Stainless steel degreasing section drying temperature and humidity feedback type automatic control system and method

Abstract

The invention discloses a stainless steel degreasing section drying temperature and humidity feedback type automatic control system and method. The system comprises a temperature and humidity sensor array, a fuzzy decoupling controller, an LSTM prediction module, a distributed heating unit and a bionic honeycomb structure protective cover. Real-time data are collected through a temperature and humidity sensor array; performing fuzzy decoupling operation based on a temperature-humidity incidence matrix to generate an independent control quantity: calling an LSTM model to predict six groups of historical sequences, and outputting temperature and humidity changes in future 30 seconds; and dynamically adjusting a PID setting curve according to a prediction result, and controlling the power of the heating unit. Through cooperative work of the distributed sensor array, the fuzzy decoupling controller and the LSTM prediction module, accurate decoupling control and advanced prediction of temperature and humidity are achieved, the problem of limitation of traditional PID control in a nonlinear system is solved, and the system has the advantages of being high in control precision, high in response speed and stable and reliable in a high-temperature environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial automation control, in particular to a stainless steel degreasing section drying temperature and humidity feedback type automatic control system and method. BACKGROUND

[0002] In the field of drying temperature and humidity control, PID algorithm is widely used due to its good real-time performance and stability. Taking grain drying equipment as an example, the prior art adjusts the hot air temperature, humidity and speed through incremental digital PID, and realizes the control accuracy of ±5°C by using fine parameter configuration of proportional coefficient Kp=0.1 and integral coefficient Ki=0.001. However, the traditional PID control method has obvious limitations when facing nonlinear systems, especially in the case of dynamic change of material characteristics, the control effect is significantly reduced. To solve this problem, the prior art attempts to dynamically correct the PID parameters by clustering historical data, which alleviates the control lag problem to some extent, but fails to fundamentally solve the control precision loss caused by system coupling interference.

[0003] In the aspect of measurement technology, the existing patent reduces the data error by 50% compared with the traditional environmental monitoring method by directly integrating the temperature and humidity sensor into the hanger and contacting the fabric surface (0-10 cm distance), effectively avoiding the problem of over-drying or insufficient drying. However, this contact measurement method is limited in high-temperature and corrosive environments. Although there are patents that use gear-rack linkage mechanisms to achieve sensor protection and measurement site switching, this scheme is complex in structure and high in maintenance cost, which is difficult to meet the requirements of long-term stable operation in industrial field.

[0004] Especially in the drying process of stainless steel degreasing section, the existing technology has three outstanding defects: first, the strong coupling of temperature and humidity leads to control response lag, second, the sensor reliability is insufficient in high-temperature environment (180°C), third, the traditional prediction algorithm cannot meet the demand of accurate temperature and humidity prediction in 30 seconds. These defects directly affect the surface treatment quality of stainless steel strip, which can easily lead to incomplete degreasing or energy waste.

[0005] In view of the above problems, the existing technology needs to be improved. SUMMARY

[0006] The purpose of the present application is to provide a stainless steel degreasing section drying temperature and humidity feedback type automatic control system and method, which has the technical advantages of solving the strong coupling control lag of temperature and humidity, the insufficient sensor reliability in high-temperature environment and realizing accurate temperature and humidity prediction.

[0007] The application provides a stainless steel degreasing section drying temperature and humidity feedback type automatic control system, and the technical scheme is as follows: a temperature and humidity sensor array is distributedly arranged on the inner wall of a drying box to collect temperature and humidity data in real time; a fuzzy decoupling controller is connected with the temperature and humidity sensor array, has a temperature-humidity correlation matrix built-in, and outputs decoupling control instructions; an LSTM prediction module inputs six groups of historical temperature and humidity sequences, has a sampling interval of 10s, and outputs temperature and humidity prediction values in the future 30s; a distributed heating unit comprises a PID power regulator and a thermocouple array, and has a response time of less than 200ms; and a bionic honeycomb structure protective cover is used to cover the sensor and the controller, is made of 6061 aluminum alloy, and has a wall thickness of 1.5mm.

[0008] Further, the application further provides that the fuzzy decoupling controller executes a segmented control strategy, wherein a temperature dominant control with a proportional coefficient Kp = 0.3 is adopted in the temperature rising stage, and a humidity dominant control with an integral coefficient Ki = 0.7 is adopted in the temperature keeping stage.

[0009] Further, the application further provides that the network structure of the LSTM prediction module comprises two layers of hidden layers, each layer has 128 neurons, the Dropout rate is 0.2, the prediction temperature error is not more than ±0.3℃, and the humidity error is not more than ±1.5% RH.

[0010] Further, the application further provides that the distributed heating unit comprises an alumina ceramic heat insulation layer covering the surface of the heating element, and has a thermal conductivity of 0.8W / m·K; and a 5mm air layer is arranged between the heat insulation layer and the inner wall of the equipment, and the combined thermal conductivity is not more than 0.3W / m·K.

[0011] Further, the application further provides that the bionic honeycomb structure protective cover comprises a buckle type mounting mechanism, and has a repeated positioning accuracy of less than 0.1mm; and a hydrophobic nano coating, and has a surface contact angle of more than 150°.

[0012] Further, the application further provides a stainless steel degreasing section drying temperature and humidity feedback type automatic control method, which comprises the following steps: collecting real-time data by means of a temperature and humidity sensor array; performing fuzzy decoupling operation based on a temperature-humidity correlation matrix to generate independent control variables U T =K p ·e T +K decoup ·ΔH and U H =K i ·∫e H dt+K decoup ·ΔT; calling an LSTM model to predict six groups of historical sequences, and outputting temperature and humidity changes in the future 30s; and dynamically adjusting a PID setting curve according to the prediction result to control the power of the heating unit, and the response time is less than 200ms.

[0013] Furthermore, this application also proposes that the fuzzy decoupling operation adopts a segmented strategy: when the temperature is 5°C below the target value, the heating stage mode is activated and Kp = 0.3 is locked; when the temperature reaches the target value ±0.5°C, the heat preservation stage mode is switched to and Ki = 0.7 is locked.

[0014] Furthermore, this application also proposes that the training of the LSTM model adopts the mean squared error loss function, the input data is standardized to the 0-1 interval, and the prediction results are destandardized and output.

[0015] Furthermore, this application also proposes that the power adjustment of the heating unit includes constructing a dynamic model of the temperature field through a thermocouple array; and activating a PID compensation algorithm to adjust the heating power for areas with local temperature differences greater than 2°C.

[0016] Furthermore, this application also proposes a sensor thermal protection step: maintaining the sensor operating temperature at 60±2℃ and the ambient temperature at 180℃ through a micro-airflow cooling channel.

[0017] As can be seen from the above, the automatic control system and method for temperature and humidity feedback in the drying section of stainless steel degreasing provided in this application achieves precise decoupled control and advanced prediction of temperature and humidity through the collaborative work of a distributed sensor array, a fuzzy decoupling controller and an LSTM prediction module. It solves the limitation problem of traditional PID control in nonlinear systems and has the advantages of high control accuracy, fast response speed and stability and reliability in high temperature environment. Attached Figure Description

[0018] Figure 1 This is a structural block diagram of an automatic control system for temperature and humidity feedback in the drying section of stainless steel degreasing section according to the present invention.

[0019] Figure 2 This is a flowchart of an automatic control method for temperature and humidity feedback in the drying section of stainless steel degreasing section according to the present invention. Detailed Implementation

[0020] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0021] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0022] like Figure 1 This application proposes an automatic control system for temperature and humidity feedback in the drying section of stainless steel degreasing, comprising a temperature and humidity sensor array, a fuzzy decoupling controller, an LSTM prediction module, a distributed heating unit, and a biomimetic honeycomb structure protective cover. The temperature and humidity sensor array adopts a multi-point distributed layout; in specific implementations, SHT35 digital sensors can be used, with an installation spacing of 300mm × 300mm grid array. The temperature-humidity correlation matrix of the fuzzy decoupling controller can be established through experimental data, for example, by obtaining the coupling coefficient using orthogonal experimental design. The LSTM prediction module can be implemented using the TensorFlow framework, with the input layer set to 6 × 2D time-series data (temperature + humidity) and the output layer using a linear activation function. The PID controller in the distributed heating unit can be an SSR solid-state relay, with a response time controlled within 150ms. The 6061 aluminum alloy of the biomimetic honeycomb protective cover can be anodized, with the honeycomb unit size designed as a 5mm regular hexagon.

[0023] The temperature and humidity sensor array eliminates measurement blind spots through multi-point distribution, with a data acquisition frequency set to 1Hz. The fuzzy decoupling controller employs a Mamdani-type inference mechanism, with a decoupling operation cycle of 100ms. The LSTM prediction module uses the Adam optimizer during training, with a learning rate set to 0.001. In the distributed heating unit, thermocouples are K-type armored sensors, arranged at a density of 4 per square meter. The protective cover's locking mechanism uses 304 stainless steel spring pins, achieving a repeatability accuracy of 0.05mm.

[0024] This technical solution employs a feedforward-feedback composite control system using a sensor array and LSTM prediction. The prediction module anticipates changes in operating conditions 30 seconds in advance, enabling the system to proactively adjust. A fuzzy decoupling controller eliminates temperature and humidity coupling interference, achieving a control command generation delay of less than 50ms. The distributed heating unit achieves precise local temperature control through PID regulation, keeping the temperature field uniformity error within ±1℃. The protective cover, with a wall thickness of 1.5mm, achieves a mechanical strength of 150MPa and extends the operating temperature range to 200℃. Compared to traditional PID control in the background technology, this solution improves control accuracy from ±5℃ to ±0.3℃, increases response speed by 4 times, and extends sensor lifespan by more than 3 times.

[0025] Furthermore, this application also proposes that the fuzzy decoupling controller executes a segmented control strategy, wherein: during the heating stage, temperature-dominated control with a proportional coefficient Kp = 0.3 is adopted; and during the heat preservation stage, humidity-dominated control with an integral coefficient Ki = 0.7 is adopted.

[0026] Specifically, in the temperature-dominated control during the heating phase, the proportional coefficient Kp = 0.3 is optimized based on experimental data, enabling the temperature rise rate to be controlled within the range of 3-5℃ / min, while limiting humidity fluctuations to within ±2%RH. As a preferred implementation, this phase limits the output weight of the humidity control loop, ensuring that the temperature control quantity accounts for more than 70% of the total control output. In the humidity-dominated control during the heat preservation phase, the integral coefficient Ki = 0.7 has been verified for stability, maintaining the steady-state humidity error within ±0.8%RH, while temperature fluctuations do not exceed ±0.5℃. For example, the control algorithm can trigger phase switching via time or temperature thresholds, automatically switching the control mode when the detected temperature reaches the set value of -0.5℃.

[0027] Therefore, this technical solution effectively solves the problem of temperature and humidity coupling interference by solidifying key control parameters and dividing the control into stages. Compared with the method of dynamically adjusting PID parameters in the background technology, the segmented control strategy avoids system oscillations caused by frequent parameter adjustments, improving control stability by approximately 40%. Specific combinations of proportional and integral coefficients have been experimentally verified, ensuring response speed while reducing transient overshoot to below 5%. Prioritization of control objectives is achieved through stage division: the heating stage ensures rapid temperature attainment of process requirements, while the holding stage focuses on maintaining humidity stability; the combined effect of these two stages improves the overall system control accuracy by approximately 30%.

[0028] Furthermore, this application also proposes that the network structure of the LSTM prediction module includes two hidden layers, each with 128 neurons, a dropout rate of 0.2, a prediction temperature error of ≤ ±0.3℃, and a humidity error of ≤ ±1.5%RH.

[0029] The number of neurons in the hidden layer can be configured between 96 and 160, with 128 being preferred. This number is determined using a grid search method to ensure feature extraction capability while avoiding gradient vanishing. The dropout probability of the Dropout layer can be adjusted between 0.1 and 0.3, with 0.2 selected as the optimal value through cross-validation. Error constraint metrics are validated through backpropagation on the test dataset, with root mean square error (RMSE) used for temperature prediction and mean absolute percentage error (MAPE) used for humidity prediction.

[0030] This technical solution captures the long-term dependencies in temperature and humidity time-series data using a two-layer LSTM structure. The first layer extracts local features, while the second layer integrates global patterns. The Dropout mechanism randomly masks 20% of neurons during training, forcing the network to learn redundant feature representations, thereby improving the model's generalization ability on unknown data. Error constraint metrics are embedded in the loss function through backpropagation, allowing the model to actively optimize prediction accuracy during training. Compared to a single-layer LSTM, this structure reduces temperature prediction error by 42% and humidity prediction fluctuation by 65% ​​on the test set. Existing prediction models using a single-layer 64-neuron structure have a temperature error of ±0.8℃ and a humidity error of ±3%RH. This solution achieves the required prediction accuracy for industrial-grade control through structural optimization and regularization strategies.

[0031] Furthermore, this application also proposes that the distributed heating unit includes an alumina ceramic insulation layer with a thermal conductivity of 0.8 W / m·K, covering the surface of the heating element; a 5 mm air gap is located between the insulation layer and the inner wall of the equipment, with a combined thermal conductivity ≤0.3 W / m·K.

[0032] The alumina ceramic insulation layer can be made of 95% or 99% ceramic material, with a preferred thickness of 1-3 mm, and is formed into a dense structure through plasma spraying or hot-pressing sintering processes. The air gaps can be maintained by a support column structure; the support columns can be made of silicon nitride or zirconium oxide ceramic, with a distribution density of 4-6 columns / cm². 2 The combined thermal conductivity was determined according to GB / T 10297-2015 standard, and tested at 200℃.

[0033] This technical solution achieves heat conduction control through a double-layer insulation structure. The alumina ceramic layer directly blocks heat radiation from the heating element, and the oxygen vacancy defects in its crystal structure can scatter phonon heat transfer. The air interlayer utilizes the characteristic that the free path of gas molecules is much greater than 5 mm to effectively suppress convective heat transfer. With the combined effect of these two layers, the heat flux density can be reduced to 1 / 3 of that of traditional mica sheet insulation solutions. Experimental data shows that at an ambient temperature of 180℃, the temperature of the inner wall of the equipment can be controlled below 65℃, avoiding thermal deformation of polymer material components. Compared to a single insulation layer design, this structure shortens the thermal response time of the heating unit by 15% while reducing the surface temperature gradient of the protective cover to within 2℃ / cm.

[0034] Furthermore, this application proposes a biomimetic honeycomb structure protective cover, including a snap-fit ​​installation mechanism and a hydrophobic nano-coating. The snap-fit ​​installation mechanism has a repeatability of less than 0.1 mm, and the hydrophobic nano-coating has a surface contact angle greater than 150 degrees.

[0035] The snap-fit ​​installation mechanism can employ a spring-assisted precision guide structure, where the clearance between the locating pin and the guide groove is controlled within 0.05 mm to ensure repeatability during installation. As a preferred embodiment, the snap-fit ​​components can be made of titanium alloy, whose coefficient of thermal expansion matches that of the 6061 aluminum alloy substrate. The hydrophobic nano-coating can be formed using a fluorosilane-modified silica nanoparticle dispersion, creating a micro / nano composite surface through vapor deposition. Specifically, the coating thickness can be controlled within the range of 200-500 nanometers, and the surface roughness Ra value can be between 0.1-0.3 micrometers.

[0036] This technical solution achieves a sealed connection between the protective cover and the equipment through a high-precision snap-fit ​​mechanism. The design, with a repeatability accuracy of less than 0.1 mm, effectively avoids seal failure caused by installation errors. The superhydrophobic properties of the hydrophobic nano-coating prevent water vapor penetration, and the surface with a contact angle greater than 150 degrees causes liquid water to form spherical droplets. Through the synergistic effect of these two elements, the honeycomb structure provides mechanical support, the snap-fit ​​mechanism maintains installation stability, and the nano-coating provides surface protection. At an ambient temperature of 180 degrees Celsius, this protective structure can maintain the internal sensors within an operating temperature range of 60±2 degrees Celsius. Compared to traditional protection methods, this solution, through the combination of structural design and surface treatment, improves protection reliability by more than 50%, effectively solving the problems of equipment corrosion and signal interference caused by high-temperature and high-humidity environments.

[0037] like Figure 2 Furthermore, this application also proposes a temperature and humidity feedback automatic control method for drying stainless steel degreasing sections, comprising the following steps: acquiring real-time data through a temperature and humidity sensor array; performing fuzzy decoupling calculations based on the temperature-humidity correlation matrix to generate an independent control quantity U. T =K p ·e T +K decoup ·ΔH and U H =K i ·∫e H dt+K decoup • ΔT; The LSTM model is called to predict the temperature and humidity changes of 6 historical sequences and output the changes in temperature and humidity in the next 30 seconds; The PID setting curve is dynamically adjusted according to the prediction results to control the power of the heating unit.

[0038] Specifically, the temperature and humidity sensor array adopts a distributed installation method, with each sensor node spaced 30cm apart and a sampling frequency of 1Hz. Data is transmitted to the controller via a CAN bus. The temperature-humidity correlation matrix is ​​an 8×8 fuzzy rule base, where the input variables are temperature deviation eT and humidity deviation eH, and the output variables are decoupling compensation amounts ΔH and ΔT. The input layer of the LSTM model receives 6 sets of historical data sequences, each containing 10 sampling points, with a time span of 100 seconds. The output layer uses a linear activation function to predict the temperature and humidity values ​​for the next 3 time steps. The dynamic adjustment of the PID setpoint curve is achieved by modifying the proportional band parameter, with an adjustment period of 200ms. The power output uses PWM modulation, achieving a duty cycle resolution of 0.1%.

[0039] As a preferred implementation, the Kdecoup coefficients in the fuzzy decoupling operation can be dynamically adjusted using an online learning algorithm, with an initial value set to 0.15 and a learning rate of 0.01. The hidden layers of the LSTM model can be replaced with a GRU structure, maintaining prediction accuracy even when the number of neurons is reduced to 64. The heating unit power control can employ a fuzzy PID composite algorithm, automatically enhancing the proportional gain during temperature abrupt changes.

[0040] This technical solution eliminates single-point monitoring errors through multi-sensor data fusion and separates the temperature and humidity control loop using a fuzzy decoupling algorithm, solving the coupling interference problem present in traditional PID control. LSTM time series prediction anticipates changes in operating conditions 30 seconds in advance, enabling the control system to have feedforward compensation capabilities and reducing the response lag time from the conventional 2-3 seconds to less than 200 milliseconds. Dynamic PID adjustment combined with the prediction results forms a closed-loop optimization, maintaining temperature control accuracy within ±0.5℃ and humidity fluctuation within ±2%RH even at a high temperature of 180℃. Compared to the traditional PID control method mentioned in the background technology, this solution significantly improves adaptability and control accuracy under nonlinear conditions, making it particularly suitable for the precise control requirements of temperature and humidity parameters during continuous drying of stainless steel strip.

[0041] Furthermore, this application also proposes that the fuzzy decoupling operation adopts a segmented strategy: when the temperature is 5℃ lower than the target value, the heating stage mode is activated and Kp = 0.3 is locked; when the temperature reaches the target value ±0.5℃, the heat preservation stage mode is switched to and Ki = 0.7 is locked.

[0042] Specifically, in the temperature threshold triggering mechanism, the 5℃ temperature difference threshold is determined through experimental data, effectively distinguishing between rapid heating requirements and steady-state regulation requirements. The control mode switching logic can be implemented using a state machine, where the state transition conditions are monitored in real time by a comparator circuit to detect temperature deviations. The parameter locking strategy fixes the control parameters through registers, prohibiting integral action during the heating phase and proportional action during the heat preservation phase, thus avoiding control coupling. As a preferred implementation, the value of Kp = 0.3 has been verified through transfer function simulation, enabling 80% of the heating requirement to be completed within 200ms; the value of Ki = 0.7 was determined through step response testing, allowing humidity fluctuations to be controlled within ±1%RH.

[0043] To address this, this technical solution establishes a mapping relationship between temperature state and control parameters, discretizing the continuous control process into distinct control stages. During the heating stage, a fixed proportional gain ensures rapid response, and overshoot is avoided by suppressing integral action. During the heat preservation stage, a fixed integral gain maintains steady-state accuracy, and disturbances are reduced by eliminating proportional action. Compared to existing methods that dynamically adjust PID parameters, this solution completely avoids mutual interference between temperature and humidity control loops through hard parameter locking and strict condition judgment. Experimental data shows that after adopting the segmented strategy, temperature control accuracy is improved to ±0.3℃, humidity control accuracy reaches ±1.2%RH, and power fluctuation amplitude during mode switching is reduced by 62%.

[0044] Furthermore, this application proposes that the training of the LSTM model adopts the mean squared error loss function, the input data is standardized to the [0,1] interval, and the prediction results are destandardized and output.

[0045] Specifically, the mathematical expression for the mean squared error loss function is MSE = 1 / nΣ(y_pred - y_true)^2, where n is the number of samples, y_pred is the predicted value, and y_true is the true value. This function amplifies the impact of larger errors through squaring, prompting the model to prioritize correcting significant biases. In terms of implementation, it can be directly called using TensorFlow's tf.keras.losses.MeanSquaredError() interface, or a weighted mean squared error variant can be implemented using a custom loss function.

[0046] Data standardization can be performed using the MinMaxScaler method, calculated as x' = (x - min) / (max - min), where min and max are the minimum and maximum values ​​of each feature in the training set, respectively. As a preferred implementation, the standardization parameters can be dynamically updated, using a sliding window to count the extreme values ​​of the most recent 100 data sets. Destandardization requires preserving the extreme value parameters of the original data, and the predicted values ​​are restored to their physical dimensions through the inverse operation x = x' * (max - min) + min.

[0047] Therefore, this technical solution solves the training efficiency problem caused by the difference in dimensions of multidimensional sensor data by establishing a closed loop of standardization-model training-destandardization data processing. The mean square error function ensures that the model converges stably in the standardized data space, and the standardization process makes temperature (on the order of 10^1) and humidity (on the order of 10^2) data comparable, ensuring consistent weight updates across all layers of the neural network. Experimental results show that the LSTM model using this solution reduces the number of training iterations by 37%, controls the predicted temperature error within ±0.28℃, and the humidity error does not exceed 1.3%RH, achieving a 42% improvement in accuracy compared to traditional unstandardized methods.

[0048] Furthermore, this application also proposes that the heating unit power adjustment includes: constructing a dynamic model of the temperature field through a thermocouple array; and activating a PID compensation algorithm to adjust the heating power for areas with a local temperature difference of -2℃.

[0049] The thermocouple array can use K-type or T-type thermocouples, uniformly distributed on the surface of the heating unit in a 5×5 matrix, with a spacing of 50 mm between adjacent thermocouples. The dynamic temperature field model is established using the finite element method, with a mesh generation accuracy of 1 mm. 3 Temperature distribution data is updated every 100ms. The PID compensation algorithm uses incremental control, with a proportional band of 20%, an integral time of 60s, and a derivative time of 5s. As a preferred implementation, when the temperature difference between 3×3 adjacent thermocouple regions exceeds a threshold, power compensation is automatically triggered. The compensation amount ΔP = Kp×ΔT + Ki×∫ΔTdt + Kd×d(ΔT) / dt, where ΔT is the maximum temperature difference value.

[0050] This technical solution constructs a temperature field model in real time using a high-density thermocouple array, accurately capturing localized temperature anomalies. When a temperature difference exceeding 2°C is detected, a PID algorithm with anti-integral saturation capability is employed for directional compensation, with the compensation response time controlled within 150ms. Compared to existing technologies, this solution, through the synergistic effect of spatial temperature field modeling and time-domain PID control, keeps local temperature differences within ±0.8°C, solving the regulation lag problem caused by thermal inertia in traditional heating systems. Specifically, the spatial resolution provided by the thermocouple array is 16 times higher than that of conventional single-point temperature measurement, and the parameter self-tuning function of the PID algorithm improves the system's adaptability to load changes by 40%.

[0051] Furthermore, this application also proposes a sensor thermal protection step to maintain the sensor operating temperature at 60±2℃ through a micro-airflow cooling channel, wherein the ambient temperature can reach 180℃.

[0052] Specifically, the micro-airflow cooling channel can be implemented as follows: an annular guide cavity is arranged around the sensor, and a forced convection airflow of 0.5-1.2 m / s is generated by a micro-pump; the cross-section of the airflow channel is designed with a gradually narrowing and expanding structure to enhance heat exchange efficiency; temperature control is achieved by adjusting the pump speed through PID control, using a PT100 temperature sensor as feedback signal, and the control cycle is set to 100 ms. As another implementation method, a three-dimensional mesh cooling channel can be constructed using porous metal materials, with a porosity controlled at 60%-70%, and the uniformity of airflow distribution is optimized through computational fluid dynamics simulation. Furthermore, the cooling medium can be dry compressed air or nitrogen, and the dew point temperature must be below -40°C to prevent condensation.

[0053] Therefore, this technical solution, by establishing a dynamically adjustable forced convection heat transfer system, stably controls the sensor core temperature within a precise range of 60±2℃ in a 180℃ high-temperature environment. Compared to traditional static thermal insulation, the micro-airflow cooling channel can actively dissipate heat and quickly respond to temperature fluctuations, solving the sensor drift problem caused by high-temperature conduction. Experimental data shows that after adopting this solution, the sensor's temperature drift at 180℃ is reduced from ±8℃ to ±0.5℃, and the measurement error is reduced by 82%. Precise temperature control ensures that the sensor always operates within its optimal temperature range, avoiding measurement inaccuracies caused by changes in the thermocouple Seebeck coefficient, and extending the sensor's lifespan.

[0054] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A temperature and humidity feedback automatic control system for drying a stainless steel degreasing section, characterized in that, include: A temperature and humidity sensor array is distributed and installed on the inner wall of the drying oven to collect temperature and humidity data in real time. A fuzzy decoupling controller is connected to the temperature and humidity sensor array, has a built-in temperature-humidity correlation matrix, and outputs decoupling control commands. The LSTM prediction module takes 6 sets of historical temperature and humidity sequences as input and outputs the predicted temperature and humidity values ​​for the next 30 seconds. Distributed heating unit, including PID power regulator and thermocouple array, with a response time of <200ms; A biomimetic honeycomb structure protective cover encloses the sensors and controllers. It is made of 6061 aluminum alloy and has a wall thickness of 1.5mm.

2. The automatic control system for temperature and humidity feedback in the stainless steel degreasing section drying process according to claim 1, characterized in that: The fuzzy decoupling controller executes a segmented control strategy, wherein: During the heating phase, temperature control with a proportional coefficient Kp = 0.3 is employed. During the heat preservation stage, humidity control is dominated by an integral coefficient Ki = 0.

7.

3. The automatic control system for temperature and humidity feedback in the stainless steel degreasing section drying process according to claim 1, characterized in that: The network structure of the LSTM prediction module includes: Two hidden layers, each with 128 neurons; The dropout rate is 0.2; Predicted temperature error ≤ ±0.3℃, humidity error ≤ ±1.5%RH.

4. The automatic control system for temperature and humidity feedback in the stainless steel degreasing section drying process according to claim 1, characterized in that: The distributed heating unit includes: An alumina ceramic insulation layer covers the surface of the heating element; A 5mm air gap is located between the insulation layer and the inner wall of the equipment, with a combined thermal conductivity of ≤0.3W / m·K.

5. The system according to claim 1, characterized in that: The biomimetic honeycomb structure protective cover includes: The snap-fit ​​installation mechanism has a repeatability accuracy of <0.1mm. Hydrophobic nano-coating with a surface contact angle >150°.

6. A method for automatic control of temperature and humidity feedback in the drying section of stainless steel degreasing, characterized in that, Includes the following steps: Real-time data is collected via an array of temperature and humidity sensors. Based on the temperature-humidity correlation matrix, fuzzy decoupling calculations are performed to generate independent control variables: UT=Kp·eT+Kdecoup·ΔH UH=Ki·∫eHdt+Kdecoup·ΔT The LSTM model is used to predict the temperature and humidity changes in the next 30 seconds. The PID setting curve is dynamically adjusted based on the prediction results to control the power of the heating unit.

7. The automatic control method for temperature and humidity feedback in the drying section of stainless steel degreasing zone according to claim 6, characterized in that: The fuzzy decoupling operation adopts a segmentation strategy: When the temperature is 5°C below the target value, the heating phase mode is activated, and Kp = 0.3 is locked. When the temperature reaches the target value ±0.5℃, switch to the heat preservation mode and lock Ki = 0.

7.

8. The automatic control method for temperature and humidity feedback in the drying section of stainless steel degreasing zone according to claim 6, characterized in that: The LSTM model is trained using the mean squared error loss function, with the input data standardized to the [0,1] interval and the prediction results output inversely standardized.

9. The automatic control method for temperature and humidity feedback in the drying section of stainless steel degreasing zone according to claim 6, characterized in that: The power adjustment of the heating unit includes: A dynamic model of the temperature field is constructed using a thermocouple array; For areas with a local temperature difference greater than 2℃, a PID compensation algorithm is activated to adjust the heating power.

10. The automatic control method for temperature and humidity feedback in the drying section of stainless steel degreasing zone according to claim 6, characterized in that: It also includes sensor thermal protection measures: The sensor's operating temperature is maintained at 60±2℃ through a micro-airflow cooling channel.

Images