Mobile emotion relieving cabin control system and method

Abstract

The invention relates to the technical field of relieving cabin control, and discloses a mobile emotion relieving cabin control system and method.The system integrates an environment monitoring module, an intelligent strategy construction module and a cloud control execution module, environment parameters in a cabin and user physiological parameters are monitored in real time, and a multi-target optimized environment regulation and control strategy model is constructed. Through a reinforcement learning algorithm, the system can dynamically adjust environment settings such as light, music, temperature, humidity and the like so as to achieve an optimal soothing effect. And meanwhile, the cloud control execution module realizes remote monitoring and intelligent regulation and control, so that the convenience and practicability of the system are further improved. The intelligent emotion relieving cabin overcomes the defects of a traditional emotion relieving cabin in the aspects of intelligence, individuation and environmental adaptability, a novel and effective technical solution is provided for emotion management in modern fast-paced life, and the intelligent emotion relieving cabin has wide application prospects and market value.

Description

Technical Field

[0001] The present invention relates to the technical field of emotional relief cabin control, and in particular to a control system and method for a mobile emotional relief cabin. Background Art

[0002] With the accelerated pace of modern life, people are facing increasing psychological pressure, and emotional management has become an urgent problem to be solved. Although traditional emotional regulation methods, such as psychological counseling and drug therapy, can help people relieve stress to a certain extent, they are often limited by time, place and the availability of professional resources. Therefore, it is particularly important to develop an environmental device that is convenient, efficient and can autonomously regulate the user's emotions. With the rapid development of the Internet of Things, artificial intelligence and environmental control technology, new technical means have been provided for emotional management in recent years. As a new type of equipment that integrates multiple environmental control functions, the mobile emotional relief cabin aims to help users relieve stress and relax by creating a comfortable and pleasant environment.

[0003] Existing mood relief cabin control systems still have many deficiencies in terms of intelligence, personalization, and environmental adaptability. Traditional mood relief cabin control systems often rely on preset fixed programs and manual adjustments by users, and lack the ability to monitor and dynamically respond to environmental parameters and user physiological parameters in real time. This results in the inability to automatically adjust the cabin environment according to the user's real-time emotional state or physiological response, thus affecting the effect of emotional relief. In addition, existing systems often lack intelligent strategy building modules, are unable to model environmental management as a multi-objective optimization problem, and dynamically adjust the cabin environment settings through algorithms to achieve the optimal state. Summary of the invention

[0004] The object of the present invention is to provide a mobile emotional relief cabin control system and method to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solution: a mobile emotion relief cabin control system, the system comprising: Environmental monitoring module, used to monitor the environmental parameters and user physiological parameters in the emotional relief cabin in real time; An intelligent strategy building module is used to build an environmental control strategy model, which models the environmental management of the emotional relief cabin as a multi-objective optimization problem and dynamically adjusts the cabin environment settings through an algorithm to achieve the optimal state; the intelligent strategy building module further includes: The state description component is used to define the current state space of the mood relief cabin and its environment, including temperature, humidity, light, sound intensity, air quality, and user physiological indicators; Action definition component, used to set executable cabin environment control action space, including adjusting light color and brightness, playing specific music or sound effects, adjusting cabin temperature and humidity, releasing aromatherapy or negative ions, starting or shutting down cabin massage equipment, and adjusting seat angle; The rule evaluation component is used to quantitatively evaluate the results of control actions based on the suitability of environmental parameters, the positivity of user physiological responses, energy efficiency ratios, and equipment operating conditions; The strategy optimization unit uses the reinforcement learning algorithm to select the optimal control action in a given state space and continuously optimizes the control strategy through continuous learning iterations; The cloud control execution module is responsible for uploading the data collected by the environmental monitoring module to the cloud server. After being processed by the intelligent strategy model, it receives the control instructions issued by the cloud and drives the actuators in the emotion relief cabin to perform corresponding environmental adjustment actions.

[0006] Preferably, the state space further includes a user's emotion recognition result, which is obtained by analyzing the user's facial expression, voice intonation or physiological signal and serves as an input parameter of the environment control strategy.

[0007] Preferably, the implementation method of the rule evaluation component includes defining an evaluation function R(s, a), where s represents the current state space and a represents the control action taken; R(s, a) performs weighted summation according to the suitability S(s, a) of environmental parameters, the positivity P(s, a) of user physiological reactions, the energy efficiency ratio E(s, a) and the stability M(s, a) of equipment operation status to obtain a quantitative evaluation value of the control action; the specific algorithm expression is: R(s, a) = w1*S(s, a) + w2*P(s, a) + w3*E(s, a) + w4*M(s, a), where w1, w2, w3, and w4 are weight coefficients of each evaluation dimension, and satisfy w1 + w2 + w3+ w4 = 1.

[0008] Preferably, the mobile mood relief cabin control system according to claim 1 also includes a data acquisition module, which collects environmental parameters and user physiological parameters in real time through various sensors deployed in the mood relief cabin, and combines the user's historical usage records and preference settings to form an initial training data set for preliminary model training of the intelligent strategy building module.

[0009] Preferably, the data acquisition module further utilizes user feedback and environmental adjustment records in the historical data to construct a historical state data set for the use of the emotion relief cabin; for each record in the historical state data set, possible control actions are simulated and their corresponding expected benefits are calculated to form a state-action-benefit triple, which is added to the initial training data set to further enrich the training data set of the model.

[0010] Preferably, the strategy optimization unit uses a deep Q network DQN algorithm to train the environment control strategy model.

[0011] Preferably, the step of the strategy optimization unit training the environment control strategy model includes: Step 1: Initialize the experience replay memory and the target Q network, where the experience replay memory is used to store the experience of the state transfer of the emotion relief cabin, and the target Q network is used to provide a stable target value during the training process; at the same time, initialize the main Q network, whose input is the current state space of the emotion relief cabin and its environment, and whose output is the Q value estimation of each control action; Step 2: Set the learning rate α, discount factor γ, exploration rate ε, and number of training rounds, where α is used to control the learning speed of the model, γ is used to determine the discount degree of future rewards, ε is used to balance the ratio of exploring new actions and using known optimal actions, and the number of training rounds determines the total number of training iterations; Step 3: For each training round, perform the following steps in sequence: Step 3.1: Randomly select an initial state s from the training data set, which includes a combination of environmental parameters in the emotion relief cabin and user physiological parameters; Step 3.2: Select a control action a according to the current state s and the ε-greedy strategy, that is, randomly select an action for exploration with a probability of ε, and select the action with the largest Q value predicted by the current main Q network with a probability of 1-ε; Step 3.3: Execute the selected action a, observe the next state s' of the emotion relief cabin and the immediate reward r calculated by the rule evaluation component; Step 3.4: Store the experience tuple (s, a, r, s') into the experience replay memory for subsequent learning; Step 3.5: Randomly extract a batch of experience tuples from the experience replay memory to update the parameters of the main Q network; Step 3.6: Use the extracted experience tuples to calculate the target Q value in combination with the target Q network, and update the weights of the main Q network according to the loss function of the DQN algorithm. The loss function is defined as the mean square error between the predicted Q value and the target Q value. Step 3.7: Copy the parameters of the main Q network to the target Q network according to the preset update frequency to maintain the stability of the target value; Step 4: Repeat step S3 until the preset number of training rounds is reached or the output Q value of the main Q network tends to converge; Step 5: After the training is completed, the trained main Q network is used as the environmental control strategy model of the emotion relief cabin. For any given current state s of the emotion relief cabin and its environment, the Q value of each control action is calculated through the forward propagation of the main Q network, and the action a with the largest Q value is selected as the optimal control action.

[0012] Preferably, it is characterized in that, in each round of iteration in step 3, the mean square error MSE is used as the performance evaluation indicator of the current round of iteration, and the specific implementation steps include: A1: Based on the sampled experience tuple (s, a, r, s'), the target Q network is used to calculate the target Q value of each experience tuple; A2: Calculate the predicted Q value of the main Q network for each state-action pair (s, a) in the experience tuple; A3: Calculate the mean square error (MSE) between the predicted Q value and the target Q value as the loss function value of the current iteration. The calculation formula of MSE is:

[0013] Where N is the number of sampled experience tuples, is the target Q value, is the predicted Q value of the main Q network; A4: Based on the calculated MSE value, the parameters of the main Q network are updated using the gradient descent algorithm to minimize the MSE value, thereby optimizing the performance of the main Q network. A5: Repeat steps A1 to A4 until the end condition of the current training round is met, and then enter the next training round.

[0014] Preferably, it is characterized in that: after each update of the main Q network parameters, the changing trend of the MSE value is monitored. If the MSE value continues to increase or fluctuates too much, the value of the learning rate α or the exploration rate ε is adjusted to balance the exploration ability and learning stability of the model.

[0015] Preferably, a method for controlling a mobile mood relief cabin comprises the following steps: S1: The environmental parameters and user physiological parameters in the mood relief cabin are monitored in real time through the environmental monitoring module, and the current state space of the mood relief cabin and its environment is defined, which includes temperature, humidity, light, sound intensity, air quality and user physiological indicators; S2: Set executable cabin environment control action space, including adjusting light color and brightness, playing specific music or sound effects, adjusting cabin temperature and humidity, releasing aromatherapy or negative ions, starting or shutting down cabin massage equipment, and adjusting seat angle; S3: Establish a rule evaluation component to quantitatively evaluate the results of the executed control actions based on the suitability of environmental parameters, the positivity of user physiological responses, energy efficiency ratio, and equipment operation status; S4: Build an environmental control strategy model and use reinforcement learning algorithms to select the optimal control action in a given state space; S5: After the training is completed, the obtained main Q network is used as the environmental control strategy model. The data collected by the environmental monitoring module is uploaded to the cloud server through the cloud control execution module. After being processed by the intelligent strategy model, the control instructions issued by the cloud are received; S6: According to the received control instructions, the actuators in the emotion relief cabin are driven to perform corresponding environmental adjustment actions to achieve intelligent regulation of the environment in the emotion relief cabin.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention can capture the cabin environment parameters (such as temperature, humidity, light, sound intensity, air quality) and user physiological parameters in real time through the environmental monitoring module, ensuring that the system can instantly perceive the changes in the user's emotional state and physiological response. Combined with the intelligent strategy building module, the system can quickly respond and dynamically adjust the cabin environment settings, such as light color and brightness, music or sound effects, temperature and humidity, aromatherapy or negative ion release, etc., so as to provide users with a personalized environment that is more in line with their current emotional needs and significantly improve the emotional relief effect. The system's built-in intelligent strategy building module models environmental management as a multi-objective optimization problem through the collaborative work of the state description component, action definition component, rule evaluation component and strategy optimization unit. Using the reinforcement learning algorithm, the system can optimize the control strategy in continuous learning iterations, and automatically select and execute the optimal environmental control action. This intelligent control method not only improves the efficiency and accuracy of emotional relief, but also can provide highly personalized emotional relief services according to the user's personal preferences and physiological characteristics to meet the diverse needs of different users.

[0017] The introduction of the cloud control execution module enables the system to have the ability of remote monitoring and intelligent regulation. The data collected by the environmental monitoring module is uploaded to the cloud server in real time. After being processed by the intelligent strategy model, the cloud sends control instructions to the system to drive the actuators in the emotional relief cabin to perform corresponding environmental adjustment actions. The cloud control execution module not only improves the convenience of the system, allowing users to remotely control and monitor the emotional relief cabin through the cloud platform at any location, but also facilitates professional institutions or service personnel to remotely participate in the emotional management process, providing more comprehensive and timely support.

[0018] The rule evaluation component takes into account the energy efficiency ratio and equipment operation status when evaluating the control action results, which helps the system reduce energy consumption and extend the service life of equipment while ensuring the emotional soothing effect. This design not only conforms to the concept of green energy saving, but also reduces the user's operating costs and improves the overall cost performance of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a working principle diagram of the control system of the mobile emotion relief cabin of the present invention; Figure 2 A flowchart for the construction of the environmental control strategy model; Figure 3 This is a flow chart for evaluating model performance using the MSE algorithm. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0021] See also Figure 1-3 The present invention provides a technical solution: a mobile emotional relief cabin control system, the system comprising: Environmental monitoring module: This module integrates various sensors to monitor the environmental parameters in the emotional relief cabin in real time, such as temperature, humidity, light intensity, sound intensity, air quality (such as PM2.5, CO2 concentration, etc.), and user physiological parameters such as heart rate, blood pressure, skin conductivity, etc. These sensors are connected to the system's data processing unit via wired or wireless means to transmit monitoring data in real time.

[0022] Intelligent strategy building module: used to build an environmental control strategy model, which models the environmental management of the emotional relief cabin as a multi-objective optimization problem, and dynamically adjusts the cabin environment settings through an algorithm to achieve the optimal state; the intelligent strategy building module further includes: State description component: defines the current state space of the emotion relief cabin and its environment, which covers all monitorable environmental parameters and user physiological indicators to form a comprehensive state description vector.

[0023] Action definition component: Set a series of executable cabin environment control actions, including but not limited to adjusting the color and brightness of lights, playing specific music or sound effects, adjusting the cabin temperature and humidity, releasing aromatherapy or negative ions, starting or shutting down massage equipment, adjusting seat angles, etc.

[0024] Rule evaluation component: Establish evaluation criteria and quantify the results of each control action based on the suitability of environmental parameters (such as whether the temperature is comfortable and the air quality is good), the positivity of the user's physiological response (such as whether the heart rate is reduced and the skin conductivity is reduced), energy efficiency ratio (such as the ratio of energy consumption to effect) and equipment operating status (such as whether the equipment is operating normally and whether there is a fault alarm).

[0025] Strategy Optimization Unit: Using reinforcement learning algorithms, we explore and select the optimal control action in a given state space. Through continuous learning iterations, the algorithm gradually optimizes the control strategy, enabling the system to automatically adjust the cabin environment according to the user's real-time status to achieve the best emotional relief effect.

[0026] Cloud control execution module: This module is responsible for uploading the data collected by the environmental monitoring module to the cloud server. The cloud server runs the intelligent strategy model, processes and analyzes the data, and generates control instructions. Subsequently, the cloud control execution module receives these control instructions and drives the actuators (such as lighting controllers, sound systems, temperature and humidity regulators, massage equipment, etc.) in the emotional relief cabin to perform corresponding environmental adjustment actions through wireless or wired communication. The cloud control execution module supports remote monitoring functions. Users or managers can view the working status, environmental parameters and user physiological data of the emotional relief cabin in real time through remote terminals such as mobile phone APP and web pages, and can manually send control instructions for remote regulation.

[0027] Through the combination of the above modules, the mobile mood relief cabin control system of this embodiment can realize intelligent and personalized control of the cabin environment, significantly improving the mood relief effect and user experience. At the same time, the application of the cloud control execution module makes the system more convenient and practical, meeting the needs of modern life for efficient and convenient mood management methods.

[0028] The present invention will be further described below in conjunction with Examples 1 to 4: Embodiment 1;

[0029] The state space further includes the user's emotion recognition results, which are obtained by analyzing the user's facial expressions, voice intonation or physiological signals (such as heart rate variability, skin conductivity changes, etc.), which are collected and processed by corresponding sensors or recognition algorithms. For example, the system can integrate facial recognition technology to identify the user's emotional state by analyzing the user's facial expressions (such as smiling, frowning, etc.); or use voice recognition technology to analyze the user's voice intonation (such as happiness, frustration, etc.) to judge his or her emotions; it can also be combined with physiological signal monitoring, such as when the user is in a tense state, the skin conductivity may increase. These emotion recognition results serve as input parameters for the environmental control strategy, allowing the system to more accurately understand the user's current emotional state, and adjust the cabin environment settings accordingly to provide emotional relief services that better meet the user's needs.

[0030] The rule evaluation component quantifies the effect of control actions by defining the evaluation function R(s, a), where s represents the current state space, including environmental parameters, user physiological parameters, and newly added user emotion recognition results; a represents the control actions taken by the system, such as adjusting lights, playing music, etc.

[0031] The evaluation function R(s, a) performs weighted summation according to four dimensions: the suitability of environmental parameters S(s, a), the positivity of user physiological reactions P(s, a), the energy efficiency ratio E(s, a), and the stability of equipment operation status M(s, a). Each dimension has a corresponding evaluation function, which is used to calculate the evaluation value under that dimension. The specific algorithm expression is: R(s, a) = w1*S(s, a) + w2*P(s, a) + w3*E(s, a) + w4*M(s, a). Among them, w1, w2, w3, and w4 are the weight coefficients of each evaluation dimension, which are used to reflect the importance of different dimensions in the overall evaluation, and satisfy w1 + w2 + w3 + w4 = 1 to ensure the normalization of the evaluation results. For example, the suitability of environmental parameters S(s, a) can be calculated based on the deviation of parameters such as cabin temperature, humidity, and light from ideal values; the positivity of user physiological reactions P(s, a) can be judged based on changes in physiological indicators such as user heart rate and skin conductivity; the energy efficiency ratio E(s, a) can consider the ratio between the energy consumed by the control action and the emotional relief effect it brings; the stability of the equipment operating status M(s, a) can be evaluated based on information such as whether the equipment is operating normally and whether there is a fault alarm.

[0032] In this way, the rule evaluation component can comprehensively consider multiple dimensions, perform quantitative evaluation on each control action, and provide accurate feedback to the strategy optimization unit, thereby continuously optimizing the environmental control strategy and improving the system's performance and user experience. Embodiment 2;

[0033] This embodiment is used to describe in detail the functions and implementation methods of the data acquisition module. The data acquisition module is responsible for collecting environmental parameters (such as temperature, humidity, light intensity, sound intensity, air quality, etc.) and user physiological parameters (such as heart rate, blood pressure, skin conductivity, facial expressions, etc.) in real time through various sensors deployed in the cabin.

[0034] In addition to real-time data collection, the data collection module also combines the user's historical usage records and preference settings to form an initial training data set. These historical usage records include the user's past mood relief cabin usage habits, preferred environmental settings, physiological reactions, etc., while the preference settings are the customized configuration of the cabin environment by the user based on personal preferences. The initial training data set provides basic data for the initial model training of the intelligent strategy building module, allowing the model to initially learn how to select appropriate control actions based on the current state based on understanding user needs and preferences.

[0035] The data collection module further uses the user feedback and environmental adjustment records in the historical data to construct a historical status data set for the use of the emotional relief cabin. User feedback includes the user's subjective evaluation of the emotional relief effect each time, such as satisfaction, degree of relaxation, etc.; environmental adjustment records are the control actions and results taken by the system in the past based on the user status and environmental parameters.

[0036] For each record in the historical state data set, the data collection module simulates possible control actions and calculates the expected benefits corresponding to these control actions based on user feedback and the evaluation function of the rule evaluation component. The expected benefit reflects the degree of improvement in the user's emotional relief effect after taking a certain control action.

[0037] In this way, the data acquisition module forms a large number of state-action-benefit triplets, which are added to the initial training data set to further enrich the training data set of the model. This enables the intelligent policy building module to train the model based on more comprehensive data, improving the accuracy and personalization of the environmental control strategy. Suppose a user has chosen to play light music to relax in a tense state many times in the past usage records. The data acquisition module will record these usage records, and when constructing the historical state data set, simulate the control actions of playing different light music in a similar tense state and calculate the corresponding expected benefits. If a simulation shows that the user's tension is significantly relieved after playing a specific light music (based on user feedback and improvement of physiological parameters), then this state-action-benefit triplet will be added to the training data set. When the intelligent policy building module performs model training, the model will learn that playing this specific light music in a similar tense state is an effective control action, so that in future actual use, it can more accurately recommend or automatically execute this action based on the user's state. Embodiment 3;

[0038] This embodiment is used to describe in detail the specific steps of the strategy optimization unit using the deep Q network (DQN) algorithm to train the environment control strategy model: Step 1: Initialize the experience replay memory to store the experience of the emotional relief cabin state transition, which includes the combination of state, action, reward and subsequent state. Experience replay technology helps to improve the sample efficiency and stability of learning.

[0039] Initialize the target Q network, which has the same structure as the main Q network, but is used to provide a stable target Q value during training to reduce fluctuations during training. The main Q network is initialized, and its input is the current state space of the emotion relief cabin and its environment (including environmental parameters, user physiological parameters, and user emotion recognition results, etc.), and its output is the Q value estimate of each control action, that is, the long-term reward that can be obtained after taking a certain action.

[0040] Step 2: Set the learning rate α to control the step size of the model parameter update, which affects the speed and stability of learning. Set the discount factor γ to determine the importance of future rewards in the current decision, that is, the degree of influence of future rewards on the current decision. Set the exploration rate ε to balance the ratio of exploring new actions and using known optimal actions, ensuring that the model can explore new possibilities and use known optimal strategies. The number of training rounds is set to determine the total number of training iterations, that is, the total number of rounds that the model needs to learn.

[0041] Step 3: The training process is: For each training round, the following steps are performed: Step 3.1: Randomly select an initial state s from the training data set. This state is a combination of the environmental parameters in the emotion relief cabin and the user's physiological parameters, representing the current state of the emotion relief cabin.

[0042] Step 3.2: Select a control action a according to the current state s and the ε-greedy strategy. Specifically, randomly select an action for exploration with a probability of ε, and select the action with the largest Q value predicted by the current main Q network with a probability of 1-ε, that is, consider this action to be the optimal action in the current state.

[0043] Step 3.3: Execute the selected action a, observe the next state s' of the emotion relief cabin, and the immediate reward r calculated by the rule evaluation component. The immediate reward reflects the degree of improvement of the environment state and the soothing effect of the user's emotions after taking the action.

[0044] Step 3.4: Store the experience tuple (s, a, r, s') in the experience replay memory for subsequent learning. The experience replay memory can store a large number of experience tuples and provide rich training samples for the model.

[0045] Step 3.5: Randomly extract a batch of experience tuples from the experience replay memory to update the parameters of the main Q network. This random sampling method helps to break the temporal correlation between experiences and improve the stability of learning.

[0046] Step 3.6: Use the extracted experience tuples and the target Q network to calculate the target Q value. The target Q value is an estimate of the long-term reward that the current action can obtain in the future. Then, update the weights of the main Q network according to the loss function of the DQN algorithm (i.e., the mean square error between the predicted Q value and the target Q value).

[0047] Step 3.7: Copy the parameters of the main Q network to the target Q network at a preset update frequency to maintain the stability of the target value. This helps reduce fluctuations and instabilities during training.

[0048] Step 4: Repeat step 3 until the preset number of training rounds is reached or the output Q value of the main Q network tends to converge. When the output Q value of the main Q network tends to be stable, it is considered that the model has learned an effective environment control strategy.

[0049] Step 5: After the training is completed, the trained main Q network is used as the environmental control strategy model of the emotional relief cabin. For any given current state s of the emotional relief cabin and its environment, the Q value of each control action is calculated through the forward propagation of the main Q network, and the action a with the largest Q value is selected as the optimal control action. In this way, the model can automatically select the most appropriate control action according to the current state to achieve the best emotional relief effect. Embodiment 4;

[0050] In each iterative training of step 3 above, the mean square error (MSE) is used as the performance evaluation indicator. The specific implementation steps include: A1. Target Q value calculation: In each iteration, the system randomly samples a batch of experience tuples (s, a, r, s') from the experience replay buffer, where s represents the current state, a represents the action taken, r represents the immediate reward after the action is performed, and s' represents the new state after the transfer. The target Q value of each experience tuple is calculated using the target Q network (which is a stable copy of the main Q network and is used to calculate the target value).

[0051] A2. Calculation of predicted Q value: Use the main Q network to forward propagate the current state-action pair (s, a) to obtain the predicted Q value. This value represents the main Q network's estimate of the expected return that can be obtained by taking a specific action in the current state.

[0052] A3. Mean square error calculation: Calculate the mean square error (MSE) between the predicted Q value and the target Q value as the loss function value of the current iteration. The calculation formula of MSE is:

[0053] Where N is the number of sampled experience tuples, is the target Q value, is the predicted Q value of the main Q network. MSE measures the accuracy of the main Q network's prediction. The smaller the value, the closer the prediction is to the true value.

[0054] A4. Parameter update: Based on the calculated MSE value, the gradient descent algorithm (such as Adam optimizer) is used to update the parameters of the main Q network, with the goal of minimizing the MSE value. Through the back propagation algorithm, the network weights are adjusted so that the predicted Q value is closer to the target Q value, thereby optimizing the performance of the main Q network.

[0055] A5, Iteration and round management: Repeat the above steps A1 to A4 until the end condition of the current training round is reached (such as reaching the preset number of iterations or the MSE value is lower than a certain threshold). After that, enter the next training round and continue the optimization process. At the end of each round, the parameters of the target Q network can be selectively updated to keep it synchronized with the main Q network to maintain the stability of the target value.

[0056] In order to further balance the exploration ability and learning stability of the model, the present invention also implements a dynamic adjustment mechanism of the learning rate α and the exploration rate ε: Monitor the MSE value trend: After each update of the main Q network parameters, the system continuously monitors the change of the MSE value. If the MSE value continues to increase or fluctuates too much, it indicates that the model may be in an unstable state, or the learning rate is too high to effectively converge.

[0057] Adjust the learning rate α: When the MSE value is abnormal, the system first tries to adjust the learning rate α. If the MSE value continues to increase, it may mean that the learning rate is too high and should be appropriately reduced; conversely, if the MSE value decreases slowly, it may mean that the learning rate is too low and should be appropriately increased.

[0058] Adjust the exploration rate ε: The system also pays attention to the adjustment of the exploration rate ε. In the early stage of learning, in order to encourage the model to explore more unknown states, the exploration rate should be set higher; as learning deepens, in order to stabilize the learning process and converge to the optimal strategy, the exploration rate should be gradually reduced. Therefore, when the MSE value fluctuates too much, you can also consider adjusting the exploration rate to balance the relationship between exploration and utilization.

[0059] The present invention also includes a method for controlling a mobile emotion relief cabin, the method comprising the following steps: S1: The environmental parameters and user physiological parameters in the mood relief cabin are monitored in real time through the environmental monitoring module, and the current state space of the mood relief cabin and its environment is defined, which includes temperature, humidity, light, sound intensity, air quality and user physiological indicators; S2: Set executable cabin environment control action space, including adjusting light color and brightness, playing specific music or sound effects, adjusting cabin temperature and humidity, releasing aromatherapy or negative ions, starting or shutting down cabin massage equipment, and adjusting seat angle; S3: Establish a rule evaluation component to quantitatively evaluate the results of the executed control actions based on the suitability of environmental parameters, the positivity of user physiological responses, energy efficiency ratio, and equipment operation status; S4: Build an environmental control strategy model and use reinforcement learning algorithms to select the optimal control action in a given state space; S5: After the training is completed, the obtained main Q network is used as the environmental control strategy model. The data collected by the environmental monitoring module is uploaded to the cloud server through the cloud control execution module. After being processed by the intelligent strategy model, the control instructions issued by the cloud are received; S6: According to the received control instructions, the actuators in the emotion relief cabin are driven to perform corresponding environmental adjustment actions to achieve intelligent regulation of the environment in the emotion relief cabin.

[0060] The implementation of this method refers to the above embodiment and will not be described in detail in the specification.

[0061] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0062] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The mobile emotional relief cabin control system is characterized by: The system comprises: Environmental monitoring module, used to monitor the environmental parameters and user physiological parameters in the emotional relief cabin in real time; An intelligent strategy building module is used to build an environmental control strategy model, which models the environmental management of the emotional relief cabin as a multi-objective optimization problem and dynamically adjusts the cabin environment settings through an algorithm to achieve the optimal state; the intelligent strategy building module further includes: The state description component is used to define the current state space of the mood relief cabin and its environment, including temperature, humidity, light, sound intensity, air quality, and user physiological indicators; Action definition component, used to set executable cabin environment control action space, including adjusting light color and brightness, playing specific music or sound effects, adjusting cabin temperature and humidity, releasing aromatherapy or negative ions, starting or shutting down cabin massage equipment, and adjusting seat angle; The rule evaluation component is used to quantitatively evaluate the results of control actions based on the suitability of environmental parameters, the positivity of user physiological responses, energy efficiency ratios, and equipment operating conditions; The strategy optimization unit uses the reinforcement learning algorithm to select the optimal control action in a given state space and continuously optimizes the control strategy through continuous learning iterations; The cloud control execution module is responsible for uploading the data collected by the environmental monitoring module to the cloud server. After being processed by the intelligent strategy model, it receives the control instructions issued by the cloud and drives the actuators in the emotion relief cabin to perform corresponding environmental adjustment actions.

2. The mobile emotional relief cabin control system according to claim 1, characterized in that: The state space further includes a user's emotion recognition result, which is obtained by analyzing the user's facial expression, voice intonation or physiological signal and serves as an input parameter of the environment control strategy.

3. The mobile emotional relief cabin control system according to claim 1, characterized in that: The implementation method of the rule evaluation component includes defining an evaluation function R(s, a), where s represents the current state space and a represents the control action taken; R(s, a) performs weighted summation according to the suitability S(s, a) of environmental parameters, the positivity P(s, a) of user physiological reactions, the energy efficiency ratio E(s, a) and the stability M(s, a) of equipment operation status to obtain a quantitative evaluation value of the control action; the specific algorithm expression is: R(s, a) = w1*S(s, a) + w2*P(s, a) + w3*E(s, a) + w4*M(s, a), where w1, w2, w3, w4 are weight coefficients of each evaluation dimension, and satisfy w1 + w2 + w3 + w4 = 1.

4. The mobile emotional relief cabin control system according to claim 1, characterized in that: It also includes a data acquisition module, which collects environmental parameters and user physiological parameters in real time through various sensors deployed in the emotion relief cabin, and combines the user's historical usage records and preference settings to form an initial training data set for preliminary model training of the intelligent strategy building module.

5. The mobile emotional relief cabin control system according to claim 4, characterized in that: The data acquisition module further utilizes user feedback and environmental adjustment records in historical data to construct a historical status data set for the use of the emotion relief cabin; for each record in the historical status data set, possible control actions are simulated and their corresponding expected benefits are calculated to form a state-action-benefit triple, which is added to the initial training data set to further enrich the training data set of the model.

6. The mobile emotional relief cabin control system according to claim 5 is characterized by: The strategy optimization unit uses a deep Q network DQN algorithm to train the environment control strategy model.

7. The mobile emotional relief cabin control system according to claim 6, characterized in that: The step of the strategy optimization unit training the environment control strategy model includes: Step 1: Initialize the experience replay memory and the target Q network, where the experience replay memory is used to store the experience of the state transfer of the emotion relief cabin, and the target Q network is used to provide a stable target value during the training process; at the same time, initialize the main Q network, whose input is the current state space of the emotion relief cabin and its environment, and whose output is the Q value estimation of each control action; Step 2: Set the learning rate α, discount factor γ, exploration rate ε, and number of training rounds, where α is used to control the learning speed of the model, γ is used to determine the discount degree of future rewards, ε is used to balance the ratio of exploring new actions and using known optimal actions, and the number of training rounds determines the total number of training iterations; Step 3: For each training round, perform the following steps in sequence: Step 3.1: Randomly select an initial state s from the training data set, which includes a combination of environmental parameters in the emotion relief cabin and user physiological parameters; Step 3.2: Select a control action a according to the current state s and the ε-greedy strategy, that is, randomly select an action for exploration with a probability of ε, and select the action with the largest Q value predicted by the current main Q network with a probability of 1-ε; Step 3.3: Execute the selected action a, observe the next state s' of the emotion relief cabin and the immediate reward r calculated by the rule evaluation component; Step 3.4: Store the experience tuple (s, a, r, s') into the experience replay memory for subsequent learning; Step 3.5: Randomly extract a batch of experience tuples from the experience replay memory to update the parameters of the main Q network; Step 3.6: Use the extracted experience tuples to calculate the target Q value in combination with the target Q network, and update the weights of the main Q network according to the loss function of the DQN algorithm. The loss function is defined as the mean square error between the predicted Q value and the target Q value. Step 3.7: Copy the parameters of the main Q network to the target Q network according to the preset update frequency to maintain the stability of the target value; Step 4: Repeat step S3 until the preset number of training rounds is reached or the output Q value of the main Q network tends to converge; Step 5: After the training is completed, the trained main Q network is used as the environmental control strategy model of the emotion relief cabin. For any given current state s of the emotion relief cabin and its environment, the Q value of each control action is calculated through the forward propagation of the main Q network, and the action a with the largest Q value is selected as the optimal control action.

8. The mobile emotional relief cabin control system according to claim 7, characterized in that: In each round of iteration of step 3, the mean square error (MSE) is used as the performance evaluation indicator of the current round of iteration. The specific implementation steps include: A1: Based on the sampled experience tuple (s, a, r, s'), the target Q network is used to calculate the target Q value of each experience tuple; A2: Calculate the predicted Q value of the main Q network for each state-action pair (s, a) in the experience tuple; A3: Calculate the mean square error (MSE) between the predicted Q value and the target Q value as the loss function value of the current iteration. The calculation formula of MSE is: Where N is the number of sampled experience tuples, is the target Q value, is the predicted Q value of the main Q network; A4: Based on the calculated MSE value, the parameters of the main Q network are updated using the gradient descent algorithm to minimize the MSE value, thereby optimizing the performance of the main Q network. A5: Repeat steps A1 to A4 until the end condition of the current training round is met, and then enter the next training round.

9. The mobile emotional relief cabin control system according to claim 8, characterized in that: After each update of the main Q network parameters, the changing trend of the MSE value is monitored. If the MSE value continues to increase or fluctuates too much, the learning rate α or the exploration rate ε is adjusted to balance the exploration ability and learning stability of the model.

10. A method for controlling a mobile mood relief cabin, characterized in that: The method comprises the following steps: S1: The environmental parameters and user physiological parameters in the mood relief cabin are monitored in real time through the environmental monitoring module, and the current state space of the mood relief cabin and its environment is defined, which includes temperature, humidity, light, sound intensity, air quality and user physiological indicators; S2: Set executable cabin environment control action space, including adjusting light color and brightness, playing specific music or sound effects, adjusting cabin temperature and humidity, releasing aromatherapy or negative ions, starting or shutting down cabin massage equipment, and adjusting seat angle; S3: Establish a rule evaluation component to quantitatively evaluate the results of the executed control actions based on the suitability of environmental parameters, the positivity of user physiological responses, energy efficiency ratio, and equipment operation status; S4: Build an environmental control strategy model and use reinforcement learning algorithms to select the optimal control action in a given state space; S5: After the training is completed, the obtained main Q network is used as the environmental control strategy model. The data collected by the environmental monitoring module is uploaded to the cloud server through the cloud control execution module. After being processed by the intelligent strategy model, the control instructions issued by the cloud are received; S6: According to the received control instructions, the actuators in the emotion relief cabin are driven to perform corresponding environmental adjustment actions to achieve intelligent regulation of the environment in the emotion relief cabin.

Images