Temperature control method and system based on flexible production line spraying and drying

Abstract

The invention provides a temperature control method and system based on flexible production line spraying and drying, and the method comprises the steps: training a lightweight LSTM network through obtaining the in-furnace temperature, the target drying temperature, the workpiece type and the heating power at the next moment under different working conditions; the temperature in the furnace and the target drying temperature are obtained in real time, the temperature error and the error change rate are calculated, according to a double-threshold switching rule, one of the trained lightweight LSTM network or PID is called, heating power is output, and the temperature is controlled; when the trained lightweight LSTM network and the PID need to be switched, judging whether a difference value between the heating power output by the lightweight LSTM network and the heating power output by the PID is greater than a preset value or not; if yes, the total transition time is calculated according to the difference value, the thermal characteristics of the workpiece and the total mass of the workpiece, then a nonlinear transition curve is defined, and transition of the heating power is conducted according to the total transition time and the nonlinear transition curve.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of temperature control, and particularly relates to a temperature control method and system based on flexible production line spraying and drying. BACKGROUND

[0002] Spraying and drying is an important unit of flexible automatic production line, therefore, intelligent monitoring of the spraying and drying unit plays a crucial role in production.

[0003] The quality of workpiece spraying and drying depends on the control of the temperature of the electric heating furnace. Currently, the temperature of the electric heating furnace is controlled by using a traditional PID controller. The traditional PID relies on the current deviation, historical deviation accumulation and deviation trend to calculate the output in real time, but it cannot predict the inertia characteristic of the inertia-lag system, which makes the system accumulate energy over time to gradually change after input, and the lag characteristic leads to time delay in the feedback of the output signal, which makes the control action of the PID always lag behind the actual state of the system and cannot respond to the deviation in time.

[0004] This lag further leads to a dilemma in control: if the proportional coefficient is increased to speed up the response, the feedback delay will cause excessive control, leading to overshoot or even sustained oscillation; if the proportional coefficient is reduced and the integral coefficient is increased to pursue stability, the system will respond slowly to the deviation and it will be difficult to meet the dynamic performance requirements. At the same time, the P (proportional), I (integral) and D (derivative) parameters of the traditional PID are fixed and cannot be dynamically adjusted according to the working condition changes of the inertia-lag system (such as changes in thermal inertia caused by different numbers of workpieces in the electric heating furnace). Once the system characteristics change, the original parameters will be invalid, leading to a decrease in control precision or destruction of stability. SUMMARY

[0005] Therefore, the present application provides a temperature control method and system based on flexible production line spraying and drying, which aims to solve the problem that the control action of the PID always lags behind the actual state of the system and cannot respond to the deviation in time.

[0006] The first aspect of the present application provides a temperature control method based on flexible production line spraying and drying, which comprises: obtaining the furnace temperature, target drying temperature, workpiece type and heating power at the next moment under different working conditions, training a lightweight LSTM network, wherein the furnace temperature, target drying temperature and workpiece type are used as inputs of the lightweight LSTM network, and the heating power at the next moment is used as the output of the lightweight LSTM network; The furnace temperature and target drying temperature are acquired in real time, the temperature error and error change rate are calculated, and the trained lightweight LSTM network or PID is called according to the dual threshold switching rule to output the heating power to control the temperature. The thresholds corresponding to the temperature error and the error change rate are calculated by fuzzy inference based on the temperature change rate, workpiece thermal characteristics and interference intensity. When the trained lightweight LSTM network and PID need to be switched, determine whether the difference between the heating power output by the lightweight LSTM network and PID is greater than a preset value. If so, the total transition time is calculated based on the difference, the thermal characteristics of the workpiece, and the total mass of the workpiece. Then, a nonlinear transition curve is defined, and the heating power is transitioned based on the total transition time and the nonlinear transition curve.

[0007] Furthermore, the lightweight LSTM network includes an input layer, a first LSTM hidden layer, a second LSTM hidden layer, and an output layer. The input layer has 12 neurons and is used to receive temporal features. The input dimensions include the furnace temperature at 10 time steps, the target drying temperature, and the workpiece type. The first LSTM hidden layer has 64 neurons and is used to learn short-term temperature-power dependence. tanh is used as the activation function and Dropout=0.2. The second LSTM hidden layer has 32 neurons and is used to learn long-term lag effects, using tanh as the activation function. The output layer has one neuron, which is used to output the heating power at the next moment, and uses sigmoid as the activation function.

[0008] Furthermore, in the step of real-time acquisition of furnace temperature and target drying temperature, calculation of temperature error and error change rate, and outputting heating power by calling either the trained lightweight LSTM network or the PID according to the dual threshold switching rule, the formula for calculating temperature error is as follows: ; For temperature error, T 实际 T represents the furnace temperature. 目标 The target drying temperature; ; The rate of change of error, Let be the temperature error at time t. The temperature error at time t-1; The trained lightweight LSTM network is invoked when any of the following conditions are met: ; ; The PID is invoked when both of the following conditions are met: ; .

[0009] Furthermore, in the step of calculating the temperature error and the threshold corresponding to the error change rate based on the temperature change rate, workpiece thermal characteristics, and interference intensity through fuzzy reasoning, the temperature change rate, workpiece thermal characteristics, and interference intensity are used as fuzzy inputs, and the temperature error and the threshold corresponding to the error change rate are used as fuzzy outputs. The threshold is calculated in real time through preset fuzzy rules, wherein the centroid method is used to convert the fuzzy output into a specific numerical threshold.

[0010] Furthermore, in the step of calculating the total transition time based on the difference, the workpiece thermal characteristics, and the total mass of the workpiece, defining a nonlinear transition curve, and then applying the heating power transition based on the total transition time and the nonlinear transition curve, the formula for calculating the total transition time is: ; ; Among them, T trans Let k1 and k2 be the total transition time, and k1 and k2 be coefficients. Let I be the system inertia, and C be the difference. p ρ is the specific heat capacity, and m is the total mass of the workpiece.

[0011] Furthermore, the expression for the nonlinear transition curve is: ; For the transition period, P PID P is the heating power output by the PID controller. LSTM To reduce the heating power output of the LSTM network, k is the curve steepness coefficient.

[0012] Furthermore, based on the system inertia, an initial value for the curve steepness coefficient is determined, and then the initial value of the curve steepness coefficient is dynamically corrected by using the temperature change trend corresponding to the heating power predicted by the lightweight LSTM network.

[0013] A second aspect of this invention provides a temperature control system for spray drying on a flexible production line, used to implement the temperature control method for spray drying on a flexible production line described in the first aspect, the system comprising: The training module is used to acquire the furnace temperature, target drying temperature, workpiece type, and heating power at the next moment under different working conditions to train a lightweight LSTM network. The furnace temperature, target drying temperature, and workpiece type are used as inputs to the lightweight LSTM network, and the heating power at the next moment is used as outputs to the lightweight LSTM network. The module is used to acquire the furnace temperature and target drying temperature in real time, calculate the temperature error and error change rate, and call one of the trained lightweight LSTM network or PID according to the dual threshold switching rule to output heating power to control the temperature. The thresholds corresponding to the temperature error and the error change rate are calculated by fuzzy inference based on the temperature change rate, workpiece thermal characteristics and interference intensity. The judgment module is used to determine whether the difference between the heating power output by the lightweight LSTM network and the PID is greater than a preset value when the trained lightweight LSTM network and PID need to be switched. The transition module is used to calculate the total transition time based on the difference between the heating power output of the lightweight LSTM network and the PID controller when the difference is greater than a preset value, the thermal characteristics of the workpiece, and the total mass of the workpiece. Then, a nonlinear transition curve is defined, and the heating power is transitioned according to the total transition time and the nonlinear transition curve.

[0014] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the temperature control method for spray drying based on a flexible production line provided in the first aspect.

[0015] A fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the temperature control method for spray drying based on a flexible production line provided in the first aspect.

[0016] This invention provides a temperature control method and system for spraying and drying on a flexible production line. The method trains a lightweight LSTM network by acquiring furnace temperature, target drying temperature, workpiece type, and heating power at the next moment under different operating conditions. It acquires the furnace temperature and target drying temperature in real time, calculates the temperature error and error change rate, and, according to a dual-threshold switching rule, calls either the trained lightweight LSTM network or a PID controller to output heating power for temperature control. Specifically, based on the temperature change rate, workpiece thermal characteristics, and interference intensity, fuzzy inference is used to calculate the thresholds corresponding to the temperature error and the error change rate. When switching between the trained lightweight LSTM network and the PID controller is required, it is determined whether the difference in heating power output between the two is greater than a preset value. If so, the total transition time is calculated based on the difference, the workpiece thermal characteristics, and the total workpiece mass. A nonlinear transition curve is then defined, and the heating power transition is performed based on the total transition time and the nonlinear transition curve. By combining the lightweight LSTM network with the PID controller, the system can automatically learn multi-factor coupling relationships, achieving both high temperature control accuracy and advantages of speed, flexibility, and adaptability. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the implementation of a temperature control method for spraying and drying on a flexible production line, as provided in Embodiment 1 of the present invention. Figure 2 This is a structural block diagram of a temperature control system for spray drying on a flexible production line, provided in Embodiment 2 of the present invention. Figure 3 This is a structural block diagram of an electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0018] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0019] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] Example 1 According to an embodiment of the present invention, a temperature control method for spray drying based on a flexible production line is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0022] This first embodiment provides a temperature control method for spray drying based on a flexible production line, which can be used in electronic devices, such as computers. Please refer to [link to relevant documentation]. Figure 1 , Figure 1 The flowchart of a temperature control method for spraying and drying on a flexible production line according to Embodiment 1 of the present invention is shown, specifically including steps S01 to S04.

[0023] Step S01: Obtain the furnace temperature, target drying temperature, workpiece type, and heating power at the next moment under different working conditions, and train a lightweight LSTM network.

[0024] The furnace temperature, the target drying temperature, and the workpiece type serve as inputs to the lightweight LSTM network, while the heating power at the next moment serves as the output of the lightweight LSTM network. It should be noted that the operating conditions should cover all scenarios. For example, normal operating conditions could be heating (room temperature rises to the target temperature), steady state (maintaining the target temperature for 1 hour), and cooling (target temperature drops to room temperature); disruptive operating conditions could be voltage fluctuations (220V±10%), batch workpiece switching (switching from 5 pieces to 20 pieces), and sudden fan start / stop; abnormal operating conditions could be heating element power decay (aging causes a 10% decrease in rated power) and temporary sensor drift (error ±0.5℃).

[0025] Specifically, the lightweight LSTM network includes an input layer, a first LSTM hidden layer, a second LSTM hidden layer, and an output layer. The input layer has 12 neurons and is used to receive temporal features. The input dimensions include the furnace temperature at 10 time steps, the target drying temperature, and the workpiece type. The first LSTM hidden layer has 64 neurons and is used to learn short-term temperature-power dependence. tanh is used as the activation function and Dropout=0.2. The second LSTM hidden layer has 32 neurons and is used to learn long-term lag effects, using tanh as the activation function. The output layer has one neuron, which is used to output the heating power at the next moment, and uses sigmoid as the activation function.

[0026] In this embodiment of the invention, data from the past 10 seconds (i.e., 10 time steps) is used as input. Since the maximum lag in spray drying is 5 seconds, the 10-step window can cover the entire lag period without increasing the computational burden. In addition, the static operating condition characteristics include the target drying temperature and the workpiece type. The target drying temperature is used to clarify the control target, allowing the model to know how much the temperature needs to be adjusted, which is the baseline information. The workpiece type is used to distinguish the heat absorption capacity of the workpiece (such as metal, plastic), allowing the model to adapt to the inertial differences of different operating conditions and avoid one-size-fits-all control. The workpiece thermal characteristic code is used to represent the workpiece type. For example, metal=1, plastic=2, ceramic=3.

[0027] Step S02: Real-time acquisition of furnace temperature and target drying temperature, calculation of temperature error and error change rate, and according to the dual threshold switching rule, calling either the trained lightweight LSTM network or PID to output heating power for temperature control.

[0028] Understandably, step S02 involves accurately calling the algorithm for different scenarios, while also addressing the issue of sudden power changes during switching to prevent temperature fluctuations from affecting coating quality.

[0029] Specifically, two key thresholds are defined to achieve intelligent switching between using a lightweight LSTM network for large errors and a PID controller for small errors. It should be noted that the formula for calculating temperature error is: ; For temperature error, T 实际 T represents the furnace temperature. 目标 The target drying temperature; ; The rate of change of error, Let be the temperature error at time t. The temperature error at time t-1; The trained lightweight LSTM network is invoked when any of the following conditions are met: ; ; Understandably, the purpose of this call is to leverage the fast response and self-learning capabilities of the lightweight LSTM network to quickly recover the temperature and combat large inertia and interference. The PID is invoked when both of the following conditions are met: ; .

[0030] Understandably, the purpose of this call is to utilize the zero steady-state error and steady-state accuracy of the PID controller to stabilize the temperature at ±0.1℃, thus meeting the requirements for spray drying. Specifically, based on the temperature change rate, workpiece thermal characteristics, and interference intensity, fuzzy inference is used to calculate the threshold corresponding to the temperature error and the error change rate. Specifically, the temperature change rate, workpiece thermal characteristics, and interference intensity are used as fuzzy inputs, and the threshold corresponding to the temperature error and the error change rate is used as fuzzy outputs. The threshold is calculated in real time using preset fuzzy rules, and the centroid method is used to convert the fuzzy output into a specific numerical threshold.

[0031] More specifically, the key factors affecting the threshold in the spraying and drying scenario are first identified. In this embodiment of the invention, the temperature change rate, workpiece thermal characteristics, and interference intensity are identified as key factors affecting the threshold. It can be understood that the temperature change rate is the rate of change of temperature within the furnace. For the temperature change rate, when the system inertia is large and the temperature change is slow, the error threshold needs to be relaxed to give the lightweight LSTM network more adjustment time; conversely, when the temperature change is fast, the threshold needs to be narrowed to avoid overshoot. For the workpiece thermal characteristics, when the workpiece has a large specific heat capacity, its heat absorption capacity is strong, and the system inertia is large, so the error threshold needs to be relaxed; when the specific heat capacity is small, heat absorption is fast, so the threshold needs to be tightened. For the interference intensity, when the interference is strong (voltage fluctuation ±10%, D=1), the error change rate threshold needs to be expanded to avoid incorrect switching caused by interference; when the interference is weak (voltage fluctuation ±2%, D=0), the threshold needs to be tightened to ensure accurate switching.

[0032] A specific implementation method for calculating thresholds in real time using preset fuzzy rules can be as follows: First, the input is fuzzified; for example, the temperature change rate is fuzzified. Divide it into three fuzzy sets: slow (0-0.2℃ / s), medium (0.2-0.4℃ / s), and fast (0.4-0.6℃ / s). Then, C... p Divide the values ​​into small (1.0-1.5), medium (1.5-2.0), and large (2.0-2.5), and divide D into weak (0-0.3), medium (0.3-0.7), and strong (0.7-1.0).

[0033] Then, fuzzy rules are designed. For example, if... Slow, C pIf the value is large and the strength is high, then the first threshold is 2℃ and the second threshold is 0.5℃ / s; if... C p In the cases of D and C, the first threshold is 1.5℃, and the second threshold is 0.3℃ / s; if Fast, C p If the value is small and D is weak, then the first threshold is 1℃ and the second threshold is 0.2℃ / s. Finally, the fuzzy output is converted into a specific numerical threshold using the centroid method, which is then directly used for switching judgments.

[0034] Step S03: When the trained lightweight LSTM network and PID need to be switched, determine whether the difference between the heating power output by the lightweight LSTM network and PID is greater than a preset value. If so, proceed to step S04.

[0035] It is understandable that if the power difference is large, using a general linear transition can easily lead to a sudden drop in temperature. A linear transition is a uniform change and cannot adapt to the inertial characteristics of the system.

[0036] In step S04, the total transition time is calculated based on the difference, the thermal characteristics of the workpiece, and the total mass of the workpiece. Then, a nonlinear transition curve is defined, and the heating power is transitioned based on the total transition time and the nonlinear transition curve.

[0037] Specifically, the formula for calculating the total transition time is: ; ; Among them, T trans Let k1 and k2 be the total transition time, and k1 and k2 be coefficients. Let I be the system inertia, and C be the difference. p Where m is the specific heat capacity and m is the total mass of the workpiece. It should be noted that when the power difference is large and the inertia is large, the transition time is long to avoid impact; when the power difference is small and the inertia is small, the transition time is short to improve the response speed.

[0038] Furthermore, the expression for the nonlinear transition curve is: ; For the transition period, P PID P is the heating power output by the PID controller. LSTMTo reduce the heating power output by the lightweight LSTM network, k is the curve steepness coefficient. It should be noted that the initial value of the curve steepness coefficient is determined based on the system inertia. For example, when the inertia is large, k=5, and the curve is gentle; when the inertia is small, k=10, and the curve is steep. Subsequently, the initial value of the curve steepness coefficient is dynamically corrected by the temperature change trend corresponding to the heating power predicted by the lightweight LSTM network.

[0039] It should be noted that during spray drying, the temperature change trend is affected by both system inertia and external disturbances (such as voltage fluctuations and fan start / stop). General solutions do not distinguish between these two types of factors, which may lead to incorrect k-value correction (e.g., a sudden drop in temperature corresponding to the heating power predicted by the lightweight LSTM network due to voltage fluctuations is mistakenly interpreted as high system inertia, leading to an increase in the k-value). Therefore, in this embodiment of the invention, the lightweight LSTM network additionally predicts the disturbance intensity, and the k-value correction strategy is dynamically adjusted based on the presence of disturbance to improve robustness.

[0040] Specifically, new interference features are added to the training data of the lightweight LSTM network, namely voltage fluctuation amplitude (collected from voltage sensors) and workshop wind speed (collected from wind speed sensors). At the same time, an interference prediction branch is added to the lightweight LSTM network to output the interference intensity level (D_pred), where D_pred=0 indicates no interference, D_pred=1 indicates weak interference (voltage ±5% / wind speed <2m / s), and D_pred=2 indicates strong interference (voltage ±10% / wind speed ≥2m / s).

[0041] Furthermore, the final expression for modifying the value of k is: k value = k0 × (1 + k_dist); k0 is the initial value of the curve steepness coefficient, and k_dist is the interference compensation coefficient. In the embodiments of the present invention, when there is no interference (D_pred=0), k_dist=0; when there is weak interference (D_pred=1), k_dist=0.1, and small compensation is performed; when there is strong interference (D_pred=2), k_dist=0.3, and large compensation is performed.

[0042] Furthermore, if the lightweight LSTM network predicts that the interference intensity decreases from 2 to 0 (interference disappears), then within 3 control cycles, k_dist is linearly reduced from 0.3 to 0 to avoid a sudden drop in the k value causing abrupt changes in the transition curve and further stabilize the temperature.

[0043] In summary, the temperature control method for spraying and drying on a flexible production line according to the above embodiments of the present invention trains a lightweight LSTM network by acquiring the furnace temperature, target drying temperature, workpiece type, and heating power at the next moment under different working conditions; it acquires the furnace temperature and target drying temperature in real time, calculates the temperature error and error change rate, and calls one of the trained lightweight LSTM network or PID according to the dual threshold switching rule to output heating power to control the temperature. Specifically, based on the temperature change rate, workpiece thermal characteristics, and interference intensity, fuzzy inference is used to calculate the thresholds corresponding to the temperature error and the error change rate. When the trained lightweight LSTM network and PID need to be switched, it is determined whether the difference between the heating power output by the lightweight LSTM network and the PID is greater than a preset value. If so, the total transition time is calculated based on the difference, the workpiece thermal characteristics, and the total mass of the workpiece, and a nonlinear transition curve is defined. The heating power transition is performed based on the total transition time and the nonlinear transition curve. By combining the lightweight LSTM network with the PID, it can automatically learn the coupling relationship of multiple factors, achieving the advantages of speed, flexibility, and adaptability while meeting temperature control accuracy requirements.

[0044] Example 2 Please see Figure 2 , Figure 2 This is a structural block diagram of a temperature control system for spray drying on a flexible production line, provided in Embodiment 2 of the present invention. This temperature control system 200 is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0045] Specifically, the temperature control system 200 based on flexible production line spraying and drying includes: a training module 21, a calling module 22, a judgment module 23, and a transition module 24, wherein: Training module 21 is used to acquire furnace temperature, target drying temperature, workpiece type, and heating power at the next moment under different working conditions, and to train a lightweight LSTM network. The furnace temperature, target drying temperature, and workpiece type are used as inputs to the lightweight LSTM network, and the heating power at the next moment is used as output. The lightweight LSTM network includes an input layer, a first LSTM hidden layer, a second LSTM hidden layer, and an output layer. The number of neurons in the input layer is 12, which are used to receive temporal features. The input dimensions include furnace temperature at 10 time steps, target drying temperature, and workpiece type. The first LSTM hidden layer has 64 neurons and is used to learn short-term temperature-power dependence. tanh is used as the activation function and Dropout=0.2. The second LSTM hidden layer has 32 neurons and is used to learn long-term lag effects, using tanh as the activation function. The output layer has one neuron, which is used to output the heating power at the next moment, and uses sigmoid as the activation function. Module 22 is invoked to acquire the furnace temperature and target drying temperature in real time, calculate the temperature error and error change rate, and, according to the dual threshold switching rule, invoke either the trained lightweight LSTM network or the PID controller to output heating power for temperature control. Specifically, based on the temperature change rate, workpiece thermal characteristics, and interference intensity, fuzzy inference is used to calculate the threshold corresponding to the temperature error and the error change rate. The formula for calculating the temperature error is: ; For temperature error, T 实际 T represents the furnace temperature. 目标 The target drying temperature; ; The rate of change of error, Let be the temperature error at time t. The temperature error at time t-1; The trained lightweight LSTM network is invoked when any of the following conditions are met: ; ; The PID is invoked when both of the following conditions are met: ; ; The temperature change rate, workpiece thermal characteristics, and interference intensity are used as fuzzy inputs, and the temperature error and the threshold corresponding to the error change rate are used as fuzzy outputs. The threshold is calculated in real time through preset fuzzy rules. The centroid method is used to convert the fuzzy output into a specific numerical threshold. The judgment module 23 is used to determine whether the difference between the heating power output by the lightweight LSTM network and the PID is greater than a preset value when the trained lightweight LSTM network and the PID need to be switched. Transition module 24 is used to calculate the total transition time based on the difference between the heating power output of the lightweight LSTM network and the PID controller when the difference is greater than a preset value, the thermal characteristics of the workpiece, and the total mass of the workpiece. Then, a nonlinear transition curve is defined, and the heating power is transitioned according to the total transition time and the nonlinear transition curve. The formula for calculating the total transition time is: ; ; Among them, T trans Let k1 and k2 be the total transition time, and k1 and k2 be coefficients. Let I be the system inertia, and C be the difference. p Where m is the specific heat capacity and m is the total mass of the workpiece; The expression for the nonlinear transition curve is: ; For the transition period, P PID P is the heating power output by the PID controller. LSTM To reduce the heating power output of the LSTM network, k is the curve steepness coefficient; Based on the system inertia, an initial value for the curve steepness coefficient is determined. Then, the initial value of the curve steepness coefficient is dynamically corrected by using the temperature change trend corresponding to the heating power predicted by the lightweight LSTM network.

[0046] Example 3 In another aspect, the present invention also proposes an electronic device, please refer to [link to relevant documentation]. Figure 3 The electronic device shown is an embodiment of the present invention, including a memory 20, a processor 10, and a computer program 30 stored in the memory and executable on the processor. When the processor 10 executes the computer program 30, it implements the temperature control method for spray drying based on a flexible production line as described above.

[0047] In some embodiments, the processor 10 may be a central processing unit (CPU), controller, microcontroller, microprocessor or other data processing chip, used to run program code stored in memory 20 or process data, such as executing access restriction programs.

[0048] The memory 20 includes at least one type of readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 20 can be an internal storage unit of an electronic device, such as the hard disk of the electronic device. In other embodiments, the memory 20 can also be an external storage device of the electronic device, such as a plug-in hard disk, SmartMediaCard (SMC), SecureDigital (SD) card, FlashCard, etc., equipped on the electronic device. Furthermore, the memory 20 can include both internal and external storage units of the electronic device. The memory 20 can be used not only to store application software and various types of data of the electronic device, but also to temporarily store data that has been output or will be output.

[0049] It should be pointed out that, Figure 3 The structure shown does not constitute a limitation on the electronic device. In other embodiments, the electronic device may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0050] This invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the temperature control method for spray drying based on a flexible production line as described above.

[0051] Those skilled in the art will understand that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0052] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0053] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0054] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0055] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A temperature control method for spray drying on a flexible production line, characterized in that, The method includes: The furnace temperature, target drying temperature, workpiece type, and heating power at the next moment are obtained under different working conditions to train a lightweight LSTM network. The furnace temperature, target drying temperature, and workpiece type are used as inputs to the lightweight LSTM network, and the heating power at the next moment is used as outputs to the lightweight LSTM network. The furnace temperature and target drying temperature are acquired in real time, the temperature error and error change rate are calculated, and the trained lightweight LSTM network or PID is called according to the dual threshold switching rule to output the heating power to control the temperature. The thresholds corresponding to the temperature error and the error change rate are calculated by fuzzy inference based on the temperature change rate, workpiece thermal characteristics and interference intensity. When the trained lightweight LSTM network and PID need to be switched, determine whether the difference between the heating power output by the lightweight LSTM network and PID is greater than a preset value. If so, the total transition time is calculated based on the difference, the thermal characteristics of the workpiece, and the total mass of the workpiece. Then, a nonlinear transition curve is defined, and the heating power is transitioned based on the total transition time and the nonlinear transition curve.

2. The temperature control method for spray drying based on a flexible production line according to claim 1, characterized in that, The lightweight LSTM network includes an input layer, a first LSTM hidden layer, a second LSTM hidden layer, and an output layer. The input layer has 12 neurons and is used to receive temporal features. The input dimensions include the furnace temperature at 10 time steps, the target drying temperature, and the workpiece type. The first LSTM hidden layer has 64 neurons and is used to learn short-term temperature-power dependence. It uses tanh as the activation function and Dropout=0.

2. The second LSTM hidden layer has 32 neurons and is used to learn long-term lag effects, using tanh as the activation function. The output layer has one neuron, which is used to output the heating power at the next moment, and uses sigmoid as the activation function.

3. The temperature control method for spray drying based on a flexible production line according to claim 2, characterized in that, In the step of real-time acquisition of furnace temperature and target drying temperature, calculation of temperature error and error change rate, and output of heating power by calling either a trained lightweight LSTM network or a PID according to the dual threshold switching rule, the formula for calculating temperature error is as follows: ； For temperature error, T 实际 T represents the furnace temperature. 目标 The target drying temperature; ； The rate of change of error, Let be the temperature error at time t. The temperature error at time t-1; The trained lightweight LSTM network is invoked when any of the following conditions are met: ； ； The PID is invoked when both of the following conditions are met: ； 。 4. The temperature control method for spray drying based on a flexible production line according to claim 3, characterized in that, In the step of calculating the temperature error and the threshold corresponding to the error change rate based on the temperature change rate, workpiece thermal characteristics and interference intensity through fuzzy inference, the temperature change rate, workpiece thermal characteristics and interference intensity are used as fuzzy inputs, and the temperature error and the threshold corresponding to the error change rate are used as fuzzy outputs. The threshold is calculated in real time through preset fuzzy rules, wherein the centroid method is used to convert the fuzzy output into a specific numerical threshold.

5. The temperature control method for spray drying based on a flexible production line according to claim 4, characterized in that, In the step of calculating the total transition time based on the difference, the thermal characteristics of the workpiece, and the total mass of the workpiece, defining a nonlinear transition curve, and then applying the heating power transition based on the total transition time and the nonlinear transition curve, the formula for calculating the total transition time is as follows: ； ； Among them, T trans Let k1 and k2 be the total transition time, and k1 and k2 be coefficients. Let I be the system inertia, and C be the difference. p ρ is the specific heat capacity, and m is the total mass of the workpiece.

6. The temperature control method for spray drying based on a flexible production line according to claim 5, characterized in that, The expression for the nonlinear transition curve is: ； For the transition period, P PID P is the heating power output by the PID controller. LSTM To reduce the heating power output of the LSTM network, k is the curve steepness coefficient.

7. The temperature control method for spray drying based on a flexible production line according to claim 6, characterized in that, Based on the system inertia, an initial value for the curve steepness coefficient is determined. Then, the initial value of the curve steepness coefficient is dynamically corrected by using the temperature change trend corresponding to the heating power predicted by the lightweight LSTM network.

8. A temperature control system for spray drying on a flexible production line, characterized in that, For implementing the temperature control method for spray drying based on a flexible production line as described in any one of claims 1-7, the system comprises: The training module is used to acquire the furnace temperature, target drying temperature, workpiece type, and heating power at the next moment under different working conditions to train a lightweight LSTM network. The furnace temperature, target drying temperature, and workpiece type are used as inputs to the lightweight LSTM network, and the heating power at the next moment is used as outputs to the lightweight LSTM network. The module is used to acquire the furnace temperature and target drying temperature in real time, calculate the temperature error and error change rate, and call one of the trained lightweight LSTM network or PID according to the dual threshold switching rule to output heating power to control the temperature. The thresholds corresponding to the temperature error and the error change rate are calculated by fuzzy inference based on the temperature change rate, workpiece thermal characteristics and interference intensity. The judgment module is used to determine whether the difference between the heating power output by the lightweight LSTM network and the PID is greater than a preset value when the trained lightweight LSTM network and PID need to be switched. The transition module is used to calculate the total transition time based on the difference between the heating power output of the lightweight LSTM network and the PID controller when the difference is greater than a preset value, the thermal characteristics of the workpiece, and the total mass of the workpiece. Then, a nonlinear transition curve is defined, and the heating power is transitioned according to the total transition time and the nonlinear transition curve.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the temperature control method for spray drying based on a flexible production line as described in any one of claims 1-7.

10. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the temperature control method for spray drying based on a flexible production line as described in any one of claims 1-7.

Images