Drying control method, apparatus and system

Abstract

The present disclosure relates to a drying control method, device and system, comprising: acquiring real-time environment state data of a clothes dryer in a drying process; determining target operating parameters according to the real-time environment state data, wherein the target operating parameters include a target compressor frequency, a target fan speed and a target drum speed; and controlling the clothes dryer to perform a drying operation according to the target operating parameters. Based on the method provided in the present example, multiple actuators can be controlled to work cooperatively, breaking through the limitations of traditional single-variable control and the rigid control of fixed programs or rules, and synchronously improving the drying speed and drying effect.

Description

Technical Field

[0001] This disclosure relates to the field of household appliances, and in particular to drying control methods, apparatus and systems. Background Technology

[0002] As an important smart home appliance in modern households, clothes dryers are widely used in scenarios requiring efficient clothing care. They not only meet the drying needs after daily laundry but also effectively protect clothing fibers and maintain their color, elasticity, and shape through precise low-temperature drying. However, current control strategies for clothes dryers mostly rely on fixed programs or rules to control a single actuator, resulting in limited control and failing to meet users' multi-dimensional needs for drying speed and effectiveness. Summary of the Invention

[0003] To overcome the problems existing in related technologies, this disclosure provides a drying control method, apparatus and system.

[0004] According to a first aspect of the present disclosure, a drying control method is provided, comprising:

[0005] Acquire real-time environmental status data of the dryer during the drying process;

[0006] The target operating parameters are determined based on real-time environmental status data. These target operating parameters include the target compressor frequency, the target fan speed, and the target barrel speed.

[0007] Control the dryer to perform drying operations according to the target operating parameters.

[0008] The beneficial effects are: the control strategy is adaptively determined based on real-time environmental status data, ensuring that the control strategy dynamically adapts to the current working conditions, and the determined control strategy can control multiple actuators to work together, breaking through the limitations of traditional single variable control and rigid control of fixed procedures or rules, and simultaneously improving drying speed and drying effect.

[0009] In one embodiment of this disclosure, determining target operating parameters based on real-time environmental state data includes:

[0010] The real-time operating status of the dryer is determined based on real-time environmental data, which includes: real-time exhaust temperature, real-time steam temperature, real-time chamber temperature, real-time chamber relative humidity, real-time relative humidity of the humidity bar, real-time drying time, real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature.

[0011] The real-time operating status is input into the preset strategy model, and forward calculation is performed on the real-time operating status based on the preset strategy model to obtain the target operating parameters corresponding to the real-time operating status.

[0012] The beneficial effects are: the target operating parameters can be obtained online in real time through the preset strategy model, which can ensure the continuity of control response.

[0013] In one embodiment of this disclosure, before acquiring real-time environmental status data of the dryer during the drying process, the method further includes:

[0014] Determine the training data;

[0015] The policy network using the preset algorithm is initialized as the initial policy model;

[0016] The initial policy model is pre-trained based on the training data to obtain a pre-trained policy model.

[0017] The pre-trained policy model is optimized and trained using a preset reward function to obtain a well-trained policy model.

[0018] The trained policy model is used as the preset policy model.

[0019] The beneficial effects are as follows: the preset strategy model has been adapted by combining historical samples and reward feedback during the training phase. Therefore, when facing different clothing loads and environmental fluctuations, the response to state changes can be more in line with the target control requirements, and the target compressor frequency, target fan speed and target drum speed generated based on real-time environmental state data have better coordination and consistency.

[0020] In one embodiment of this disclosure, determining training data includes:

[0021] The drying process is divided into multiple drying stages based on the relative humidity inside the dryer's compartment;

[0022] Determine the preset operating parameters corresponding to multiple drying stages, including preset compressor frequency, preset fan speed and preset drum speed;

[0023] The dryer is controlled to perform drying operations according to the corresponding preset operating parameters in multiple drying stages, and environmental status data and operating parameters of the dryer are collected at preset time intervals during the drying process. The operating parameters include compressor frequency, fan speed and drum speed.

[0024] Environmental state data and operating parameters collected at the same time are constructed into data pairs, and all data pairs are used to construct training data.

[0025] The beneficial effects are as follows: the obtained training data can correspond to the actual control behavior under different drying stages, so that a time-series correlation is formed between environmental state data and operating parameters. The preset strategy model trained based on the training data can learn the parameter mapping rules of each drying stage, so as to output the target compressor frequency, target fan speed and target drum speed that are adapted to the current state in different drying stages.

[0026] In one embodiment of this disclosure, the drying process is divided into multiple drying stages based on the relative humidity inside the dryer, including:

[0027] The drying stage where the relative humidity inside the chamber is greater than the first preset relative humidity is defined as the preheating period.

[0028] The drying stage in which the relative humidity inside the chamber is less than or equal to the first preset relative humidity and greater than or equal to the second preset relative humidity is defined as the constant-rate drying period.

[0029] The drying stage in which the relative humidity inside the chamber is less than the second preset relative humidity and greater than the third preset relative humidity is defined as the falling-rate drying period.

[0030] The drying stage where the relative humidity inside the chamber is less than or equal to the third preset relative humidity is defined as the slow-release period.

[0031] The beneficial effects are as follows: By dividing the drying process into a preheating period, a constant-rate drying period, a falling-rate drying period, and a tempering period, this example enables the determination of subsequent training data to be based on a more clearly defined drying stage, thereby improving the consistency and stage adaptability of drying control.

[0032] In one embodiment of this disclosure, pre-training an initial policy model based on training data to obtain a pre-trained policy model includes:

[0033] Input the training data into the initial policy model;

[0034] The pre-defined behavior cloning algorithm is deployed into the initial policy model so that the initial policy model can learn by imitation from the training data based on the behavior cloning algorithm.

[0035] The difference between the network weight parameters in the initial policy model and the training data is determined based on the preset loss function.

[0036] The gradient corresponding to the difference is determined according to the preset backpropagation algorithm, and the network weight parameters in the initial policy model are updated according to the preset gradient descent algorithm until the network weight parameters reach the first preset convergence condition. Then the pre-training of the initial policy model ends to obtain the pre-trained policy model.

[0037] The beneficial effects are: it can learn from the historical drying control trajectory pointed out by the expert control strategy, which can make the pre-trained strategy model closer to the effective control strategy of the expert control strategy, thereby improving the accuracy of the model.

[0038] In one embodiment of this disclosure, a pre-trained policy model is optimized and trained using a preset reward function to obtain a trained policy model, including:

[0039] The basic reward value between environmental status data and operating parameters is determined based on a preset reward function;

[0040] Keeping the environmental status data unchanged, the operating parameters are randomly adjusted to obtain the adjusted operating parameters;

[0041] The actual reward value between environmental state data and adjusted operating parameters is determined based on a preset reward function;

[0042] Based on the base reward value and the actual reward value, the pre-trained policy model is optimized and trained until the policy network of the pre-trained policy model reaches the second preset convergence condition, so as to obtain the trained policy model.

[0043] The beneficial effects are as follows: the strategy model trained in this example can more accurately reflect the influence of operating parameters on the reward function, reduce the control deviation caused by single parameter adjustment, make the subsequent online control more adaptable to different loads and environmental fluctuations, and improve the synergy and stability of the drying process.

[0044] In one embodiment of this disclosure, determining the basic reward value between environmental state data and operating parameters according to a preset reward function includes:

[0045] The first energy efficiency reward value, first time reward value, first dryness reward value, and first stability reward value are determined based on the reward function, corresponding to the environmental state data and operating parameters.

[0046] Based on environmental status data, the target weight coefficients corresponding to energy efficiency reward, time reward, dryness reward and stability reward are determined respectively. Among them, the first energy efficiency reward value is the first value of energy efficiency reward, the first time reward value is the first value of time reward, the first dryness reward value is the first value of dryness reward, and the first stability reward value is the first value of stability reward.

[0047] Based on the target weight coefficient, the first energy efficiency reward value, the first time reward value, the first dryness reward value, and the first stability reward value are weighted and summed to obtain the basic reward value.

[0048] The beneficial effects are as follows: The reward function in this example no longer relies on a single indicator, but adaptively allocates the weights of each reward item through environmental state, so that the model can still form a stable optimization objective under complex working conditions, thereby improving the convergence consistency of policy training and the overall balance of control results.

[0049] In one embodiment of this disclosure, determining the actual reward value between environmental state data and adjusted operating parameters according to a preset reward function includes:

[0050] The second energy efficiency reward value, second time reward value, second dryness reward value, and second stability reward value are determined based on the reward function, corresponding to the environmental state data and adjusted operating parameters.

[0051] Based on environmental status data, the target weight coefficients corresponding to energy efficiency reward, time reward, dryness reward and stability reward are determined respectively. Among them, the second energy efficiency reward value is the second value of energy efficiency reward, the second time reward value is the second value of time reward, the second dryness reward value is the second value of dryness reward, and the second stability reward value is the second value of stability reward.

[0052] Based on the target weight coefficient, the second energy efficiency reward value, the second time reward value, the second dryness reward value, and the second stability reward value are weighted and summed to obtain the actual reward value.

[0053] The beneficial effects are as follows: the actual reward value obtained in this example no longer depends on a single indicator, but combines environmental conditions to uniformly quantify multiple control objectives, enabling the strategy model to form more suitable optimization feedback in different drying stages, thereby improving the discrimination accuracy of the adjusted operating parameters and the consistency of training convergence.

[0054] In one embodiment of this disclosure, the target weight coefficients corresponding to energy efficiency bonus, time bonus, dryness bonus, and stability bonus are determined based on environmental state data, including:

[0055] The target drying stage is determined based on the relative humidity inside the chamber.

[0056] Based on the preset weight coefficient table, determine the target weight coefficients for energy efficiency reward, time reward, dryness reward, and stability reward corresponding to the target drying stage. The weight coefficient table includes the weight coefficients for energy efficiency reward, time reward, dryness reward, and stability reward corresponding to multiple drying stages.

[0057] The beneficial effect is that the reward allocation is consistent with the actual drying stage, thereby improving the pertinence and consistency of the control parameter solution.

[0058] In one embodiment of this disclosure, it further includes:

[0059] During the optimization training of the pre-trained policy model, if the environmental state data meets the preset termination condition, the optimization training of the pre-trained policy model will end.

[0060] The beneficial effects are as follows: by setting termination conditions during the optimization training process, the model training can end in a timely manner when the preset convergence or task completion characteristics are met, so that the trained strategy model is consistent with the actual end state of the drying process, reducing the disturbance of model parameters by invalid iterations, and making the subsequent control output more in line with the actual drying conditions.

[0061] In one embodiment of this disclosure, the preset termination conditions include: the relative humidity of the humidity bar being less than a preset relative humidity threshold for a duration exceeding a first preset duration, or the drying time exceeding a second preset duration.

[0062] The beneficial effects are: the termination timing can be constrained by both the humidity change results and the time limit, so that the termination judgment can reflect the actual degree of drying and limit the longest operating boundary, thereby making the termination judgment of the drying process more stable and reducing premature termination or ineffective extension caused by humidity fluctuations.

[0063] In one embodiment of this disclosure, the method further includes:

[0064] If the real-time compressor temperature exceeds the preset compressor safe temperature, a compressor protection strategy will be generated.

[0065] If the real-time compressor frequency reaches the preset resonance frequency, a compressor frequency resonance protection strategy will be generated.

[0066] If the real-time condenser temperature exceeds the preset safe condenser temperature, a condenser protection strategy will be generated.

[0067] Control the dryer to implement compressor protection strategies and / or compressor frequency resonance protection strategies and / or condenser protection strategies.

[0068] The beneficial effects are as follows: The protection strategy in this example is triggered by real-time detection values, thus enabling timely intervention before the equipment enters overheating, resonance, or abnormal heat load states, maintaining a stable operating relationship between the compressor, condenser, and overall machine control. Adopting this approach allows for timely response to abnormal operating conditions without altering the main drying control framework, reducing damage to critical components while maintaining the continuity and consistency of the drying process and control.

[0069] According to a second aspect of the present disclosure, a drying control device is provided, comprising:

[0070] The acquisition module is used to acquire real-time environmental status data of the dryer during the drying process;

[0071] The processing module is used to determine the target operating parameters based on real-time environmental status data. The target operating parameters include the target compressor frequency, the target fan speed, and the target barrel speed.

[0072] The control module is used to control the dryer to perform drying operations according to the target operating parameters.

[0073] Optional real-time environmental status data include: real-time exhaust temperature, real-time steam temperature, real-time chamber temperature, real-time chamber relative humidity, real-time humidity bar relative humidity, real-time drying time, real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature.

[0074] In one embodiment of this disclosure, the processing module is specifically used to determine the real-time operating status of the dryer based on real-time environmental status data;

[0075] The processing module is also used to input the real-time running status into the preset strategy model, perform forward calculation on the real-time running status based on the preset strategy model, and obtain the target running parameters corresponding to the real-time running status.

[0076] In one embodiment of this disclosure, the processing module is further configured to determine training data;

[0077] The processing module is also used to initialize the policy network using a preset algorithm as an initial policy model;

[0078] The processing module is also used to pre-train the initial policy model based on the training data to obtain a pre-trained policy model.

[0079] The processing module is also used to optimize the pre-trained policy model through a preset reward function to obtain a trained policy model.

[0080] The processing module is also used to use the trained policy model as a preset policy model.

[0081] In one embodiment of this disclosure, the processing module is further configured to divide the drying process into multiple drying stages based on the relative humidity inside the dryer.

[0082] The processing module is also used to determine the preset operating parameters corresponding to multiple drying stages, including preset compressor frequency, preset fan speed and preset drum speed.

[0083] The processing module is also used to control the dryer to perform drying operations according to the corresponding preset operating parameters in multiple drying stages, and to collect environmental status data and operating parameters of the dryer at preset time intervals during the drying process. The operating parameters include compressor frequency, fan speed and drum speed.

[0084] The processing module is specifically used to construct data pairs from environmental state data and operating parameters collected at the same time, and to construct training data from all data pairs.

[0085] In one embodiment of this disclosure, the processing module is further configured to determine the drying stage in which the relative humidity inside the chamber is greater than a first preset relative humidity as the preheating period;

[0086] The processing module is also specifically used to determine the drying stage in which the relative humidity inside the chamber is less than or equal to the first preset relative humidity and the relative humidity inside the chamber is greater than or equal to the second preset relative humidity as a constant-rate drying period.

[0087] The processing module is also specifically used to determine the drying stage in which the relative humidity inside the chamber is less than the second preset relative humidity and greater than the third preset relative humidity as the falling rate drying period.

[0088] The processing module is also specifically used to determine the drying stage where the relative humidity inside the chamber is less than or equal to the third preset relative humidity as the tempering period.

[0089] In one embodiment of this disclosure, the processing module is further configured to input training data into the initial policy model;

[0090] The processing module is also specifically used to deploy the preset behavior cloning algorithm into the initial policy model, so that the initial policy model can learn by imitation of the training data based on the behavior cloning algorithm.

[0091] The processing module is also specifically used to determine the difference between the network weight parameters in the initial policy model and the training data based on the preset loss function;

[0092] The processing module is further used to determine the gradient corresponding to the difference according to the preset backpropagation algorithm, and to update the network weight parameters in the initial policy model according to the preset gradient descent algorithm until the network weight parameters reach the first preset convergence condition, and then end the pre-training of the initial policy model to obtain the pre-trained policy model.

[0093] In one embodiment of this disclosure, the processing module is further configured to determine a basic reward value between environmental state data and operating parameters based on a preset reward function;

[0094] The processing module is also used to keep the environmental status data unchanged and randomly adjust the operating parameters to obtain the adjusted operating parameters;

[0095] The processing module is also specifically used to determine the actual reward value between the environmental state data and the adjusted operating parameters based on the preset reward function;

[0096] The processing module is specifically used to optimize the pre-trained policy model based on the base reward value and the actual reward value until the policy network of the pre-trained policy model reaches the second preset convergence condition, so as to obtain the trained policy model.

[0097] In one embodiment of this disclosure, the processing module is further configured to determine, based on the reward function, a first energy efficiency reward value, a first time reward value, a first dryness reward value, and a first stability reward value corresponding to the environmental state data and operating parameters;

[0098] The processing module is further used to determine the target weight coefficients corresponding to energy efficiency reward, time reward, dryness reward and stability reward respectively based on environmental status data. The first energy efficiency reward value is the first value of energy efficiency reward, the first time reward value is the first value of time reward, the first dryness reward value is the first value of dryness reward and the first stability reward value is the first value of stability reward.

[0099] The processing module is also used to perform a weighted summation operation on the first energy efficiency reward value, the first time reward value, the first dryness reward value, and the first stability reward value based on the target weight coefficient, so as to obtain the basic reward value.

[0100] In one embodiment of this disclosure, the processing module is further configured to determine, based on the reward function, a second energy efficiency reward value, a second time reward value, a second dryness reward value, and a second stability reward value corresponding to the environmental state data and the adjusted operating parameters;

[0101] The processing module is further used to determine the target weight coefficients corresponding to energy efficiency reward, time reward, dryness reward and stability reward respectively based on environmental status data. The second energy efficiency reward value is the second value of energy efficiency reward, the second time reward value is the second value of time reward, the second dryness reward value is the second value of dryness reward and the second stability reward value is the second value of stability reward.

[0102] The processing module is also specifically used to perform a weighted summation operation on the second energy efficiency reward value, the second time reward value, the second dryness reward value, and the second stability reward value based on the target weight coefficient, so as to obtain the actual reward value.

[0103] In one embodiment of this disclosure, the processing module is further configured to determine the target drying stage based on the relative humidity inside the chamber;

[0104] The processing module is further used to determine the target weight coefficients for energy efficiency rewards, time rewards, dryness rewards, and stability rewards corresponding to the target drying stage based on a preset weight coefficient table. The weight coefficient table includes the weight coefficients for energy efficiency rewards, time rewards, dryness rewards, and stability rewards corresponding to multiple drying stages.

[0105] In one embodiment of this disclosure, the processing module is further configured to terminate the optimization training of the pre-trained policy model if the environmental state data meets a preset termination condition during the optimization training of the pre-trained policy model.

[0106] In one embodiment of this disclosure, the preset termination conditions include: the relative humidity of the humidity bar being less than a preset relative humidity threshold for a duration exceeding a first preset duration, or the drying time exceeding a second preset duration.

[0107] In one embodiment of this disclosure, the processing module is further configured to generate a compressor protection strategy if the real-time compressor temperature exceeds a preset compressor safety temperature.

[0108] The processing module is also used to generate a compressor frequency resonance protection strategy if the real-time compressor frequency reaches the preset resonance frequency.

[0109] The processing module is also used to generate a condenser protection strategy if the real-time condenser temperature exceeds the preset condenser safety temperature.

[0110] The processing module is also used to control the dryer to execute compressor protection strategies and / or compressor frequency resonance protection strategies and / or condenser protection strategies.

[0111] According to a third aspect of the present disclosure, a drying control system is provided, comprising: a clothes dryer and a cloud server, wherein the clothes dryer includes a sensor mechanism and a control mechanism, and the cloud server includes a decision module;

[0112] The sensor mechanism is used to collect real-time environmental status data of the dryer during the drying process and send the real-time environmental status data to the decision module;

[0113] The decision module is used to determine the target operating parameters based on real-time environmental status data and send the target operating parameters to the control mechanism. The target operating parameters include the target compressor frequency, the target fan speed, and the target barrel speed.

[0114] The control mechanism is used to control the dryer to perform drying operations according to the target operating parameters.

[0115] In one embodiment of this disclosure, the cloud server further includes a security monitoring module; real-time environmental status data includes: real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature;

[0116] The safety monitoring module is used to generate a compressor protection strategy when the real-time compressor temperature exceeds the preset compressor safety temperature; to generate a compressor frequency resonance protection strategy when the real-time compressor frequency reaches the preset resonance frequency; and to generate a condenser protection strategy when the real-time condenser temperature exceeds the preset condenser safety temperature.

[0117] The safety monitoring module is also used to send compressor protection strategies and / or compressor frequency resonance protection strategies and / or condenser protection strategies to the control mechanism, so that the control mechanism executes the compressor protection strategies and / or compressor frequency resonance protection strategies and / or condenser protection strategies.

[0118] According to a fourth aspect of the present disclosure, an air conditioner is provided, comprising: a processor, a memory communicatively connected to the processor, and a compressor;

[0119] The memory stores the instructions that the computer executes;

[0120] The processor executes computer execution instructions stored in memory to implement the drying control method as described in any of the first aspects above.

[0121] According to a fifth aspect of the present disclosure, a computer-readable storage medium is provided, which stores computer-executable instructions that, when executed by a processor, are used to implement the drying control method as described in any of the first aspects above.

[0122] According to a sixth aspect of the present disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the drying control method as described in any of the first aspects above.

[0123] The technical solutions provided by the embodiments of this disclosure may include the following beneficial effects: This disclosure adaptively determines the control strategy based on real-time environmental state data, ensuring that the control strategy dynamically adapts to the current working conditions, and the determined control strategy can control multiple actuators to work together, breaking through the limitations of traditional single variable control and rigid control of fixed procedures or rules, and simultaneously improving drying speed and drying effect.

[0124] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0125] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0126] Figure 1 This is a schematic diagram of the scenario involved in this disclosure;

[0127] Figure 2 This is a schematic flowchart of a drying control method disclosed herein. Figure 1 ;

[0128] Figure 3 This is a schematic flowchart of a drying control method disclosed herein. Figure 2 ;

[0129] Figure 4 This is a schematic flowchart of a drying control method disclosed herein. Figure 3 ;

[0130] Figure 5 This is a schematic flowchart of a drying control method disclosed herein. Figure 4 ;

[0131] Figure 6 This is a schematic flowchart of a drying control method disclosed herein. Figure 5 ;

[0132] Figure 7 This is a schematic diagram illustrating the training process of an example model;

[0133] Figure 8 This is a schematic flowchart of a drying control method disclosed herein. Figure 6 ;

[0134] Figure 9 This is a schematic flowchart of a drying control method disclosed herein. Figure 7 ;

[0135] Figure 10 This is a network architecture diagram of an example TD3 agent;

[0136] Figure 11 This is a schematic flowchart of a drying control method disclosed herein. Figure 8 ;

[0137] Figure 12 This is a schematic flowchart of a drying control method disclosed herein. Figure 9 ;

[0138] Figure 13 This is a schematic flowchart of a drying control method disclosed herein. Figure 10 ;

[0139] Figure 14This is a schematic flowchart of a drying control method disclosed herein. Figure 10 one;

[0140] Figure 15 This is a schematic diagram of the structure of a drying control device disclosed herein;

[0141] Figure 16 This is a schematic diagram of the drying control system;

[0142] Figure 17 This is a schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

[0143] Some embodiments of this disclosure will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. Various changes, modifications, and equivalents of the methods, apparatus, and / or systems described herein will become apparent upon understanding this disclosure. For example, the order of operations described herein is merely illustrative and is not limited to those orders set forth herein, but can be changed as will become apparent upon understanding this disclosure, except for operations that must be performed in a particular order. Furthermore, for clarity and brevity, descriptions of features known in the art may be omitted.

[0144] First, the application scenarios of this disclosure will be explained. This disclosure relates to the field of household appliances. As an important smart home appliance in modern households, clothes dryers are widely used in scenarios requiring efficient clothing care. They not only meet the drying needs after daily laundry in a timely manner, but also effectively protect clothing fibers and maintain their color, elasticity, and shape through precise low-temperature drying. However, most current control strategies for clothes dryers control a single actuator through fixed programs or rules, resulting in a single control dimension and failing to meet users' multi-dimensional needs for drying speed and drying effect.

[0145] To address the aforementioned problems, this disclosure provides a drying control method, apparatus, and system. A schematic diagram of the scenario involved in this disclosure is shown below. Figure 1As shown, this disclosure relates to a local dryer and a cloud server. The local dryer can be a heat pump dryer or a washer-dryer combo, etc. During the drying process, the dryer collects real-time environmental status data and uploads it to the cloud server. The cloud server processes the real-time environmental status data to extract the current operating status of the dryer and processes it online through a strategy network to obtain a control strategy for the dryer. The control strategy consists of target operating parameters for controlling the operation of the compressor, fan, and drum, such as the target compressor frequency, target fan speed, and target drum speed. These target operating parameters are then sent back to the dryer to achieve closed-loop control, that is, controlling the operation of the compressor, fan, and drum based on the target compressor frequency, target fan speed, and target drum speed. This disclosure adaptively determines the control strategy based on real-time environmental status data, ensuring that the control strategy dynamically adapts to the current operating conditions. Furthermore, the determined control strategy can control multiple actuators to work collaboratively, breaking through the limitations of traditional single-variable control and rigid control based on fixed procedures or rules, simultaneously improving drying speed and drying effect.

[0146] The present disclosure will now be described with reference to specific embodiments. The embodiments described in the following examples do not represent all embodiments consistent with the present disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0147] Figure 2 This is a schematic flowchart of a drying control method disclosed herein. Figure 1 ,like Figure 2 As shown, it includes:

[0148] S201. Obtain real-time environmental status data of the dryer during the drying process.

[0149] For example, in this example, the drying control method is applied to a cloud server, which acts as a computing and decision-making center and is connected to the dryer via network communication. The dryer, as the controlled object, includes a variable frequency compressor, a continuously variable speed fan, a bidirectional drum motor, and internal temperature and humidity sensors, a humidity strip, and a temperature sensor. The temperature sensor collects exhaust temperature, condenser temperature, and steam temperature. An embedded main control device with an integrated IoT communication module is also included. The user places wet clothes into the dryer and starts the drum drying program. Real-time environmental status data is used to characterize the current drying condition of the dryer. Specifically, this can be status information collected by detection components inside the dryer during drying operation, including at least one or more data reflecting the current thermal and / or wet conditions. The dryer uploads the real-time environmental status data to the cloud server via network communication, enabling the cloud server to access the real-time environmental status data.

[0150] S202. Determine the target operating parameters based on real-time environmental status data, including the target compressor frequency, target fan speed, and target barrel speed.

[0151] For example, the target operating parameters are determined based on multivariate collaborative outputs derived from real-time environmental status data. The target compressor frequency controls the compressor's operating frequency to adjust heating capacity and system energy consumption; the target fan speed controls the fan's airflow intensity and the air circulation speed in the duct; and the target tub speed controls the tub's tumbling rhythm and the mechanical tumbling of clothing. The cloud server can process the real-time environmental status data acquired by S201 based on preset control logic, rule algorithms, lookup table methods, or other parameter determination methods to obtain the target compressor frequency, target fan speed, and target tub speed. Alternatively, model algorithms can be used to assist in determining the target operating parameters.

[0152] Optional, Figure 3 This is a schematic flowchart of a drying control method disclosed herein. Figure 2 ,like Figure 3 As shown, in S202, the target operating parameters are determined based on real-time environmental status data, including:

[0153] S301. Determine the real-time operating status of the dryer based on real-time environmental status data, including: real-time exhaust temperature, real-time steam temperature, real-time chamber temperature, real-time chamber relative humidity, real-time humidity bar relative humidity, real-time drying time, real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature.

[0154] For example, the main control unit of the dryer collects environmental status data in real time through temperature and humidity sensors and / or temperature sensors. Real-time exhaust temperature refers to the real-time temperature of the hot air blown out by the fan at the air outlet of the duct; real-time steam temperature refers to the real-time temperature of the high-temperature steam generated by the dryer; real-time chamber temperature refers to the ambient air temperature in the drying chamber where the clothes are located; real-time chamber relative humidity refers to the percentage of water vapor content in the air inside the drum, a core data point for determining the dryness of the clothes; the relative humidity of the real-time humidity bar is similar to the real-time chamber relative humidity, presented in numerical form; real-time drying time refers to the cumulative running time since the dryer's drying process started; real-time compressor temperature is the real-time operating temperature of the heat pump cycle compressor body; real-time compressor frequency refers to the real-time operating frequency of the compressor; and real-time condenser temperature refers to the real-time temperature of the condenser heat exchanger surface. After obtaining the real-time environmental status data, the cloud server processes and extracts features to determine the real-time operating status of the dryer. The real-time operating status of the dryer is equivalent to its real-time environmental status data.

[0155] S302. Input the real-time operating status into the preset strategy model, perform forward calculation on the real-time operating status based on the preset strategy model, and obtain the target operating parameters corresponding to the real-time operating status.

[0156] For example, a cloud server deploys a pre-set strategy model containing a trained, performance-compliant Twin Delayed Deep Deterministic Policy Gradient (TD3) network model. The real-time operating state of the dryer is input into the pre-set strategy model, and forward computation is performed through the TD3 network to obtain the control action matching the dryer's current operating state. The real-time environmental state data collected in this example includes comprehensive state data during the dryer's operation, accurately reflecting its status. Therefore, determining the target operating parameters based on this real-time environmental state data improves the accuracy and comprehensiveness of these parameters. Thus, the target operating parameters obtained in this example include the target compressor frequency, target fan speed, and target drum speed, enhancing the comprehensiveness and accuracy of dryer control. Furthermore, this example obtains the target operating parameters online in real-time through the pre-set strategy model, ensuring continuous control response.

[0157] S203. Control the dryer to perform drying operations according to the target operating parameters.

[0158] For example, after obtaining the target operating parameters, the target compressor frequency, target fan speed, and target drum speed are sent to the dryer's main control unit. This allows the main control unit to adjust the compressor's operating frequency to the target compressor frequency, the fan speed to the target fan speed, and the drum speed to the target drum speed. This example demonstrates how by implementing the determined target operating parameters into the control of the compressor, fan, and drum, the dryer can operate according to the target operating parameters corresponding to the real-time operating state throughout the drying process. This allows for dynamic adjustment of the dryer's control for different operating states.

[0159] This example adaptively determines the control strategy based on real-time environmental status data, ensuring that the control strategy dynamically adapts to the current working conditions. Furthermore, the determined control strategy can control multiple actuators to work collaboratively, breaking through the limitations of traditional single-variable control and rigid control of fixed procedures or rules, and simultaneously improving drying speed and drying effect.

[0160] Optional, Figure 4 This is a schematic flowchart of a drying control method disclosed herein. Figure 3 ,like Figure 4 As shown, before S201, it also includes:

[0161] S401. Determine the training data.

[0162] For example, during the offline training phase, the dryer controller first builds training data based on historical drying records. The training data is then used to train the strategy model to obtain the preset strategy model.

[0163] Optional, Figure 5 This is a schematic flowchart of a drying control method disclosed herein. Figure 4 ,like Figure 5 As shown, S401 includes:

[0164] S501. The drying process is divided into multiple drying stages based on the relative humidity inside the dryer.

[0165] For example, based on the drying kinetics principle of heat pump drying, the drying process is divided into drying stages according to the relative humidity inside the chamber.

[0166] Optional, Figure 6 This is a schematic flowchart of a drying control method disclosed herein. Figure 5 ,like Figure 6 As shown, S501 includes:

[0167] S601, The drying stage in which the relative humidity inside the chamber is greater than the first preset relative humidity is defined as the preheating period.

[0168] For example, the first preset relative humidity can be determined according to the actual situation. For instance, the first preset relative humidity can be defined as H1%. When the relative humidity inside the chamber is > H1%, the humidity inside the chamber rises rapidly. The control target is to rapidly raise the temperature to the preset target evaporation temperature. Therefore, this drying stage can be defined as the preheating period, where the target evaporation temperature can be 50°C.

[0169] S602. The drying stage in which the relative humidity inside the chamber is less than or equal to the first preset relative humidity and greater than or equal to the second preset relative humidity is defined as the constant-rate drying period.

[0170] For example, the second preset relative humidity can be determined according to the actual situation. For instance, the second preset relative humidity can be defined as H2% and H1% > H2%. When H2% ≤ relative humidity inside the chamber ≤ H1%, the moisture on the surface of the clothes continues to evaporate. The control target is to maintain the maximum evaporation rate that the dryer can achieve, so this drying stage can be defined as the constant-rate drying period.

[0171] S603. The drying stage in which the relative humidity inside the chamber is less than the second preset relative humidity and greater than the third preset relative humidity is defined as the falling-rate drying period.

[0172] For example, the third preset relative humidity can be determined according to the actual situation. For instance, the third preset relative humidity can be defined as H3%, H2% > H3%. When H3% < relative humidity inside the chamber < H2%, the diffusion of moisture inside the clothes is dominant. The control objective is to balance energy consumption and drying efficiency, so this drying stage can be defined as the slow-down drying period.

[0173] S604. The drying stage in which the relative humidity inside the chamber is less than or equal to the third preset relative humidity is defined as the warming period.

[0174] For example, when the relative humidity inside the chamber is ≤H3%, the residual moisture diffuses, and the control objective is to minimize energy consumption to complete the final drying. Therefore, this drying stage can be defined as the tempering period.

[0175] For example, by dividing the drying process into a preheating period, a constant-rate drying period, a falling-rate drying period, and a tempering period, this example enables the determination of subsequent training data to be based on a more clearly defined drying stage, thereby improving the consistency and stage adaptability of drying control.

[0176] S502. Determine the preset operating parameters corresponding to multiple drying stages, including preset compressor frequency, preset fan speed and preset drum speed.

[0177] For example, the preset operating parameters for the drying stage are designed based on the expert control strategy. Table 1 shows the preset operating parameters for each drying stage.

[0178] Table 1

[0179]

[0180] The specific control quantities for different drying stages satisfy f1>f2>f3, v2>v1, u1>u2>u3.

[0181] S503 controls the dryer to perform drying operations according to the corresponding preset operating parameters in multiple drying stages, and collects the environmental status data and operating parameters of the dryer at preset time intervals during the drying process. The operating parameters include compressor frequency, fan speed and drum speed.

[0182] For example, based on the preset operating parameters for each drying stage in Table 1, the compressor, fan, and drum are controlled. The preset time interval can be determined according to actual conditions, for example, it can be 10 seconds. The actual environmental state data of the dryer collected during operation according to the preset operating parameters can be expressed as follows:

[0183] s t exp , .

[0184] Among them, T cp T represents the exhaust temperature. cd T represents the condenser temperature. ep T represents the temperature during steaming. ntc This indicates the cabin temperature measured by the temperature and humidity sensor, RH. ntc The value represents the relative humidity of the drying chamber as measured by the temperature and humidity sensor. RH represents the relative humidity of the humidity bar at the bottom of the drying chamber where the clothes are in direct contact. t represents the drying time.

[0185] The collected operating parameters are the control actions of the expert control strategy, which can be expressed as:

[0186] a t exp , .

[0187] Where f represents the compressor frequency, v represents the fan speed, and u represents the barrel speed.

[0188] S504. Construct environmental state data and operating parameters collected at the same time into data pairs, and construct training data from all data pairs.

[0189] For example, by controlling the dryer's operation according to preset operating parameters, the control trajectory of the dryer under expert control strategy can be obtained. The actual environmental state data and actual operating parameter changes of the dryer at the same moment are constructed into data pairs, demonstrating the effectiveness of the expert control strategy in the drying process, resulting in a demonstration dataset. The demonstration dataset can be used as the training data. The demonstration dataset is shown in Formula 1 below:

[0190] Formula 1:

[0191] Based on the method provided in this example, the obtained training data can correspond to the actual control behavior under different drying stages, so that a temporal correlation is formed between environmental state data and operating parameters. The preset strategy model trained based on this training data can learn the parameter mapping rules of each drying stage, so as to output the target compressor frequency, target fan speed and target drum speed that are adapted to the current state in different drying stages.

[0192] S402. Initialize the policy network using the preset algorithm as the initial policy model.

[0193] For example, the preset algorithm used by the policy network can be the TD3 algorithm. After initializing the policy network of the TD3 algorithm, an initial policy model is obtained.

[0194] S403. Pre-train the initial policy model based on the training data to obtain a pre-trained policy model.

[0195] Example, Figure 7 This is a schematic diagram of the example model training process, such as... Figure 7 As shown, the model training process includes two stages: offline imitation training and online reinforcement learning. The offline imitation training stage involves pre-training the initial policy model. This initial policy model is an imitation learning model. Using training data, supervised learning training is performed on the initial policy model with the objective of minimizing the mean squared error between the model's predicted operating parameters and the actions of the expert control policy in the training data. This results in a pre-trained policy model, which yields a high-performance initial policy network.

[0196] Optional, Figure 8 This is a schematic flowchart of a drying control method disclosed herein. Figure 6 ,like Figure 8 As shown, S403 includes:

[0197] S801. Input the training data into the initial policy model.

[0198] For example, the initial policy model is trained by feeding training data into it.

[0199] S802. Deploy the preset behavior cloning algorithm into the initial policy model so that the initial policy model can learn by imitation of the training data based on the behavior cloning algorithm.

[0200] For example, the pre-training of the initial policy model uses the Behavioral Cloning (BC) algorithm. BC is a method that learns control policies by imitating data pairs in the training data; it is essentially a form of supervised learning. For the drying control problem of a clothes dryer, the BC algorithm implements the learning process of the TD3 agent policy network through regression.

[0201] S803. Determine the difference between the network weight parameters in the initial policy model and the training data based on the preset loss function.

[0202] For example, before training the initial policy model, the data pairs in the training data are first constructed into tuples X, as shown in Formula 2 below:

[0203] Formula 2:

[0204] Where m is the number of data pairs in the training data, s j exp and a j exp These correspond to the input data and training reference results in the supervised learning process, respectively.

[0205] The BC algorithm encourages the policy network to make action decisions consistent with the expert control policy, that is, it encourages the policy network to output actions consistent with a. j exp Therefore, given the similar results, the loss function is defined as shown in Formula 3 below:

[0206] Formula 3:

[0207] Among them, L BC Let φ0 be the loss function, and φ0 be the policy network π. BC The weighting parameter of (•).

[0208] S804. Determine the gradient corresponding to the difference according to the preset backpropagation algorithm, and update the network weight parameters in the initial policy model according to the preset gradient descent algorithm until the network weight parameters reach the first preset convergence condition, then end the pre-training of the initial policy model to obtain the pre-trained policy model.

[0209] For example, the policy network then uses back propagation (BP) to compute the gradient and gradient descent to update the network weight parameters until the network weight parameters meet a first preset convergence condition. The first preset convergence condition is shown in Equation 4 below:

[0210] Formula 4:

[0211] Where β is the learning rate of the behavioral clone; (•) is the gradient of the loss function. Therefore, once Equation 4 converges, it means that the pre-training of the initial policy model has been completed, and the final trained model result is determined as the pre-trained policy model.

[0212] Based on the method provided in this example, this example can learn the historical drying control trajectory indicated by the expert control strategy, which can make the pre-trained strategy model closer to the effective control strategy of the expert control strategy, thereby improving the accuracy of the model.

[0213] S404. Optimize and train the pre-trained policy model using a preset reward function to obtain a well-trained policy model.

[0214] For example, such as Figure 7 As shown, after the pre-trained policy model meets the transfer requirements, the pre-trained policy model is optimized and trained through an online reinforcement learning phase to obtain a well-trained policy model.

[0215] Optional, Figure 9 This is a schematic flowchart of a drying control method disclosed herein. Figure 7 ,like Figure 9 As shown, S404 includes:

[0216] S901. Determine the basic reward value between environmental status data and operating parameters based on the preset reward function.

[0217] Example, Figure 10 This is a network architecture diagram of the example TD3 agent, including an Actor network and a Critic network. During the online reinforcement learning phase, the Actor network of the TD3 algorithm is initialized with the policy network from the pre-trained policy model. The Actor network's policy network includes an input layer, hidden layers, and an output layer. The input layer receives the dryer's operating state, processes it through multiple linear layers in the hidden layers, and outputs the control action through the output layer. The Actor network uses backpropagation (BP) to calculate gradients and uses gradient descent to update the network weight parameters until the network weight parameters reach a first preset convergence condition. In the Critic network, the TD3 algorithm mitigates overestimation of the value function through smoothing, improving stability. The agent maximizes the cumulative reward function r. t To optimize the strategy, the reward function integrates energy efficiency rewards, time rewards, dryness rewards, and stability rewards.

[0218] Optional, Figure 11 This is a schematic flowchart of a drying control method disclosed herein. Figure 8 ,like Figure 11 As shown, S1001 includes:

[0219] S1101. Determine the first energy efficiency reward value, first time reward value, first dryness reward value, and first stability reward value corresponding to the environmental state data and operating parameters based on the reward function.

[0220] Example, combination Figure 10 The Q1 value, obtained from processing on the expert 1 side of the Q-network (Critic), is the base reward value derived from the operating state and control actions corresponding to the operating parameters based on the environmental state data. The policy model generates a control policy based on the observed environmental state data, which directly affects the accuracy of the decision. To ensure the effectiveness of the decision, the environmental state data should reflect the dynamic changes of the dryer environment as comprehensively as possible, while avoiding invalid and redundant information. During the online fine-tuning phase, the state of the TD3 agent is consistent with the state acquired from the expert control policy during the pre-training phase. The state of the TD3 agent is defined as: s t , The actions output by the TD3 agent are consistent with the action settings of the expert control strategy during the pre-training phase, defined as a. t, .

[0221] Energy efficiency rewards incentivize the TD3 agent to adopt low-power operating strategies. The TD3 agent reduces the cumulative value of control actions by lowering compressor frequency, optimizing fan speed, and drum speed, thereby obtaining higher rewards. This directly guides the dryer system towards energy-saving optimization. The energy efficiency reward can be obtained using the following formula:

[0222] Formula 5:

[0223] Where, r e For energy efficiency rewards, k e This represents the energy efficiency conversion factor, used to map the cumulative value of control actions to an appropriate reward value scale.

[0224] The time reward applies a constant negative reward to the passage of time, incentivizing the TD3 agent to complete the drying task of the current stage as quickly as possible, thereby shortening the overall drying time. The time reward can be obtained through the following formula 6:

[0225] Formula 6:

[0226] Where, r ti As a time reward, k t This represents a time coefficient, the value of which is dynamically adjusted according to different drying stages. Different values ​​of k are set for different stages. t The value can reflect the different time urgency requirements at each stage.

[0227] Dryness reward is the core of control accuracy. Based on the absolute deviation between the current humidity and the target humidity, it drives the TD3 agent to accurately maintain the system humidity near the preset optimal trajectory, which is crucial to ensuring the drying effect. The dryness reward can be obtained through the following formula 7:

[0228] Formula 7:

[0229] Where, r d As a dryness bonus, RH s This represents the target relative humidity to be achieved in the current drying stage, a setpoint determined by drying kinetics, k. d This represents the dryness weight coefficient, used to adjust the importance of humidity deviation in the total reward. In the early stages with fewer iterations, the weight of the dryness reward can be increased to guide the agent to converge quickly to a reliable policy. As the number of iterations increases, the weight of the dryness reward needs to be appropriately reduced, allowing the TD3 agent to explore more broadly and obtain the optimal control policy.

[0230] The stability reward encourages the TD3 agent to learn a smooth control strategy based on drastic temperature fluctuations. This helps protect clothing fibers from thermal shock, reduces mechanical wear on actuators, and improves system reliability and control quality. The stability reward can be obtained using the following formula:

[0231] Formula 8:

[0232] Where, r s As a stability reward, k represents the variance of the cabin temperature measured by the temperature and humidity sensor within 5 minutes. s This represents the stability coefficient.

[0233] Based on the above formulas 5, 6, 7 and 8, the first energy efficiency bonus value, the first time bonus value, the first dryness bonus value and the first stability bonus value are obtained respectively.

[0234] S1102. Determine the target weight coefficients corresponding to energy efficiency reward, time reward, dryness reward and stability reward respectively based on environmental status data. Among them, the first energy efficiency reward value is the first value of energy efficiency reward, the first time reward value is the first value of time reward, the first dryness reward value is the first value of dryness reward and the first stability reward value is the first value of stability reward.

[0235] For example, the first energy efficiency bonus value, the first time bonus value, the first dryness bonus value, and the first stability bonus value are the first values ​​of the energy efficiency bonus item, the time bonus item, the dryness bonus item, and the stability bonus item, respectively.

[0236] Optional, Figure 12 This is a schematic flowchart of a drying control method disclosed herein. Figure 9 ,like Figure 12 As shown, the process of determining the target weight coefficients includes:

[0237] S1201. Determine the target drying stage based on the relative humidity inside the chamber.

[0238] For example, the weight coefficients of each reward item are not fixed, but are determined based on the relative humidity inside the cabin in the environmental status data, according to Table 1.

[0239] S1202. Based on the preset weight coefficient table, determine the target weight coefficients for energy efficiency reward, time reward, dryness reward, and stability reward corresponding to the target drying stage. The weight coefficient table includes the weight coefficients for energy efficiency reward, time reward, dryness reward, and stability reward corresponding to multiple drying stages.

[0240] For example, the target weight coefficients for energy efficiency rewards, time rewards, dryness rewards, and stability rewards are determined based on the current target drying stage. The weight coefficients for energy efficiency rewards, time rewards, dryness rewards, and stability rewards at different drying stages are determined by the weight coefficient table, as shown in Table 2 below.

[0241] Table 2

[0242]

[0243] in, , , , .

[0244] This example uses a weighted coefficient table to match reward weights, enabling reward calculations to dynamically switch according to the drying process. This allows for coordinated constraints on four types of indicators—energy efficiency, time, dryness, and stability—during strategy optimization, and outputs target operating parameters adapted to the current drying state. This approach ensures that reward allocation remains consistent with the actual drying stage, thereby improving the relevance and consistency of control parameter solutions.

[0245] S1103. Based on the target weight coefficient, perform a weighted summation operation on the first energy efficiency reward value, the first time reward value, the first dryness reward value, and the first stability reward value to obtain the basic reward value.

[0246] For example, the reward r obtained at each step t It is obtained by weighted summation of the four sub-reward items, and the weighted summation formula is as follows: Formula 9:

[0247] Formula 9:

[0248] Among them, w1, w2, w3, and w4 represent the weighting coefficients of energy efficiency bonus, time bonus, dryness bonus, and stability bonus, respectively.

[0249] Therefore, in this example, the first energy efficiency reward value, the first time reward value, the first dryness reward value, and the first stability reward value determined above are weighted and summed using the target weight coefficients of each reward to obtain the basic reward value through Formula 9.

[0250] The reward function in this example no longer relies on a single metric, but instead adaptively allocates the weights of each reward item based on the environmental state, enabling the model to form a stable optimization objective even under complex conditions, thereby improving the convergence consistency of policy training and the overall balance of control results.

[0251] S902. Keep the environmental status data unchanged and randomly adjust the operating parameters to obtain the adjusted operating parameters.

[0252] For example, if the environmental state data remains unchanged, that is, if the running state is fixed, the running parameters are fine-tuned, which means adjusting the control actions of the expert control strategy to obtain the adjusted running parameters, which in turn means obtaining the adjusted control actions of the expert control strategy.

[0253] S903. Determine the actual reward value between the environmental state data and the adjusted operating parameters based on the preset reward function.

[0254] Example, combination Figure 10 The Q2 value obtained from the processing on the expert 2 side of the Q network (Critic) is the second stability reward value obtained from the control action of the expert control strategy corresponding to the operating state corresponding to the environmental state data and the adjusted operating parameters.

[0255] Optional, Figure 13 This is a schematic flowchart of a drying control method disclosed herein. Figure 10 ,like Figure 13 As shown, S903 includes:

[0256] S1301. Determine the second energy efficiency reward value, second time reward value, second dryness reward value, and second stability reward value corresponding to the environmental state data and adjusted operating parameters based on the reward function.

[0257] For example, based on the above formulas 5, 6, 7 and 8, the environmental state data and the adjusted operating parameters are processed respectively to obtain the second energy efficiency bonus value, the second time bonus value, the second dryness bonus value and the second stability bonus value.

[0258] S1302. Determine the target weight coefficients corresponding to energy efficiency reward, time reward, dryness reward and stability reward respectively based on environmental status data. Among them, the second energy efficiency reward value is the second value of energy efficiency reward, the second time reward value is the second value of time reward, the second dryness reward value is the second value of dryness reward, and the second stability reward value is the second value of stability reward.

[0259] For example, the second energy efficiency bonus value, the second time bonus value, the second dryness bonus value, and the second stability bonus value are the second values ​​for the energy efficiency bonus item, the time bonus item, the dryness bonus item, and the stability bonus item, respectively. The target weight coefficients for each bonus item are determined according to Table 2.

[0260] S1303. Based on the target weight coefficient, perform a weighted summation operation on the second energy efficiency reward value, the second time reward value, the second dryness reward value, and the second stability reward value to obtain the actual reward value.

[0261] The actual reward value obtained in this example no longer depends on a single indicator, but combines environmental conditions to uniformly quantify multiple control objectives, enabling the strategy model to form more suitable optimization feedback in different drying stages, thereby improving the discrimination accuracy of the adjusted operating parameters and the consistency of training convergence.

[0262] For example, this example uses Formula 9 to weight and sum the second energy efficiency reward value, the second time reward value, the second dryness reward value, and the second stability reward value determined above, using the target weight coefficients of each reward to obtain the actual reward value.

[0263] S904. Based on the base reward value and the actual reward value, optimize the pre-trained policy model until the policy network of the pre-trained policy model reaches the second preset convergence condition, so as to obtain the trained policy model.

[0264] For example, the TD3 optimization algorithm is a further development of Deep Deterministic Policy Gradient (DDPG), aiming to address the limitations of DDPG in continuous action spaces and its susceptibility to Q-value overestimation. Q-value refers to the expected sum of all rewards. Improvements were also made to address Q-value prediction bias and learning instability. TD3 introduces three key improvements: a dual-action network, delayed policy network updates, and target policy smoothing to achieve these goals. The TD3 algorithm uses six networks in its update process, including an Actor network (policy network) with parameter φ and its corresponding target Actor network, and two other networks with parameter φ. and The Critic network (double Q network) and its corresponding target Critic network are described. The goal of TD3 is to obtain the target policy parameterized as φ. The expected return of the reward is maximized using the following formula 10:

[0265] Formula 10:

[0266] Where T is the total step size of the Markov decision process. Indicates the discount factor. This represents the current policy function (which is being updated and is not optimal). This represents the state transition probability under this strategy.

[0267] In a continuous action space, the strategy The deterministic policy gradient (DPG) algorithm, as shown in Equation 11, can be used to directly update the policy:

[0268] Formula 11:

[0269] in, It is the gradient of the reinforcement learning objective function corresponding to equation (15). It is the gradient of the policy function. It is the gradient of the Critic function (or Q function).

[0270] In practice, The equation can be updated based on the Bellman equation shown in Equation 12:

[0271] Formula 12:

[0272] Considering that in settings like DDPG, directly using Equation 11 to update the policy network parameters φ might lead to an overestimation of the Q-value, TD3 introduces a target network mechanism and employs a truncated double Q-learning technique to mitigate this overestimation. Specifically, TD3 uses two target Critic networks and takes the minimum of their outputs as the target Q-value. Therefore, the temporal difference error of TD3 is rewritten using Equation 13 from the example:

[0273] Formula 13:

[0274] At this point, the parameterization is as follows: and The optimization objective of the Critic network can be calculated using Formula 14:

[0275] Formula 14:

[0276] Objective functions Equations 10 and 14 are used to evaluate and update the policy network and Critic network, respectively. However, the above update method still suffers from Q-function overfitting. To address this issue, TD3 introduces an objective policy smoothing technique, adding Gaussian noise to the actions and pruning them before calculating the gradient. Therefore, the action generation process is restated as shown in Equations 15 and 16 below:

[0277] Formula 15:

[0278] Formula 16:

[0279] in, It is the strategy of the target Actor network; It is Gaussian noise; and are different standard deviations; c and c' are different noise boundaries; clip(·) is a cutoff function used to ensure that the noisy action still conforms to the action space specification. In addition, the small variance noise introduced by equations (20) and (21) can also enhance the agent's exploration ability.

[0280] Subsequently, the derivatives of the objective functions (10) and (14) can be calculated using the DPG algorithm and batch sampling data, as shown in Equations 17 and 18 below:

[0281] Formula 17:

[0282] Formula 18:

[0283] To prevent degradation loops during training, TD3 introduces a delayed policy network update mechanism on top of the standard Actor-Critic architecture, reducing the update frequency of the Actor network. This mechanism reduces the computational variance of the Q-function, helping the agent learn better policies. The update frequency in TD3 is controlled by the delay parameter d, which can be adjusted according to specific scenarios. Furthermore, TD3 introduces a target network mechanism to stabilize the training process. The target network follows a soft update rule every d time steps, gradually updating its parameters, thereby further improving training stability, as shown in Equation 19 below:

[0284] Formula 19:

[0285] Where λ is the soft update step size of the target network.

[0286] According to Formula 19, once the parameter update results converge, the second preset convergence condition is met. At this point, the model training result is determined as the trained policy model.

[0287] The strategy model trained based on this example can more accurately reflect the influence of operating parameters on the reward function, reduce the control deviation caused by adjusting a single parameter, make it more adaptable to different loads and environmental fluctuations during subsequent online control, and improve the synergy and stability of the drying process.

[0288] S405. Use the trained policy model as the preset policy model.

[0289] For example, once the trained strategy model is formed, it is used as a preset strategy model. When acquiring real-time environmental state data, it is directly used for forward calculation and outputs the target operating parameters. Because this preset strategy model has been adapted by combining historical samples and reward feedback during the training phase, it can respond to state changes more closely to the target control requirements when facing different clothing loads and environmental fluctuations. This also makes the target compressor frequency, target fan speed, and target drum speed generated based on real-time environmental state data have better coordination and consistency.

[0290] Optionally, it also includes: during the optimization training of the pre-trained policy model, if the environmental state data meets the preset termination condition, then the optimization training of the pre-trained policy model is terminated.

[0291] For example, when optimizing a pre-trained policy model, the termination condition for environmental state data can be determined based on the overall state. By setting a termination condition during the optimization training process, the model training can end in a timely manner when the preset convergence or task completion characteristics are met. This ensures that the trained policy model is consistent with the actual end state of the drying process, reduces the disturbance of model parameters by invalid iterations, and makes subsequent control outputs more closely match the actual drying conditions.

[0292] Optionally, the preset termination conditions include: the relative humidity of the humidity bar being less than the preset relative humidity threshold for a duration exceeding a first preset duration, or the drying time exceeding a second preset duration.

[0293] For example, in reinforcement learning, the IsDone flag is typically used as an explicit termination criterion. During optimization training or drying control, the relative humidity of the humidity bar is continuously read, and its relationship with a preset relative humidity threshold is determined. When the relative humidity of the humidity bar is detected to be continuously lower than the preset relative humidity threshold, the duration of this low humidity is simultaneously accumulated; when the accumulated duration exceeds a first preset duration, the drying termination condition is considered triggered. For example, this can be determined using the following formula 20.

[0294] Formula 20:

[0295] That is, when the relative humidity RH on the humidity bar is below the preset relative humidity threshold RH for L consecutive minutes. t When the condition is met, the termination condition is determined to be met.

[0296] At the same time, considering the possibility of undried products during the process, the drying time is also timed. When the cumulative drying time exceeds the second preset time, the termination condition is also determined to be met. For example, this can be determined using the following formula 21.

[0297] Formula 21:

[0298] When the drying time t reaches the preset second preset time t max At that time, the termination condition is also determined to be met.

[0299] Based on this example, the termination timing can be constrained by both the humidity change result and the time limit, so that the termination judgment can reflect the actual degree of drying and limit the longest operating boundary, thereby making the termination judgment of the drying process more stable and reducing premature termination or invalid extension caused by humidity fluctuations.

[0300] Optional, Figure 14 This is a schematic flowchart of a drying control method disclosed herein. Figure 10 First, such as Figure 14 As shown, it also includes:

[0301] S1401. If the real-time compressor temperature exceeds the preset compressor safety temperature, a compressor protection strategy is generated.

[0302] For example, the cloud server continuously receives real-time compressor temperatures and compares them with a preset safe compressor temperature, which can be determined based on actual conditions. When the compressor temperature exceeds the preset safe compressor temperature, the generated compressor protection strategy includes reducing the compressor frequency, short-term shutdown, or limiting the compressor duty cycle.

[0303] S1402. If the real-time compressor frequency reaches the preset resonance frequency, a compressor frequency resonance protection strategy is generated.

[0304] For example, the cloud server continuously receives the real-time compressor frequency and compares it with a preset resonant frequency, which can be determined based on actual conditions. When the real-time compressor frequency reaches the preset resonant frequency, the generated compressor frequency resonance protection strategy includes switching the compressor frequency to a target frequency point within the avoidance range.

[0305] S1403. If the real-time condenser temperature exceeds the preset condenser safety temperature, a condenser protection strategy is generated.

[0306] For example, the end server continuously receives the real-time condenser temperature and compares it with a preset safe condenser temperature, which can be determined based on actual conditions. When the real-time condenser temperature exceeds the preset safe condenser temperature, the generated condenser protection strategy includes reducing the compressor output and simultaneously increasing the fan heat exchange capacity.

[0307] S1404. Control the dryer to implement compressor protection strategy and / or compressor frequency resonance protection strategy and / or condenser protection strategy.

[0308] For example, the three protection strategies can be executed individually, in parallel, or in combination, depending on their trigger status. The protection strategy in this example is triggered by real-time detection values, thus enabling timely intervention before the equipment enters overheating, resonance, or abnormal heat load states, maintaining a stable operational relationship between the compressor, condenser, and overall system control. Using this approach, abnormal operating conditions can be responded to promptly without altering the main drying control framework, reducing damage to critical components while maintaining the continuity and consistency of the drying process and control.

[0309] Figure 15 This is a schematic diagram of the structure of a drying control device disclosed herein, such as... Figure 15 As shown, it includes:

[0310] The acquisition module 151 is used to acquire real-time environmental status data of the dryer during the drying process;

[0311] The processing module 152 is used to determine the target operating parameters based on real-time environmental status data, wherein the target operating parameters include the target compressor frequency, the target fan speed, and the target barrel speed;

[0312] The control module 153 is used to control the dryer to perform drying operations according to the target operating parameters.

[0313] Optional real-time environmental status data include: real-time exhaust temperature, real-time steam temperature, real-time chamber temperature, real-time chamber relative humidity, real-time humidity bar relative humidity, real-time drying time, real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature.

[0314] Optionally, the processing module 152 is specifically used to determine the real-time operating status of the dryer based on real-time environmental status data;

[0315] The processing module 152 is further used to input the real-time running status into the preset strategy model, perform forward calculation on the real-time running status based on the preset strategy model, and obtain the target running parameters corresponding to the real-time running status.

[0316] Optionally, the processing module 152 is also used to determine the training data;

[0317] The processing module 152 is also used to initialize the policy network using a preset algorithm as an initial policy model;

[0318] The processing module 152 is also used to pre-train the initial policy model based on the training data to obtain the pre-trained policy model.

[0319] The processing module 152 is also used to optimize the pre-trained policy model through a preset reward function to obtain a trained policy model.

[0320] The processing module 152 is also used to use the trained policy model as a preset policy model.

[0321] Optionally, the processing module 152 is further used to divide the drying process into multiple drying stages based on the relative humidity inside the dryer.

[0322] The processing module 152 is further used to determine the preset operating parameters corresponding to multiple drying stages, including preset compressor frequency, preset fan speed and preset drum speed.

[0323] The processing module 152 is specifically used to control the dryer to perform drying operations according to the corresponding preset operating parameters in multiple drying stages, and to collect the environmental status data and operating parameters of the dryer at preset time intervals during the drying process. The operating parameters include compressor frequency, fan speed and drum speed.

[0324] The processing module 152 is also used to construct data pairs from environmental state data and operating parameters collected at the same time, and to construct training data from all data pairs.

[0325] Optionally, the processing module 152 is further used to determine the drying stage where the relative humidity inside the chamber is greater than the first preset relative humidity as the preheating period;

[0326] The processing module 152 is further used to determine the drying stage in which the relative humidity inside the chamber is less than or equal to the first preset relative humidity and the relative humidity inside the chamber is greater than or equal to the second preset relative humidity as a constant speed drying period.

[0327] The processing module 152 is further used to determine the drying stage in which the relative humidity inside the chamber is less than the second preset relative humidity and the relative humidity inside the chamber is greater than the third preset relative humidity as the falling rate drying period.

[0328] The processing module 152 is also specifically used to determine the drying stage where the relative humidity inside the chamber is less than or equal to the third preset relative humidity as the tempering period.

[0329] Optionally, the processing module 152 is further used to input the training data into the initial policy model;

[0330] The processing module 152 is further configured to deploy a preset behavior cloning algorithm into the initial policy model, so that the initial policy model can learn by imitation of the training data based on the behavior cloning algorithm.

[0331] The processing module 152 is further used to determine the difference between the network weight parameters in the initial policy model and the training data according to the preset loss function;

[0332] The processing module 152 is further configured to determine the gradient corresponding to the difference according to the preset backpropagation algorithm, update the network weight parameters in the initial policy model according to the preset gradient descent algorithm, and terminate the pre-training of the initial policy model to obtain the pre-trained policy model when the network weight parameters reach the first preset convergence condition.

[0333] Optionally, the processing module 152 is further used to determine the basic reward value between the environmental state data and the operating parameters according to the preset reward function;

[0334] The processing module 152 is also used to keep the environmental state data unchanged and randomly adjust the operating parameters to obtain the adjusted operating parameters;

[0335] The processing module 152 is further used to determine the actual reward value between the environmental state data and the adjusted operating parameters according to the preset reward function;

[0336] The processing module 152 is further used to optimize the training of the pre-trained policy model based on the basic reward value and the actual reward value until the policy network of the pre-trained policy model reaches the second preset convergence condition, so as to obtain the trained policy model.

[0337] Optionally, the processing module 152 is further configured to determine the first energy efficiency reward value, the first time reward value, the first dryness reward value, and the first stability reward value corresponding to the environmental state data and operating parameters based on the reward function;

[0338] The processing module 152 is further used to determine the target weight coefficients corresponding to the energy efficiency reward, time reward, dryness reward and stability reward respectively based on the environmental status data. The first energy efficiency reward value is the first value of the energy efficiency reward, the first time reward value is the first value of the time reward, the first dryness reward value is the first value of the dryness reward and the first stability reward value is the first value of the stability reward.

[0339] The processing module 152 is further used to perform a weighted summation operation on the first energy efficiency reward value, the first time reward value, the first dryness reward value and the first stability reward value based on the target weight coefficient, so as to obtain the basic reward value.

[0340] Optionally, the processing module 152 is further used to determine the second energy efficiency reward value, the second time reward value, the second dryness reward value, and the second stability reward value corresponding to the environmental state data and the adjusted operating parameters based on the reward function;

[0341] The processing module 152 is further used to determine the target weight coefficients corresponding to the energy efficiency reward, time reward, dryness reward and stability reward respectively based on the environmental status data. The second energy efficiency reward value is the second value of the energy efficiency reward, the second time reward value is the second value of the time reward, the second dryness reward value is the second value of the dryness reward and the second stability reward value is the second value of the stability reward.

[0342] The processing module 152 is further used to perform a weighted summation operation on the second energy efficiency reward value, the second time reward value, the second dryness reward value and the second stability reward value based on the target weight coefficient, so as to obtain the actual reward value.

[0343] Optionally, the processing module 152 is further used to determine the target drying stage based on the relative humidity inside the chamber;

[0344] The processing module 152 is further configured to determine the target weight coefficients for energy efficiency rewards, time rewards, dryness rewards, and stability rewards corresponding to the target drying stage based on a preset weight coefficient table. The weight coefficient table includes the weight coefficients for energy efficiency rewards, time rewards, dryness rewards, and stability rewards corresponding to multiple drying stages.

[0345] Optionally, the processing module 152 is also used to terminate the optimization training of the pre-trained policy model if the environmental state data meets the preset termination condition during the optimization training of the pre-trained policy model.

[0346] Optionally, the preset termination conditions include: the relative humidity of the humidity bar being less than the preset relative humidity threshold for a duration exceeding a first preset duration, or the drying time exceeding a second preset duration.

[0347] Optionally, the processing module 152 is also used to generate a compressor protection strategy if the real-time compressor temperature exceeds the preset compressor safe temperature.

[0348] The processing module 152 is also used to generate a compressor frequency resonance protection strategy if the real-time compressor frequency reaches the preset resonance frequency.

[0349] The processing module 152 is also used to generate a condenser protection strategy if the real-time condenser temperature exceeds the preset condenser safety temperature.

[0350] The processing module 152 is also used to control the dryer to perform compressor protection strategies and / or compressor frequency resonance protection strategies and / or condenser protection strategies.

[0351] The drying control device provided in this embodiment can execute the drying control method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0352] Figure 16 This is a schematic diagram of the drying control system, such as... Figure 16 As shown, it includes: a clothes dryer and a cloud server, wherein the clothes dryer includes a sensor mechanism and a control mechanism, and the cloud server includes a decision module;

[0353] The sensor mechanism is used to collect real-time environmental status data of the dryer during the drying process and send the real-time environmental status data to the decision module; the decision module is used to determine the target operating parameters based on the real-time environmental status data and send the target operating parameters to the control mechanism, wherein the target operating parameters include the target compressor frequency, the target fan speed and the target drum speed; the control mechanism is used to control the dryer to perform drying operations according to the target operating parameters.

[0354] For example, the sensor mechanism includes a temperature and humidity sensor and / or a temperature sensor for collecting real-time environmental status data, which can be defined as raw signals. Before processing, the decision module can first perform data processing on the real-time environmental status data through the data processing module, such as feature extraction and drying stage judgment, to determine the real-time operating status of the dryer. Then, the decision module obtains the target operating parameters through the TD3 network. The target operating parameters are sent to the control mechanism of the dryer, which includes a compressor driver, a fan driver, and a drum electrode driver. The compressor driver can control the compressor operation according to the target compressor frequency in the target operating parameters, the fan driver can control the fan operation according to the target fan speed in the target operating parameters, and the drum electrode driver can control the drum operation according to the target drum speed in the target operating parameters.

[0355] Simultaneously, the timer in the sensor mechanism can accumulate the duration of low humidity when the relative humidity detected by the temperature and humidity sensor and / or the temperature sensor is continuously lower than a preset relative humidity threshold. When the accumulated duration exceeds a first preset duration, the termination condition is determined to be met. The drying time is also timed, and when the accumulated drying time exceeds a second preset duration, the termination condition is also determined to be met. When the termination condition is met, the dryer is controlled to end the drying process.

[0356] Optionally, the cloud server also includes a security monitoring module; real-time environmental status data includes: real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature; the security monitoring module is used to generate a compressor protection strategy when the real-time compressor temperature exceeds a preset compressor safety temperature; generate a compressor frequency resonance protection strategy when the real-time compressor frequency reaches a preset resonance frequency; and generate a condenser protection strategy when the real-time condenser temperature exceeds a preset condenser safety temperature; the security monitoring module is also used to send the compressor protection strategy and / or the compressor frequency resonance protection strategy and / or the condenser protection strategy to the control mechanism, so that the control mechanism executes the compressor protection strategy and / or the compressor frequency resonance protection strategy and / or the condenser protection strategy.

[0357] For example, the safety monitoring module detects the safety of real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature. When the real-time compressor temperature exceeds the preset safe compressor temperature, the generated compressor protection strategy includes reducing the compressor frequency, short-term shutdown, or limiting the compressor duty cycle. When the real-time compressor frequency reaches the preset resonant frequency, the generated compressor frequency resonance protection strategy includes switching the compressor frequency to the target frequency point within the avoidance range. When the real-time condenser temperature exceeds the preset safe condenser temperature, the generated condenser protection strategy includes reducing the compressor output and simultaneously increasing the fan heat exchange capacity. The safety monitoring module sends the compressor protection strategy and / or the compressor frequency resonance protection strategy and / or the condenser protection strategy to the control mechanism, so that the compressor driver and / or fan driver and / or tank electrode driver in the control mechanism control the operation of the compressor and / or fan and / or tank. The triggering state of the three types of protection strategies can be selected to be executed individually, in parallel, or superimposed. When the real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature are normal, a safety signal can be sent to the control mechanism so that the compressor driver and / or fan driver and / or tank electrode driver in the control mechanism can control the compressor and / or fan and / or tank to operate normally according to the target operating parameters.

[0358] The drying control system provided in this embodiment can execute the drying control method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0359] Figure 17 This is a schematic diagram of the structure of the electronic device provided in this application. For example... Figure 17 As shown, the air conditioner 50 provided in this embodiment includes at least one processor 501 and a memory 502. Optionally, the electronic device 50 further includes a communication component 503. The processor 501, memory 502, and communication component 503 are connected via a bus.

[0360] In the specific implementation process, at least one processor 501 executes computer execution instructions stored in memory 502, causing at least one processor 501 to execute the above-described method embodiment for adjusting the operating frequency of the air conditioner compressor.

[0361] The specific implementation process of processor 501 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0362] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0363] The memory may include random access memory (RAM) and non-volatile memory (NVM), such as at least one disk storage device.

[0364] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0365] Embodiments of this application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above-described front-end resource allocation method embodiments when running.

[0366] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0367] Embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above-described front-end resource allocation method embodiments.

[0368] Embodiments of this application also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in any of the above-described front-end resource allocation method embodiments.

[0369] It should be noted that the division of units is merely a logical functional division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0370] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0371] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0372] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.

[0373] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0374] Those skilled in the art will also understand that the various illustrative logic blocks and steps listed in the embodiments of this disclosure can be implemented by electronic hardware, computer software, or a combination of both. Whether such functionality is implemented in hardware or software depends on the specific application and the overall system design requirements. Those skilled in the art can implement the functionality using various methods for each specific application, but such implementation should not be construed as exceeding the scope of protection of the embodiments of this disclosure.

[0375] In the above detailed description, reference has been made to the accompanying drawings, which illustrate specific aspects of this disclosure by way of illustration. In this regard, terms indicating direction or positional relationship, such as “center,” “longitudinal,” “lateral,” “length,” “width,” “thickness,” “upper,” “lower,” “front,” “rear,” “left,” “right,” “vertical,” “horizontal,” “top,” “bottom,” “inner,” “outer,” “clockwise,” “counterclockwise,” “axial,” “radial,” and “circumferential,” are used with reference to the orientation of the described figures. Since components of the described device can be positioned in multiple different orientations, directional terms are used for illustrative purposes and not for limitation. It should be understood that other aspects can be utilized and structural or logical changes can be made without departing from the concept of this disclosure. Therefore, the following detailed description should not be considered limiting.

[0376] It should be understood that, unless otherwise specifically indicated, features of various embodiments of this disclosure described herein can be combined with each other. As used herein, the term "and / or" includes any one of the relevant listed items and any combination of any two or more.

[0377] It should be understood that, unless otherwise expressly specified and limited, the terms "joining," "attaching," "installing," "connecting," "linking," "fixing," etc., used in the embodiments of this disclosure 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 connection that allows communication between them; 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, unless otherwise expressly limited. Those skilled in the art can understand the specific meaning of the above terms herein based on the specific circumstances.

[0378] Furthermore, the term "above" as used herein with respect to components, elements, or material layers formed or located "above" a surface may be used to indicate that the component, element, or material layer is "indirectly" positioned (e.g., placed, formed, deposited, etc.) on the surface such that one or more additional components, elements, or layers are arranged between the surface and the component, element, or material layer. However, the term "above" as used with respect to components, elements, or material layers formed or located "above" a surface may also optionally have a specific meaning: that the component, element, or material layer is "directly" positioned (e.g., placed, formed, deposited, etc.) on the surface, for example, in direct contact with the surface.

[0379] Although terms such as “first,” “second,” and “third” may be used herein to describe various components, parts, regions, layers, or sections, these components, parts, regions, layers, or sections are not limited to these terms. Rather, these terms are used only to distinguish one component, part, region, layer, or section from another. Therefore, without departing from the teachings of the examples described herein, the first component, part, region, layer, or section mentioned in the examples may also be referred to as the second component, part, region, layer, or section. Furthermore, the terms “first” and “second” are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as “first” or “second” may explicitly or implicitly include at least one of that feature. In the description herein, “a plurality” means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0380] It should be understood that spatial relative terms, such as “above,” “upper,” “below,” and “lower,” are used herein to describe the relationship between one element and another shown in the figures. In addition to the orientation depicted in the figures, these spatial relative terms are also intended to encompass different orientations of the device in use or operation. For example, if the device in the figures is flipped, an element described as “above” or “upper” relative to another element would be “below” or “lower” relative to that other element. Thus, depending on the spatial orientation of the device, the term “above” encompasses both above and below orientations. Devices may have other orientations (e.g., rotated 90 degrees or in other orientations), and the spatial relative terms used herein should be interpreted accordingly.

[0381] Furthermore, the term “exemplary” is used herein to mean serving as an example, instance, or illustration. Any aspect or design described herein as “exemplary” is not necessarily to be construed as advantageous compared to other aspects or designs. Rather, the use of the term “exemplary” is intended to present the concept in a concrete manner. As used herein, the term “or” is intended to mean an inclusive “or” rather than an exclusive “or.” That is, unless otherwise specified or clear from the context, “X applies A or B” is intended to mean any of the natural inclusive arrangements. That is, “X applies A or B” satisfies any of the foregoing instances if X applies A; X applies B; or both X applies A and B. Additionally, unless otherwise specified or clear from the context to refer to the singular form, the articles “a” and “an” as used in this disclosure and the appended claims are generally understood to mean “one or more.”

[0382] Similarly, although this disclosure has been shown and described with respect to one or more implementations, equivalent variations and modifications will occur to those skilled in the art upon reading and understanding this specification and the accompanying drawings. This disclosure includes all such modifications and variations and is limited only by the scope of the claims. In particular, with respect to the various functions performed by the components described above (e.g., elements, resources, etc.), unless otherwise indicated, the terminology used to describe such components is intended to correspond to any component (functionally equivalent) that performs the specific function of the described component, even if structurally not equivalent to the disclosed structure. Furthermore, although specific features of this disclosure may have been disclosed with respect to only one of several implementations, such features may be combined with one or more other features of other implementations, as may be desired and advantageous to any given or particular application. Moreover, with regard to the terms “comprising,” “owning,” “having,” “having,” or variations thereof as used in the detailed description or claims, such terms are intended to be inclusive in a manner similar to the term “including.”

[0383] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0384] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A drying control method, characterized in that, include: Acquire real-time environmental status data of the dryer during the drying process; The target operating parameters are determined based on the real-time environmental status data, wherein the target operating parameters include the target compressor frequency, the target fan speed, and the target barrel speed; The dryer is controlled to perform drying operations according to the target operating parameters.

2. The method according to claim 1, characterized in that, The step of determining the target operating parameters based on the real-time environmental status data includes: The real-time operating status of the dryer is determined based on the real-time environmental status data, which includes: real-time exhaust temperature, real-time steam temperature, real-time chamber temperature, real-time chamber relative humidity, real-time relative humidity of the humidity bar, real-time drying time, real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature. The real-time operating status is input into a preset strategy model, and forward calculation is performed on the real-time operating status based on the preset strategy model to obtain the target operating parameters corresponding to the real-time operating status.

3. The method according to claim 2, characterized in that, Before acquiring the real-time environmental status data of the dryer during the drying process, the method further includes: Determine the training data; The policy network using the preset algorithm is initialized as the initial policy model; The initial policy model is pre-trained based on the training data to obtain a pre-trained policy model. The pre-trained policy model is optimized and trained using a preset reward function to obtain a well-trained policy model. The trained policy model is used as the preset policy model.

4. The method according to claim 3, characterized in that, The determination of training data includes: The drying process is divided into multiple drying stages based on the relative humidity inside the dryer. Determine preset operating parameters corresponding to multiple drying stages, wherein the preset operating parameters include preset compressor frequency, preset fan speed and preset drum speed; The dryer is controlled to perform drying operations according to the corresponding preset operating parameters in multiple drying stages, and the environmental status data and operating parameters of the dryer are collected at preset time intervals during the drying process. The operating parameters include compressor frequency, fan speed and drum speed. The environmental state data and the operating parameters collected at the same time are constructed into data pairs, and all the data pairs are constructed into the training data.

5. The method according to claim 4, characterized in that, The drying process is divided into multiple drying stages based on the relative humidity inside the dryer, including: The drying stage in which the relative humidity inside the cabin is greater than the first preset relative humidity is defined as the preheating period. The drying stage in which the relative humidity inside the chamber is less than or equal to the first preset relative humidity and the relative humidity inside the chamber is greater than or equal to the second preset relative humidity is defined as the constant-rate drying period. The drying stage in which the relative humidity inside the chamber is less than the second preset relative humidity and greater than the third preset relative humidity is defined as the slow-down drying period. The drying stage in which the relative humidity inside the chamber is less than or equal to the third preset relative humidity is defined as the warming period.

6. The method according to claim 3, characterized in that, The step of pre-training the initial policy model based on the training data to obtain a pre-trained policy model includes: The training data is input into the initial policy model; A preset behavior cloning algorithm is deployed into the initial policy model, so that the initial policy model can learn by imitation from the training data based on the behavior cloning algorithm. The difference between the network weight parameters in the initial policy model and the training data is determined according to the preset loss function; The gradient corresponding to the difference is determined according to the preset backpropagation algorithm, and the network weight parameters in the initial policy model are updated according to the preset gradient descent algorithm until the network weight parameters reach the first preset convergence condition. Then the pre-training of the initial policy model ends to obtain the pre-trained policy model.

7. The method according to claim 4, characterized in that, The step of optimizing and training the pre-trained policy model using a preset reward function to obtain a trained policy model includes: The basic reward value between the environmental state data and the operating parameters is determined according to the preset reward function; Keeping the environmental state data unchanged, the operating parameters are randomly adjusted to obtain the adjusted operating parameters; The actual reward value between the environmental state data and the adjusted operating parameters is determined according to the preset reward function; Based on the base reward value and the actual reward value, the pre-trained policy model is optimized and trained until the policy network of the pre-trained policy model reaches the second preset convergence condition, so as to obtain the trained policy model.

8. The method according to claim 7, characterized in that, The step of determining the basic reward value between the environmental state data and the operating parameters according to the preset reward function includes: The first energy efficiency reward value, the first time reward value, the first dryness reward value, and the first stability reward value corresponding to the environmental state data and the operating parameters are determined according to the reward function. Based on the environmental state data, target weight coefficients are determined for energy efficiency reward, time reward, dryness reward and stability reward respectively, wherein the first energy efficiency reward value is the first value of the energy efficiency reward, the first time reward value is the first value of the time reward, the first dryness reward value is the first value of the dryness reward and the first stability reward value is the first value of the stability reward. Based on the target weight coefficient, the first energy efficiency reward value, the first time reward value, the first dryness reward value, and the first stability reward value are weighted and summed to obtain the basic reward value.

9. The method according to claim 7, characterized in that, Determining the actual reward value between the environmental state data and the adjusted operating parameters according to the preset reward function includes: The second energy efficiency reward value, the second time reward value, the second dryness reward value, and the second stability reward value corresponding to the environmental state data and the adjusted operating parameters are determined according to the reward function. Based on the environmental state data, target weight coefficients are determined for the energy efficiency reward, the time reward, the dryness reward, and the stability reward, respectively, wherein the second energy efficiency reward is a second value of the energy efficiency reward, the second time reward is a second value of the time reward, the second dryness reward is a second value of the dryness reward, and the second stability reward is a second value of the stability reward; Based on the target weighting coefficient, the second energy efficiency reward value, the second time reward value, the second dryness reward value, and the second stability reward value are weighted and summed to obtain the actual reward value.

10. The method according to claim 8 or 9, characterized in that, The step of determining the target weight coefficients corresponding to energy efficiency bonus, time bonus, dryness bonus, and stability bonus based on the environmental state data includes: The target drying stage is determined based on the relative humidity inside the chamber. Based on a preset weighting coefficient table, the target weighting coefficients for energy efficiency rewards, time rewards, dryness rewards, and stability rewards corresponding to the target drying stage are determined. The weighting coefficient table includes multiple weighting coefficients for energy efficiency rewards, time rewards, dryness rewards, and stability rewards corresponding to the drying stage.

11. The method according to claim 4, characterized in that, Also includes: During the optimization training of the pre-trained policy model, if the environmental state data meets the preset termination condition, the optimization training of the pre-trained policy model ends.

12. The method according to claim 11, characterized in that, The preset termination conditions include: the relative humidity of the humidity bar being less than the preset relative humidity threshold for a duration exceeding a first preset duration, or the drying time exceeding a second preset duration.

13. The method according to claim 2, characterized in that, The method further includes: If the real-time compressor temperature exceeds the preset compressor safety temperature, a compressor protection strategy is generated. If the real-time compressor frequency reaches the preset resonance frequency, a compressor frequency resonance protection strategy is generated. If the real-time condenser temperature exceeds the preset condenser safety temperature, a condenser protection strategy is generated. Control the dryer to execute the compressor protection strategy and / or the compressor frequency resonance protection strategy and / or the condenser protection strategy.

14. A drying control device, characterized in that, include: The acquisition module is used to acquire real-time environmental status data of the dryer during the drying process; The processing module is used to determine target operating parameters based on the real-time environmental status data, wherein the target operating parameters include target compressor frequency, target fan speed and target barrel speed; The control module is used to control the dryer to perform drying operations according to the target operating parameters.

15. A drying control system, characterized in that, include: A clothes dryer and a cloud server, wherein the clothes dryer includes a sensor mechanism and a control mechanism, and the cloud server includes a decision module; The sensor mechanism is used to collect real-time environmental status data of the dryer during the drying process and send the real-time environmental status data to the decision module; The decision module is used to determine the target operating parameters based on the real-time environmental status data and send the target operating parameters to the control mechanism. The target operating parameters include the target compressor frequency, the target fan speed, and the target barrel speed. The control mechanism is used to control the dryer to perform drying operations according to the target operating parameters.

16. The system according to claim 15, characterized in that, The cloud server also includes a security monitoring module; the real-time environmental status data includes: real-time compressor temperature, real-time compressor frequency, and real-time condenser temperature. The safety monitoring module is used to generate a compressor protection strategy when the real-time compressor temperature exceeds a preset compressor safety temperature; to generate a compressor frequency resonance protection strategy when the real-time compressor frequency reaches a preset resonance frequency; and to generate a condenser protection strategy when the real-time condenser temperature exceeds a preset condenser safety temperature. The safety monitoring module is also used to send the compressor protection strategy and / or the compressor frequency resonance protection strategy and / or the condenser protection strategy to the control mechanism, so that the control mechanism executes the compressor protection strategy and / or the compressor frequency resonance protection strategy and / or the condenser protection strategy.

Images