Intelligent control method and system for flour workshop

Abstract

The invention relates to the technical field of flour processing, and discloses an intelligent control method and system for a flour workshop, and the method comprises the following steps: collecting raw material characteristic data and equipment operation state data in real time through a sensor, calculating the state variable of the current system according to the collected data, and calculating the state variable of the current system according to the current state variable. Performing rolling optimization on a control strategy in a future time domain, dynamically adjusting grinding parameters according to a rolling optimization result, correcting a prediction model according to a data deviation between real-time acquired data and nonlinear model prediction data, and calculating an optimal control parameter combination based on a stochastic optimization theory; the system comprises a sensing module, a control module, an execution module, a feedback correction module and a random optimization module. By adopting the technical scheme that a high-precision sensor is combined with intelligent control, comprehensive automation and accurate control of the flour production process are achieved. The stability and consistency of flour quality are remarkably improved, and quality fluctuation is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of flour processing, and in particular to an intelligent control method and system for a flour workshop. Background Art

[0002] Flour production is a crucial component of the grain processing industry. Traditional flour mills typically rely on manual labor to complete multiple steps, including raw material processing, grinding, and screening. With technological advancements, flour production processes are gradually moving towards automation and intelligence. Despite this, many traditional production lines still rely heavily on manual intervention and face technical bottlenecks in the control of various links. To improve production efficiency and ensure consistent product quality, flour mills need to introduce intelligent control technologies for flour workshops.

[0003] Existing intelligent control technology for flour mills relies on fixed process flows and manual adjustments, combined with simple sensors and mechanical devices for production management. Some high-end flour mills are beginning to adopt PLCs (programmable logic controllers) and other automated equipment to achieve simple process control. These devices can achieve a certain degree of automation, reducing the need for human intervention.

[0004] However, existing intelligent control technologies for milling plants are overly simplistic, relying on experience-based adjustments rather than data-driven ones. This makes it difficult to adjust parameters during the production process in real time, leading to large fluctuations in flour quality. Furthermore, the coordination mechanism between existing equipment is poor, and equipment often cannot adapt to each other's operating conditions, resulting in an unsmooth production process and even energy waste and severe equipment wear. The prediction and handling of equipment failures in existing intelligent control systems for milling plants primarily rely on manual inspections, which often fail to detect potential problems in a timely manner, leading to production interruptions and equipment damage, affecting overall production efficiency and cost control. Therefore, the present invention provides an intelligent control method and system for a flour milling plant to address the shortcomings of the existing technology. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides an intelligent control method and system for a flour workshop, which solves the problem that the existing workshop intelligent control technology is too single in control mode, relies on experience adjustment rather than data-driven, and is difficult to adjust parameters in the production process in real time, resulting in large fluctuations in flour quality.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A flour workshop intelligent control method, comprising the following steps: Collect raw material characteristic data and equipment operation status data in real time through sensors; According to the collected data, the state variables of the current system are calculated through the nonlinear grinding dynamics model; According to the current state variables, the control strategy in the future time domain is optimized through the nonlinear model predictive control algorithm; According to the results of rolling optimization, the grinding parameters are dynamically adjusted through the actuator; Correct the prediction model based on the data deviation between the real-time collected data and the nonlinear model prediction data; Based on stochastic optimization theory, the optimal control parameter combination is calculated by simulating the random fluctuation of raw material characteristic data.

[0007] Preferably, the raw material characteristic data includes raw material moisture and hardness, and the equipment operation status data includes grinding roller speed, grinding roller pressure and equipment temperature.

[0008] Preferably, the state variables include flour particle size, grinding roller speed, grinding roller pressure and equipment temperature, and the nonlinear grinding dynamics model describes the relationship between state variables, control variables and random variables.

[0009] Preferably, the nonlinear model predictive control algorithm is implemented in the following manner: Predict future states based on nonlinear grinding dynamics model and stochastic process model; At each time step, an optimization problem is solved to minimize the expected total energy consumption while satisfying the probabilistic constraint of flour particle size distribution.

[0010] Preferably, the objective function of the optimization problem is the expected total energy consumption, which is defined as the integral of the instantaneous power function in the prediction time domain, and the instantaneous power function is related to the roller spacing adjustment amount, the motor voltage and the air circuit valve opening.

[0011] Preferably, the grinding parameters include grinding roller speed, roller spacing and air circuit valve opening, which are dynamically adjusted through the motor drive and speed regulation system, roller spacing adjustment device and air circuit valve adjustment device.

[0012] Preferably, the correction prediction model calculates the deviation between the actual measured value and the predicted value, and uses a correction gain matrix to correct the prediction model.

[0013] Preferably, the stochastic optimization theory is implemented in the following way: Generate multiple random scenarios based on the random process model; Solve the optimization problem for each scenario and obtain the control strategy; Calculate the expected control strategy for all scenarios to ensure the robustness of the system under random fluctuations in raw material characteristics.

[0014] Preferably, the results of the rolling optimization include the following: Roller gap adjustment, used to adjust the gap between the grinding rollers to achieve the target flour particle size; Motor voltage, used to control the grinding roller speed to match current production needs; The air path valve opening is used to adjust the air flow speed to optimize the screening efficiency.

[0015] A flour workshop intelligent control system is also provided, including: The sensing module is used to collect raw material characteristic data and equipment operation status data in real time through sensors; A control module is used to estimate the state variables of the current system based on the nonlinear grinding dynamics model and to optimize the control strategy in the future time domain through a nonlinear model predictive control algorithm; An execution module for dynamically adjusting grinding parameters according to the results of rolling optimization; The feedback correction module is used to correct the prediction model according to the data deviation between the real-time collected data and the nonlinear model prediction data; the random optimization module is used to calculate the optimal control parameter combination by simulating the random fluctuation of the raw material characteristic data.

[0016] The present invention provides an intelligent control method and system for a flour workshop, which has the following beneficial effects: 1. This invention combines high-precision sensors with intelligent control to achieve comprehensive automation and precise control of the flour production process. Compared to traditional flour mills that rely on manual operation and empirical judgment, this method eliminates the errors and inconsistencies caused by manual operation, significantly improves the stability and consistency of flour quality, and avoids quality fluctuations.

[0017] 2. This invention optimizes the grinding process through dynamic adjustment of grinding parameters and real-time feedback control, ensuring precision and uniformity in each grinding. Unlike existing technologies that fix grinding parameters, this invention automatically adjusts the operating conditions of the grinding rollers based on actual production conditions and flour particle size requirements, significantly reducing energy consumption during the grinding process and improving production efficiency.

[0018] 3. This invention incorporates intelligent fault prediction and maintenance technology to proactively identify potential equipment issues and automatically issue warnings, thus avoiding downtime caused by sudden equipment failures. Compared to traditional flour mills, which rely primarily on manual inspections and passive maintenance, this invention enables the development of intelligent maintenance plans in advance, reducing downtime and improving equipment reliability and production continuity.

[0019] 4. This invention achieves energy-saving control of production equipment through an intelligent energy management system. By precisely controlling motor operating status and equipment start and stop timing, energy waste is effectively reduced. Compared to the energy waste caused by idling or inefficient equipment operation in traditional production models, this invention can flexibly adjust energy distribution according to production needs, significantly reducing production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of the method steps of the present invention; Figure 2 This is a system architecture diagram of the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0022] Please see the attached Figure 1 The embodiment of the present invention provides a flour workshop intelligent control method, comprising the following steps: S1, collect raw material characteristic data and equipment operation status data in real time through sensors; S2. Calculate the state variables of the current system through a nonlinear grinding dynamics model based on the collected data; S3. Based on the current state variables, the control strategy in the future time domain is optimized through the nonlinear model predictive control algorithm; S4. Dynamically adjust the grinding parameters through the actuator according to the results of the rolling optimization; S5. Correcting the prediction model based on the data deviation between the real-time collected data and the data predicted by the nonlinear model; S6. Based on stochastic optimization theory, the optimal control parameter combination is calculated by simulating the random fluctuation of raw material characteristic data.

[0023] Regarding step S1, in this embodiment, the sensing module mainly includes a series of industrial sensors provided on the raw material entry section, the grinding unit and the transmission mechanism.

[0024] In some embodiments, the collection of raw material characteristic data is achieved by installing an online moisture sensor and hardness measuring device at the raw material feeding section. The online moisture sensor can be a microwave or near-infrared reflective moisture detection device, whose output signal is a continuous voltage or current signal, indicating the real-time moisture content of the raw material w r (t). The hardness measuring device can measure the crushing response of the raw material based on acoustic emission analysis or mechanical impact response method, thereby estimating the instantaneous hardness coefficient h of the raw material. r (t).

[0025] Specifically, the moisture content of raw materials can be collected and standardized in real time using the following formula: Among them, wn (t) represents the normalized moisture signal; w r (t) is the original moisture measurement value (%); w min and w max The lower and upper acceptable limits for raw material moisture are set for the equipment respectively.

[0026] As an option, the hardness signal can be extracted based on the characteristic spectrum parameters of the acoustic emission generated by the raw material under pressure and converted into an equivalent hardness index h through a pattern recognition algorithm. e (t), which is used as an input variable in the subsequent nonlinear modeling module.

[0027] The equipment operating status data mainly include: grinding roller speed ω(t), grinding roller pressure P(t) and equipment temperature T(t).

[0028] In some embodiments, the grinding roller speed is measured in real time by a rotary encoder mounted on the roller shaft. The encoder outputs a pulse frequency signal, which is then converted to angular velocity (in rad / s or rpm). The grinding roller pressure can be measured using a piezoelectric or resistance strain gauge sensor mounted on the roller pressure adjustment mechanism to monitor the change in normal pressure generated by the rolling gap in real time. The output signal can be a continuous analog quantity, typically in N or MPa.

[0029] The temperature acquisition device can be a thermocouple or a PT100 thermistor, which can be arranged on the roller body, bearing or motor housing surface of the grinding device to reflect the changes in thermal load during the grinding process.

[0030] In general, to avoid high-frequency noise and sudden interference in the measurement signal, in this embodiment, the sensor signal is preprocessed in real time by combining low-pass filtering and sliding mean filtering: x f (t) = α·x(t) + (1-α)·x f (t-1); Among them, x(t) is the original signal collected at the current time; x f (t) is the filtered signal at the current moment; x f (t-1) is the filtered signal at the previous moment; α∈(0,1) is the smoothing factor, which is generally adjusted between 0.1 and 0.3.

[0031] In one possible implementation, in order to improve the response sensitivity of the sensor system to sudden changes in the flow of raw materials, the system can set a boundary threshold trigger mechanism: if the moisture fluctuation exceeds the set change rate, the model correction module is triggered to respond in advance.

[0032] In addition, in some extended solutions, the sensor acquisition frequency can be set in the range of 1Hz to 10Hz, and adaptively adjusted according to the grinding roller speed and raw material feeding speed, thereby achieving data integrity in high-frequency dynamic scenarios.

[0033] It should be noted that all the above-mentioned signal acquisition uses the industrial field bus to access the system main control unit to achieve millisecond-level data aggregation and provide a unified time-scale data stream for subsequent model identification and control optimization.

[0034] During the specific implementation process, sensor interfaces can also be reserved for subsequent expansion, such as online particle size measuring instruments, equipment vibration monitoring modules, etc., to support more complex model construction and multi-dimensional status monitoring.

[0035] In step S2, in this embodiment, the system processes sensor-collected data on raw material moisture, hardness, grinding roller speed, and pressure, and constructs a dynamic prediction model for the flour production process based on nonlinear dynamic modeling. This model considers the impact of multiple factors on final product quality, including raw material characteristics, production equipment status, and environmental changes, and accurately predicts the changing trends of flour quality parameters such as particle size and ash content during production.

[0036] Specifically, the dynamic prediction model adopts a nonlinear regression model or a neural network model, wherein the neural network model can learn through training samples to approximate the nonlinear characteristics of the flour production process. As an option, this embodiment adopts a multi-layer perceptron (MLP) neural network model, which is based on the input features (such as raw material moisture w r (t), hardness h r (t), grinding roller speed ω(t), pressure P(t), equipment temperature T(t), etc.) to obtain output prediction values, such as flour particle size and production efficiency.

[0037] In this model, the neural network's input layer corresponds to the collected raw material characteristics and equipment status data, while the output layer outputs prediction results based on demand, such as key indicators such as particle size distribution and ash content. The model training is optimized using the error backpropagation algorithm to minimize the prediction error. The error function can be defined as: Where E represents the prediction error; N is the number of samples; y pred (t i ) is the model at time t i The predicted value of y true (t i ) is the actual collected value.

[0038] Specifically, the structure of a neural network model consists of an input layer, multiple hidden layers, and an output layer. The number of nodes in each layer and the choice of activation function (such as ReLU or Sigmoid) can be optimized through experimentation. During training, by learning from a large amount of historical production data, the network gradually adjusts its weights and biases, ultimately reaching an optimized state.

[0039] To improve the robustness of the model, in some embodiments, other machine learning methods such as support vector machines (SVM) can be used to further optimize the accuracy and stability of the model. By comparing different algorithms, the modeling method that best suits the specific production process can be selected.

[0040] In some cases, this embodiment may further combine fuzzy control with optimization methods such as genetic algorithms to post-process the results of the prediction model, optimize the model prediction error, and improve the adaptability of the system in actual operation.

[0041] Typically, this nonlinear modeling process is dynamic and can be updated in real time. By continuously learning from data collected in real time during the production process, the model is continuously optimized to adapt to changes in raw material composition, equipment status, and environmental conditions.

[0042] In this implementation, the system does not rely solely on static empirical formulas, but instead acquires and adapts to changes in the production process in real time through dynamic learning and data-driven methods, thereby improving overall control accuracy and flexibility.

[0043] In step S3, in this embodiment, calculations are performed based on the established dynamic prediction model to predict the changing trends of various production indicators over a period of time. Based on these predictions and real-time data, the system then uses an optimization algorithm to adjust key equipment parameters such as the grinding roller speed, roller spacing, and temperature control to ensure that the final flour quality meets predetermined standards.

[0044] Specifically, the control module utilizes a model predictive control (MPC) algorithm to perform multi-step predictions of future production states based on a predictive model and select an optimal set of control parameters through a rolling optimization process. These control parameters include, but are not limited to, grinding roller speed ω(t), roller spacing d(t), and equipment temperature T(t), which affect quality indicators such as particle size distribution and ash content of the final product.

[0045] In this control mode, the model predictive control algorithm optimizes the production process through the following steps: Data input and prediction: The control module first inputs real-time production data, including raw material moisture, hardness, equipment status, etc., and uses the prediction model to predict the subsequent production process to obtain the expected change trend of various quality indicators.

[0046] Objective function construction: According to the production target, set an objective function that includes quality control and energy efficiency optimization. The objective function can be expressed as: Among them, J is the optimization objective function; N is the number of samples; y pred (t i ) is at time t i The prediction quality indicator of y target (t i ) is the predetermined quality target value; u(t i ) is the adjustment amount of the control parameters (such as speed, roller gap, temperature, etc.); λ1 and λ2 are the control target weight coefficients.

[0047] Rolling Optimization: The optimization algorithm solves the objective function to obtain a set of optimal control parameters. These control parameters can minimize quality errors while avoiding excessive control fluctuations to ensure the smoothness of the production process.

[0048] Feedback, Correction, and Implementation: The control module generates control instructions based on the calculation results and adjusts the equipment's operating parameters. To enhance system adaptability, the optimization process is repeated within each control cycle, employing a rolling optimization strategy. This process ensures that the system can adapt instantly to actual changes, maintaining efficient and stable production.

[0049] Typically, the control algorithm performs adjustments periodically. This period can be set based on actual production needs, typically ranging from a few minutes to tens of minutes. In some embodiments, the control system also dynamically adjusts the optimization frequency based on the load conditions at each stage of the production process to improve the real-time response of the system.

[0050] Alternatively, control systems can incorporate global optimization algorithms, such as genetic algorithms or particle swarm optimization, to conduct a more comprehensive search of the control parameter space and find the globally optimal solution. These algorithms can effectively avoid local minima, thus achieving better control results in complex environments.

[0051] During implementation, this embodiment also incorporates a real-time feedback mechanism. When unexpected changes occur during production or in the external environment, the system promptly updates the control model and recalculates the optimal control strategy. For example, if the system detects a temperature change exceeding a set threshold, it automatically adjusts the roller spacing and speed to stabilize product quality.

[0052] Specifically, the control module not only relies on the prediction model but also uses a real-time data correction mechanism to further improve control accuracy. When there is a significant deviation between the data collected by the sensor and the output of the prediction model, the system automatically performs parameter correction to ensure the accuracy of the control strategy.

[0053] Regarding step S4, in this embodiment, the feedback correction module uses a dynamic correction mechanism to modify the model based on the deviation between real-time data and the prediction model output. The system calculates the error value in real time and adjusts the relevant parameters in the prediction model based on the error feedback. This allows the system to respond promptly to any changes in the production process, thereby maintaining high control accuracy and production efficiency.

[0054] Specifically, the key to the feedback correction process is to calculate the error and update the model by monitoring the difference between the real-time data and the prediction model results. Assume that at a certain time t, the real-time data collected by the system is y real (t), and the output of the prediction model is y pred (t), the error Δy(t) can be expressed as: Δy(t)=y real (t)-y pred (t); Among them, Δy(t) is the prediction error; y real (t) is the actual collected data value; y pred (t) is the value output by the prediction model. Once a significant error is detected, the system will initiate a correction mechanism. In order to improve the real-time performance and accuracy of the correction, the system can adopt an adaptive gain adjustment mechanism to weight the error and adjust the model parameters according to the size of the error. In some embodiments, the error correction gain factor γ can be automatically adjusted by the system according to the degree of change in real-time data, and the formula is expressed as: γ(t) = α·|Δy(t)|+β; Where γ(t) is the current gain coefficient; α is the sensitivity coefficient for gain adjustment; β is the base gain value; and |Δy(t)| is the absolute value of the error. The gain coefficient γ(t) influences the adjustment range of subsequent control parameters. Specifically, larger errors result in larger corrections, ensuring the system quickly returns to the target state; smaller errors, on the other hand, require only minor adjustments to avoid instability caused by overcorrection.

[0055] Alternatively, the feedback correction module can be combined with filtering algorithms such as Kalman filters to further optimize the error correction process. Kalman filters can effectively suppress measurement noise and improve system stability. In some embodiments, the Kalman filter fuses the collected real-time data with the output of the prediction model to obtain a weighted average correction value as the new input parameter.

[0056] in, is the revised estimate; is the output value of the prediction model; K(t) is the Kalman gain; y real (t) represents the actual collected data. The Kalman gain K(t) is dynamically calculated based on the system's noise characteristics and measurement errors, typically optimized using the minimum mean square error criterion. This approach enables the system to correct errors in real time and improve control accuracy in complex production environments.

[0057] For example, if a certain type of prediction error occurs repeatedly within a certain time window, the correction module will pre-correct the model based on this historical data to reduce the probability of future errors. Specifically, the global optimization of the model can be performed by minimizing the sum of squared errors: Where Δy i The difference between historical data and the output of the prediction model; Δy i is the difference between the i-th historical collected data and the output of the prediction model; n is the number of historical data samples.

[0058] In this way, the feedback correction module can not only cope with immediate errors but also identify long-term trend deviations, thereby further improving the accuracy and robustness of the system.

[0059] In some embodiments, the feedback correction module also has a self-diagnostic function. Based on the corrected control data, the system can analyze whether there are equipment failures or other abnormal conditions. For example, if the error of a device exceeds a set threshold, the system will automatically trigger a device check or alarm mechanism to ensure that the device status remains within the normal range during production.

[0060] In step S5, in this embodiment, when there is a significant deviation between actual production data and the prediction model output, the prediction model is promptly corrected to maintain consistent prediction accuracy and control effectiveness. This mechanism is directly based on the constructed nonlinear grinding dynamics model and uses real-time data as a correction reference to achieve dynamic model updates and adaptive adjustments.

[0061] In this embodiment, the system implements parameter correction of the prediction model through an error feedback mechanism. Specifically, the system continuously calculates the error between the actual measurement value and the predicted value. If the error exceeds a threshold ∈, the model correction process is triggered. Assuming the current time is t, the model error can be expressed as: Δy(t)=y real (t)-y model (t); Among them, Δy(t) represents the current prediction error; y real (t) is the actual value collected by the sensor in real time; y model (t) is the predicted output value of the nonlinear model at the same time.

[0062] In general, if |Δy(t)|>∈ in multiple consecutive sampling periods, the system considers that the model has deviated from the actual working conditions and needs to perform parameter update. In order to achieve rapid model adjustment, this embodiment adopts an incremental parameter optimization algorithm. i For example, the update formula is: Among them, θ i (t) is the current model parameter value; θ i (t+1) is the updated model parameter at the next moment; η is the learning rate (step coefficient); is the partial derivative of the error with respect to the parameter; Δy(t) is the model error at the current moment.

[0063] This formula essentially embodies a gradient descent-based error minimization update mechanism. By solving the first-order derivative of the error function, the system can quickly adjust the direction of the model parameters so that the model prediction results tend to the actual sampled values.

[0064] In some embodiments, to improve the convergence speed and global stability of the parameter correction process, the system also combines the recursive method of weighted historical error to construct the cost function: Where J(θ) represents the weighted sum of squared historical errors of the model under the current parameters; θ represents the model parameter vector; w k is the time decay weight, satisfying w k >w k+1 >0; M is the length of the historical lookback window; y real (tk) is the actual output value at time tk; y model (tk) is the predicted output calculated by the model at time tk.

[0065] This objective function is used to introduce historical trend changes during optimization, so that the model correction not only reflects the current error, but also covers recent system state changes, effectively suppressing model oscillations.

[0066] As an option, the system can also introduce a regularization term in the correction process to prevent parameter overfitting. Specifically, a regularized cost function of the following form can be constructed: J reg (θ)=J(θ)+λ·||θ|| 2 ; Among them, J reg (θ) is the cost function after regularization; λ is the regularization coefficient, which controls the weight of the penalty term; J reg (θ) is the original cost function; ||θ|| 2 is the L2 norm squared of the model parameter vector.

[0067] Specifically, the correction module can use algorithms such as recursive least squares (RLS) or extended Kalman filtering (EKF) to re-estimate the nonlinear model online. In some embodiments, for nonlinear models with complex parameters and strong coupling, a neural network structure can be introduced to replace the black box modeling, and the network weights can be continuously updated through error feedback. For example, if a feedforward neural network is used as an alternative modeling framework, the loss function can be set as: And update the neural network weight W using the back propagation algorithm i : Among them, L is the loss function; y real (t) is the actual output value at the current moment; y NN (t) is the output value of the neural network model at the current moment; W i (t) The value of the connection weight of the i-th neuron at the current moment; η is the learning rate; is the gradient of the loss function with respect to this weight.

[0068] In one possible implementation, the system also sets up a dynamic tolerance adjustment module to automatically adjust the model correction threshold ∈ and update step size η according to the complexity and stability of the real-time working conditions to prevent false corrections or frequent corrections in a high-noise environment, thereby ensuring the robustness of the overall control process.

[0069] In some embodiments, the model correction module will also automatically generate a correction log to record the parameter adjustment path, error change trend and output error range after model correction, which will be used for subsequent system learning and long-term optimization of the model structure.

[0070] In step S6, in this embodiment, after dynamically calibrating the prediction model, these optimized prediction results are used to guide control decisions during the production process. Specifically, the system uses the actuators to precisely adjust control parameters based on the previously calibrated prediction values ​​and current operating conditions, combined with production targets. This ensures that all production indicators remain within the set range, ensuring the stability of flour quality and production efficiency.

[0071] In this process, the system first receives the corrected prediction data y pred(t), these data have been error-corrected to more accurately reflect the actual production status. The system then calculates the difference in control targets based on the deviation between these data and the target values ​​and adjusts key parameters in the production process. Specifically, the following steps are involved: Obtain the deviation between the target value and the current forecast value: In this embodiment, the system first calculates the forecast error at the current moment. Assuming that the current moment is t, the system will calculate the deviation between the target value and the current forecast value according to the production target y. target (t) and the optimized predicted value y pred (t), calculate its error value: Δy(t) = y target (t)-y pred (t); Among them, Δy(t) is the deviation at the current moment, y target (t) is the preset target value or ideal production parameter, y pred (t) is the revised prediction result.

[0072] Calculating a control strategy based on the deviation: In this embodiment, if the calculated deviation Δy(t) exceeds a preset tolerance threshold, the system will adjust the control strategy. Typically, the system uses a model predictive control (MPC) algorithm to input the current deviation into an optimization model for prediction and optimization to obtain the optimal control instructions.

[0073] For example, if the goal is to adjust the operating state of a grinding machine, the controller can compensate for deviations by adjusting parameters such as the grinding roller's speed and pressure. Based on the MPC algorithm, the system dynamically calculates the adjustment amount over a period of time to minimize the overall system error.

[0074] Where u(t) is the control input (e.g., the adjustment amount of the grinding roller); y target (k) is the expected target value at the kth moment in the prediction interval; y pred (k) is the output value at time k predicted by the system model; u(k) is the control input at time k; t is the current time; λ is the penalty coefficient for the control input; and u is the future prediction horizon. This optimization objective function minimizes the deviation while avoiding over-adjustment of control parameters.

[0075] Actuators adjust control parameters: In this embodiment, based on the calculated optimal control strategy, the system uses actuators to adjust equipment parameters. These actuators, including electric adjustment devices and drive motors, receive signals from the control layer to precisely adjust key parameters such as grinding roller spacing, pressure, and motor speed. This real-time adjustment ensures that every step in the flour production process operates within the target range, thereby achieving stable and consistent flour quality.

[0076] In some embodiments, if the system error is small or the production process is stable, the control strategy can adopt a gradual adjustment, that is, making small adjustments to the control parameters to avoid excessive control inputs that cause system oscillations. This approach helps improve system robustness, especially when the raw material characteristics fluctuate significantly during the production process.

[0077] In one possible implementation, to further enhance the adaptability of the control process, the system could also incorporate an adaptive control mechanism, automatically adjusting the control strategy and adjustment step size based on real-time data and changes in production objectives. For example, if flour production requirements change, the system can quickly adjust the control strategy to ensure a balance between production quality and efficiency.

[0078] The intelligent control system for a flour workshop described below and the intelligent control method for a flour workshop described above can refer to each other.

[0079] Please see the attached Figure 2 The present invention also provides a flour workshop intelligent control system, comprising: The perception module collects real-time data on raw material properties and equipment operating status through sensors. It monitors the state of raw materials and equipment operating conditions on the production line using a variety of high-precision sensors (such as moisture, hardness, temperature, pressure, and flow), ensuring the system can grasp raw material properties (such as moisture and hardness) and equipment operating conditions (such as speed, temperature, and pressure) in real time. This data serves as the basis for subsequent processing and decision-making, providing accurate input for the control module.

[0080] For example, as raw materials enter the production line, the perception module uses sensors to monitor the moisture and hardness of the wheat in real time, ensuring that the raw material's condition meets the requirements of subsequent processing. Furthermore, the equipment's operating status, such as the speed and temperature of the grinding rollers, is also monitored in real time, providing data support for subsequent model optimization and adjustment.

[0081] The control module estimates the current system state variables based on a nonlinear grinding dynamics model and uses a nonlinear model predictive control algorithm to continuously optimize the control strategy for the future time domain. This optimization is performed using a nonlinear model predictive control (MPC) algorithm. Specifically, the control module uses a nonlinear model to predict the optimal control strategy for the future time domain based on the current raw material characteristics and equipment status, thereby guiding the optimization and adjustment of each link in the production process.

[0082] This module enables the system to adjust parameters such as the grinding roller speed and pressure in real time, ensuring the flour production process is always optimal. For example, based on real-time monitoring data, the control module can precisely adjust grinding parameters to achieve the set flour particle size target.

[0083] The execution module dynamically adjusts grinding parameters based on the results of rolling optimization. It controls the operating state of production equipment by adjusting the equipment's actuators, such as the electric adjustment device and drive motor. By adjusting equipment parameters in real time, the execution module ensures that every step in the production process remains within the set control range, thereby achieving precise control of flour quality and production efficiency.

[0084] For example, the execution module optimizes the grinding process by adjusting the spacing or pressure of the grinding rollers to ensure that the flour particle size is uniform and meets quality standards.

[0085] The feedback correction module is used to correct the prediction model based on data deviations between real-time collected data and the nonlinear model's predictions. Specifically, when significant deviations occur between the actual data collected in real time during production and the predicted data, the feedback correction module activates a model correction mechanism, dynamically adjusting the prediction model parameters through error feedback. This process is performed using an incremental parameter optimization algorithm to minimize prediction errors, ensuring that the model can adaptively adjust and accurately reflect current production conditions.

[0086] For example, if the flour particle size measured at a certain moment deviates from the predicted result, the feedback correction module will adjust the parameters of the control model through the error feedback mechanism, making the next prediction result more accurate and improving the prediction accuracy.

[0087] The stochastic optimization module simulates random fluctuations in raw material properties to calculate the optimal control parameter combination. Based on stochastic optimization theory, it optimizes control parameters while taking into account random fluctuations in raw material properties (such as moisture and hardness), thereby ensuring that the production process remains efficient under various uncertainties. By simulating fluctuations in raw material properties, the system can calculate the optimal control parameter combination in advance, providing a more stable control solution for the production process.

[0088] For example, the random optimization module can automatically adjust the drying time or temperature according to the moisture fluctuations of the raw materials, ensuring that the raw materials are always in the best condition, improving production efficiency and reducing energy consumption.

[0089] The system of this embodiment can be used to execute the above method embodiments, and its principles and technical effects are similar, so they will not be repeated here.

[0090] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A flour workshop intelligent control method, characterized in that: The following steps are involved: Collect raw material characteristic data and equipment operation status data in real time through sensors; According to the collected data, the state variables of the current system are calculated through the nonlinear grinding dynamics model; According to the current state variables, the control strategy in the future time domain is optimized through the nonlinear model predictive control algorithm; According to the results of rolling optimization, the grinding parameters are dynamically adjusted through the actuator; Correct the prediction model based on the data deviation between the real-time collected data and the nonlinear model prediction data; Based on stochastic optimization theory, the optimal control parameter combination is calculated by simulating the random fluctuation of raw material characteristic data.

2. A flour workshop intelligent control method according to claim 1, characterized in that: The raw material characteristic data includes raw material moisture and hardness, and the equipment operation status data includes grinding roller speed, grinding roller pressure and equipment temperature.

3. A flour workshop intelligent control method according to claim 1, characterized in that: The state variables include flour particle size, grinding roller speed, grinding roller pressure and equipment temperature, and the nonlinear grinding dynamics model describes the relationship between state variables, control variables and random variables.

4. A flour workshop intelligent control method according to claim 1, characterized in that: The nonlinear model predictive control algorithm is implemented in the following way: Predict future states based on nonlinear grinding dynamics model and stochastic process model; At each time step, an optimization problem is solved to minimize the expected total energy consumption while satisfying the probabilistic constraint of flour particle size distribution.

5. A flour workshop intelligent control method according to claim 4, characterized in that: The objective function of the optimization problem is the expected total energy consumption, which is defined as the integral of the instantaneous power function in the prediction time domain. The instantaneous power function is related to the roller spacing adjustment amount, the motor voltage and the air circuit valve opening.

6. A flour workshop intelligent control method according to claim 1, characterized in that: The grinding parameters include grinding roller speed, roller spacing and air circuit valve opening, which are dynamically adjusted through the motor drive and speed regulation system, roller spacing adjustment device and air circuit valve adjustment device.

7. A flour workshop intelligent control method according to claim 1, characterized in that: The correction prediction model calculates the deviation between the actual measured value and the predicted value, and uses the correction gain matrix to correct the prediction model.

8. The intelligent control method for a flour workshop according to claim 1, characterized in that: The stochastic optimization theory is implemented in the following way: Generate multiple random scenarios based on the random process model; Solve the optimization problem for each scenario and obtain the control strategy; Calculate the expected control strategy for all scenarios to ensure the robustness of the system under random fluctuations in raw material characteristics.

9. A flour workshop intelligent control method according to claim 1, characterized in that: The results of the rolling optimization include the following: Roller gap adjustment, used to adjust the gap between the grinding rollers to achieve the target flour particle size; Motor voltage, used to control the grinding roller speed to match current production needs; The air path valve opening is used to adjust the air flow speed to optimize the screening efficiency.

10. A flour workshop intelligent control system, applied to a flour workshop intelligent control method according to any one of claims 1 to 9, characterized in that: include: The perception module is used to collect raw material characteristic data and equipment operation status data in real time through sensors; A control module is used to estimate the state variables of the current system based on the nonlinear grinding dynamics model and to optimize the control strategy in the future time domain through a nonlinear model predictive control algorithm; An execution module for dynamically adjusting grinding parameters according to the results of rolling optimization; A feedback correction module is used to correct the prediction model according to the data deviation between the real-time collected data and the nonlinear model prediction data; The stochastic optimization module is used to calculate the optimal control parameter combination by simulating the random fluctuation of raw material characteristic data.

Images