command into a power control amount for the proportional burner, a rotation speed control amount for the circulating pump and an opening control amount for the electric flow control valve; the heating power of the proportional burner is regulated by adopting a pulse width modulation and pulse frequency modulation mixing technology, the rotating speed of a motor is regulated by a vector control algorithm for the circulating pump, and the opening of the electric flow control valve is regulated based on a fuzzy control rule; the method comprises the steps of introducing a multi-target particle swarm optimization algorithm, calculating optimal weights of the power control quantity, the rotating speed control quantity and the opening control quantity by taking temperature stability, energy consumption minimization and oil protection as optimization targets, initializing a particle swarm by the multi-target particle swarm optimization algorithm, iteratively updating the position and the speed of each particle to find a Pareto optimal solution set by calculating a fitness value, wherein each particle represents a group of control weights; And sending a control signal carrying the optimal weight to the physical equipment device.
5. The intelligent temperature control system of claim 1, wherein the adaptive learning module is configured to: Receiving the multi-dimensional dataset from the sensing acquisition module, and constructing a digital twin model of a system physical entity based on the multi-dimensional dataset; analyzing the thermodynamic behavior of the simulated oil flow through finite elements, and solving a Navier-Stokes equation and an energy equation to predict oil temperature distribution and flow characteristics; On-line estimating heat conduction coefficient and flow resistance parameters by adopting a recursive least square method, and updating parameters of the digital twin model; Carrying out state estimation by using an extended Kalman filtering algorithm, linearizing a nonlinear model of the system, and detecting system abnormality through a prediction and correction step; based on a long-term and short-term memory neural network, a control experience library is established, an LSTM unit is adopted to capture the long-term dependence of a time sequence, and offline training and online fine tuning are carried out through a back propagation algorithm; Global parameter optimizing is carried out through a genetic algorithm, a population is initialized, control parameters are encoded, and an optimal parameter combination is found through selection, crossing and mutation operation of the population; And feeding back an optimization result to the prediction decision module for correcting a prediction model in the model prediction control framework, and feeding back a learning result to the collaborative optimization module for updating a control strategy.
6. The intelligent temperature control system for hydronic heating of frying oil of claim 1, further comprising: The prediction decision module packages the preliminary control instruction into a structured data packet, wherein the structured data packet comprises a time stamp, an instruction type, a control parameter and a priority identifier; The prediction decision module transmits the structured data packet to the collaborative optimization module through a real-time data bus; The collaborative optimization module receives the structured data packet and then performs instruction analysis and conflict detection to verify the feasibility of the instruction; the collaborative optimization module feeds back an execution result to the prediction decision module, wherein the execution result comprises an instruction execution state, an actual output value and a deviation of an expected target; and the prediction decision module adjusts control parameters according to the execution result, and corrects a membership function of a fuzzy rule base or a reinforcement learning reward function.
7. The intelligent temperature control system for hydronic heating of frying oil of claim 1, further comprising: Acquiring real-time operation data from the sensing acquisition module, acquiring a control instruction sequence from the prediction decision module, and acquiring an actuator state from the collaborative optimization module; Constructing a digital twin model input based on the real-time operation data, the control instruction sequence and the actuator state, performing simulation verification and completing model correction; Generating a virtual simulation result through the digital twin model, wherein the virtual simulation result comprises a temperature prediction track and system response characteristics under different control strategies; the virtual simulation result is sent to the prediction decision module and used for correcting the rolling optimization parameters in the model prediction control framework; Generating an actuator optimization suggestion based on historical data analysis, wherein the actuator optimization suggestion comprises a valve opening and temperature relation curve; And sending the actuator optimization suggestion to the collaborative optimization module for guiding weight distribution in the multi-objective particle swarm optimization algorithm.
8. The intelligent temperature control system for hydronic heating of frying oil of claim 1, further comprising: When any one of the sensing acquisition module, the prediction decision module, the collaborative optimization module or the self-adaptive learning module detects an abnormal state, an alarm is sent to the other three modules through a priority interrupt channel; The prediction decision module starts an emergency control mode after receiving the abnormal alarm, and a preset simplified control strategy is adopted to maintain the basic operation of the system; The collaborative optimization module redistributes control tasks after receiving the abnormal alarm, reduces dependence on an abnormal executor and adjusts control weights; The self-adaptive learning module carries out fault influence simulation after receiving the abnormal alarm, predicts the system behavior change and generates a coping scheme; The sensing acquisition module, the prediction decision module, the collaborative optimization module and the self-adaptive learning module maintain communication state monitoring through a heartbeat detection mechanism, and periodically exchange state information; When a module communication interruption is detected, the system automatically downgrades to a standby control mode, and the basic control function is maintained by the remaining normal modules.
9. The intelligent temperature control system for hydronic heating of frying oil of claim 1, further comprising: Under the impact of feeding, the collaborative optimization module receives a temperature change signal and simultaneously generates a power adjustment instruction for the proportional burner, a rotating speed adjustment instruction for the circulating pump and an opening adjustment instruction for the electric flow control valve; Adopting a multi-target particle swarm optimization algorithm, taking temperature stability as a priority target, combining energy consumption minimization and oil protection as optimization targets, and calculating the optimal weight combination of the power adjustment instruction, the rotating speed adjustment instruction and the opening adjustment instruction; applying the optimal weight combination to control execution by executing the power adjustment instruction using a pulse width modulation and pulse frequency modulation hybrid technique to adjust heating power of the proportional burner; Executing the rotating speed regulating instruction by using a vector control algorithm, decoupling three-phase current into a direct axis component and a quadrature axis component through Clarke transformation and Park transformation, respectively controlling excitation and torque, and accurately regulating the motor rotating speed of the circulating pump; And executing the opening adjustment instruction by using a fuzzy control rule, and dynamically adjusting the opening of the electric flow control valve according to the temperature deviation and the oil flow characteristic to finish the mixing of the high-temperature oil and the return oil.
10. The intelligent temperature control method for circulation heating of frying oil, which is applied to the intelligent temperature control system for circulation heating of frying oil according to any one of claims 1 to 9, is characterized by comprising the following steps: the method comprises the steps of 1, synchronously collecting oil temperature analog signals at millisecond sampling frequency through a temperature sensor array, and monitoring pipeline oil pressure change through a pressure sensor to generate pressure analog signals; Step 2, performing dimension reduction processing on the multidimensional data set to extract key characteristic parameters, performing rolling optimization calculation on the basis of a model prediction control frame and the key characteristic parameters, and predicting an oil temperature change track; Step 3, decomposing the preliminary control instruction into a power control quantity aiming at a proportional combustor, a rotating speed control quantity aiming at a circulating pump and an opening control quantity aiming at an electric flow control valve; Step 4, based on the optimal weight, adopting a pulse width modulation and pulse frequency modulation mixing technology to adjust the heating power of the proportional burner, adjusting the motor rotating speed of the circulating pump through a vector control algorithm, and adjusting the opening of the electric flow control valve based on a fuzzy control rule; Step 5, constructing a digital twin model of the system, analyzing the thermodynamic behavior of the simulated oil flow through finite elements, adopting a recursive least square method to update model parameters on line, and carrying out state estimation and anomaly detection by using an extended Kalman filtering algorithm; step 6, establishing a control experience library based on the long-short-term memory neural network, and performing global parameter optimization through a genetic algorithm; Step 7, packaging the preliminary control instruction into a structured data packet comprising a time stamp, an instruction type, a control parameter and a priority identifier, transmitting the preliminary control instruction through a real-time data bus, and carrying out instruction analysis and conflict detection; Step 8, acquiring real-time operation data, a control instruction sequence and an actuator state, constructing a digital twin model based on the data for input and simulation verification, generating a virtual simulation result and sending the virtual simulation result to correct rolling optimization parameters, analyzing and generating an actuator optimization suggestion based on historical data and sending the actuator optimization suggestion to guide weight distribution; And 9, sending an alarm when an abnormal state is detected, starting an emergency control mode to adopt a simplified control strategy, reassigning a control task to adjust a control weight, performing fault influence simulation prediction system behavior change, maintaining communication state monitoring through a heartbeat detection mechanism, and automatically degrading to a standby control mode when communication interruption is detected.

Description

Intelligent temperature control system and method for circulation heating of frying oil Technical Field The invention relates to the technical field of food industrial processing and intelligent manufacturing, in particular to an intelligent temperature control system and method for circulating heating of frying oil. Background The frying oil circulation heating system adopts an intelligent temperature control technology, acquires heat data of frying oil in real time through a temperature sensor and transmits signals to a microprocessor, and the microprocessor carries out logic analysis based on a preset temperature threshold value to generate a control instruction to drive a heating element to adjust output power, so that the oil temperature is dynamically balanced in the circulation process, and a stable heating state is maintained to adapt to requirements of different frying technologies. The existing intelligent temperature control of frying oil circulation heating has the technical pain point that the on-off control strategy based on the bimetallic strip temperature controller cuts off a heating source when the temperature reaches a set upper limit because of inherent defects of the working principle of the on-off control strategy, and starts heating with full power when the temperature falls back to a set lower limit. The discontinuous step control mode causes obvious control lag and inertia of the system, so that the oil temperature continuously shows periodical and large fluctuation above and below a set value. In continuous industrial frying production, when a large amount of low-temperature food materials are put into, the system cannot provide smooth gradual power to perform temperature compensation, and the unstable state of the oil temperature is aggravated due to severe start-stop operation. The fluctuation directly causes the difference of color, crispness and oil content of foods in different batches, influences the consistency of the products, accelerates the cracking and oxidation of the edible oil when the oil temperature frequently spans the optimal temperature interval, increases the generation risk of harmful substances, and forms a potential threat to the safety of the foods. Disclosure of Invention Aiming at the defects of the prior art, the invention provides an intelligent temperature control system and an intelligent temperature control method for circulation heating of frying oil, which solve the technical problem that the oil temperature greatly fluctuates near a set value due to the adoption of an on-off type temperature control strategy in the conventional frying equipment, and the continuous industrial production with strict requirements on the consistency of the quality of fried foods and the safety of the foods can not be met. In order to solve the technical problems, the invention comprises the following specific contents: in a first aspect, the intelligent temperature control system for circulating heating of frying oil comprises a physical equipment device and a control device; The physical equipment device comprises a fryer, an oil storage tank, an oil outlet pipeline and an oil return pipeline which are connected with the fryer and the oil storage pipeline, a circulating pump, an oil filter, a heating boiler, a proportional burner, a temperature sensor array, a pressure sensor, an electric flow control valve, a frequency converter and an electric executing mechanism; The control device establishes real-time communication connection with the physical equipment device through an industrial bus, and comprises a sensing acquisition module, a prediction decision module, a collaborative optimization module and a self-adaptive learning module which are sequentially used for data interaction; The sensing acquisition module acquires an oil temperature analog signal through the temperature sensor array, monitors the oil pressure change of a pipeline through the pressure sensor, converts the acquired analog signal into a digital signal, and generates a multidimensional data set through filtering and feature extraction; The prediction decision module receives the multi-dimensional data set, performs dimension reduction processing on the multi-dimensional data set to extract key characteristic parameters, performs rolling optimization calculation based on a model prediction control framework and the key characteristic parameters, predicts an oil temperature change track, and generates a preliminary control instruction by adopting a fuzzy PID composite control algorithm; The collaborative optimization module receives the preliminary control command, decomposes the preliminary control command into a power control quantity aiming at the proportional burner, a rotating speed control quantity aiming at the circulating pump and an opening control quantity aiming at the electric flow control valve, adjusts the heating power of the proportional burner by adopting a pulse width modulation a