Control model training method, loose moisture recovery machine control method and device

Abstract

The present invention discloses a control model training method, a control method and a device for a loose moisture conditioning machine. The control model training method includes: training the control model based on the first sample state variable at each moment and the historical exogenous variable at each moment to obtain a target control model; wherein the first sample state variable at each moment includes the historical state variable at the starting moment and the predicted state variable at other moments, and the predicted state variable at other moments is obtained during the training process of the control model, based on the predicted control variables at each moment predicted by the control model, and is predicted based on a pre-trained simulation model of a loose moisture conditioning machine. The present invention fully reflects the coupling relationship between multiple control mechanisms and the quality of the input material and the environmental state through a pre-trained simulation model of a loose moisture conditioning machine, and obtains a control model of the loose moisture conditioning machine under the coupling relationship, so that the moisture content and temperature of the output material of the loose moisture conditioning machine can be stabilized at the set values.

Description

Technical Field

[0001] The invention relates to the technical field of tobacco processing equipment control, and in particular to a control model training method, and a control method and device for a loosening and rehumidifying machine. Background Art

[0002] After the raw tobacco leaves are harvested and dried in the production area, and then packaged and bundled and transported to the production workshop, in order to facilitate the subsequent shredding process to be completed stably, the dried and brittle tobacco leaves need to be re-moistened to improve their flexibility and avoid the tobacco leaves from breaking during the subsequent shredding process, which leads to problems such as the tobacco shreds being too short and too many broken pieces after shredding. The loose rehumidification machine sprays high-temperature and high-humidity water mist on the tobacco leaves and constantly turns the tobacco leaves in the drum to achieve tobacco leaf rehumidification and increase the moisture content of the tobacco leaves. The control target of the loose rehumidification machine is the moisture content of the discharge (18% to 22%). There are four control mechanisms, namely the drum circulation fan, the heat exchanger steam valve, the dual-medium nozzle steam valve, and the dual-medium nozzle pure water valve.

[0003] At present, the mainstream control methods of loose moisture conditioning machines are (1) feedback control based on the moisture content of the discharge material; (2) subjective control based on expert experience. However, these control methods ignore the impact of input material quality and environmental factors on control, and do not fully consider the coupling relationship between the four control mechanisms. Instead, they view the relationship between each control mechanism and the moisture content of the discharge material in isolation and make adjustments. This can easily lead to insufficient control of the moisture content of the discharge material, resulting in excessive or low moisture content. Summary of the invention

[0004] The present invention provides a control model training method, a loosening and rehumidifying machine control method and device to solve the above technical problems.

[0005] According to one aspect of the present invention, a control model training method is provided, wherein the control model is applied to a loose conditioning machine, comprising:

[0006] Taking any historical moment as the starting moment, obtaining the historical variable data of the loose moisture recovery machine, the historical variable data including the historical state variables and historical exogenous variables at each moment starting from the starting moment;

[0007] The control model is trained based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain a target control model; wherein the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted by the control model during the training process of the control model, based on the predicted control variables at each moment predicted by the control model, and are predicted based on the pre-trained simulation model of the loose rehumidification machine.

[0008] According to another aspect of the present invention, a control method for a loosening and moistening machine is provided, characterized in that a controller applied to the loosening and moistening machine comprises:

[0009] Obtain the state variables and exogenous variables of the loose rehumidification machine at the current moment;

[0010] The state variables and exogenous variables are input into the control model for prediction to obtain the control variables at the current moment, and the operation of the execution components of the loose rehumidification machine is controlled based on the control variables at the current moment; wherein, the control model is trained based on the training method of the control model described in any embodiment of the present invention.

[0011] According to another aspect of the present invention, a training device for a control model is provided, wherein the control model is applied to a loose conditioning machine and comprises:

[0012] The historical variable data acquisition module is used to acquire the historical variable data of the loose moisture recovery machine with any historical moment as the starting moment, and the historical variable data includes the historical state variables and historical exogenous variables at each moment starting from the starting moment.

[0013] A control model training module is used to train the control model based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain a target control model; wherein the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted by the control model during the training process of the control model, based on the predicted control variables at each moment predicted by the control model, and are predicted based on the pre-trained simulation model of the loose rehumidification machine.

[0014] According to another aspect of the present invention, a control device for a loosening and moistening machine is provided, characterized in that the controller applied to the loosening and moistening machine comprises:

[0015] The current variable acquisition module is used to obtain the state variables and exogenous variables of the loose moisture recovery machine at the current moment;

[0016] A control variable prediction module is used to input the state variables and exogenous variables into the control model for prediction, obtain the control variables at the current moment, and control the operation of the execution components of the loose rehumidification machine based on the control variables at the current moment; wherein the control model is trained based on the training method of the control model described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0018] at least one processor; and

[0019] a memory communicatively connected to the at least one processor; wherein,

[0020] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the control model and / or the control method of the loose rehumidification machine described in any embodiment of the present invention.

[0021] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the training method of the control model and / or the control method of the loose rehumidification machine described in any embodiment of the present invention when executed.

[0022] The technical solution of the embodiment of the present invention is to obtain the historical variable data of the loose moisture conditioning machine with any historical moment as the starting moment, and the historical variable data includes the historical state variables and historical exogenous variables at each moment starting from the starting moment; the control model is trained based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain the target control model; wherein, the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted based on the simulation model of the loose moisture conditioning machine trained in advance for the predicted control variables at each moment predicted by the control model during the training process of the control model. The simulation model of the loose moisture conditioning machine trained in advance fully reflects the coupling relationship between multiple control mechanisms and the input material quality and environmental status, and the control model of the loose moisture conditioning machine is obtained by training under the coupling relationship, so that the moisture content and temperature of the output material of the loose moisture conditioning machine can be stabilized at the set values, thereby effectively improving the output material quality.

[0023] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 is a flow chart of training a control model provided in the first embodiment of the present invention;

[0026] Figure 2 is a flow chart of a control method of a loosening and moistening machine provided in Embodiment 2 of the present invention;

[0027] Figure 3 It is a structural schematic diagram of a training device for a control model provided in Embodiment 3 of the present invention;

[0028] Figure 4 It is a structural schematic diagram of a control device of a loosening and moistening machine provided in a fourth embodiment of the present invention;

[0029] Figure 5 It is a structural schematic diagram of an electronic device provided in Embodiment 5 of the present invention. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first sample state variable", "second sample state variable", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] Embodiment 1

[0033] Figure 1 This is a flow chart of a control model training provided by the first embodiment of the present invention. This embodiment is applicable to the case of training the control model of a loose rehumidification machine. The method can be executed by a control model training device. The control model training device can be implemented in the form of hardware and / or software. The control model training device can be configured in an electronic device for training the control model. Figure 1 As shown, the method includes:

[0034] S110, taking any historical moment as a starting moment, obtaining historical variable data of the loose moisture recovery machine, wherein the historical variable data includes historical state variables and historical exogenous variables at each moment starting from the starting moment.

[0035] Among them, historical variable data refers to the historical variables at each moment in the operation process of the loose rehumidification machine. Specifically, the historical variable data includes three types of variables: historical state variables, historical exogenous variables and historical control variables; state variables include the circulating wind speed in the drum, the circulating wind temperature in the drum, the moisture content of the discharge material and the discharge temperature; exogenous variables include the feed temperature and the moisture content of the feed; control variables include the circulating fan frequency, the opening of the heat exchanger steam valve, the opening of the dual-medium nozzle steam valve and the opening of the dual-medium nozzle pure water valve.

[0036] In this embodiment, the historical variable data of the loose moisture recovery machine is obtained by taking any historical moment as the starting moment, and the historical state variables, historical exogenous variables and historical control variables at each moment starting from the starting moment can be obtained. Among them, the starting moment and the time unit are selected by those skilled in the art according to needs, and are not limited here. For example, one time unit can be 1 minute.

[0037] S120. Train the control model based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain a target control model; wherein the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted by the control model during the training process of the control model, based on the predicted control variables at each moment predicted by the control model, and are predicted based on the pre-trained simulation model of the loose moisture regaining machine.

[0038] Among them, the first sample state variable is the state variable used to train the control model, and the first sample state variable includes the historical state variable at the starting moment and the predicted state variable at other moments. It can be understood that except for the starting moment, the state variables at other moments are predicted during the training process of the control model for the predicted control variables of the control model at each moment based on the simulation model of the pre-trained loose rehumidification machine. In this embodiment, the first sample state variable at each moment starting from the starting moment and the historical exogenous variable at each moment are used as training data to train the pre-constructed control model to obtain the target control model; wherein the control model is used to predict the control variable at the current moment based on the state variable and exogenous variable at the current moment. For example, if the input of the control model is the state variable and control variable (S at moment t) at moment t t ,F t ), then the output is the control variable A at time t tThe simulation model is used to predict the state variables at the next moment based on the state variables, exogenous variables and control variables at the current moment. For example, if the input of the simulation model is the state variables, exogenous variables and control variables at time t (S t ,F t ,A t ), then the output is the state variable S at time t+1 t+1 , which is the state variable at the next moment. It should be noted that one time unit here is 1 minute.

[0039] It should be noted that the control model and simulation model of the loose moisture recovery machine are both neural network models.

[0040] On the basis of the above embodiment, optionally, the control model is trained based on the first sample state variable at each moment and the historical exogenous variable at each moment to obtain the target control model; including:

[0041] The following steps are iteratively performed until the first stop condition is met to obtain the target control model:

[0042] 1) determining the control trajectory and state trajectory of the control model based on the first sample state variable at each moment, the historical exogenous variable at each moment, and the control model;

[0043] 2) scoring the control trajectory and state trajectory of the control model based on a preset scoring function to obtain a comprehensive scoring result;

[0044] 3) Using a preset reinforcement learning algorithm, the model parameters of the control model are updated based on the comprehensive scoring result.

[0045] Among them, the first stop condition can be that the number of iterations reaches a preset iteration threshold, or that the control effect of the control model reaches a preset effect, for example, the moisture content and temperature of the discharge material are accurately controlled within a preset numerical range. In this embodiment, the single iterative training process is: based on the first sample state variable at each moment and the historical exogenous variables and control model at each moment, determine the control trajectory and state trajectory of the control model; score the control trajectory and state trajectory of the control model based on a preset scoring function to obtain a comprehensive scoring result; use a preset reinforcement learning algorithm to update the model parameters of the control model based on the comprehensive scoring result. Among them, the control trajectory refers to the change trajectory of the control variables predicted by the control model starting from a certain moment. For example, the control trajectory can be The state trajectory refers to the change trajectory of the state variable from a certain moment. For example, the state trajectory can be The preset scoring function is used to calculate the comprehensive score of this training. Specifically, the target control variables and target state variables can be set by those skilled in the art based on experience. If the control variables and state variables in the state trajectory and control trajectory are closer to the target control variables and target state variables, the higher the comprehensive score. The preset reinforcement learning algorithm can be a proximal policy optimization (PPO) reinforcement learning algorithm based on a policy.

[0046] On the basis of the above embodiment, optionally, determining the control trajectory and state trajectory of the control model based on the first sample state variable at each moment, the historical exogenous variable at each moment, and the control model includes:

[0047] At any moment, the first sample state variable at the current moment and the historical exogenous variable at the current moment are input into the control model for prediction to obtain the predicted control variable at the current moment;

[0048] And, inputting the first sample state variable at the current moment, the historical exogenous variable at the current moment and the predicted control variable at the current moment into the simulation model of the loose moisture recovery machine for prediction, and obtaining the predicted state variable at the next moment;

[0049] The above steps are iteratively executed based on the time dimension until the number of executions reaches the first target number of executions, a control trajectory is determined based on the predicted control variables at each moment, and a state trajectory is determined based on the predicted state variables at each moment.

[0050] Among them, the first target execution number refers to the number of times the control trajectory and state trajectory are obtained by iterative execution based on the time dimension. It can be understood that the first target execution number is associated with the length of the control trajectory and the state trajectory. The larger the first target execution number, the longer the control trajectory and the state trajectory. Specifically, the first target execution number is set by technical personnel in this field based on experience and needs, and is not limited here.

[0051] Exemplarily, the training process of the control model is as follows:

[0052] 1) Randomly select a time t from the historical variable data as the starting time, and read the historical state variables at time t and the historical exogenous variables at each moment

[0053] 2) Input to the control model Get the predicted control variables

[0054] 3) Input to the trained simulation model of the loose rehydration machine Get the predicted state variable for the next moment

[0055] 4) Input to the control model Get the predicted control variables

[0056] 5) Input to the trained simulation model Get the predicted state variable for the next moment

[0057] 5) Iterate steps 4) and 5) until the execution times reach the first target execution times and obtain the control trajectory and state trajectory

[0058] 6) Scoring function based on preset scoring function Calculate the comprehensive score of this simulation control (i.e. this training)

[0059] 7) Based on the preset reinforcement learning algorithm, update the control model according to the comprehensive score The parameter values ​​of the neural network in .

[0060] 8) Iterate the above steps 1) to 7) and update the neural network parameters of the control model according to the comprehensive score obtained from each simulation control, which can gradually improve the control model. The control effect is obtained and the target control model is obtained. The target control model can be applied to the controller of the loose rehumidification machine.

[0061] On the basis of the above embodiment, optionally, the historical variable data also includes historical control variables at each moment starting from the starting moment; the training method of the simulation model of the loose moisture recovery machine includes:

[0062] The following steps are iteratively performed until the second stop condition is met, thereby obtaining a trained simulation model of the loose moisture regaining machine:

[0063] 1) Determine the predicted state trajectory based on the second sample state variable at each moment, the historical exogenous variable at each moment, the historical control variable at each moment and the simulation model of the loose moisture recovery machine; wherein the second sample state variable at each moment includes the historical state variable at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted based on the simulation model of the loose moisture recovery machine;

[0064] 2) determining the error between the predicted state trajectory and the historical state trajectory based on a preset error function; the historical state trajectory is determined based on the historical state variables at each moment;

[0065] 3) Using a reverse gradient propagation algorithm, the model parameters of the simulation model of the loose moisture recovery machine are updated based on the error between the predicted state trajectory and the historical state trajectory.

[0066] Among them, the second stop condition can be that the number of iterations reaches a preset iteration number threshold, or that the prediction effect of the simulation model reaches a preset performance index. In this embodiment, the process of a single iteration training is: based on the second sample state variables at each moment, the historical exogenous variables at each moment, the historical control variables at each moment and the simulation model of the loose rehumidification machine, the predicted state trajectory is determined; based on the preset error function, the error between the predicted state trajectory and the historical state trajectory is determined; the historical state trajectory is determined based on the historical state variables at each moment; using the reverse gradient propagation algorithm, the model parameters of the simulation model of the loose rehumidification machine are updated based on the error between the predicted state trajectory and the historical state trajectory. Among them, the second sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments. The predicted state variables at other moments are predicted based on the simulation model of the loose rehumidification machine. It can be understood that the simulation model of the loose rehumidification machine can predict the state variables at the next moment through the state variables, exogenous variables and control variables at the current moment. Therefore, the predicted state variables at other moments are predicted by the simulation model based on the state variables, exogenous variables and control variables at the previous moment of the moment. The preset error function includes but is not limited to mean square error (MSE) and cross-entropy error (Cross-Entropy Error), etc., which are not limited here. The predicted state trajectory refers to the change trajectory of the predicted state variable predicted by the simulation model from a certain moment. For example, the predicted state trajectory can be The historical state trajectory refers to the change trajectory of the historical state variables in the historical variable data from a certain moment. For example, the historical state trajectory can be

[0067] On the basis of the above embodiment, optionally, the method of determining the predicted state trajectory based on the second sample state variable at each moment, the historical exogenous variable at each moment, the historical control variable at each moment and the simulation model of the loose moisture recovery machine includes:

[0068] At any moment, the second sample state variable at the current moment, the historical exogenous variable at the current moment, and the historical control variable at the current moment are input into the simulation model of the loose moisture recovery machine for prediction, and the predicted state variable at the next moment is obtained;

[0069] The above steps are iteratively executed based on the time dimension until the number of executions reaches the second target number of executions, and the predicted state trajectory is determined based on the predicted state variables at each moment.

[0070] Among them, the second target execution number refers to the number of iterative executions of the predicted state trajectory obtained by iterative execution based on the time dimension. The second target execution number is associated with the predicted state trajectory. The larger the second target execution number, the longer the predicted state trajectory.

[0071] Exemplarily, the training process of the simulation model is as follows:

[0072] 1) Randomly select a time t from the historical variable data as the starting time, and read the historical state variables at the starting time And the historical exogenous variables at each moment since the starting moment and historical control variables

[0073] 2) Input to simulation model The predicted state variable for the next moment is obtained

[0074]

[0075] 3) Input to simulation model The predicted state variable for the next moment is obtained

[0076] 4) Iterate step 3) based on the time dimension until the number of predictions reaches the second target execution number N, so that the predicted state trajectory can be obtained Extract historical state trajectories of the same time period from historical data According to the preset error function Calculate the error between the predicted trajectory and the actual trajectory

[0077]

[0078] 5) Based on this error, use the back gradient propagation algorithm to update the simulation model The neural network parameter values.

[0079] 6) Iterate the above steps 1) to 5) until the stop condition is met to obtain the simulation model of the loose moisture recovery machine after training. It can be understood that the neural network parameters of the simulation model are updated according to the scoring error obtained in each training, that is, the simulation model can be gradually improved. accuracy, thus obtaining a usable simulation model.

[0080] The technical solution of this embodiment is to obtain the historical variable data of the loose moisture-replenishing machine with any historical moment as the starting moment, and the historical variable data includes the historical state variables and historical exogenous variables at each moment starting from the starting moment; the control model is trained based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain the target control model; wherein, the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted based on the simulation model of the loose moisture-replenishing machine trained in advance during the training process of the control model, with respect to the predicted control variables at each moment predicted by the control model. The simulation model of the loose moisture-replenishing machine trained in advance fully reflects the coupling relationship between multiple control mechanisms and the quality of the input material and the environmental state, and the control model of the loose moisture-replenishing machine is obtained by training under this coupling relationship, so that the moisture content and temperature of the output material of the loose moisture-replenishing machine can be stabilized at the set values, thereby effectively improving the output material quality.

[0081] Embodiment 2

[0082] Figure 2 This is a flow chart of a control method for a loose moisture recovery machine provided by the second embodiment of the present invention. This embodiment applies the target control model obtained by the control model training method in the above embodiment to the controller of the loose moisture recovery machine to obtain the control method for the loose moisture recovery machine. The explanations of the terms that are the same or corresponding to the above embodiments are not repeated here. Figure 2 As shown, the method is applied to the controller of the loose rehumidification machine, including:

[0083] S210, obtaining the state variables and exogenous variables of the loose moisture conditioning machine at the current moment.

[0084] S220. Input the state variables and exogenous variables into the control model for prediction to obtain the control variables at the current moment, and control the operation of the control mechanism of the loose rehumidification machine based on the control variables at the current moment; wherein, the control model is trained based on the training method of the control model described in any embodiment of the present invention.

[0085] In this embodiment, after the trained control model is configured in the controller of the loose moisture-replenishing machine, the loose moisture-replenishing machine is operated, the state variables and exogenous variables of the loose moisture-replenishing machine at the current moment are obtained, the exogenous variables are input into the control model, and the control variables of the loose moisture-replenishing machine at the current moment are predicted; further, the control mechanisms of the loose moisture-replenishing machine are controlled by the predicted control variables of the loose moisture-replenishing machine at the current moment. Among them, the control mechanisms of the loose moisture-replenishing machine include a drum circulation fan, a heat exchanger steam valve, a dual-medium nozzle steam valve, and a dual-medium nozzle pure water valve.

[0086] The technical solution of this embodiment is to obtain the state variables and exogenous variables of the loose moisture regaining machine at the current moment; input the state variables and exogenous variables into the control model for prediction, obtain the control variables at the current moment, and control the operation of the control mechanism of the loose moisture regaining machine based on the control variables at the current moment; wherein the control model is trained by the training method of the control model of any embodiment of the present invention. By controlling the control mechanism of the loose moisture regaining machine with the pre-trained control model, the moisture content and temperature of the discharge of the loose moisture regaining machine can be stabilized at the set values, thereby effectively improving the discharge quality.

[0087] Embodiment 3

[0088] Figure 3 FIG. 1 is a schematic diagram of a control model training device provided in Embodiment 3 of the present invention. Figure 3 As shown, the control model obtained by training the device is applied to the loose conditioning machine, including:

[0089] The historical variable data acquisition module 310 is used to acquire the historical variable data of the loose moisture recovery machine with any historical moment as the starting moment, and the historical variable data includes the historical state variables and historical exogenous variables at each moment starting from the starting moment.

[0090] The control model training module 320 is used to train the control model based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain a target control model; wherein the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted by the control model during the training process of the control model, based on the predicted control variables at each moment predicted by the control model, and are predicted based on the pre-trained simulation model of the loose rehumidification machine.

[0091] The technical solution of this embodiment is to obtain the historical variable data of the loose moisture-replenishing machine with any historical moment as the starting moment, and the historical variable data includes the historical state variables and historical exogenous variables at each moment starting from the starting moment; the control model is trained based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain the target control model; wherein, the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted based on the simulation model of the loose moisture-replenishing machine trained in advance during the training process of the control model, with respect to the predicted control variables at each moment predicted by the control model. The simulation model of the loose moisture-replenishing machine trained in advance fully reflects the coupling relationship between multiple control mechanisms and the quality of the input material and the environmental state, and the control model of the loose moisture-replenishing machine is obtained by training under this coupling relationship, so that the moisture content and temperature of the output material of the loose moisture-replenishing machine can be stabilized at the set values, thereby effectively improving the output material quality.

[0092] Based on the above embodiment, optionally, the control model training module 320 is specifically used to iteratively execute the following steps until the first stop condition is met to obtain the target control model: determine the control trajectory and state trajectory of the control model based on the first sample state variable at each moment and the historical exogenous variables and control model at each moment; score the control trajectory and state trajectory of the control model based on a preset scoring function to obtain a comprehensive scoring result; use a preset reinforcement learning algorithm to update the model parameters of the control model based on the comprehensive scoring result.

[0093] On the basis of the above embodiment, optionally, the control model training module 320 also includes a trajectory determination module, which is used to input the first sample state variable at the current moment and the historical exogenous variables at the current moment into the control model for prediction at any moment, so as to obtain the predicted control variables at the current moment; and input the first sample state variable at the current moment, the historical exogenous variables at the current moment and the predicted control variables at the current moment into the simulation model of the loose moisture retaining machine for prediction, so as to obtain the predicted state variables at the next moment; iteratively execute the above steps based on the time dimension until the number of executions reaches the first target number of executions, determine the control trajectory based on the predicted control variables at each moment, and determine the state trajectory based on the predicted state variables at each moment.

[0094] On the basis of the above embodiment, optionally, the device also includes a training module of the simulation model, which is used to iteratively execute the following steps until two stop conditions are met to obtain a trained simulation model of the loose moisture recovery machine: determine the predicted state trajectory based on the second sample state variables at each moment, the historical exogenous variables at each moment, the historical control variables at each moment and the simulation model of the loose moisture recovery machine; wherein the second sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted based on the simulation model of the loose moisture recovery machine; determine the error between the predicted state trajectory and the historical state trajectory based on a preset error function; the historical state trajectory is determined based on the historical state variables at each moment; use the reverse gradient propagation algorithm to update the model parameters of the simulation model of the loose moisture recovery machine based on the error between the predicted state trajectory and the historical state trajectory.

[0095] On the basis of the above embodiment, optionally, the training module of the simulation model also includes a predicted state trajectory determination unit, which is used to input the second sample state variable at the current moment, the historical exogenous variable at the current moment, and the historical control variable at the current moment into the simulation model of the loose moisture recovery machine for prediction at any moment, and obtain the predicted state variable at the next moment; iteratively execute the above steps based on the time dimension until the number of executions reaches the second target number of executions, and determine the predicted state trajectory based on the predicted state variables at each moment.

[0096] The control model training device provided in the embodiment of the present invention can execute the control model training method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0097] Embodiment 4

[0098] Figure 4 Schematic diagram of the structure of a control device for a loosening and moisture-replenishing machine provided in the fourth embodiment of the present invention. Figure 4 As shown, the device is applied to the controller of the loose rehumidification machine, including:

[0099] The current variable acquisition module 410 is used to acquire the state variables and exogenous variables of the loosening and moistening machine at the current moment;

[0100] The control variable prediction module 420 is used to input the state variables and exogenous variables into the control model for prediction, obtain the control variables at the current moment, and control the operation of the control mechanism of the loose rehumidification machine based on the control variables at the current moment; wherein the control model is trained based on the training method of the control model described in any embodiment of the present invention.

[0101] The technical solution of this embodiment is to obtain the state variables and exogenous variables of the loose moisture regaining machine at the current moment; input the state variables and exogenous variables into the control model for prediction, obtain the control variables at the current moment, and control the operation of the control mechanism of the loose moisture regaining machine based on the control variables at the current moment; wherein the control model is trained by the training method of the control model of any embodiment of the present invention. By controlling the control mechanism of the loose moisture regaining machine with the pre-trained control model, the moisture content and temperature of the discharge of the loose moisture regaining machine can be stabilized at the set values, thereby effectively improving the discharge quality.

[0102] The control device of the loosening and moistening machine provided in the embodiment of the present invention can execute the control method of the loosening and moistening machine provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0103] Embodiment 5

[0104] Figure 5 1 is a schematic diagram of the structure of an electronic device provided in Embodiment 5 of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0105] like Figure 5 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0106] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0107] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the training method of the control model and / or the control method of the loose rehumidification machine.

[0108] In some embodiments, the training method of the control model and / or the control method of the loose rehumidifier can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the training method of the control model and / or the control method of the loose rehumidifier described above can be executed. Alternatively, in other embodiments, the processor 11 can be configured to execute the training method of the control model and / or the control method of the loose rehumidifier in any other appropriate manner (for example, by means of firmware).

[0109] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0110] The computer programs for implementing the training method of the control model and / or the control method of the loose rehumidification machine of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or the block diagram are implemented. The computer program can be executed completely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or completely on a remote machine or server.

[0111] Embodiment 6

[0112] Embodiment 6 of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute a control model training method, the method comprising:

[0113] Taking any historical moment as the starting moment, the historical variable data of the loose re-dehumidification machine is obtained. The historical variable data includes the historical state variables and historical exogenous variables at each moment starting from the starting moment.

[0114] The control model is trained based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain the target control model; wherein the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted by the control model during the training process of the control model, based on the predicted control variables at each moment predicted by the control model, and are predicted based on the pre-trained simulation model of the loose rehumidification machine.

[0115] And / or, performing a control method of a loose conditioning machine, the method comprising:

[0116] Obtain the state variables and exogenous variables of the loose rehumidification machine at the current moment;

[0117] The state variables and exogenous variables are input into the control model for prediction to obtain the control variables at the current moment, and the operation of the control mechanism of the loose rehumidification machine is controlled based on the control variables at the current moment; wherein the control model is trained by the training method of the control model based on any embodiment of the present invention.

[0118] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0119] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0120] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0121] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0122] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0123] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A training method for a control model, characterized in that: The control model is applied to the loose conditioning machine, including: Taking any historical moment as the starting moment, obtaining the historical variable data of the loose moisture recovery machine, the historical variable data including the historical state variables and historical exogenous variables at each moment starting from the starting moment; The control model is trained based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain a target control model; wherein the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted by the control model during the training process of the control model, based on the predicted control variables of each moment predicted by the control model, and are predicted based on the pre-trained simulation model of the loose moisture retaining machine. The control model is used to predict the control variables at the current moment based on the state variables and exogenous variables at the current moment, and the simulation model is used to predict the state variables at the next moment based on the state variables, exogenous variables and control variables at the current moment.

2. The method according to claim 1, characterized in that The control model is trained based on the first sample state variable at each moment and the historical exogenous variable at each moment to obtain the target control model; including: The following steps are iteratively performed until the first stop condition is met to obtain the target control model: Determine the control trajectory and state trajectory of the control model based on the first sample state variable at each moment, the historical exogenous variable at each moment, and the control model; Scoring the control trajectory and state trajectory of the control model based on a preset scoring function to obtain a comprehensive scoring result; Using a preset reinforcement learning algorithm, the model parameters of the control model are updated based on the comprehensive scoring result.

3. The method according to claim 2, characterized in that The determining the control trajectory and state trajectory of the control model based on the first sample state variable at each moment, the historical exogenous variable at each moment, and the control model includes: At any moment, the first sample state variable at the current moment and the historical exogenous variable at the current moment are input into the control model for prediction to obtain the predicted control variable at the current moment; And, inputting the first sample state variable at the current moment, the historical exogenous variable at the current moment and the predicted control variable at the current moment into the simulation model of the loose moisture recovery machine for prediction, and obtaining the predicted state variable at the next moment; The above steps are iteratively executed based on the time dimension until the number of executions reaches the first target number of executions, a control trajectory is determined based on the predicted control variables at each moment, and a state trajectory is determined based on the predicted state variables at each moment.

4. The method according to claim 1, characterized in that: The historical variable data also includes historical control variables at each moment starting from the starting moment; the training method of the simulation model of the loose moisture recovery machine includes: The following steps are iteratively performed until the second stop condition is met, thereby obtaining a trained simulation model of the loose moisture regaining machine: Determine the predicted state trajectory based on the second sample state variables at each moment, the historical exogenous variables at each moment, the historical control variables at each moment, and the simulation model of the loose moisture recovery machine; wherein the second sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are obtained based on the prediction of the simulation model of the loose moisture recovery machine; Determine the error between the predicted state trajectory and the historical state trajectory based on a preset error function; the historical state trajectory is determined based on the historical state variables at each moment; The model parameters of the simulation model of the loose moisture recovery machine are updated based on the error between the predicted state trajectory and the historical state trajectory using a back gradient propagation algorithm.

5. The method according to claim 4, characterized in that The method of determining the predicted state trajectory based on the second sample state variable at each moment, the historical exogenous variable at each moment, the historical control variable at each moment, and the simulation model of the loose moisture recovery machine includes: At any moment, the second sample state variable at the current moment, the historical exogenous variable at the current moment, and the historical control variable at the current moment are input into the simulation model of the loose moisture recovery machine for prediction, and the predicted state variable at the next moment is obtained; The above steps are iteratively executed based on the time dimension until the number of executions reaches the second target number of executions, and the predicted state trajectory is determined based on the predicted state variables at each moment.

6. A control method for a loosening and moistening machine, characterized in that: Controllers for loose rehumidification machines include: Obtain the state variables and exogenous variables of the loose rehumidification machine at the current moment; The state variables and exogenous variables are input into the control model for prediction to obtain the control variables at the current moment, and the operation of the control mechanism of the loose rehumidification machine is controlled based on the control variables at the current moment; wherein, the control model is trained based on the training method of the control model described in any one of claims 1-5.

7. A training device for a control model, characterized in that: The control model is applied to the loose conditioning machine, including: A historical variable data acquisition module, used to acquire the historical variable data of the loose moisture recovery machine with any historical moment as the starting moment, wherein the historical variable data includes the historical state variables and historical exogenous variables at each moment starting from the starting moment; A control model training module is used to train the control model based on the first sample state variables at each moment and the historical exogenous variables at each moment to obtain a target control model; wherein the first sample state variables at each moment include the historical state variables at the starting moment and the predicted state variables at other moments, and the predicted state variables at other moments are predicted by the control model during the training process of the control model, based on the predicted control variables of each moment predicted by the control model, and are predicted based on the pre-trained simulation model of the loose moisture recovery machine. The control model is used to predict the control variables at the current moment based on the state variables and exogenous variables at the current moment, and the simulation model is used to predict the state variables at the next moment based on the state variables, exogenous variables and control variables at the current moment.

8. A control device for a loosening and rehumidifying machine, characterized in that: Controllers for loose rehumidification machines include: The current variable acquisition module is used to obtain the state variables and exogenous variables of the loose moisture recovery machine at the current moment; A control variable prediction module is used to input the state variables and exogenous variables into the control model for prediction, obtain the control variables at the current moment, and control the operation of the control mechanism of the loose rehumidification machine based on the control variables at the current moment; wherein the control model is trained based on the training method of the control model described in any one of claims 1-5.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the control model described in any one of claims 1 to 5 and / or the control method of the loose rehumidification machine described in claim 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to implement the training method of the control model described in any one of claims 1 to 5 and / or the control method of the loose moisture regaining machine described in claim 6 when executed by a processor.

Images