Intelligent cooling system for thickness gauge

Abstract

The utility model discloses a kind of intelligent cooling systems of thickness gauge, including thickness gauge body, device cabin, multimodal sensor group, intelligent controller, active cooling system and drive circuit board.Multimodal sensor group is set in device cabin, and intelligent controller is set in device cabin, and it include artificial intelligence processing module and control signal output unit, and artificial intelligence processing module generates control instruction, and control signal output unit is used to convert control instruction into standardized drive signal.Active cooling system includes fan assembly and liquid cooling component.Drive circuit board receives control signal and drives cooling equipment to execute.The system is through multimodal data fusion and artificial intelligence prediction algorithm, realizes the early prediction to the temperature change trend of thickness gauge, changes from traditional passive response cooling into active predictive cooling, significantly improves temperature control precision, reduces energy consumption, prolongs the service life of equipment, improves measurement precision and working stability.

Description

Technical Field

[0001] This utility model relates to the field of precision measuring instrument technology, specifically to an intelligent cooling system for a thickness gauge. Background Technology

[0002] Traditional thickness gauges typically use physical principles such as ultrasound, eddy current, or magnetic induction to measure thickness. Taking an ultrasonic thickness gauge as an example, its core components include a high-frequency ultrasonic generator, a precision receiver, a signal conditioning circuit, a digital signal processor, and a display control module. These electronic components generate a significant amount of heat during operation. In particular, the ultrasonic generator typically operates in the 1-15MHz range, with a drive power reaching tens of watts; the signal processing circuit requires extensive digital calculations to ensure measurement accuracy, often resulting in a processor frequency exceeding 100MHz; the backlight and drive circuitry of the display module also generate considerable heat. In continuous operating environments, the internal temperature of the equipment can often rise to 60-80 degrees Celsius, severely impacting measurement accuracy and equipment lifespan. Currently, the main solutions on the market for thermal management of thickness gauges are as follows. The most common is a passive cooling solution, using aluminum alloy heat sinks or a heat dissipation shell to dissipate heat into the environment through natural convection and heat conduction. This solution is simple in structure and low in cost, but its heat dissipation efficiency is limited, making it suitable only for low-power, low-end products. For high-end thickness gauges with high power consumption, active air cooling solutions are typically used, with one or more fans installed inside the device for forced convection cooling. However, traditional air cooling systems usually operate at constant speed or use simple temperature control switches, which suffer from problems such as slow response, high energy consumption, and high noise. The fans only start working or adjust their speed when the temperature exceeds a set threshold, and this passive response method is often too late, as the device temperature has already risen to a level that affects performance.

[0003] Some products are beginning to adopt liquid cooling technology, which uses circulating coolant to remove heat, resulting in significantly improved heat dissipation efficiency compared to air cooling. However, existing liquid cooling systems typically use a fixed circulation flow rate, which cannot be dynamically adjusted according to actual heat dissipation needs, leading to energy waste and overcooling issues. Furthermore, most existing cooling control systems are based on simple PID controllers, which can only adjust based on current temperature feedback and lack the ability to predict temperature change trends, making proactive preventative cooling difficult.

[0004] In summary, the main problems with existing technologies include: First, limited temperature monitoring. Existing systems typically monitor only one or a few temperature points inside the equipment, failing to comprehensively reflect the equipment's thermal state or detect the impact of environmental factors on the equipment's temperature. Second, passive control strategies. Temperature threshold-based switching control or PID regulation are passive response modes, only adjusting when the temperature is abnormal, exhibiting significant lag. Third, lack of predictive capability. Existing systems cannot predict temperature trends based on equipment operating status, environmental changes, etc., making proactive preventative cooling difficult. Fourth, insufficient energy consumption optimization. Fixed-speed fans and fixed-flow liquid cooling systems cannot dynamically adjust according to actual needs, resulting in unnecessary energy waste. Fifth, low system intelligence. Lacking self-learning and adaptive capabilities, they cannot optimize control strategies based on long-term operating data. Utility Model Content

[0005] The purpose of this invention is to develop a cooling system that possesses multi-dimensional environmental sensing capabilities, enabling real-time monitoring of various environmental parameters such as temperature, humidity, vibration, and pressure; intelligent prediction capabilities, allowing analysis of multimodal data based on artificial intelligence algorithms to predict equipment temperature change trends; proactive control capabilities, enabling advance adjustment of cooling intensity based on prediction results to achieve predictive cooling; and adaptive optimization capabilities, continuously improving prediction models and control strategies based on historical operating data.

[0006] To achieve the above objectives, this utility model provides an intelligent cooling system for a thickness gauge, which mainly includes the following technical features: A multimodal sensor array, installed inside the device compartment, is used to synchronously collect temperature data and other environmental parameter data; The device compartment, located inside or outside the thickness gauge body, is used to house the cooling system components; The intelligent controller, located within the device compartment, includes: an artificial intelligence processing module for receiving sensor data, executing predictive models, formulating cooling strategies, and generating control commands; and a control signal output unit for converting the control commands into standardized drive signals. An active cooling system includes: a fan assembly disposed within the device compartment, with adjustable speed; and a liquid cooling assembly including copper pipes and coolant, wherein the flow rate of the coolant in the copper pipes is controllable. The thickness gauge body is used to measure the thickness of the object to be measured; The drive circuit board, located inside the device compartment, is electrically connected to the intelligent controller and the active cooling system. It receives control signals and drives the cooling equipment to perform predictive cooling.

[0007] The intelligent controller includes a microprocessor chip, a memory, an analog-to-digital converter, and a communication interface circuit. The microprocessor chip is used to run the algorithm program of the artificial intelligence processing module.

[0008] The multimodal sensor group includes at least two of the following: temperature sensor, humidity sensor, vibration sensor, and pressure sensor.

[0009] The artificial intelligence processing module includes a data preprocessing unit, a feature extraction unit, a prediction model execution unit, and a control strategy generation unit; the prediction model is a machine learning model, including at least one of neural networks, support vector machines, or random forests.

[0010] The control strategy generation unit calculates the optimal fan speed and coolant flow rate parameters based on the prediction results and generates the corresponding PWM control signal.

[0011] The driver circuit board includes a signal amplification circuit, a power drive circuit, and a protection circuit.

[0012] The drive circuit board also includes a status monitoring circuit, which monitors the operating status of the cooling equipment and feeds back fault information to the intelligent controller.

[0013] The system also includes a display screen, which is set on the surface of the thickness gauge body, to display the current temperature status, predicted temperature changes, and the working status of the cooling system.

[0014] The memory is used to store historical sensor data, prediction model parameters, and cooling control effect data, providing data support for the adaptive optimization of the prediction model.

[0015] The artificial intelligence processing module has a self-learning function, continuously optimizing the prediction model and control strategy based on historical cooling effect data.

[0016] This utility model has the following beneficial effects: To address the issue of single-mode temperature monitoring in existing technologies, this invention employs a multi-modal sensor array, significantly improving the system's ability to identify abnormal temperature trends early. Addressing the core problem of passive control strategies, the system can predict temperature change trends in advance based on multi-modal sensor data and historical operating experience, adjusting cooling intensity before the temperature reaches a critical value. This predictive cooling control method completely changes the traditional passive mode of cooling only after the temperature rises, achieving an active mode of preventing cooling by anticipating temperature increases, effectively avoiding excessive temperature rises, and making equipment temperature control smoother and more stable.

[0017] To address the technical bottleneck of insufficient predictive capabilities, this invention's artificial intelligence processing module integrates multiple predictive models such as neural networks, support vector machines, and random forests. Through data preprocessing and feature extraction, it identifies key patterns and regularities in temperature changes. Simultaneously, intelligent dynamic adjustment significantly reduces unnecessary energy waste and extends the device's lifespan.

[0018] This invention features real-time intelligent control and long-term adaptive learning capabilities. The artificial intelligence processing module continuously analyzes system operating data, automatically adjusts predictive model parameters and control strategy weights, and also has a fault early warning function, enabling early detection of potential fault risks. This significantly reduces the frequency of overheat protection activation and unexpected downtime, improving equipment availability and reliability. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments or examples of this utility model, the drawings used in the embodiments or examples will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this utility model. For those skilled in the art, other drawings can be obtained according to these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the overall structure of an intelligent cooling system for a thickness gauge; Figure 2 This is a diagram of the intelligent system architecture of an intelligent cooling system for a thickness gauge. Figure 3 This is a flowchart of the intelligent control system for an intelligent cooling system of a thickness gauge.

[0021] Reference numerals: 1-Multimodal sensor group; 2-Fan assembly; 3-Liquid cooling assembly; 4-Thickness gauge body. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, 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 application pertains.

[0024] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0025] Example 1 Reference Figures 1 to 3 The thickness gauge body 4 adopts the ultrasonic thickness measurement principle, equipped with a high-frequency ultrasonic probe with a working frequency of 5MHz. The measurement range covers 0.75 mm to 300 mm, and the measurement accuracy reaches ±0.01 mm. The body integrates an ultrasonic generator, receiver, signal conditioning circuit, digital signal processor, and LCD display module. These electronic components generate approximately 35 watts of heat power during operation, requiring an effective cooling system to maintain normal operating temperature.

[0026] The device compartment is constructed of 6061 aluminum alloy, with external dimensions of 200 mm long, 150 mm wide, and 80 mm high, a wall thickness of 2 mm, and a weight of approximately 1.2 kg. The compartment surface undergoes anodizing treatment, which not only improves corrosion resistance but also enhances heat dissipation. The compartment is secured to the side of the thickness gauge body 4 with four M6 bolts made of stainless steel to ensure reliable connection. The interior of the compartment features a partition structure, dividing the space into three functional areas: a sensor installation area, a controller installation area, and a cooling equipment installation area. These areas are electrically connected via wiring channels, resulting in a rational layout and convenient maintenance.

[0027] The multimodal sensor group 1 is the core component of the system's environmental sensing, comprising four types: temperature sensor, humidity sensor, vibration sensor, and pressure sensor. The temperature sensor is a DS18B20 digital temperature sensor, which uses a one-wire bus interface, has a measurement accuracy of ±0.5 degrees Celsius, a measurement range from -55 degrees Celsius to +125 degrees Celsius, and a response time of less than 750 milliseconds. The sensor is encapsulated in a stainless steel probe and makes close contact with the key heat-generating parts of the thickness gauge body 4 via thermally conductive silicone grease, ensuring accurate temperature measurement. The humidity sensor is a DHT22 type digital temperature and humidity integrated sensor, with a humidity measurement accuracy of ±2%RH, a measurement range covering 0 to 100% relative humidity, and also provides temperature information as a supplement. This sensor is installed in the central position inside the device compartment, accurately reflecting changes in the humidity of the compartment environment.

[0028] The vibration sensor is an ADXL345 triaxial digital accelerometer with a measurement range of ±16g, a resolution of 13 bits, and a maximum sampling frequency of 3200Hz. This sensor is fixed to the outer shell of the thickness gauge body 4 using double-sided adhesive tape and is used to detect the vibration state during equipment operation, as vibration is often related to the heat generation of internal components. The pressure sensor is a BMP280 digital barometric pressure sensor with a measurement accuracy of ±0.12 hPa. Besides monitoring atmospheric pressure changes, it can also detect pressure changes inside the chamber caused by temperature variations, providing additional environmental parameter information to the system. All sensors are connected to the intelligent controller via an I2C bus with a frequency set to 400kHz to ensure real-time and reliable data transmission.

[0029] The intelligent controller is the core processing unit of the entire system, using an STM32F407VGT6 microcontroller as the main processing chip. This chip is based on the ARM Cortex-M4 core, with a clock speed of 168MHz, and integrates a single-precision floating-point unit, 512KB of Flash program memory, and 192KB of SRAM data memory. To meet the demands of large-scale data processing and model calculations, an additional 32MB of SDRAM is added as a data cache, and 16MB of SPI Flash memory is used to store historical data and model parameters. The controller also integrates a 12-bit precision analog-to-digital converter with a conversion speed of 2.4MSPS, supporting simultaneous sampling of 16 channels, providing the hardware foundation for the digitization of sensor signals.

[0030] The artificial intelligence processing module is the core software of the intelligent controller. It adopts a modular design, comprising four main functional modules: a data preprocessing unit, a feature extraction unit, a prediction model execution unit, and a control strategy generation unit. The data preprocessing unit is responsible for filtering, calibrating, and normalizing the raw data collected by the sensors. A fifth-order Butterworth low-pass filter is used to remove high-frequency noise, with a cutoff frequency set to 50Hz. The feature extraction unit extracts key feature parameters from the preprocessed data, including temperature change trends, correlations with environmental parameters, and equipment vibration characteristics. The extracted feature vector has a 32-dimensional dimension and an update period of 100 milliseconds. The prediction model uses a lightweight feedforward neural network with an input layer of 32 nodes, two hidden layers of 64 and 32 nodes respectively, and an output layer of 8 nodes. The total number of parameters is approximately 15KB, and the single inference time is less than 5 milliseconds, enabling it to predict temperature change trends within the next 30 seconds.

[0031] The control signal output unit is responsible for converting the control strategies generated by the artificial intelligence processing module into standardized drive signals. This unit integrates a multi-channel PWM signal generator with a PWM frequency set to 25kHz and a duty cycle adjustment accuracy of 0.1%, enabling precise control of fan speed and water pump flow. Simultaneously, this unit provides multiple communication interfaces such as UART, SPI, and I2C, supporting data exchange with the drive circuit board and display screen, with a maximum communication baud rate of 115200bps, ensuring timely delivery of control commands.

[0032] The active cooling system, comprising two subsystems—fan assembly 2 and liquid cooling assembly 3—is the actuator for temperature control. Fan assembly 2 uses three 12V brushless DC fans, measuring 80mm × 80mm × 25mm, with a rated speed of 2000 rpm, a maximum airflow of 35 CFM, and a noise level below 32 decibels. The fans support PWM speed control, with a speed range from 20% to 100%, dynamically adjusting the speed according to cooling requirements. The three fans are mounted in an equilateral triangle layout on the top of the unit compartment. Optimized airflow design ensures airflow covers all heat-generating areas, improving cooling efficiency. Each fan is equipped with a speed feedback signal, using a conventional Hall effect sensor to detect the actual speed, forming a closed-loop control system.

[0033] The liquid cooling assembly 3 comprises four main parts: a water pump, a radiator, copper tubing, and coolant. The water pump is a 12V DC brushless pump with a flow rate ranging from 200 liters per hour to 800 liters per hour, a head of 1.5 meters, and supports PWM speed control. The radiator is made of aluminum alloy, measuring 120 mm × 120 mm × 27 mm, and features a dense fin structure, achieving a total heat dissipation area of ​​0.8 square meters. The copper tubing uses 4 mm inner diameter copper tubing, with a total length of 1.2 meters and a wall thickness of 0.5 mm, forming a closed-loop circulation from the water pump to the radiator and back to the water tank. The coolant is a mixture of 30% ethylene glycol and 70% deionized water. The thermal resistance of the entire liquid cooling system is approximately 0.15 degrees Celsius per watt, resulting in significantly higher cooling efficiency than traditional air-cooled systems.

[0034] The driver circuit board employs a four-layer PCB design, with a thickness of 1.6 mm and dimensions of 100 mm × 80 mm. The board integrates four functional modules: signal amplification circuit, power drive circuit, protection circuit, and status monitoring circuit. The signal amplification circuit uses an LM358 dual operational amplifier to amplify the 3.3V control signal output from the intelligent controller to 12V, providing sufficient drive capability. The power drive circuit uses an N-type MOSFET IRF540N as the power switch, with a maximum continuous current of 33A, an on-resistance of 44 milliohms, and the ability to withstand high-frequency PWM switching with a switching frequency up to 25kHz. The protection circuit includes four protection mechanisms: overcurrent protection, overtemperature protection, short-circuit protection, and reverse connection protection. The overcurrent protection threshold is set to 15A, the overtemperature protection temperature is set to 85 degrees Celsius, and the short-circuit protection response time is less than 1 microsecond. The status monitoring circuit uses an ACS712 Hall effect current sensor to monitor the load current and an NTC thermistor to monitor the driver board temperature, providing real-time feedback on the device's operating status to the intelligent controller, forming a complete monitoring closed loop.

[0035] Example 2 In this embodiment, the system's data stream begins with multimodal sensor group 1. The sensors synchronously collect environmental data at a fixed frequency of 100Hz. Within each sampling period, the temperature sensor, humidity sensor, vibration sensor, and pressure sensor operate simultaneously, generating four parallel data streams. Due to the different data formats and accuracies of the various sensors, the system first performs data standardization processing, converting all sensor data into a 32-bit floating-point format, and then performs linear calibration according to preset calibration parameters to eliminate individual sensor differences and the effects of environmental drift.

[0036] Data preprocessing is the first crucial step in the entire data flow. Raw sensor data often contains interference factors such as high-frequency noise, anomalous abrupt changes, and transmission errors. The system employs a multi-stage filtering strategy. First, a median filter is used to remove impulse noise, with the filter window size set to 5 sampling points. Next, a fifth-order Butterworth low-pass filter is used for smoothing, with a cutoff frequency set to 50Hz, effectively preserving the main trend information of temperature changes while removing irrelevant high-frequency components. For anomaly detection, the system uses the statistically based 3σ criterion. When a data point deviates from the recent mean by more than 3 standard deviations, it is marked as an anomaly and corrected using linear interpolation. The data fusion process uses a weighted average algorithm, assigning different weight coefficients based on the reliability and correlation of each sensor: temperature sensor weight is 0.4, humidity sensor weight is 0.3, vibration sensor weight is 0.2, and pressure sensor weight is 0.1, ultimately generating a comprehensive environmental state vector.

[0037] The feature extraction unit extracts key feature parameters for prediction from the preprocessed data stream. Time-domain features include statistical characteristics such as mean, variance, skewness, and kurtosis of each sensor data point, as well as dynamic characteristics such as rate of change and acceleration. Frequency-domain features are obtained through Fast Fourier Transform (FFT) and include parameters such as main frequency components, spectral energy distribution, and spectral centroid; these features reflect the periodic variation of the system. Correlation features are obtained by calculating the cross-correlation coefficients between different sensor data points, including the correlation between temperature and humidity, and the correlation between temperature and vibration; these features reveal the coupling relationships within the system. Time-series features are extracted using an autoregressive (AR) model, including parameters such as AR model coefficients and prediction error variance, to capture the time dependence of the data. All feature parameters form a 32-dimensional feature vector. The feature extraction time window is set to 10 seconds, with a sliding step of 1 second to ensure the timeliness and continuity of the features.

[0038] The prediction model is the core algorithm of the system, employing a multi-layer feedforward neural network to achieve nonlinear mapping relationships. The network's input layer receives a 32-dimensional feature vector. The first hidden layer contains 64 neurons, using the ReLU activation function to handle complex nonlinear relationships between features. The second hidden layer contains 32 neurons, also using the ReLU activation function, primarily for further feature abstraction and dimensionality reduction. The output layer contains 8 neurons, corresponding to the predicted temperature values ​​for 8 future time steps (3.75 seconds each, for a total of 30 seconds), using a linear activation function to ensure output continuity. The network is trained using the backpropagation algorithm, with mean squared error as the loss function, the Adam algorithm as the optimizer, a learning rate of 0.001, and a batch size of 32. To prevent overfitting, dropout layers are added between the hidden layers with a dropout rate of 0.2. The model has approximately 15,000 parameters, with weights stored as 32-bit floating-point numbers, and the model file size is approximately 60KB.

[0039] The control strategy generation unit formulates specific cooling control strategies based on the output of the prediction model. Strategy generation employs a multi-objective optimization method, simultaneously considering three objective functions: cooling effect, energy consumption, and noise. The cooling effect objective function is defined as the weighted error between the predicted temperature and the target temperature, with weights decreasing over time; more recent predictions have larger weights. The energy consumption objective function is defined as the weighted sum of fan and pump power consumption, encouraging the system to select low-power control strategies. The noise objective function is defined as a nonlinear function of fan speed, as the relationship between noise and speed is typically nonlinear. Multi-objective optimization uses the concept of Pareto optimality, merging multiple objective functions into a single objective function through weighted summation. The weight coefficients can be adjusted according to actual application requirements. The optimization algorithm uses gradient descent to search for the optimal combination of fan PWM duty cycle and pump PWM duty cycle.

[0040] The system's timing design ensures the coordinated operation of all functional modules. After system startup, a 3-second initialization phase occurs, including hardware self-tests, sensor calibration, and model parameter loading. During normal operation, a 10-millisecond basic clock cycle is used. Sensor data acquisition is completed in the first 2 milliseconds of each cycle, data preprocessing and feature extraction are completed in the next 5 milliseconds, prediction model inference takes 2 milliseconds, and control strategy generation takes 1 millisecond. To reduce computational load, the prediction model is not executed every cycle, but rather once every 100 milliseconds, i.e., prediction is performed once every 10 basic cycles. The output frequency of control commands is consistent with the prediction frequency, and the PWM duty cycle is updated every 100 milliseconds. The frequency of status monitoring and feedback is set to 10Hz, meaning that device operating status information is collected every 100 milliseconds.

[0041] The data storage employs a circular buffer structure, allocating 16MB of space in SDRAM for storing historical sensor data. At a sampling frequency of 100Hz, this allows for the storage of approximately 24 hours of complete data. Historical data is stored chronologically, and when the buffer is full, the oldest data is overwritten using a first-in, first-out (FIFO) principle. Important statistical information and model parameters, including sensor calibration parameters, neural network weights, and control strategy parameters, are stored in Flash memory. This data is retained even after a power outage, ensuring the system can be restored to its previous state upon restart. The system also implements data compression, using differential encoding for slowly changing parameters and frequency domain compression for periodic data, achieving an average compression ratio of 3:1, effectively saving storage space.

[0042] The system continuously optimizes the prediction model and control strategy based on historical operating data. The learning process employs online learning, eliminating the need for system downtime and retraining; instead, it incrementally updates model parameters during normal operation. Every 24 hours of system operation, a fine-tuning of model parameters is automatically triggered. By comparing the error between predicted and actual measurements, a gradient descent algorithm is used to make small adjustments to the weights of the neural network. The learning rate is set to 1 / 10 of the initial training rate to ensure the stability of the learning process. After each week of operation, the system analyzes historical control performance data and optimizes the control strategy parameters using a reinforcement learning algorithm. The reward function is defined as a comprehensive evaluation of control performance, energy consumption, and user satisfaction. Abnormal operating condition detection is achieved through statistical methods. When the system detects significant changes in environmental conditions, it increases the learning frequency to accelerate the model's adaptation process. The entire self-learning process employs a conservative update strategy, limiting single parameter adjustments to within 5% to avoid instability introduced during the learning process.

Claims

1. An intelligent cooling system for a thickness gauge, characterized in that, include: The thickness gauge body (4) is used to measure the thickness of the object to be measured. The device compartment, located outside the thickness gauge body, is used to house the cooling system components; A multimodal sensor group (1) is installed inside the device cabin to synchronously collect temperature data and other environmental parameter data; The intelligent controller includes: an artificial intelligence processing module for receiving sensor data, executing predictive models, formulating cooling strategies, and generating control commands; and a control signal output unit for converting control commands into drive signals. An active cooling system includes: a fan assembly (2) installed in the device compartment with adjustable speed; and a liquid cooling assembly (3) including copper pipes and coolant, wherein the flow rate of the coolant in the copper pipes is controllable. The drive circuit board is electrically connected to the intelligent controller and the active cooling system, receives control signals, and drives the cooling equipment to perform operations.

2. The intelligent cooling system for a thickness gauge according to claim 1, characterized in that, The intelligent controller includes a microprocessor chip, a memory, an analog-to-digital converter, and a communication interface circuit. The microprocessor chip is used to run the algorithm program of the artificial intelligence processing module.

3. The intelligent cooling system for a thickness gauge according to claim 1, characterized in that, The multimodal sensor group (1) includes at least two of the following: a temperature sensor, a humidity sensor, a vibration sensor, and a pressure sensor.

4. The intelligent cooling system for a thickness gauge according to claim 1, characterized in that, The artificial intelligence processing module includes a data preprocessing unit, a feature extraction unit, a prediction model execution unit, and a control strategy generation unit; the prediction model is a machine learning model, including at least one of neural networks, support vector machines, or random forests.

5. The intelligent cooling system for a thickness gauge according to claim 4, characterized in that, The control strategy generation unit calculates the optimal fan speed and coolant flow rate parameters based on the prediction results, and generates the corresponding PWM control signal.

6. The intelligent cooling system for a thickness gauge according to claim 1, characterized in that, The drive circuit board includes a signal amplification circuit, a power drive circuit, and a protection circuit.

7. The intelligent cooling system for a thickness gauge according to claim 6, characterized in that, The drive circuit board also includes a status monitoring circuit for monitoring the operating status of the cooling equipment and feeding back fault information to the intelligent controller.

8. The intelligent cooling system for a thickness gauge according to claim 1, characterized in that, The system also includes a display screen, which is disposed on the surface of the thickness gauge body, for displaying the current temperature status, predicted temperature changes, and the working status of the cooling system.

9. The intelligent cooling system for a thickness gauge according to claim 2, characterized in that, The memory is used to store historical sensor data, prediction model parameters, and cooling control effect data, providing data support for the adaptive optimization of the prediction model.

10. The intelligent cooling system for a thickness gauge according to claim 1, characterized in that, The artificial intelligence processing module has a self-learning function, continuously optimizing the prediction model and control strategy based on historical cooling effect data.

Images