Anti-explosion elevator control cabinet cooling device and temperature prediction control method
Through the ConvLSTM model combined with the internal and external circulation cooling device, efficient temperature management of explosion-proof elevator control cabinet is achieved, the problems of temperature management lag and low prediction accuracy are solved, and the safety and reliability of the elevator are improved.
Patent Information
- Application Number
- CN202510340442.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-20
- Publication Date
- 2025-07-08
AI Technical Summary
In the prior art, the temperature management response of explosion-proof elevator control cabinets is lagging, the prediction accuracy is low, and the excessive manual intervention leads to attenuation of electronic components and safety hazards.
The temperature prediction is performed using a hybrid neural network model based on ConvLSTM, combined with the internal and external circulation cooling device, dynamic heat dissipation control is performed through the fins of the internal circulation device and the air-cooling system and the external circulation heat pipes and liquid-cooling system to achieve preventive regulation.
It improves the response speed and accuracy of temperature management, optimizes resource allocation, improves fault warning capabilities, and ensures the safety and reliability of explosion-proof elevators.
Smart Images
Figure CN120277524A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of explosion-proof elevator control auxiliary systems, and relates to a temperature reduction device and a temperature prediction control method for an explosion-proof elevator control cabinet. Background Art
[0002] Explosion-proof elevators are widely used in places such as petrochemical, mines, and military industries where explosive gases, vapors, or dust and other hazardous substances may exist. In such environments, the design of the internal devices of explosion-proof elevators needs to consider the sealing to isolate the intrusion of external explosive gases or dust. The design of the sealing may exacerbate the retention of internal heat. To avoid internal high temperature triggering an explosion accident, it is necessary to cooperate with efficient heat dissipation technology to ensure heat dissipation and balance explosion-proof safety and temperature control efficiency.
[0003] As the core control unit of the explosion-proof elevator system, the explosion-proof elevator control cabinet integrates high-power electronic components such as inverters, contactors, and control boards. During operation, a large amount of heat is generated due to the conversion of electrical energy and component losses. If the internal temperature of the control cabinet continues to be too high, it will lead to the attenuation of the performance of electronic components, the shortening of their service life, and even cause system failures or local overheating and ignition, seriously threatening the safe operation of the explosion-proof elevator. Therefore, temperature management is a key link to ensure the stability and reliability of the explosion-proof elevator system.
[0004] In recent years, with the rapid development of artificial intelligence technology, neural networks have shown significant advantages in the field of time series prediction. By constructing a deep learning model, it is possible to learn the law of temperature change from historical data and combine multi-dimensional environmental and operating parameters to accurately predict the future temperature trend of the control cabinet. Compared with traditional real-time monitoring technologies, the temperature prediction method based on neural networks can achieve preventive regulation, optimize resource allocation, and improve the fault warning ability.
[0005] Therefore, in view of the above technical problems, it is necessary to provide a temperature reduction device and a temperature prediction control method for an explosion-proof elevator control cabinet. Compared with traditional real-time monitoring technologies, the temperature prediction method based on neural networks can achieve preventive regulation, optimize resource allocation, and improve the fault warning ability. The temperature reduction device combines the prediction results and realizes dynamic heat dissipation control matching the predicted temperature change trend through the coordinated operation of multiple heat dissipation modules. Summary of the Invention
[0006] The purpose of the present invention is to provide a temperature reduction device and a temperature prediction control method for an explosion-proof elevator control cabinet to solve the problems of lagging response, low prediction accuracy, and excessive manual intervention in the temperature management of the existing elevator control cabinet.
[0007] To achieve the above purpose, the present invention provides the following solutions:
[0008] An explosion-proof elevator control cabinet cooling device and a temperature prediction control method, comprising the following steps:
[0009] S1: Cooling device design and deployment: Design a cooling device with sealing performance according to the requirements of the explosion-proof and flammable industrial environment of the explosion-proof elevator, and install it.
[0010] S2: Dataset construction and preprocessing: Collect and organize the temperature data of the explosion-proof elevator control cabinet under different working conditions, clean and normalize the original data, and enhance the quality of the dataset through time series analysis.
[0011] S3: Neural network model design: Construct a hybrid neural network model based on ConvLSTM.
[0012] S4: Model training and verification: Input the preprocessed temperature dataset into the ConvLSTM model for training.
[0013] S5: Temperature prediction and threshold warning: Use the trained model to predict the newly input time series temperature data, and generate a sequence of predicted temperature values. Set temperature thresholds and fluctuation tolerance thresholds, and implement two-stage logical judgment.
[0013] S6: Cooling device control: Deploy the trained temperature prediction model to edge computing devices or cloud servers, and implement a hierarchical control strategy based on the safety margin design.
[0014] As a further improvement of the present invention, in the S1, through an internal and external double-loop collaborative architecture: the internal loop device is attached to the surface of the heating element through thermal grease, and heat dissipation is achieved through fins and a wind cooling system; the external loop device uses an explosion-proof aluminum alloy housing, is fixed to the control cabinet housing through a bolt array, and is equipped with heat pipes and a liquid cooling system to achieve heat dissipation.
[0015] As a further improvement of the present invention, in the S2, collect the temperature data under different working conditions, and increase the quantity and diversity of data samples through methods such as time series extension and outlier processing of the original dataset, and perform normalization processing to ensure the consistency and stability of the data. The specific linear interpolation formula is as follows
[0016] In the formula, t represents the time point where the outlier to be repaired is located. t1 represents the time index of the nearest valid data point before the outlier time point t. t2 represents the time index of the nearest valid data point after the outlier time point t. T raw (t1) and T raw (t2) respectively represent the original temperature values collected at time points t1 and t2. T clean(t) represents the repaired temperature value at time point t calculated by linear interpolation, which is used to replace the outlier.
[0017] As a further improvement of the present invention, in step S3, the neural network model design adopts a ConvLSTM hybrid architecture. Combining the characteristics of time series, a model including three layers of ConvLSTM, a fully connected layer and an output layer is designed.
[0018] As a further improvement of the present invention, the steps of model training and verification in step S4 include
[0019] S41 Set the input data format as a time series tensor, with a time window size of 10, a sensor number of 5, an input channel number Cin = 5, and an output channel number Cout = 32;
[0020] S42 Dataset division: Divide it into a training set, a validation set and a test set according to the ratio of 70%:15%:15%;
[0021] S43 Set the initial learning rate to 0.0005, the weight decay coefficient to 0.0001, the batch size to 64, and the training iteration period Epoch to 200;
[0022] S44 Input the preprocessed temperature dataset into the ConvLSTM model for training to generate a model weight file;
[0023] S45 Set the loss function as the mean square error (MSE), select the Adam optimizer, and dynamically adjust the learning rate;
[0024] S46 Use 5-fold cross-validation to evaluate the model performance, and calculate the mean square error (MSE) and the mean absolute error (MAE) as evaluation indicators;
[0025] S47 After 200 times of cyclic iterative training, retain the best weight file for temperature prediction, and terminate the training in combination with the validation set loss convergence condition (the decrease amplitude in 5 consecutive epochs < 1%).
[0026] As a further improvement of the present invention, the steps of temperature prediction and threshold warning in step S5 include:
[0027] S51: Input the real-time collected time series temperature data into the trained ConvLSTM model to generate a sequence of predicted temperature values at future time points
[0028] S52: Calculate the short-term fluctuation and the long-term fluctuation;
[0029] S53: Set the temperature threshold (T th ) and the fluctuation tolerance threshold (ΔT tol );
[0030] S54: Determine whether the temperature is close to the critical value and implement a two-stage logical judgment;
[0031] S55: Dynamically adjust the operation strategy of the cooling device according to the warning result.
[0032] As a further improvement of the present invention, in S5, the predicted temperature value sequence at future time points is generated The average temperature sequence and the fluctuation value are calculated through the formula, and the specific formula is as follows.
[0033] In the formula, w s = 5 represents the time window. Represents the short-term moving average temperature sequence. Short-term fluctuation value.
[0034] In the formula, w l = 20 represents the time window. Represents the long-term moving average temperature sequence. Represents the long-term fluctuation value.
[0035] As a further improvement of the present invention, in S6, the intelligent temperature control device control considers the safety margin design. When the predicted temperature is close to 0.8T th Start the internal circulation device. If the temperature continues to rise, start the external circulation device.
[0036] Compared with the prior art, the advantages of the present invention are as follows:
[0037] This solution proposes a cooling device for an explosion-proof elevator control cabinet and a temperature prediction control method. Based on the ConvLSTM model, it captures the non-linear change trend of temperature, combines the sliding window and the two-stage threshold judgment to predict temperature fluctuations, upgrades the traditional passive monitoring to active preventive control, and eliminates the risk of response lag. In the heat dissipation design, the internal circulation and external circulation devices are used to jointly ensure the explosion-proof sealing requirements and are started dynamically according to the predicted temperature. In addition, the model realizes the high-precision prediction and adaptive optimization of the heat dissipation strategy through dynamic learning rate adjustment and multi-condition data training. This method improves the temperature management efficiency, energy consumption optimization and safety by combining the cooling device with the ConvLSTM neural network. Description of the Drawings
[0038] Figure 1 Is a flowchart of a cooling device for an explosion-proof elevator control cabinet and a temperature prediction control method
[0039] Figure 2 Is the internal circulation cooling device of the elevator control cabinet
[0040] Figure 3 For the external circulation cooling device of the elevator control cabinet Specific implementation manner
[0041] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention; obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0042] Embodiment:
[0043] Please refer to Figure 1 , an explosion-proof elevator control cabinet cooling device and temperature prediction control method, including the following steps
[0044] S1: Design and deployment of the cooling device: Design a cooling device with sealing performance and install it according to the requirements of the explosion-proof elevator's flammable and explosive industrial environment.
[0045] S2: Dataset construction and preprocessing: Collect and organize the temperature data of the explosion-proof elevator control cabinet under different working conditions, clean and normalize the original data, and enhance the quality of the dataset through time series analysis.
[0046] S3: Neural network model design: Construct a hybrid neural network model based on ConvLSTM.
[0047] S4: Model training and verification: Input the preprocessed temperature dataset into the ConvLSTM model for training.
[0048] S5: Temperature prediction and threshold warning: Use the trained model to predict the newly input time series temperature data to generate a sequence of predicted temperature values. Set temperature thresholds and fluctuation tolerance thresholds, and implement two-stage logical judgment.
[0049] S6: Cooling device control: Deploy the trained temperature prediction model to edge computing devices or cloud servers, and implement a hierarchical control strategy based on the safety margin design.
[0050] The device in S1 includes an internal circulation device and an external circulation device
[0051] Please refer to Figure 2The internal circulation cooling device in the device shown in S1, where (1) is the substrate that is attached to the surface of the heating element through thermal conductive silicone grease, and uses its high thermal conductivity to evenly conduct heat to the (4) fins; (2) is the metal shell used to fix the (5) fan; (3) is the base that fixes the contact between the (4) fins and the (1) substrate to prevent displacement or deformation, and efficiently disperses heat to the surface of the (4) fins; (4) is the fin that expands the heat dissipation surface area and accelerates heat transfer through convection and conduction; (5) is the auxiliary fan that speeds up gas flow through forced air circulation and enhances the heat exchange efficiency between the fins and the air.
[0052] Please refer to Figure 3 The external circulation cooling device in the device shown in S1, where (6) is the substrate installed in the metal shell of the control cabinet, and evenly conducts heat to the (12) pulsating heat pipe through high thermal conductivity materials; (7) is the metal shell that ensures the tightness between the control cabinet and the heat gas conduction of the external circulation cooling device; (8) is the air outlet that transfers the cooled gas back into the control cabinet; (9) is the bolt that fixes the (7) metal shell to ensure tightness; (10) is the air inlet that introduces the hot air in the control cabinet into the external circulation cooling device; (11) is the water tank that stores the cooling medium; (12) is the pulsating heat pipe that can quickly spread heat from the local high-temperature area to the entire condensation area and improve the temperature uniformity; (13) is the Z-shaped pipe that connects the water tank to the (20) connection box; (14) is the heat sink that increases the contact area, accelerates heat conduction and diffusion, and efficiently releases the heat absorbed by the heat pipe to the surrounding medium; (15) is the base used to fix the (21) connection box and the (21) condenser; (16) is the water inlet pipe that connects the (18) water pump to the (11) water tank; (17) is the bellows that absorbs the vibration and thermal expansion and contraction displacement between the (18) water pump and the (16) water inlet pipe to ensure tightness; (18) is the water pump that drives the coolant to circulate forcibly in the liquid cooling loop; (19) is the connecting pipe that connects the (18) water pump and the (21) condenser; (20) is the connection box that coordinates the coolant flow path between the water tank and the condenser and buffers the pressure fluctuation; (21) is the condenser that releases the heat of the gaseous working medium in the heat pipe to the coolant or the external environment, promotes its condensation and reflux, and maintains the heat pipe cycle.
[0053] As a further improvement of the present invention, the temperature data of the control cabinet in S2 is obtained by collecting in real time through temperature sensors distributed on the surfaces of the core heating components and the inlet and outlet areas of the heat dissipation air ducts, and collecting the historical temperature data of the elevator control cabinet under different working conditions (such as high load, low load, ambient temperature change, etc.), including time stamps, temperature values, and associated sensor data. The original data set is processed by methods such as time series extension and outlier processing to increase the quantity and diversity of data samples, and is normalized to ensure the consistency and stability of the data. The method formula for outlier processing using linear interpolation repair is as follows:
[0054] In the formula, t represents the time point where the outlier to be repaired is located. t1 represents the time index of the nearest valid data point before the outlier time point t. t2 represents the time index of the nearest valid data point after the outlier time point t. T raw (t1) and T raw (t2) respectively represent the original temperature values collected at time points t1 and t2. T clean (t) represents the repaired temperature value at time point t calculated by linear interpolation, which is used to replace the outlier.
[0055] As a further improvement of the present invention, in step S3, the neural network model design adopts a ConvLSTM hybrid architecture. Combining with the characteristics of time series, a model including three layers of ConvLSTM, a fully connected layer and an output layer is designed.
[0056] As a further improvement of the present invention, the steps in step S4 model training and verification include:
[0057] S41: Set the input data format as a time series tensor, the time window size is 10, the number of sensors is 5, the number of input channels Cin = 5, and the number of output channels Cout = 32;
[0058] S42: Dataset division: Divide it into a training set, a validation set and a test set according to the ratio of 70%:15%:15%;
[0059] S43: Set the initial learning rate to 0.0005, the weight decay coefficient to 0.0001, the batch size to 64, and the training iteration period Epoch to 200;
[0060] S44: Input the preprocessed temperature dataset into the ConvLSTM model for training to generate a model weight file;
[0061] S45: Set the loss function as the mean square error (MSE), select the Adam optimizer, and dynamically adjust the learning rate;
[0062] S46: Use 5-fold cross-validation to evaluate the model performance, and calculate the mean square error (MSE) and the mean absolute error (MAE) as evaluation indicators;
[0063] S47: After 200 times of cyclic iterative training, retain the best weight file for temperature prediction, and terminate the training in combination with the validation set loss convergence condition (the decline amplitude in 5 consecutive epochs < 1%).
[0064] As a further improvement of the present invention, the steps in step S4 real-time prediction fluctuation analysis include:
[0065] S51: Input the time series temperature data collected in real time into the trained ConvLSTM model to generate a sequence of predicted temperature values at future time points.
[0066] S52: Calculate the short-term moving average temperature sequence and the short-term fluctuation value as well as the long-term moving average temperature sequence and the long-term fluctuation value
[0067] S53: Set the temperature threshold (T th ) and the fluctuation tolerance threshold (ΔT tol );
[0068] S54: Determine whether the temperature is close to the critical value and implement a two-stage logical judgment. When , a preliminary warning is triggered, and when the short-term fluctuation or the long-term fluctuation , a formal warning is confirmed;
[0069] S55: Dynamically adjust the operation strategy of the cooling device according to the warning result.
[0070] As a further improvement of the present invention, in S5, according to the sequence of predicted temperature values at future time points , the average temperature sequence and the fluctuation value are calculated through formulas. The specific formulas are as follows.
[0071] In the formula, w s = 5 represents the time window. represents the short-term moving average temperature sequence. Short-term fluctuation value.
[0072] In the formula, w l = 20 represents the time window. represents the long-term moving average temperature sequence. represents the long-term fluctuation value.
[0073] As a further improvement of the present invention, in S6, when designing the control of the cooling device considering the safety margin, when the predicted temperature is close to 0.8T th , the inner circulation device is started. If the temperature continues to rise, the outer circulation device is started.
[0074] It is obvious to those skilled in the art that the present invention is not limited to the details of the above-described exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, in any aspect, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Accordingly, all changes that fall within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claim concerned.
[0075] In addition, it should be understood that although this specification is described in terms of embodiments, not every embodiment only contains an independent technical solution. This narrative manner of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.
Claims
1. An explosion-proof elevator control cabinet cooling device and a temperature prediction control method, characterized in that, It includes the following steps: S1: Design and deployment of the cooling device: In response to the requirements of the explosion-proof elevator in the flammable and explosive industrial environment, design a cooling device with tightness and install it. S2: Construction and preprocessing of the dataset: Collect and organize the temperature data of the explosion-proof elevator control cabinet under different working conditions, clean and normalize the original data, and enhance the quality of the dataset through time series analysis. S3: Design of the neural network model: Construct a hybrid neural network model based on ConvLSTM. S4: Model training and verification: Input the preprocessed temperature dataset into the ConvLSTM model for training. S5: Temperature prediction and threshold warning: Use the trained model to predict the newly input time series temperature data and generate a sequence of predicted temperature values. Set temperature thresholds and fluctuation tolerance thresholds and implement a two-stage logical judgment. S6: Control of the cooling device: Deploy the trained temperature prediction model to edge computing devices or cloud servers and implement a hierarchical control strategy based on the safety margin design.
2. The explosion-proof elevator control cabinet cooling device and temperature prediction control method according to claim 1, characterized in that In the step S1 of the design and deployment of the cooling device, through an internal and external double-loop collaborative architecture: The internal loop device is attached to the surface of the heating element through thermal grease, and heat dissipation is achieved through fins and a air cooling system; The external loop device uses an explosion-proof aluminum alloy housing, which is fixed to the control cabinet housing through a bolt array, and is equipped with heat pipes and a liquid cooling system to achieve heat dissipation.
3. An explosion-proof elevator control cabinet cooling device and temperature prediction control method according to claim 1, characterized in that In the step S2 of the construction and preprocessing of the dataset, the original dataset is processed by methods such as time series extension and outlier processing to increase the number and diversity of data samples, and normalization processing is performed to ensure the consistency and stability of the data.
4. An explosion-proof elevator control cabinet cooling device and temperature prediction control method according to claim 3, characterized in that, In the outlier repair of the S2 data preprocessing, the linear interpolation method is used, and the formula is: Where, t represents the time point where the outlier to be repaired is located. t1 represents the time index of the nearest valid data point before the outlier time point t. t2 represents the time index of the nearest valid data point after the outlier time point t. T raw (t1) and T raw (t2) respectively represent the original temperature values collected at the time points t1 and t2. T clean (t) represents the repaired temperature value at the time point t calculated by linear interpolation, which is used to replace the outlier.
5. An explosion-proof elevator control cabinet cooling device and temperature prediction control method according to claim 1, characterized in that, The step S3 of the design of the neural network model includes: Considering the time series characteristics of the temperature data of the explosion-proof elevator control cabinet, select the ConvLSTM (Convolutional Long Short-Term Memory) neural network architecture. Build a neural network training environment based on PyTorch, configure the ConvLSTM hybrid model architecture, including 3 ConvLSTM layers, a fully connected layer, and an output layer; 6. The explosion-proof elevator control cabinet cooling device and temperature prediction control method according to claim 1, wherein The step S4 of model training and verification includes the following steps: S41: Set the input data format as a time series tensor, the time window size is 10, the number of sensors is 5, the input channel number Cin = 5, and the output channel number Cout = 32; S42: Dataset division: Divide it into a training set, a validation set, and a test set according to the ratio of 70%:15%:15%; S43: Set the initial learning rate to 0.0005, the weight decay coefficient to 0.0001, the batch size to 64, and the training iteration period Epoch to 200; S44: Input the preprocessed temperature dataset into the ConvLSTM model for training to generate a model weight file; S45: Set the loss function as the mean square error (MSE), select the Adam optimizer, and dynamically adjust the learning rate; S46: Use 5-fold cross-validation to evaluate the model performance, and calculate the mean squared error (MSE) and mean absolute error (MAE) as evaluation metrics; S47: After 200 cycles of iterative training, retain the best weight file for temperature prediction, and terminate the training in combination with the convergence condition of the validation set loss (the decline rate in 5 consecutive epochs < 1%).
7. An explosion-proof elevator control cabinet cooling device and temperature prediction control method according to claim 1, characterized in that, The step S4 temperature prediction and threshold warning described above includes the following steps: S51: Input the time series temperature data collected in real time into the trained ConvLSTM model to generate a sequence of predicted temperature values at future time points S52: Calculate the short-term moving average temperature series and the short-term fluctuation value and the long-term moving average temperature series and the long-term fluctuation value S53: Set the temperature threshold (T th ) and the fluctuation tolerance threshold (ΔT tol ); S54: Judge whether the temperature is close to the critical value and implement a two-stage logical judgment; S55: Dynamically adjust the operation strategy of the cooling device according to the warning result.
8. An explosion-proof elevator control cabinet cooling device and temperature prediction control method according to claim 7, characterized in that, The step S52 calculates the average temperature sequence and the fluctuation value, and the formula is as follows: where w s = 5 represents the time window. represents the short-term moving average temperature series. Short-term fluctuation value. where w l = 20 represents the time window. represents the long-term moving average temperature series. represents the long-term fluctuation value.
9. An explosion-proof elevator control cabinet cooling device and temperature prediction control method according to claim 7, characterized in that The step S54 implements a two-stage logical judgment, where: The first stage: When the predicted temperature value exceeds the temperature threshold T th , a preliminary warning is triggered; The second stage: When there is a short-term fluctuation or a long-term fluctuation , a formal warning is confirmed; 10. The explosion-proof elevator control cabinet cooling device and temperature prediction control method according to claim 1, characterized in that, In the step S6 of controlling the temperature reduction device, considering the safety margin, when the predicted temperature approaches 0.8T th Start the internal circulation device. If the temperature continues to rise, start the external circulation device.