Intelligent precision temperature and humidity control drying machine system based on internet of things
By leveraging IoT technology and machine learning algorithms, combined with multi-dimensional sensor data and user feedback, personalized drying control for clothes dryers has been achieved. This solves the problems of high energy consumption and uneven drying effects in traditional dryers, improving drying efficiency and garment protection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- JIANGSU XIANDAO DRYING TECH CO LTD
- Filing Date
- 2026-03-26
- Publication Date
- 2026-07-03
Smart Images
Figure CN122327520A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data analysis technology, and more specifically to an intelligent precision temperature and humidity control dryer system based on the Internet of Things. Background Technology
[0002] In existing technologies, traditional clothes dryers mostly adopt fixed drying modes based on time and preset programs. They lack the ability to perceive and adaptively adjust to individual differences in clothing materials, weight, and initial humidity, resulting in high energy consumption, uneven drying effects, and easy damage to clothes. Existing intelligent solutions often rely on single temperature and humidity feedback control, failing to integrate multi-dimensional sensor data and user subjective feedback for closed-loop learning and optimization. They cannot achieve truly personalized and precise drying. The systems generally lack intelligent decision-making mechanisms based on real-time status prediction and dynamic adjustment of safety boundaries, and still have limitations in terms of efficiency, safety, and user experience. Summary of the Invention
[0003] To address the aforementioned technical challenges, an intelligent and precise temperature and humidity control dryer system based on the Internet of Things is provided. This technical solution resolves the issues related to efficiency, safety, and user experience.
[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The IoT-based intelligent precision temperature and humidity control dryer system includes: The module includes a dataset creation module, a prediction and control instruction module, and an optimal drying state trajectory module. Among them, the prediction and control instruction module is electrically connected to the dataset establishment module, and the optimal dry state trajectory module is electrically connected to the prediction and control instruction module; The dataset building module performs a baseline drying process for the target garment, collects multi-dimensional sensor data in real time during the drying process, generates time-series data of the target garment drying process, receives user feedback ratings on the drying effect and maps them to the time-series data, and builds a target garment drying process-feedback effect dataset. The predictive control instruction module, based on the target clothing drying process-feedback effect dataset, uses machine learning algorithms to establish a personalized drying strategy model for the target clothing, and obtains predictive control instructions for the drying execution components. The optimal drying state trajectory module identifies the static attributes of the clothes to be dried, acquires the current drying process status data in real time, and inputs them into the personalized drying strategy model for the target clothes to obtain the predictive control instructions of the drying execution components of the clothes to be dried. It then dynamically adjusts the drying execution parameters in real time to generate the optimal drying state trajectory of the clothes to be dried.
[0005] Preferably, the dataset creation module specifically includes: The target clothing drying process time sequence data generation unit, based on temperature and humidity sensors, weight sensors, infrared sensors, and current sensors, acquires temperature and humidity data of the target clothing, weight data of the target clothing throughout the drying process, monitors changes in the color reflectance of the target clothing, and calculates the power of each component and the whole machine in the drying process. It sets a fixed data acquisition frequency and a synchronization timestamp, collects multi-dimensional sensor data in real time during the drying process, preprocesses the data, and generates target clothing drying process time sequence data. The trend feature extraction unit, based on the generated time-series data of the target clothing drying process, sets a fixed sliding time window length of 60 seconds. It calculates the mean, standard deviation, extreme values, and kurtosis of the multi-dimensional sensor data during the drying process within the time window. It performs a first-order linear regression on the temperature and humidity data of the target clothing within the time window, and performs linear fitting with the rate of temperature and humidity rise and fall of the target clothing within the time window as the slope. It calculates the second-order difference of the time-series data of the target clothing drying process. If the sign of the difference value changes abruptly and exceeds the 0.95 threshold, it is marked as an inflection point. The occurrence time and the time-series value of the target clothing drying process are recorded to obtain the trend features of the time-series data of the target clothing drying process within the time window.
[0006] Preferably, the dataset creation module also includes: The moisture content prediction unit calculates the predicted moisture content of the target garment within a time window based on the weight data of the target garment throughout the entire drying process. The heat exchange efficiency calculation unit calculates the temperature rise per unit energy of each component of the drying process within a time window based on the temperature and humidity data of the target clothing and the power of each component and the whole machine during the drying process, and obtains its heat exchange efficiency. The physical feature acquisition unit establishes the overall power curve of the dryer within a time window based on the power of each component and the whole machine during the drying process. It performs numerical integration to obtain the cumulative power integral of the dryer. Combined with the estimated moisture content and heat exchange efficiency of the target clothing, it obtains the physical features of the time-series data of the drying process of the target clothing within the time window. The event feature acquisition unit takes the decrease of more than 5% in the internal humidity of the target clothing within 10 seconds as the start event of the rapid drying stage, and takes preheating, constant speed drying, slow speed drying and cooling as the drying process states, to obtain the time sequence data event features of the drying process of the target clothing within the time window.
[0007] Preferably, the dataset creation module also includes: The drying process time sequence recording unit integrates the trend characteristics, physical characteristics, and event characteristics of the target clothing drying process time sequence data within a time window, associates them with the timestamp at the center of the time window, obtains the time sequence characteristics of the target clothing drying process within the time window, acquires the complete target clothing drying process time sequence data characteristics, arranges them in chronological order, generates a complete target clothing drying process time sequence data characteristic sequence matrix, associates it with metadata such as program ID, clothing type, initial weight, and ambient temperature and humidity, obtains a unique drying process time sequence record for the target clothing, and stores it in the time sequence database; The drying process-feedback effect dataset establishment unit is based on the immediate triggering of a feedback questionnaire to a mobile APP after the drying process is completed. Feedback scores are given based on the dryness, softness, smoothness, energy consumption satisfaction, and overall satisfaction of the target clothing. The data is then associated and mapped with the time-series data of the target clothing drying process through the program ID to establish the target clothing drying process-feedback effect dataset.
[0008] Preferably, the predictive control instruction module specifically includes: The encoder processing unit, based on the target clothing drying process-feedback effect dataset, divides it into training and testing sets. It uses the complete temporal data features of the target clothing drying process as input to train a long short-term memory network with an encoder-decoder architecture. In the long short-term memory network, the encoder processes the complete temporal data features of the target clothing drying process and outputs the final hidden state.
[0009] Preferably, the predictive control instruction module also includes: The transformation unit, through 2-3 fully connected layers, each followed by a ReLU activation function, processes the static properties of the target clothing and encodes them into numerical vectors; The static attributes include: material, initial weight, color, and initial humidity. Among them, the material and color of the clothing are independently coded, and the initial weight and initial humidity are Z-Score normalized. By combining the joint feature vector unit and setting up a feature fusion layer, the final hidden state output by the encoder in the long short-term memory network is concatenated with the static attribute feature vector to obtain the joint feature vector of the target clothing drying process.
[0010] Preferably, the predictive control instruction module also includes: The predictive control instruction unit takes the joint feature vector of the target clothing drying process as the initial state, inputs it into the long short-term memory network decoder, introduces a fully connected layer, and generates predictive control instructions for the drying execution component for the next K time steps. The model building unit uses the real component control instruction sequences recorded in the historical drying process of the target clothing drying process-feedback effect dataset, in which the overall user satisfaction and dryness score are both higher than the set threshold of 0.95, as training labels. The mean squared error-binary cross-entropy weighted sum is used as the loss function, and the Adam optimizer is used to optimize the model parameters to establish a personalized drying strategy model for the target clothing.
[0011] Preferably, the optimal drying state trajectory module specifically includes: The static attribute recognition unit for clothes to be dried is equipped with multiple input interfaces to identify the static attributes of the clothes to be dried. For the manual input channel, the user selects the clothes to be dried via a mobile APP, and the system provides material, weight, and color options for the user to input. For the RFID reading channel, the system is equipped with an RFID tag reader that automatically scans the RFID tags on the clothes to be dried to obtain static attribute data. For the visual recognition channel, the dryer is equipped with an image recognition function that captures images of the clothes through a camera and transmits them to the edge computing unit for real-time analysis to identify the material and color of the clothes to be dried. Furthermore, when the results from multiple channels are inconsistent, the static attributes of the clothes to be dried are identified according to the priority of RFID reading channel > visual recognition channel > manual input channel. The reference value acquisition unit uses temperature and humidity sensors and weight sensors to measure and record the temperature and humidity of the internal environment of the dryer and the initial weight of the clothes to be dried as reference values when the dryer is started. The status data acquisition unit configures a sensor network after the dryer is started. The sensors collect real-time data at a fixed frequency to obtain the internal temperature and humidity of the dryer, the current weight of the clothes to be dried, and the air flow speed. It also acquires the current drying process status data in real time and performs data preprocessing.
[0012] Preferably, the optimal drying state trajectory module also includes: The comprehensive feature acquisition unit, for each prediction and control command cycle of the drying execution component, uses sliding window technology to calculate the trend features, physical features and event features of the current drying process status data within the time window, and combines them with the static attributes of the clothes to be dried to perform Z-score standardization processing to obtain the comprehensive features of the current drying process status data. The predictive control command acquisition unit takes the comprehensive characteristics of the current drying process status data and the static attributes of the clothes to be dried as input, associates and matches them with the personalized drying strategy model for the target clothes, and obtains the predictive control command of the drying execution component of the clothes to be dried.
[0013] Preferably, the optimal drying state trajectory module also includes: The optimal drying state trajectory unit is based on the predicted control instructions of the drying execution components of the clothes to be dried, including the target temperature, target humidity and fan speed control parameters of the dryer. It takes the predicted control instructions of the drying execution components of the clothes to be dried, the comprehensive features of the current drying process state data and the static attributes of the clothes to be dried as input, checks whether the predicted control instructions violate the safety boundary. If they violate the safety boundary, they are adjusted to the safe range according to the known rule base. The optimal control instructions under different feature combinations of the optimal target temperature, humidity and fan speed are used as output to generate the optimal drying state trajectory of the clothes to be dried. The safety boundaries are: the temperature exceeds the material's ignition point, the absolute value of the target temperature change in adjacent command cycles is less than a set threshold t, and a sudden temperature rise is prohibited during specific high humidity periods to prevent condensation. Compared with the prior art, the beneficial effects of the present invention are: This invention proposes an intelligent and precise temperature and humidity control dryer solution based on the Internet of Things. This solution establishes a target clothing drying process-feedback effect dataset and constructs a personalized drying strategy model based on machine learning algorithms. It realizes intelligent predictive control of the drying execution components, dynamically generates optimal drying parameters based on the static attributes and real-time drying status of different clothing, improves the accuracy and adaptability of the drying process, effectively improves energy utilization efficiency and meets the personalized needs of users while ensuring clothing safety and drying quality. Attached Figure Description
[0014] Figure 1 This is a system framework diagram of an IoT-based intelligent precision temperature and humidity control dryer. Detailed Implementation
[0015] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.
[0016] Reference Figure 1 As shown, the IoT-based intelligent precision temperature and humidity control dryer system includes: The module includes a dataset creation module, a prediction and control instruction module, and an optimal drying state trajectory module. Among them, the prediction and control instruction module is electrically connected to the dataset establishment module, and the optimal dry state trajectory module is electrically connected to the prediction and control instruction module; The dataset building module performs a baseline drying process for the target garment, collects multi-dimensional sensor data in real time during the drying process, generates time-series data of the target garment drying process, receives user feedback ratings on the drying effect and maps them to the time-series data, and builds a target garment drying process-feedback effect dataset. The predictive control instruction module, based on the target clothing drying process-feedback effect dataset, uses machine learning algorithms to establish a personalized drying strategy model for the target clothing, and obtains predictive control instructions for the drying execution components. The optimal drying state trajectory module identifies the static attributes of the clothes to be dried, acquires the current drying process status data in real time, and inputs them into the personalized drying strategy model for the target clothes to obtain the predictive control instructions of the drying execution components of the clothes to be dried. It then dynamically adjusts the drying execution parameters in real time to generate the optimal drying state trajectory of the clothes to be dried.
[0017] The dataset creation module specifically includes: The target clothing drying process time sequence data generation unit, based on temperature and humidity sensors, weight sensors, infrared sensors, and current sensors, acquires temperature and humidity data of the target clothing, weight data of the target clothing throughout the drying process, monitors changes in the color reflectance of the target clothing, and calculates the power of each component and the whole machine in the drying process. It sets a fixed data acquisition frequency and a synchronization timestamp, collects multi-dimensional sensor data in real time during the drying process, preprocesses the data, and generates target clothing drying process time sequence data. The trend feature extraction unit, based on the generated time-series data of the target clothing drying process, sets a fixed sliding time window length of 60 seconds. It calculates the mean, standard deviation, extreme values, and kurtosis of the multi-dimensional sensor data during the drying process within the time window. It performs a first-order linear regression on the temperature and humidity data of the target clothing within the time window, and performs linear fitting with the rate of temperature and humidity rise and fall of the target clothing within the time window as the slope. It calculates the second-order difference of the time-series data of the target clothing drying process. If the sign of the difference value changes abruptly and exceeds the 0.95 threshold, it is marked as an inflection point. The occurrence time and the time-series value of the target clothing drying process are recorded to obtain the trend features of the time-series data of the target clothing drying process within the time window.
[0018] The dataset creation module also includes: The moisture content prediction unit calculates the predicted moisture content of the target garment within a time window based on the weight data of the target garment throughout the entire drying process. The heat exchange efficiency calculation unit calculates the temperature rise per unit energy of each component of the drying process within a time window based on the temperature and humidity data of the target clothing and the power of each component and the whole machine during the drying process, and obtains its heat exchange efficiency. The physical feature acquisition unit establishes the overall power curve of the dryer within a time window based on the power of each component and the whole machine during the drying process. It performs numerical integration to obtain the cumulative power integral of the dryer. Combined with the estimated moisture content and heat exchange efficiency of the target clothing, it obtains the physical features of the time-series data of the drying process of the target clothing within the time window. The event feature acquisition unit takes the decrease of more than 5% in the internal humidity of the target clothing within 10 seconds as the start event of the rapid drying stage, and takes preheating, constant speed drying, slow speed drying and cooling as the drying process states, to obtain the time sequence data event features of the drying process of the target clothing within the time window.
[0019] The dataset creation module also includes: The drying process time sequence recording unit integrates the trend characteristics, physical characteristics, and event characteristics of the target clothing drying process time sequence data within a time window, associates them with the timestamp at the center of the time window, obtains the time sequence characteristics of the target clothing drying process within the time window, acquires the complete target clothing drying process time sequence data characteristics, arranges them in chronological order, generates a complete target clothing drying process time sequence data characteristic sequence matrix, associates it with metadata such as program ID, clothing type, initial weight, and ambient temperature and humidity, obtains a unique drying process time sequence record for the target clothing, and stores it in the time sequence database; The drying process-feedback effect dataset establishment unit is based on the immediate triggering of a feedback questionnaire to a mobile APP after the drying process is completed. Feedback scores are given based on the dryness, softness, smoothness, energy consumption satisfaction, and overall satisfaction of the target clothing. The data is then associated and mapped with the time-series data of the target clothing drying process through the program ID to establish the target clothing drying process-feedback effect dataset.
[0020] When using it, combine the content of the above modules: Traditional garment drying process monitoring typically relies on static data collection from a single sensor, lacking the synchronous fusion and real-time analysis of multi-dimensional sensor data. Drying stage segmentation depends on preset fixed thresholds, failing to dynamically identify drying inflection points and state transitions. Drying effect evaluation is often based on subjective human experience, failing to accurately correlate multi-dimensional user feedback with time-series data of the drying process. This results in a lack of data-driven basis for drying strategy optimization, hindering personalized drying control and energy efficiency improvements. This step, by fusing data from multiple sensors such as temperature, humidity, weight, infrared, and current, extracts trend, physical, and event characteristics of time-series data based on a sliding time window, achieving a full-dimensional dynamic representation of the drying process. Furthermore, by mapping user feedback ratings for dryness, softness, and other dimensions to time-series data, a drying process-feedback effect dataset is constructed, providing a precise data foundation for data-driven drying strategy optimization, energy efficiency improvement, and personalized drying program development.
[0021] The predictive control instruction module specifically includes: The encoder processing unit, based on the target clothing drying process-feedback effect dataset, divides it into training and testing sets. It uses the complete temporal data features of the target clothing drying process as input to train a long short-term memory network with an encoder-decoder architecture. In the long short-term memory network, the encoder processes the complete temporal data features of the target clothing drying process and outputs the final hidden state.
[0022] The predictive control instruction module also includes: The transformation unit, through 2-3 fully connected layers, each followed by a ReLU activation function, processes the static properties of the target clothing and encodes them into numerical vectors; The static attributes include: material, initial weight, color, and initial humidity. Among them, the material and color of the clothing are independently coded, and the initial weight and initial humidity are Z-Score normalized. By combining the joint feature vector unit and setting up a feature fusion layer, the final hidden state output by the encoder in the long short-term memory network is concatenated with the static attribute feature vector to obtain the joint feature vector of the target clothing drying process.
[0023] The predictive control instruction module also includes: The predictive control instruction unit takes the joint feature vector of the target clothing drying process as the initial state, inputs it into the long short-term memory network decoder, introduces a fully connected layer, and generates predictive control instructions for the drying execution component for the next K time steps. The model building unit uses the real component control instruction sequences recorded in the historical drying process of the target clothing drying process-feedback effect dataset, in which the overall user satisfaction and dryness score are both higher than the set threshold of 0.95, as training labels. The mean squared error-binary cross-entropy weighted sum is used as the loss function, and the Adam optimizer is used to optimize the model parameters to establish a personalized drying strategy model for the target clothing.
[0024] When using it, combine the content of the above modules: Existing technologies for clothing drying control largely rely on preset general programs or linear adjustments based on simple sensor feedback. They lack deep integration of the static properties of clothing and the temporal characteristics of the dynamic drying process. Furthermore, they typically employ a single model or rule-based system, which struggles to adapt to the personalized drying needs of different garments, resulting in low energy efficiency, unstable drying effects, or a higher risk of clothing damage. They also fail to effectively utilize user satisfaction feedback data from historical drying processes to optimize control strategies. This approach constructs an encoder-decoder long short-term memory network, integrating static clothing properties with temporal data from the drying process. It introduces a loss function labeled with user satisfaction and dryness level for model training, enabling personalized predictive control of the drying execution components. This allows the network to dynamically learn and adapt to the characteristics of the target garment and the patterns of the drying process, generating more precise control command sequences, improving drying efficiency and clothing protection. Simultaneously, data-driven personalized strategies optimize overall user satisfaction, reducing energy consumption and operational risks.
[0025] The optimal drying state trajectory module specifically includes: The static attribute recognition unit for clothes to be dried is equipped with multiple input interfaces to identify the static attributes of the clothes to be dried. For the manual input channel, the user selects the clothes to be dried via a mobile APP, and the system provides material, weight, and color options for the user to input. For the RFID reading channel, the system is equipped with an RFID tag reader that automatically scans the RFID tags on the clothes to be dried to obtain static attribute data. For the visual recognition channel, the dryer is equipped with an image recognition function that captures images of the clothes through a camera and transmits them to the edge computing unit for real-time analysis to identify the material and color of the clothes to be dried. Furthermore, when the results from multiple channels are inconsistent, the static attributes of the clothes to be dried are identified according to the priority of RFID reading channel > visual recognition channel > manual input channel. The reference value acquisition unit uses temperature and humidity sensors and weight sensors to measure and record the temperature and humidity of the internal environment of the dryer and the initial weight of the clothes to be dried as reference values when the dryer is started. The status data acquisition unit configures a sensor network after the dryer is started. The sensors collect real-time data at a fixed frequency to obtain the internal temperature and humidity of the dryer, the current weight of the clothes to be dried, and the air flow speed. It also acquires the current drying process status data in real time and performs data preprocessing.
[0026] The optimal drying state trajectory module also includes: The comprehensive feature acquisition unit, for each prediction and control command cycle of the drying execution component, uses sliding window technology to calculate the trend features, physical features and event features of the current drying process status data within the time window, and combines them with the static attributes of the clothes to be dried to perform Z-score standardization processing to obtain the comprehensive features of the current drying process status data. The predictive control command acquisition unit takes the comprehensive characteristics of the current drying process status data and the static attributes of the clothes to be dried as input, associates and matches them with the personalized drying strategy model for the target clothes, and obtains the predictive control command of the drying execution component of the clothes to be dried.
[0027] The optimal drying state trajectory module also includes: The optimal drying state trajectory unit is based on the predicted control instructions of the drying execution components of the clothes to be dried, including the target temperature, target humidity and fan speed control parameters of the dryer. It takes the predicted control instructions of the drying execution components of the clothes to be dried, the comprehensive features of the current drying process state data and the static attributes of the clothes to be dried as input, checks whether the predicted control instructions violate the safety boundary. If they violate the safety boundary, they are adjusted to the safe range according to the known rule base. The optimal control instructions under different feature combinations of the optimal target temperature, humidity and fan speed are used as output to generate the optimal drying state trajectory of the clothes to be dried. The safety boundaries are: the temperature exceeds the material's ignition point, the absolute value of the target temperature change in adjacent command cycles is lower than the set threshold t, and a sudden temperature rise is prohibited during a specific high humidity period to prevent condensation.
[0028] When using it, combine the content of the above modules: In existing technologies, clothing drying processes typically rely on preset fixed programs or simple temperature and humidity feedback control. They lack multi-channel, accurate identification of static attributes such as clothing material, weight, and color. The drying strategies are simplistic and unable to dynamically adapt to the personalized needs of different garments, leading to energy waste, uneven drying effects, or the risk of clothing damage. Traditional methods often rely on single sensors for data acquisition, resulting in delayed responses to real-time changes and a lack of prioritization mechanisms for conflicting multi-source data. Safety controls are also often based on simple thresholds, failing to effectively prevent problems such as condensation or material overheating caused by sudden temperature increases. Overall, the system lacks intelligence and adaptability. This approach introduces multi-channel input interfaces and sets priority rules to ensure the accuracy and robustness of static attribute identification. It combines a multi-sensor network for temperature, humidity, and weight with sliding window feature extraction to capture dynamic changes in the drying process in real time. By associating personalized drying strategy models with safety boundary checks, it achieves predictive control and optimization of the drying execution components, generating a safe and efficient optimal drying trajectory. This improves the adaptability and intelligence of the drying system, ensuring clothing safety while increasing drying efficiency and effectiveness and reducing energy consumption.
[0029] Based on the above, the specific implementation method is as follows: When a user first uses the smart dryer to dry a dark gray wool sweater, the system initiates a baseline drying program. The dryer's built-in temperature and humidity sensors, weight sensor, infrared sensor, and current sensor begin working synchronously, collecting data once per second. The system records the sweater's surface temperature and humidity changes, weight decrease curve, infrared reflectivity, and the power of the heater, fan, and the entire machine. The system analyzes the data in 60-second time windows, calculating the mean, standard deviation, and trend of temperature and humidity within each window, including the rate of humidity decrease. It identifies an inflection point in the humidity curve at the 25-minute drying stage, marking the transition from constant-rate drying to falling-rate drying. Based on the weight data, the system estimates the moisture content in real time and calculates the heat exchange efficiency by combining the power data, automatically dividing the process into preheating, constant-rate drying, falling-rate drying, and cooling stages. After the program ends, the user's mobile app receives a satisfaction questionnaire. The user rates the dryness, softness, smoothness, energy consumption, and overall satisfaction of the sweater based on their actual experience. The overall satisfaction score is 5 points, and the user feedback score is 4.8 points. The system associates this score with the previously collected complete drying time series data through the program ID to form a sweater drying process-feedback effect data record, which is stored in the cloud time series database. After accumulating multiple wool sweater drying records, the system activates the predictive control instruction module. The system selects records from historical data with user satisfaction scores higher than 4.5 and excellent dryness ratings as training samples and uses an encoder-decoder long short-term memory network for learning. The encoder inputs the temporal features of the drying process, humidity decrease rate, heat exchange efficiency, and event markers. The decoder combines the static attributes of the clothing, such as material: wool, color: dark gray, initial weight: 0.6 kg, and initial humidity: 45%, to generate a control instruction sequence. The model generates a personalized drying strategy model for wool sweaters by iteratively training by minimizing the error between the predicted instructions and the actual satisfactory records, combined with energy consumption indicators. The model learns that wool materials need to be dried at low temperature and slowly, and dark clothes should avoid being overheated to prevent fading. The optimal drying trajectory should include low temperature preheating - step heating - slow cooling stages. The user put the same wool sweater back in to dry. The system reads the garment tag via RFID to confirm the material is wool and the weight is 0.58kg; the camera visually confirms the color is dark gray; the user manually selects the wool sweater mode in the APP, and the results from all three channels are consistent, so the system adopts the RFID data. After the drying process is started, the sensors collect real-time data on the temperature and humidity inside the chamber, as well as changes in the weight of the clothing. The system calculates real-time characteristics every 60 seconds, including the current rate of humidity decrease and cumulative energy consumption. Combined with the static attributes of the clothing, the system inputs the pre-trained personalized strategy model for wool sweaters. The model outputs a predictive control command for the next 5 minutes, with a target temperature of 42℃, humidity of 30%, and fan speed at medium. The system checked that the command met the safety boundaries, the wool ignition point was greater than 150℃, and the temperature change did not exceed 10℃ / cycle, so the command was executed. During the process, the system dynamically adjusted the command based on real-time data. If the humidity dropped too quickly, the temperature was lowered to ensure that the clothes were not over-dried. A dynamically optimized temperature-humidity-fan speed time-series trajectory was generated throughout the process and recorded as the optimal drying state trajectory. The wool sweater was dried within 40 minutes, soft and smooth. Energy consumption was reduced by 15% compared to the traditional program, and user satisfaction remained high. New data will be fed back to the system for continuous model optimization.
[0030] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.
Claims
1. An intelligent precision temperature and humidity control dryer system based on the Internet of Things, characterized in that, include: The module includes a dataset creation module, a prediction and control instruction module, and an optimal drying state trajectory module. Among them, the prediction and control instruction module is electrically connected to the dataset establishment module, and the optimal dry state trajectory module is electrically connected to the prediction and control instruction module; The dataset building module performs a baseline drying process for the target garment, collects multi-dimensional sensor data in real time during the drying process, generates time-series data of the target garment drying process, receives user feedback ratings on the drying effect and maps them to the time-series data, and builds a target garment drying process-feedback effect dataset. The predictive control instruction module, based on the target clothing drying process-feedback effect dataset, uses machine learning algorithms to establish a personalized drying strategy model for the target clothing, and obtains predictive control instructions for the drying execution components. The optimal drying state trajectory module identifies the static attributes of the clothes to be dried, acquires the current drying process status data in real time, and inputs them into the personalized drying strategy model for the target clothes to obtain the predictive control instructions of the drying execution components of the clothes to be dried. It then dynamically adjusts the drying execution parameters in real time to generate the optimal drying state trajectory of the clothes to be dried.
2. The IoT-based smart precision temperature and humidity controlled drying machine system as claimed in claim 1, wherein, The dataset creation module specifically includes: The target clothing drying process time sequence data generation unit, based on temperature and humidity sensors, weight sensors, infrared sensors, and current sensors, acquires temperature and humidity data of the target clothing, weight data of the target clothing throughout the drying process, monitors changes in the color reflectance of the target clothing, and calculates the power of each component and the whole machine in the drying process. It sets a fixed data acquisition frequency and a synchronization timestamp, collects multi-dimensional sensor data in real time during the drying process, preprocesses the data, and generates target clothing drying process time sequence data. The trend feature extraction unit, based on the generated time-series data of the target clothing drying process, sets a fixed sliding time window length of 60 seconds. It calculates the mean, standard deviation, extreme values, and kurtosis of the multi-dimensional sensor data during the drying process within the time window. It performs a first-order linear regression on the temperature and humidity data of the target clothing within the time window, and performs linear fitting with the rate of temperature and humidity rise and fall of the target clothing within the time window as the slope. It calculates the second-order difference of the time-series data of the target clothing drying process. If the sign of the difference value changes abruptly and exceeds the 0.95 threshold, it is marked as an inflection point. The occurrence time and the time-series value of the target clothing drying process are recorded to obtain the trend features of the time-series data of the target clothing drying process within the time window. 3.The IoT-based smart precision temperature and humidity controlled drying machine system according to claim 2, wherein, The dataset creation module also includes: The moisture content prediction unit calculates the predicted moisture content of the target garment within a time window based on the weight data of the target garment throughout the entire drying process. The heat exchange efficiency calculation unit calculates the temperature rise per unit energy of each component of the drying process within a time window based on the temperature and humidity data of the target clothing and the power of each component and the whole machine during the drying process, and obtains its heat exchange efficiency. The physical feature acquisition unit establishes the overall power curve of the dryer within a time window based on the power of each component and the whole machine during the drying process. It performs numerical integration to obtain the cumulative power integral of the dryer. Combined with the estimated moisture content and heat exchange efficiency of the target clothing, it obtains the physical features of the time-series data of the drying process of the target clothing within the time window. The event feature acquisition unit takes the decrease of more than 5% in the internal humidity of the target clothing within 10 seconds as the start event of the rapid drying stage, and takes preheating, constant speed drying, slow speed drying and cooling as the drying process states, to obtain the time sequence data event features of the drying process of the target clothing within the time window.
4. The IoT-based smart precision temperature and humidity controlled drying machine system as claimed in claim 3, wherein, The dataset creation module also includes: The drying process time sequence recording unit integrates the trend characteristics, physical characteristics, and event characteristics of the target clothing drying process time sequence data within a time window, associates them with the timestamp at the center of the time window, obtains the time sequence characteristics of the target clothing drying process within the time window, acquires the complete target clothing drying process time sequence data characteristics, arranges them in chronological order, generates a complete target clothing drying process time sequence data characteristic sequence matrix, associates it with metadata such as program ID, clothing type, initial weight, and ambient temperature and humidity, obtains a unique drying process time sequence record for the target clothing, and stores it in the time sequence database; The drying process-feedback effect dataset establishment unit is based on the immediate triggering of a feedback questionnaire to a mobile APP after the drying process is completed. Feedback scores are given based on the dryness, softness, smoothness, energy consumption satisfaction, and overall satisfaction of the target clothing. The data is then associated and mapped with the time-series data of the target clothing drying process through the program ID to establish the target clothing drying process-feedback effect dataset. 5.The IoT-based smart precision temperature and humidity controlled drying machine system according to claim 4, wherein, The predictive control instruction module specifically includes: The encoder processing unit, based on the target clothing drying process-feedback effect dataset, divides it into training and testing sets. It uses the complete temporal data features of the target clothing drying process as input to train a long short-term memory network with an encoder-decoder architecture. In the long short-term memory network, the encoder processes the complete temporal data features of the target clothing drying process and outputs the final hidden state. 6.The IoT-based smart precision temperature and humidity controlled drying machine system according to claim 5, wherein, The predictive control instruction module also includes: The transformation unit, through 2-3 fully connected layers, each followed by a ReLU activation function, processes the static properties of the target clothing and encodes them into numerical vectors; The static attributes include: material, initial weight, color, and initial humidity. Among them, the material and color of the clothing are independently coded, and the initial weight and initial humidity are Z-Score normalized. By combining the joint feature vector unit and setting up a feature fusion layer, the final hidden state output by the encoder in the long short-term memory network is concatenated with the static attribute feature vector to obtain the joint feature vector of the target clothing drying process.
7. The IoT-based smart precision temperature and humidity controlled dryers system as claimed in claim 6 wherein, The predictive control instruction module also includes: The predictive control instruction unit takes the joint feature vector of the target clothing drying process as the initial state, inputs it into the long short-term memory network decoder, introduces a fully connected layer, and generates predictive control instructions for the drying execution component for the next K time steps. The model building unit uses the real component control instruction sequences recorded in the historical drying process of the target clothing drying process-feedback effect dataset, in which the overall user satisfaction and dryness score are both higher than the set threshold of 0.95, as training labels. The mean squared error-binary cross-entropy weighted sum is used as the loss function, and the Adam optimizer is used to optimize the model parameters to establish a personalized drying strategy model for the target clothing. 8.The IoT-based smart precision temperature and humidity controlled drying machine system according to claim 1, wherein, The optimal drying state trajectory module specifically includes: The static attribute recognition unit for clothes to be dried is equipped with multiple input interfaces to identify the static attributes of the clothes to be dried. For the manual input channel, the user selects the clothes to be dried via a mobile APP, and the system provides material, weight, and color options for the user to input. For the RFID reading channel, the system is equipped with an RFID tag reader that automatically scans the RFID tags on the clothes to be dried to obtain static attribute data. For the visual recognition channel, the dryer is equipped with an image recognition function that captures images of the clothes through a camera and transmits them to the edge computing unit for real-time analysis to identify the material and color of the clothes to be dried. Furthermore, when the results from multiple channels are inconsistent, the static attributes of the clothes to be dried are identified according to the priority of RFID reading channel > visual recognition channel > manual input channel. The reference value acquisition unit uses temperature and humidity sensors and weight sensors to measure and record the temperature and humidity of the internal environment of the dryer and the initial weight of the clothes to be dried as reference values when the dryer is started. The status data acquisition unit configures a sensor network after the dryer is started. The sensors collect real-time data at a fixed frequency to obtain the internal temperature and humidity of the dryer, the current weight of the clothes to be dried, and the air flow speed. It also acquires the current drying process status data in real time and performs data preprocessing. 9.The IoT-based smart precision temperature and humidity controlled drying machine system according to claim 8, wherein, The optimal drying state trajectory module also includes: The comprehensive feature acquisition unit, for each prediction and control command cycle of the drying execution component, uses sliding window technology to calculate the trend features, physical features and event features of the current drying process status data within the time window, and combines them with the static attributes of the clothes to be dried to perform Z-score standardization processing to obtain the comprehensive features of the current drying process status data. The predictive control command acquisition unit takes the comprehensive characteristics of the current drying process status data and the static attributes of the clothes to be dried as input, associates and matches them with the personalized drying strategy model for the target clothes, and obtains the predictive control command of the drying execution component of the clothes to be dried.
10. The IoT-based intelligent precision temperature and humidity control dryer system according to claim 9, characterized in that, The optimal drying state trajectory module also includes: The optimal drying state trajectory unit is based on the predicted control instructions of the drying execution components of the clothes to be dried, including the target temperature, target humidity and fan speed control parameters of the dryer. It takes the predicted control instructions of the drying execution components of the clothes to be dried, the comprehensive features of the current drying process state data and the static attributes of the clothes to be dried as input, checks whether the predicted control instructions violate the safety boundary. If they violate the safety boundary, they are adjusted to the safe range according to the known rule base. The optimal control instructions under different feature combinations of the optimal target temperature, humidity and fan speed are used as output to generate the optimal drying state trajectory of the clothes to be dried. The safety boundaries are: the temperature exceeds the material's ignition point, the absolute value of the target temperature change in adjacent command cycles is lower than the set threshold t, and a sudden temperature rise is prohibited during a specific high humidity period to prevent condensation.