Intelligent AI temperature control electronic valve with multi-working condition adaptive adjustment

Abstract

This invention relates to the field of temperature-controlled electronic valve technology and discloses an intelligent AI temperature-controlled electronic valve with multi-condition adaptive adjustment. The valve includes a valve body, a sleeve fixedly connected to the inner wall of the valve body, a valve core slidably connected to the inner wall of the sleeve, a valve hole on the side wall of the valve core, a drive cylinder fixedly connected to the upper end of the valve body, and a firing rod fixedly connected to the upper end of the valve core. The side wall of the firing rod is rotatably and permeably connected to the side walls of the valve body and the sleeve. Two rectangular rods are fixedly connected to the top of the drive cylinder, and a sliding plate is slidably connected to the side walls of the two rectangular rods. This invention collects multi-dimensional data of temperature, pressure, and flow rate through a sensor array, and combines this with the condition recognition algorithm built into the AI ​​control chip. It can automatically identify multiple operating conditions such as different fluid media, different pressure levels, different temperature ranges, and dynamic load changes, without manual intervention, automatically adapting and adjusting parameters to achieve seamless switching of operating conditions.

Description

Technical Field

[0001] This invention relates to the field of temperature control electronic valve technology, and in particular to an intelligent AI temperature control electronic valve with multi-condition adaptive adjustment. Background Technology

[0002] As a core component for fluid temperature and flow control, temperature-controlled electronic valves are widely used in various fields such as industrial production, household appliances, and high-end equipment.

[0003] Most existing temperature-controlled electronic valves employ a single temperature sensor and fixed algorithm control mode, which has the following drawbacks: Poor adaptability to operating conditions: Most valves are designed only for specific operating conditions (such as fixed medium and fixed pressure). When the operating conditions change (such as medium replacement, pressure fluctuation, or sudden change in target temperature), the control parameters cannot be automatically adapted, resulting in decreased temperature control accuracy, delayed response, and even valve jamming and leakage. It is difficult to meet the needs of multiple scenarios for reuse. The low level of intelligence, lack of AI intelligent learning and autonomous decision-making capabilities, inability to analyze historical operating data and predict future changes in operating conditions, requires frequent manual intervention to adjust control parameters, increases operation and maintenance costs, and makes it difficult to achieve efficient and stable operation throughout the entire life cycle. Summary of the Invention

[0004] The technical problem to be solved by this invention is that the existing temperature control electronic valves have the disadvantages of poor adaptability to working conditions and low level of intelligence. To this end, we propose an intelligent AI temperature control electronic valve with multi-condition adaptive adjustment.

[0005] To achieve the above objectives, this application adopts the following technical solution: an intelligent AI temperature control electronic valve with multi-condition adaptive adjustment, comprising a valve body, a sleeve fixedly connected to the inner wall of the valve body, a valve core slidably connected to the inner wall of the sleeve, a valve hole opened on the side wall of the valve core, a drive cylinder fixedly connected to the upper end of the valve body, a firing rod fixedly connected to the upper end of the valve core, the side wall of the firing rod being rotatably and sealingly connected to the side wall of the valve body and the sleeve, two rectangular rods fixedly connected to the top of the drive cylinder, a sliding plate slidably connected to the side wall of the two rectangular rods, a threaded cylinder fixedly connected to the lower end of the sliding plate, the inner wall of the threaded cylinder being threadedly connected to the side wall of the firing rod, and the top of the drive cylinder being elastically connected to the upper end of the sliding plate through multiple magnetic springs.

[0006] Preferably, a sensor group is fixedly connected to the inner wall of the valve body via a bracket, and an AI intelligent controller is fixedly connected to the side wall of the drive cylinder. The sensor group, the AI ​​intelligent controller, and multiple magnetic springs are electrically connected. The sensor group can sense various operating conditions such as temperature, pressure, and flow rate within the valve body.

[0007] Preferably, the AI ​​intelligent controller includes an AI control chip, a data storage module, a wireless communication module, and a power supply module. The AI ​​control chip is used for adaptive regulation of the temperature control electronic valve. The AI ​​control chip has built-in operating condition recognition algorithm, adaptive regulation algorithm, and fault diagnosis algorithm. The data storage module is used for long-term storage, classification management, and traceability of various types of data. The wireless communication module is used to establish bidirectional communication between the temperature control electronic valve and external devices.

[0008] Preferably, the working condition identification algorithm uses a combination of a BP neural network and a Softmax classifier, and its feature vector is defined as:

[0009] in, For the inlet temperature, For the outlet temperature, The ambient temperature of the valve body. The pressure on both sides of the temperature control electronic valve, This represents the fluid flow rate.

[0010] Preferably, the working condition recognition algorithm is divided into an input layer, a hidden layer, and an output layer, wherein: The hidden layer uses ReLU as the activation function to achieve non-linear feature extraction, and the formula is as follows:

[0011] in, Let be the connection weight from the i-th node in the input layer to the j-th node in the hidden layer. Let be the bias of the j-th node in the hidden layer. This is the output value of the j-th node in the hidden layer; The output layer uses the Softmax activation function to output the probability that the current working condition belongs to each preset working condition category, as shown in the formula:

[0012] in, The connection weights from the j-th node in the hidden layer to the k-th node in the output layer are: This is the bias of the k-th node in the output layer. Let be the probability that the current working condition belongs to the k-th type of working condition, and satisfy . (M represents the total number of preset working condition categories); The operating condition identification algorithm selects the operating condition category with the highest probability as the current operating condition, using the following formula:

[0013] in, This represents the final identified current operating condition.

[0014] Preferably, the adaptive control algorithm uses the deviation between the target temperature and the real-time outlet temperature as the core input, combines the operating condition identification results to achieve adaptive correction of the control parameters, and introduces LSTM temperature prediction for feedforward compensation to avoid temperature overshoot and lag. The specific control steps are as follows: S1. Definition of Temperature Deviation: The difference between the target temperature and the real-time outlet temperature is calculated and used as the basis for control. The formula is:

[0015] in, The target temperature set by the user, Let be the real-time outlet temperature at time t. Temperature deviation at any given time; S2. Improved Adaptive PID Calculation: PID parameters (proportional coefficient, integral coefficient, derivative coefficient) are identified according to operating conditions. Dynamic adjustment avoids the poor adaptability to operating conditions caused by fixed parameters. The PID formula is:

[0016] in, , and For working conditions The corresponding adaptive PID parameters and their relationship with the baseline parameters are as follows:

[0017] in, , and The preset baseline PID parameters, , and For working conditions The corresponding parameter correction coefficients are learned by the AI ​​control chip through historical operating data; S3. Integral Limiting: To avoid overshoot caused by the accumulation of the integral term, the integral term is limited. The formula is as follows:

[0018] in, The preset maximum score is... For the amplitude-limiting function, ensure that the integral term is within [ Within the range; S4. LSTM Temperature Prediction and Feedforward Compensation: Using an LSTM neural network to predict future temperature t+ The outlet temperature at time t is predicted to anticipate temperature change trends, enabling feedforward compensation and reducing lag errors. The formula is as follows: Forgotten Gate:

[0019] Input Gate:

[0020] Candidate state:

[0021] Cell state:

[0022] Output gate:

[0023] Hidden state:

[0024] in, It is the Sigmoid activation function. The hyperbolic tangent activation function is used. For element-wise multiplication, The hidden state at time t-1 Let be the input features at time t. and Here is the weight matrix for each gate. and These are the bias vectors for each gate; LSTM hidden state-based output of future temperature prediction values:

[0025] in, To predict the output weight matrix, To predict the output bias, for Predicted exit temperature at any given time; Calculation of prediction bias and feedforward compensation term:

[0026]

[0027] in, To predict the deviation between the predicted temperature and the target temperature, To predict the compensation coefficient, This is the feedforward compensation control quantity; S5. Final Control Quantity and Coil Power Output: Integrating the PID output and feedforward compensation term, the total control quantity is obtained. Then, the total control quantity is limited to obtain the final drive power of the electromagnetic coil. The formula is:

[0028] in, This is the total control quantity. The driving power of the electromagnetic coil, These are the minimum and maximum limits for coil power, respectively, to ensure stable operation of the drive components.

[0029] Preferably, the fault diagnosis algorithm is based on residual analysis and threshold judgment of multi-parameter sensor data, combined with fault probability calculation, to achieve accurate identification and judgment of various valve faults. Its control steps are as follows: S1. Definition of Sensor Group Residual: The difference between the real-time value and the model prediction value of each sensor is calculated and used as the core indicator for fault identification. The formula is:

[0030] in, Let be the real-time acquired value of the i-th sensor at time t. This represents the predicted value of the i-th sensing quantity obtained by the AI ​​control chip through a prediction model. Let be the residual value of the i-th sensor; S2. Fault Probability Calculation: For various preset faults, the probability of fault occurrence is calculated based on the residual values ​​of the corresponding sensors. The formula is as follows:

[0031] in, Let be the probability of occurrence of the m-th type of fault. Let m be the set of sensors associated with the m-th type of fault. Let be the noise standard deviation of the i-th sensor. As a normalization factor, it ensures that the sum of all failure probabilities is 1; S3. Final Fault Judgment: Select the fault type with the highest probability as the current fault judgment result. The formula is:

[0032] in, The final fault type is determined.

[0033] Preferably, the AI ​​control chip adopts an embedded ARM chip, integrates a neural network processor, has a computing speed of ≥1GHz, and a control response time of ≤50ms; the data storage module adopts a non-volatile memory chip with a storage capacity of ≥16GB; the wireless communication module supports Bluetooth, WiFi or industrial Ethernet communication, and has edge computing capabilities.

[0034] The technical effects and advantages of this invention are as follows: In this invention, multi-dimensional data such as temperature, pressure, and flow are collected by a sensor group. Combined with the working condition recognition algorithm built into the AI ​​control chip, it can automatically identify multiple working conditions such as different fluid media, different pressure levels, different temperature ranges, and dynamic load changes. Without manual intervention, it can automatically adapt and adjust the control parameters to achieve seamless switching of working conditions, thus solving the problems of poor working condition adaptability and insufficient reusability in multiple scenarios of existing temperature control electronic valves.

[0035] In this invention, an AI adaptive control algorithm is used in combination with a magnetic spring drive to achieve real-time optimization and precise output of control parameters. Compared with traditional temperature control electronic valves, the control accuracy and response speed are significantly improved.

[0036] This invention employs AI intelligent learning, autonomous decision-making, and fault diagnosis capabilities to analyze historical operating data, predict changes in operating conditions, and continuously optimize control performance. It achieves remote control, data query, and fault alarm through a wireless communication module, eliminating the need for frequent manual intervention to adjust parameters, thus reducing operation and maintenance costs. The fault diagnosis function can quickly identify fault types and provide feedback, facilitating timely maintenance by users and improving the valve's operation and maintenance efficiency and overall lifecycle stability. Attached Figure Description

[0037] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts: Figure 1 This is a schematic diagram of the structure of an intelligent AI temperature control electronic valve with multi-condition adaptive adjustment according to the present invention. Figure 2 for Figure 1 A schematic diagram of the vertical sectional structure; Figure 3 This is a schematic diagram of the AI ​​intelligent controller framework in this invention.

[0038] Legend: 1. Valve body; 2. Sleeve; 3. Valve core; 4. Valve hole; 5. Drive cylinder; 6. Rifling rod; 7. Rectangular rod; 8. Slide plate; 9. Threaded cylinder; 10. Magnetic spring; 11. Sensor group; 12. AI intelligent controller. Detailed Implementation

[0039] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0040] Reference Figures 1-3 As shown, the present invention provides a technical solution: an intelligent AI temperature control electronic valve with multi-condition adaptive adjustment, including a valve body 1, a sleeve 2 fixedly connected to the inner wall of the valve body 1, a valve core 3 slidably connected to the inner wall of the sleeve 2, a valve hole 4 opened on the side wall of the valve core 3, a drive cylinder 5 fixedly connected to the upper end of the valve body 1, a bolt 6 fixedly connected to the upper end of the valve core 3, the side wall of the bolt 6 being rotatably connected to the side wall of the valve body 1 and the sleeve 2 through sealing, two rectangular rods 7 fixedly connected to the top of the drive cylinder 5, a slide plate 8 slidably connected to the side wall of the two rectangular rods 7, a threaded cylinder 9 fixedly connected to the lower end of the slide plate 8, the inner wall of the threaded cylinder 9 being threadedly connected to the side wall of the bolt 6, and the top of the drive cylinder 5 being elastically connected to the upper end of the slide plate 8 through multiple magnetic springs 10.

[0041] A sensor group 11 is fixedly connected to the inner wall of the valve body 1 via a bracket, and an AI intelligent controller 12 is fixedly connected to the side wall of the drive cylinder 5. The sensor group 11, the AI ​​intelligent controller 12, and multiple magnetic springs 10 are electrically connected. The sensor group 11 can sense the temperature, pressure, and flow rate of the valve body 1 under various operating conditions.

[0042] The AI ​​intelligent controller 12 includes an AI control chip, a data storage module, a wireless communication module, and a power supply module. The AI ​​control chip is used for adaptive regulation of the temperature control electronic valve. The AI ​​control chip has built-in operating condition recognition algorithm, adaptive regulation algorithm, and fault diagnosis algorithm. The data storage module is used for long-term storage, classification management, and traceability of various types of data. The wireless communication module is used to establish bidirectional communication between the temperature control electronic valve and external devices.

[0043] The working condition identification algorithm uses a combination of a BP neural network and a Softmax classifier, and its feature vector is defined as:

[0044] in, For the inlet temperature, For the outlet temperature, The ambient temperature of the valve body. The pressure on both sides of the temperature control electronic valve, This represents the fluid flow rate.

[0045] The working condition recognition algorithm consists of an input layer, a hidden layer, and an output layer, wherein: The hidden layer uses ReLU as the activation function to achieve non-linear feature extraction, as shown in the formula:

[0046] in, Let be the connection weight from the i-th node in the input layer to the j-th node in the hidden layer. Let be the bias of the j-th node in the hidden layer. This is the output value of the j-th node in the hidden layer; The output layer uses the Softmax activation function to output the probability that the current working condition belongs to each preset working condition category. The formula is as follows:

[0047] in, The connection weights from the j-th node in the hidden layer to the k-th node in the output layer are: This is the bias of the k-th node in the output layer. Let be the probability that the current working condition belongs to the k-th type of working condition, and satisfy . (M represents the total number of preset working condition categories); The operating condition identification algorithm selects the operating condition category with the highest probability as the current operating condition, using the following formula:

[0048] in, This represents the final identified current operating condition.

[0049] The adaptive control algorithm uses the deviation between the target temperature and the real-time outlet temperature as the core input, and combines the operating condition identification results to achieve adaptive correction of the control parameters. Simultaneously, it introduces LSTM temperature prediction for feedforward compensation to avoid temperature overshoot and lag. The specific control steps are as follows: S1. Definition of Temperature Deviation: The difference between the target temperature and the real-time outlet temperature is calculated and used as the basis for control. The formula is:

[0050] in, The target temperature set by the user, Let be the real-time outlet temperature at time t. Temperature deviation at any given time; S2. Improved Adaptive PID Calculation: PID parameters (proportional coefficient, integral coefficient, derivative coefficient) are identified according to operating conditions. Dynamic adjustment avoids poor adaptability to operating conditions caused by fixed parameters. The PID formula is:

[0051] in, and For working conditions The corresponding adaptive PID parameters and their relationship with the baseline parameters are as follows:

[0052] in, and The preset baseline PID parameters, and For working conditions The corresponding parameter correction coefficients are learned by the AI ​​control chip through historical operating data; S3. Integral Limiting: To avoid overshoot caused by the accumulation of the integral term, the integral term is limited. The formula is as follows:

[0053] in, The preset maximum score is... For the amplitude-limiting function, ensure that the integral term is within [ Within the range; S4. LSTM Temperature Prediction and Feedforward Compensation: Using an LSTM neural network to predict future temperature t+ The outlet temperature at time t is predicted to anticipate temperature change trends, enabling feedforward compensation and reducing lag errors. The formula is as follows: Forgotten Gate:

[0054] Input Gate:

[0055] Candidate state:

[0056] Cell state:

[0057] Output gate:

[0058] Hidden state:

[0059] in, It is the Sigmoid activation function. The hyperbolic tangent activation function is used. For element-wise multiplication, The hidden state at time t-1 Let be the input features at time t. and Here is the weight matrix for each gate. and These are the bias vectors for each gate; LSTM hidden state-based output of future temperature prediction values:

[0060] in, To predict the output weight matrix, To predict the output bias, for Predicted exit temperature at any given time; Calculation of prediction bias and feedforward compensation term:

[0061] in, To predict the deviation between the predicted temperature and the target temperature, To predict the compensation coefficient, This is the feedforward compensation control quantity; S5. Final Control Quantity and Coil Power Output: Integrating the PID output and feedforward compensation term, the total control quantity is obtained. Then, the total control quantity is limited to obtain the final drive power of the electromagnetic coil. The formula is:

[0062] in, This is the total control quantity. The driving power of the electromagnetic coil, These are the minimum and maximum limits for coil power, respectively, to ensure stable operation of the drive components.

[0063] The fault diagnosis algorithm, based on residual analysis and threshold judgment of multi-parameter sensor data, combined with fault probability calculation, achieves accurate identification and judgment of various valve faults. Its control steps are as follows: S1. Definition of Sensor Group Residual: The difference between the real-time value and the model prediction value of each sensor is calculated and used as the core indicator for fault identification. The formula is:

[0064] in, Let be the real-time acquired value of the i-th sensor at time t. This represents the predicted value of the i-th sensing quantity obtained by the AI ​​control chip through a prediction model. The residual value of the i-th sensor; S2. Fault Probability Calculation: For various preset faults, the probability of fault occurrence is calculated based on the residual values ​​of the corresponding sensors. The formula is as follows:

[0065] in, Let be the probability of occurrence of the m-th type of fault. Let m be the set of sensors associated with the m-th type of fault. Let be the noise standard deviation of the i-th sensor. As a normalization factor, ensure that the sum of all fault probabilities is 1; S3, Final Fault Judgment: Select the fault type with the highest fault probability as the current fault judgment result, the formula is: in, The final fault type is determined.

[0066] The AI ​​control chip uses an embedded ARM chip with an integrated neural network processor, a computing speed of ≥1GHz, and a control response time of ≤50ms; the data storage module uses a non-volatile memory chip with a storage capacity of ≥16GB; the wireless communication module supports Bluetooth, WiFi, or industrial Ethernet communication and has edge computing capabilities.

[0067] Working principle: During the use of this intelligent AI temperature control electronic valve, the temperature sensor, pressure sensor and flow sensor in sensor group 11 are activated simultaneously to collect data such as inlet and outlet temperature, pressure and flow rate of the fluid in real time. The collected data is transmitted to the AI ​​control chip in real time. The AI ​​control chip uses the working condition recognition algorithm and combines it with the pre-stored working condition feature parameter library to automatically identify the current operating condition, determine the fluid medium type, pressure level, temperature range and load status, and identify the trend of working condition changes. Subsequently, the AI ​​control chip analyzes, optimizes PID, and predicts the real-time temperature data based on the operating condition identification results and the target temperature set by the user through an adaptive control algorithm. It automatically calculates the optimal control parameters and sends control commands to the adaptive drive component. The electromagnetic coil adjusts its power according to the control commands, regulates its current, and then adjusts the extension and retraction of multiple magnetic springs 10, causing the slide plate 8 to move up or down. Through the threaded cylinder 9, the bolt rod 6 rotates forward or backward, causing the valve core 3 to rotate and adjusting the position of the valve hole 4, thereby autonomously adjusting the temperature control electronic valve.

[0068] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A smart AI temperature control electronic valve with multi-condition adaptive adjustment, characterized in that, The device includes a valve body 1, a sleeve 2 fixedly connected to the inner wall of the valve body 1, a valve core 3 slidably connected to the inner wall of the sleeve 2, a valve hole 4 opened on the side wall of the valve core 3, a drive cylinder 5 fixedly connected to the upper end of the valve body 1, a bolt 6 fixedly connected to the upper end of the valve core 3, the side wall of the bolt 6 being rotatably connected to the side walls of the valve body 1 and the sleeve 2 through a sealing connection, two rectangular rods 7 fixedly connected to the top of the drive cylinder 5, a slide plate 8 slidably connected to the side walls of the two rectangular rods 7, a threaded cylinder 9 fixedly connected to the lower end of the slide plate 8, the inner wall of the threaded cylinder 9 being threadedly connected to the side wall of the bolt 6, and the top of the drive cylinder 5 being elastically connected to the upper end of the slide plate 8 by multiple magnetic springs 10.

2. The intelligent AI temperature control electronic valve with multi-condition adaptive adjustment according to claim 1, characterized in that: A sensor group 11 is fixedly connected to the inner wall of the valve body 1 via a bracket, and an AI intelligent controller 12 is fixedly connected to the side wall of the drive cylinder 5. The sensor group 11, the AI ​​intelligent controller 12, and multiple magnetic springs 10 are electrically connected. The sensor group 11 can sense various operating conditions such as temperature, pressure, and flow rate inside the valve body 1.

3. The intelligent AI temperature control electronic valve with multi-condition adaptive adjustment according to claim 2, characterized in that: The AI ​​intelligent controller 12 includes an AI control chip, a data storage module, a wireless communication module, and a power supply module. The AI ​​control chip is used for adaptive regulation of the temperature control electronic valve. The AI ​​control chip has built-in working condition recognition algorithm, adaptive regulation algorithm, and fault diagnosis algorithm. The data storage module is used for long-term storage, classification management, and traceability of various types of data. The wireless communication module is used to establish bidirectional communication between the temperature control electronic valve and external devices.

4. The intelligent AI temperature control electronic valve with multi-condition adaptive adjustment according to claim 3, characterized in that: The operating condition identification algorithm uses a combination of a BP neural network and a Softmax classifier. Its eigenvector is defined as: ； in, For the inlet temperature, For the outlet temperature, The ambient temperature of the valve body. The pressure on both sides of the temperature control electronic valve, This represents the fluid flow rate.

5. The intelligent AI temperature control electronic valve with multi-condition adaptive adjustment according to claim 4, characterized in that: The working condition recognition algorithm is divided into an input layer, a hidden layer, and an output layer, wherein: The hidden layer uses ReLU as the activation function to achieve non-linear feature extraction, and the formula is as follows: ; in, Let be the connection weight from the i-th node in the input layer to the j-th node in the hidden layer. Let be the bias of the j-th node in the hidden layer. This is the output value of the j-th node in the hidden layer; The output layer uses the Softmax activation function to output the probability that the current working condition belongs to each preset working condition category, as shown in the formula: ; in, The connection weights from the j-th node in the hidden layer to the k-th node in the output layer are: This is the bias of the k-th node in the output layer. Let be the probability that the current working condition belongs to the k-th type of working condition, and satisfy . (M represents the total number of preset working condition categories); The operating condition identification algorithm selects the operating condition category with the highest probability as the current operating condition, using the following formula: ; in, This represents the final identified current operating condition.

6. The intelligent AI temperature control electronic valve with multi-condition adaptive adjustment according to claim 5, characterized in that: The adaptive control algorithm uses the deviation between the target temperature and the real-time outlet temperature as the core input, and combines the operating condition identification results to achieve adaptive correction of the control parameters. Simultaneously, it introduces LSTM temperature prediction for feedforward compensation to avoid temperature overshoot and lag. The specific control steps are as follows: S1. Definition of Temperature Deviation: The difference between the target temperature and the real-time outlet temperature is calculated and used as the basis for control. The formula is: ; in, The target temperature set by the user, Let be the real-time outlet temperature at time t. Temperature deviation at any given time; S2. Improved Adaptive PID Calculation: PID parameters (proportional coefficient, integral coefficient, derivative coefficient) are identified according to operating conditions. Dynamic adjustment avoids the poor adaptability to operating conditions caused by fixed parameters. The PID formula is: ； in, , and For working conditions The corresponding adaptive PID parameters and their relationship with the baseline parameters are as follows: in, and The preset baseline PID parameters, For working conditions The corresponding parameter correction coefficients are learned by the AI ​​control chip through historical operating data; S3. Integral Limiting: To avoid overshoot caused by the accumulation of the integral term, the integral term is limited. The formula is as follows: ; in, The preset maximum score is... For a limit function, ensure that the integral term is within the limit. Within the range; S4. LSTM Temperature Prediction and Feedforward Compensation: Using an LSTM neural network to predict future temperatures. The outlet temperature at time t is predicted to anticipate temperature change trends, enabling feedforward compensation and reducing lag errors. The formula is as follows: in, It is the Sigmoid activation function. The hyperbolic tangent activation function is used. For element-wise multiplication, The hidden state at time t-1 Let be the input features at time t. Here is the weight matrix for each gate. These are the bias vectors for each gate; LSTM hidden state-based output of future temperature prediction values: ； in, To predict the output weight matrix, To predict the output bias, for Predicted exit temperature at any given time; Calculation of prediction bias and feedforward compensation term: in, To predict the deviation between the predicted temperature and the target temperature, To predict the compensation coefficient, This is the feedforward compensation control quantity; S5. Final Control Quantity and Coil Power Output: Integrating the PID output and feedforward compensation term, the total control quantity is obtained. Then, the total control quantity is limited to obtain the final drive power of the electromagnetic coil. The formula is: in, This is the total control quantity. The driving power of the electromagnetic coil, These are the minimum and maximum limits for coil power, respectively, to ensure stable operation of the drive components.

7. The intelligent AI temperature control electronic valve with multi-condition adaptive adjustment according to claim 5, characterized in that: The fault diagnosis algorithm is based on residual analysis and threshold judgment of multi-parameter sensor data, combined with fault probability calculation, to achieve accurate identification and judgment of various valve faults. Its control steps are as follows: S1. Definition of Sensor Group Residual: The difference between the real-time value and the model prediction value of each sensor is calculated and used as the core indicator for fault identification. The formula is: in, Let be the real-time acquired value of the i-th sensor at time t. This represents the predicted value of the i-th sensing quantity obtained by the AI ​​control chip through a prediction model. Let be the residual value of the i-th sensor; S2. Fault Probability Calculation: For various preset faults, the probability of fault occurrence is calculated based on the residual values ​​of the corresponding sensors. The formula is as follows: in, Let be the probability of occurrence of the m-th type of fault. Let m be the set of sensors associated with the m-th type of fault. Let be the noise standard deviation of the i-th sensor. As a normalization factor, it ensures that the sum of all failure probabilities is 1; S3. Final Fault Judgment: Select the fault type with the highest probability as the current fault judgment result. The formula is: in, The final fault type is determined.

8. The intelligent AI temperature control electronic valve with multi-condition adaptive adjustment according to claim 3, characterized in that: The AI ​​control chip uses an embedded ARM chip with an integrated neural network processor, a computing speed of ≥1GHz, and a control response time of ≤50ms; the data storage module uses a non-volatile memory chip with a storage capacity of ≥16GB; the wireless communication module supports Bluetooth, WiFi, or industrial Ethernet communication and has edge computing capabilities.

Images