A method and system for coordinated control of a curing barn heat pump based on edge computing
By combining edge computing and LSTM neural networks, the temperature distribution inside the curing barn can be monitored and adjusted in real time, solving the problem of temperature monitoring blind spots under centralized control, realizing the uniformity and stability of the temperature inside the curing barn, and improving energy utilization efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHONGNENG XINAN TECH CO LTD
- Filing Date
- 2026-01-19
- Publication Date
- 2026-06-12
AI Technical Summary
The existing heat pump control method for drying ovens adopts centralized control, which leads to blind spots in temperature monitoring, making it difficult to detect local temperature anomalies in a timely manner and adjust the heat pump's operating status, thus affecting the uniformity and stability of temperature distribution inside the drying oven.
By employing edge computing-based distributed temperature sensors and LSTM neural networks, a three-dimensional thermal field distribution map is generated in real time. The LSTM neural network is used to predict changes in the thermal field, control the working status of the heat transfer pipe network circulation pump, and adjust the air outlet angle and fan speed to improve temperature distribution.
It enables precise monitoring and dynamic adjustment of the temperature inside the curing oven, avoids monitoring blind spots, improves the uniformity and stability of temperature distribution, and optimizes energy utilization.
Smart Images
Figure CN122191949A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of drying room heat pump control technology, specifically a drying room heat pump collaborative control method and system based on edge computing. Background Technology
[0002] Curing barns are widely used in numerous industrial production and agricultural product processing sectors. Whether for tobacco curing, agricultural product drying, or heat treatment of industrial materials, curing barns play an indispensable role. Among these, heat pumps, as highly efficient and energy-saving heat exchange devices, can absorb a large amount of low-grade heat energy from the surrounding environment while consuming only a small amount of high-grade energy such as electricity, converting it into high-temperature heat energy usable in the curing barn. Their heating efficiency ratio is far higher than that of traditional electric heating equipment. By coordinating the control of heat pumps with the existing heating system of the curing barn, not only can the energy-saving advantages of heat pumps be fully utilized, but the complementarity of the two heating methods can also be achieved, further improving the curing barn's performance. To ensure the stability and reliability of heating, the intelligent control system utilizes advanced sensor technology, automated control technology, and intelligent algorithms to comprehensively and in real-time monitor and analyze the temperature, humidity, material drying status, and operating parameters of the heat pump and traditional heating equipment within the drying chamber. Based on the preset drying process curve and real-time monitoring data, the intelligent control system can automatically adjust the operating mode, power output, and operating time of the heat pump and traditional heating equipment to ensure that the temperature and humidity within the drying chamber are always at their optimal levels, meeting the process requirements for material drying, while minimizing energy consumption and reducing material loss due to uneven or excessive drying.
[0003] Existing curing oven heat pumps typically employ centralized control, judging temperature changes within the curing oven based on specific locations. However, because temperature monitoring points are scattered throughout the oven, blind spots can easily appear, making it difficult to detect localized temperature anomalies in a timely manner. Furthermore, it is difficult to predict real-time temperature changes within the oven and adjust the operation of the heat transfer network circulation pumps to improve the temperature distribution within the oven. Therefore, improvements are needed. Summary of the Invention
[0004] To overcome the above-mentioned shortcomings, the present invention provides that existing curing oven heat pumps generally adopt a centralized control method when controlling the temperature. The temperature changes inside the curing oven are judged based on the temperature at a specific location. However, since the temperature monitoring locations are relatively dispersed inside the curing oven, blind spots are easily formed. This makes it difficult to detect abnormal local temperatures inside the curing oven in a timely manner. At the same time, it is difficult to predict the real-time temperature changes inside the curing oven and adjust the working status of the heat conduction pipe network circulation pump in a timely manner to improve the temperature distribution inside the curing oven.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a heat pump collaborative control system for a drying oven based on edge computing, comprising an edge computing terminal, wherein the input end of the edge computing terminal is connected to a distributed temperature sensor, which is embedded inside the wall of the drying oven to monitor the temperature data at different locations inside the drying oven in real time, and transmits the temperature data monitored by the temperature sensors at different locations inside the drying oven to the edge computing terminal for aggregation, and after processing and calculation by the edge computing terminal, generates a real-time three-dimensional thermal field distribution map, which is displayed on the screen to intuitively show the spatial distribution of the temperature inside the drying oven;
[0006] Based on the three-dimensional thermal field distribution map generated in real time by the edge computing terminal, the thermal field uniformity index inside the baking oven is calculated in real time. The calculated thermal field uniformity index is then aggregated to the cloud platform to train an LSTM neural network to predict thermal field changes until the value of the thermal field change predicted by the LSTM neural network corresponds to the actual value. Finally, the trained LSTM neural network is connected to the heat conduction pipe network circulation pump control system.
[0007] The working state of the heat transfer network circulation pump is controlled based on the thermal field uniformity index predicted by the LSTM neural network, and the air outlet angle and fan speed are adjusted in a timely manner when the heat transfer network circulation pump is working.
[0008] The temperature data monitored in real time by the temperature sensor is transmitted to the edge computing terminal and numbered according to the temperature sensor's location coordinates and time. This refers to the position of an object. and time Temperature;
[0009] The temperature data transmitted to the edge computing terminal is processed, including removing data that exceeds three times the standard deviation of the mean as outliers. For missing data, the average of the temperature data from adjacent time points is used as the filler value. A standard thermometer corresponding to a specific temperature sensor location is installed inside the drying oven, and the temperature of the drying oven at different time points is recorded in real time by the standard thermometer and the temperature sensor, and the error of the standard thermometer reading is calculated. ,in The displayed value is from the temperature control device in the drying room. The temperature is measured using a standard thermometer, and then the oven temperature data is calibrated based on the error; that is, the calibrated temperature. ; The heat conduction calculation formula for generating a real-time three-dimensional thermal field distribution map is as follows: , in This refers to the position of an object. and time Temperature; It is the thermal diffusivity of an object, defined as ,in It is thermal conductivity. It's density. It is specific heat capacity; In position and time A heat source per unit volume.
[0010] The formula for calculating the thermal field uniformity index is as follows: ,in Indicates the first Temperature values measured by a temperature sensor This represents the average temperature measured by all temperature sensors. This represents the total number of temperature sensors. The value is between 0 and 1. The closer the value is to 1, the more uniform the surface thermal field.
[0011] The training process for the LSTM neural network to predict thermal field changes includes: Define the LSTM neural network structure: Choose an appropriate number of network layers and neurons per layer; Defining the loss function: The mean squared error is usually used as the loss function; Select an optimizer to train the network: use historical data for training, calculate the error between the predicted value and the true value based on the loss function, backpropagate to update the parameters, until the network's prediction error reaches a level that basically matches the true value; The trained LSTM neural network is directly applied to the control of the circulating pump of the heat conduction pipe network in the drying oven. By comparing the heat field uniformity index inside the drying oven with the set threshold in real time through the LSTM neural network, the working status of the circulating pump of the heat conduction pipe network can be controlled in a timely manner to maintain the uniformity of the heat field.
[0012] The formula for calculating the loss function is as follows: ,in Indicates the first The true value of each sample Indicates the first The predicted value for each sample, This represents the total number of samples.
[0013] A control method for a heat pump collaborative control system for a drying room based on edge computing, the control method comprising the following steps: When the drying room is in operation, distributed temperature sensors embedded in the walls collect temperature data from different locations inside the drying room to monitor the temperature in real time. The distance between adjacent temperature sensors is less than 1 meter. The temperature data monitored by the temperature sensors is transmitted to the edge computing terminal in real time. The edge computing terminal labels the temperature data monitored by the temperature sensors in real time, and then processes and calculates the data to generate a real-time three-dimensional thermal field distribution map. The generated three-dimensional thermal field distribution map is displayed on the screen in real time, which makes it convenient for staff to observe the thermal field distribution inside the drying room in real time.
[0014] Based on the temperature data displayed in the real-time 3D thermal field distribution map, the thermal field uniformity index inside the drying oven is calculated in real time. Multiple sets of calculated thermal field uniformity index data are aggregated to the cloud platform. An appropriate number of thermal field uniformity index data points are selected, and an LSTM neural network is trained through the cloud platform. After a certain training period, until the preset number of training rounds is reached, the LSTM neural network's prediction of thermal field changes inside the drying oven corresponds to the actual thermal field changes inside the drying oven. The trained LSTM neural network is then connected to the heat transfer pipe network circulation pump control system. The LSTM neural network predicts the thermal field changes inside the drying oven in real time and controls the working state of the heat transfer pipe network circulation pump. Simultaneously, the edge computing terminal continues to generate a 3D thermal field distribution map at regular intervals based on the real-time monitored temperature data. The actual thermal field uniformity index inside the drying oven is calculated based on the 3D thermal field distribution map. If the error between the thermal field uniformity index predicted by the LSTM neural network and the actual thermal field uniformity index does not exceed ±0.01, the working state of the heat transfer pipe network circulation pump continues to be controlled according to the thermal field uniformity index predicted by the LSTM neural network. If the error exceeds ±0.01, an error is promptly reported and corrected.
[0015] Meanwhile, the thermal uniformity index is set to 0.85 as the threshold for controlling the operation of the heat conduction pipe network circulation pump. When the thermal uniformity index predicted by the LSTM neural network in real time is less than the set threshold, the heat conduction pipe network circulation pump is started to enhance heat conduction. When the thermal uniformity index predicted by the LSTM neural network in real time recovers to above the set threshold, the heat conduction pipe network circulation pump is stopped to enhance heat conduction. The outlet angle and fan speed of the heat conduction pipe network circulation pump are dynamically adjusted in a timely manner to improve the thermal field distribution, optimize the three-dimensional thermal field distribution map, and make the temperature distribution in the drying room more uniform.
[0016] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0017] In this invention, temperature data from the drying oven monitored by temperature sensors are numbered and aggregated, then processed in real-time at an edge computing terminal to generate a real-time three-dimensional thermal field distribution map. This allows for precise control of the temperature distribution throughout the drying oven. Since the temperature sensors are distributed across different locations within the oven, it effectively avoids the difficulty in detecting localized temperature anomalies. Simultaneously, a thermal field uniformity index is calculated based on the real-time displayed three-dimensional thermal field distribution map. This calculated index is then uploaded to a cloud platform for aggregation. A certain number of thermal field uniformity indices are selected to train an LSTM neural network until the network can predict changes in the thermal field within the drying oven. The operating status of the heat transfer pipe circulation pump is controlled based on the predicted thermal field changes. By predicting temperature changes within the drying oven, the outlet angle and fan speed of the heat transfer pipe circulation pump are controlled in a timely manner to improve the temperature distribution within the oven, precisely adjusting heat delivery and making energy utilization more efficient. Attached Figure Description
[0018] Figure 1 This is a schematic diagram of the system of the present invention; Detailed Implementation
[0019] The technical solution of this application will be further described in detail below with reference to specific embodiments.
[0020] As shown in the figure, the present invention provides a technical solution: a heat pump collaborative control system for a drying oven based on edge computing, including an edge computing terminal. The input end of the edge computing terminal is connected to a distributed temperature sensor. The distributed temperature sensor is embedded inside the wall of the drying oven to monitor the temperature data at different locations inside the drying oven in real time. The temperature data monitored by the temperature sensors at different locations inside the drying oven is transmitted to the edge computing terminal in real time for aggregation. After processing and calculation by the edge computing terminal, a real-time three-dimensional thermal field distribution map is generated, which is displayed on the screen to intuitively show the spatial distribution of the temperature inside the drying oven.
[0021] Based on the three-dimensional thermal field distribution map generated in real time by the edge computing terminal, the thermal field uniformity index inside the baking oven is calculated in real time. The calculated thermal field uniformity index is then aggregated to the cloud platform to train an LSTM neural network to predict thermal field changes until the value of the thermal field change predicted by the LSTM neural network corresponds to the actual value. Finally, the trained LSTM neural network is connected to the heat conduction pipe network circulation pump control system.
[0022] The working state of the heat transfer network circulation pump is controlled based on the thermal field uniformity index predicted by the LSTM neural network, and the air outlet angle and fan speed are adjusted in a timely manner when the heat transfer network circulation pump is working.
[0023] The temperature data monitored in real time by the temperature sensor is transmitted to the edge computing terminal and numbered according to the temperature sensor's location coordinates and time. This refers to the position of an object. and time Temperature;
[0024] The temperature data transmitted to the edge computing terminal is processed, including removing data that exceeds three times the standard deviation of the mean as outliers. For missing data, the average of the temperature data from adjacent times is used as the filler value. A standard thermometer corresponding to a specific temperature sensor location is installed inside the drying chamber, and the temperature of the drying chamber at different time points is recorded in real time by the standard thermometer and the temperature sensor. The error of the standard thermometer reading is calculated. ,in The displayed value is from the temperature control device in the drying room. The temperature is measured using a standard thermometer, and then the oven temperature data is calibrated based on the error; that is, the calibrated temperature. ; The heat conduction calculation formula for generating a real-time three-dimensional thermal field distribution map is as follows: , in This refers to the position of an object. and time Temperature; It is the thermal diffusivity of an object, defined as ,in It is thermal conductivity. It's density. It is specific heat capacity; In position and time A heat source per unit volume.
[0025] The formula for calculating the thermal field uniformity index is: ,in Indicates the first Temperature values measured by a temperature sensor This represents the average temperature measured by all temperature sensors. This represents the total number of temperature sensors. The value is between 0 and 1. The closer the value is to 1, the more uniform the surface thermal field.
[0026] The training process for LSTM neural networks to predict thermal field changes includes:
[0027] Define the LSTM neural network structure: Choose an appropriate number of network layers and neurons per layer;
[0028] Defining the loss function: The mean squared error is usually used as the loss function;
[0029] Select an optimizer to train the network: use historical data for training, calculate the error between the predicted value and the true value based on the loss function, backpropagate to update the parameters, until the network's prediction error reaches a level that basically matches the true value;
[0030] The trained LSTM neural network is directly applied to the control of the circulating pump of the heat conduction pipe network in the drying oven. By comparing the heat field uniformity index inside the drying oven with the set threshold in real time through the LSTM neural network, the working status of the circulating pump of the heat conduction pipe network can be controlled in a timely manner to maintain the uniformity of the heat field.
[0031] The formula for calculating the loss function is: ,in Indicates the first The true value of each sample Indicates the first The predicted value for each sample, This represents the total number of samples.
[0032] A control method for a heat pump collaborative control system for a drying room based on edge computing, the control method comprising the following steps:
[0033] When the drying room is in operation, distributed temperature sensors embedded in the walls collect temperature data from different locations inside the drying room to monitor the temperature in real time. The distance between adjacent temperature sensors is less than 1 meter. The temperature data monitored by the temperature sensors is transmitted to the edge computing terminal in real time. The edge computing terminal labels the temperature data monitored by the temperature sensors in real time, and then processes and calculates the data to generate a real-time three-dimensional thermal field distribution map. The generated three-dimensional thermal field distribution map is displayed on the screen in real time, which makes it convenient for staff to observe the thermal field distribution inside the drying room in real time.
[0034] Based on the temperature data displayed in the real-time 3D thermal field distribution map, the thermal field uniformity index inside the drying oven is calculated in real time. Multiple sets of calculated thermal field uniformity index data are aggregated to the cloud platform. An appropriate number of thermal field uniformity index data points are selected, and an LSTM neural network is trained through the cloud platform. After a certain training period, until the preset number of training rounds is reached, the LSTM neural network's prediction of thermal field changes inside the drying oven corresponds to the actual thermal field changes inside the drying oven. The trained LSTM neural network is then connected to the heat transfer pipe network circulation pump control system. The LSTM neural network predicts the thermal field changes inside the drying oven in real time and controls the working state of the heat transfer pipe network circulation pump. Simultaneously, the edge computing terminal continues to generate a 3D thermal field distribution map at regular intervals based on the real-time monitored temperature data. The actual thermal field uniformity index inside the drying oven is calculated based on the 3D thermal field distribution map. If the error between the thermal field uniformity index predicted by the LSTM neural network and the actual thermal field uniformity index does not exceed ±0.01, the working state of the heat transfer pipe network circulation pump continues to be controlled according to the thermal field uniformity index predicted by the LSTM neural network. If the error exceeds ±0.01, an error is promptly reported and corrected.
[0035] Meanwhile, the thermal uniformity index is set to 0.85 as the threshold for controlling the operation of the heat conduction pipe network circulation pump. When the thermal uniformity index predicted by the LSTM neural network in real time is less than the set threshold, the heat conduction pipe network circulation pump is started to enhance heat conduction. When the thermal uniformity index predicted by the LSTM neural network in real time recovers to above the set threshold, the heat conduction pipe network circulation pump is stopped to enhance heat conduction. The outlet angle and fan speed of the heat conduction pipe network circulation pump are dynamically adjusted in a timely manner to improve the thermal field distribution, optimize the three-dimensional thermal field distribution map, and make the temperature distribution in the drying room more uniform.
[0036] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, an electrical connection, or a communication connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.
[0037] In this invention, unless otherwise expressly specified and limited, the first feature "on" or "below" the second feature may be in direct contact with the first and second features, or indirect contact through an intermediate medium. In the description of this specification, references to terms such as "an embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
Claims
1. A heat pump collaborative control system for a drying room based on edge computing, comprising an edge computing terminal, characterized in that: The input terminal of the edge computing terminal is connected to a distributed temperature sensor, which is embedded in the wall of the drying oven. The distributed temperature sensor monitors the temperature data at different locations inside the drying oven in real time and transmits the temperature data monitored by the temperature sensor at different locations in the drying oven to the edge computing terminal for aggregation. After processing and calculation by the edge computing terminal, a real-time three-dimensional thermal field distribution map is generated, which is displayed on the screen to intuitively show the spatial distribution of the temperature inside the drying oven. Based on the three-dimensional thermal field distribution map generated in real time by the edge computing terminal, the thermal field uniformity index inside the baking oven is calculated in real time. The calculated thermal field uniformity index is then aggregated to the cloud platform to train an LSTM neural network to predict thermal field changes until the value of the thermal field change predicted by the LSTM neural network corresponds to the actual value. Finally, the trained LSTM neural network is connected to the heat conduction pipe network circulation pump control system. The working state of the heat transfer network circulation pump is controlled based on the thermal field uniformity index predicted by the LSTM neural network, and the air outlet angle and fan speed are adjusted in a timely manner when the heat transfer network circulation pump is working.
2. The edge computing-based heat pump collaborative control system for drying rooms according to claim 1, characterized in that: The temperature data monitored in real time by the temperature sensor is transmitted to the edge computing terminal and numbered according to the temperature sensor's location coordinates and time. This refers to the position of an object. and time Temperature; The temperature data transmitted to the edge computing terminal is processed, including removing data that exceeds three times the standard deviation of the mean as outliers. For missing data, the average of the temperature data from adjacent time points is used as the filler value. A standard thermometer corresponding to a specific temperature sensor location is installed inside the drying oven, and the temperature of the drying oven at different time points is recorded in real time by the standard thermometer and the temperature sensor, and the error of the standard thermometer reading is calculated. ,in The displayed value is from the temperature control device in the drying room. The temperature is measured using a standard thermometer, and then the oven temperature data is calibrated based on the error; that is, the calibrated temperature. ; The heat conduction calculation formula for generating a real-time three-dimensional thermal field distribution map is as follows: ， in This refers to the position of an object. and time Temperature; It is the thermal diffusivity of an object, defined as ,in It is thermal conductivity. It's density. It is specific heat capacity; In position and time A heat source per unit volume.
3. The edge computing-based heat pump collaborative control system for drying rooms according to claim 1, characterized in that: The formula for calculating the thermal field uniformity index is as follows: ,in Indicates the first Temperature values measured by a temperature sensor This represents the average temperature measured by all temperature sensors. This represents the total number of temperature sensors. The value is between 0 and 1. The closer the value is to 1, the more uniform the surface thermal field.
4. The edge computing-based heat pump collaborative control system for drying ovens according to claim 1, characterized in that: The training process for the LSTM neural network to predict thermal field changes includes: Define the LSTM neural network structure: Choose an appropriate number of network layers and neurons per layer; Defining the loss function: The mean squared error is usually used as the loss function; Select an optimizer to train the network: use historical data for training, calculate the error between the predicted value and the true value based on the loss function, backpropagate to update the parameters, until the network's prediction error reaches a level that basically matches the true value; The trained LSTM neural network is directly applied to the control of the circulating pump of the heat conduction pipe network in the drying oven. By comparing the heat field uniformity index inside the drying oven with the set threshold in real time through the LSTM neural network, the working status of the circulating pump of the heat conduction pipe network can be controlled in a timely manner to maintain the uniformity of the heat field.
5. The edge computing-based heat pump collaborative control system for drying rooms according to claim 4, characterized in that: The formula for calculating the loss function is as follows: ,in Indicates the first The true value of each sample Indicates the first The predicted value for each sample, This represents the total number of samples.
6. A control method for a heat pump collaborative control system for a drying room based on edge computing, wherein the heat pump collaborative control system for a drying room based on edge computing according to any one of claims 1-5 is characterized in that: The control method includes the following steps: When the drying room is in operation, distributed temperature sensors embedded in the walls collect temperature data from different locations inside the drying room to monitor the temperature in real time. The distance between adjacent temperature sensors is less than 1 meter. The temperature data monitored by the temperature sensors is transmitted to the edge computing terminal in real time. The edge computing terminal labels the temperature data monitored by the temperature sensors in real time, and then processes and calculates the data to generate a real-time three-dimensional thermal field distribution map. The generated three-dimensional thermal field distribution map is displayed on the screen in real time, which makes it convenient for staff to observe the thermal field distribution inside the drying room in real time. Based on the temperature data displayed in the real-time 3D thermal field distribution map, the thermal field uniformity index inside the drying oven is calculated in real time. Multiple sets of calculated thermal field uniformity index data are aggregated to the cloud platform. An appropriate number of thermal field uniformity index data points are selected, and an LSTM neural network is trained through the cloud platform. After a certain training period, until the preset number of training rounds is reached, the LSTM neural network's prediction of thermal field changes inside the drying oven corresponds to the actual thermal field changes inside the drying oven. The trained LSTM neural network is then connected to the heat transfer pipe network circulation pump control system. The LSTM neural network predicts the thermal field changes inside the drying oven in real time and controls the working state of the heat transfer pipe network circulation pump. Simultaneously, the edge computing terminal continues to generate a 3D thermal field distribution map at regular intervals based on the real-time monitored temperature data. The actual thermal field uniformity index inside the drying oven is calculated based on the 3D thermal field distribution map. If the error between the thermal field uniformity index predicted by the LSTM neural network and the actual thermal field uniformity index does not exceed ±0.01, the working state of the heat transfer pipe network circulation pump continues to be controlled according to the thermal field uniformity index predicted by the LSTM neural network. If the error exceeds ±0.01, an error is promptly reported and corrected. Meanwhile, the thermal uniformity index is set to 0.85 as the threshold for controlling the operation of the heat conduction pipe network circulation pump. When the thermal uniformity index predicted by the LSTM neural network in real time is less than the set threshold, the heat conduction pipe network circulation pump is started to enhance heat conduction. When the thermal uniformity index predicted by the LSTM neural network in real time recovers to above the set threshold, the heat conduction pipe network circulation pump is stopped to enhance heat conduction. The outlet angle and fan speed of the heat conduction pipe network circulation pump are dynamically adjusted in a timely manner to improve the thermal field distribution, optimize the three-dimensional thermal field distribution map, and make the temperature distribution in the drying room more uniform.