Intelligent door and window control method and system

Abstract

The application provides a kind of intelligent door and window control method and system, it is related to the technical field of regulating intelligent door and window, the method comprises: adding at least one probability adjustment label for each control strategy;Set initial executable probability for the control strategy of each intelligent door and window, collect real-time data, the real-time data includes: real-time environmental data, real-time user data and the real-time state of each intelligent door and window, update the initial executable probability of the control strategy of each intelligent door and window based on real-time data, the updated initial executable probability is recorded as real-time executable probability, and the control strategy of real-time executable probability greater than preset probability threshold is executed, the application can evaluate the executable probability of each control strategy according to real-time environmental data and real-time user data, realizes the intelligent management and optimization control of intelligent door and window system.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of regulating intelligent doors and windows, and in particular to an intelligent door and window control method and system. BACKGROUND

[0002] With the rapid progress of technology and the continuous improvement of people's living standards, smart home systems have gradually become an important part of modern families. Smart home aims to provide users with a more convenient, comfortable, safe and energy-saving living environment by integrating various smart devices such as smart doors and windows, smart lighting, smart home appliances, etc. Among them, smart doors and windows, as an important part of the smart home system, not only can realize the basic opening and closing function, but also can be intelligently adjusted according to different environments and user needs, such as automatic opening and closing, adjusting the opening degree, etc., so as to improve the user's life quality.

[0003] The Chinese invention patent with the application publication date of February 13, 2024 provides a door and window control method and system for smart home, which obtains a door and window control instruction by calculating the time sequence difference feature vector of indoor and outdoor light intensity and the temperature difference feature vector.

[0004] However, the actual environment is constantly changing, and user needs may change at any time. The originally set door and window control instruction may not be suitable for execution when the user has a special need (such as needing a quiet environment for rest). Due to the lack of dynamic evaluation and adjustment of the execution probability of the control strategy, the above-mentioned patent cannot flexibly select a more appropriate control strategy according to real-time conditions, thereby affecting the control effect of the intelligent door and window and the user experience. SUMMARY

[0005] In order to be able to evaluate the executable probability of each control strategy according to real-time environmental data and real-time user data and improve user satisfaction, the application provides an intelligent door and window control method and system.

[0006] In a first aspect, the application provides an intelligent door and window control method, which adopts the following technical solution:

[0007] An intelligent door and window control method, comprising the following steps:

[0008] A database storing control strategies of each intelligent door and window is established, at least one probability adjustment label is added to each control strategy, the control strategy includes the opening mode, closing mode and opening degree of each intelligent door and window, and the probability adjustment label includes an environment label and a user label.

[0009] An initial executable probability is set for the control strategy of each smart door and window. Real-time data is collected, including real-time environmental data, real-time user data, and the real-time status of each smart door and window. The initial executable probability of the control strategy of each smart door and window is updated based on the real-time data. The updated initial executable probability is recorded as the real-time executable probability, and the control strategy with a real-time executable probability greater than a preset probability threshold is executed.

[0010] This application establishes a database storing control strategies, enabling centralized storage and management of all control strategies, facilitating rapid retrieval and invocation. Subsequently, this application adds environment and user tags to each control strategy, allowing for the updating of the initial executable probability of each strategy based on different environmental and user data. Next, this application sets an initial executable probability for each control strategy, providing a benchmark for subsequent real-time probability updates, enabling the control system to dynamically adjust the execution probability of control strategies according to actual conditions. Then, this application collects real-time environmental data, real-time user data, and the real-time status of each smart door and window to understand changes in the current environment and user needs. Based on this real-time data, the control system can update the initial executable probability of each control strategy to obtain the real-time executable probability. The collection and probability updates of real-time data allow the control system to adapt to constantly changing environments and user needs, improving its adaptability and robustness. Finally, this application only executes control strategies with a real-time executable probability greater than a preset probability threshold, ensuring that only control strategies with high execution value under the current environment and user needs are executed, thereby reducing resource waste caused by executing inefficient or ineffective control strategies. Meanwhile, this application improves user satisfaction by updating the initial executable probability of the control strategy using real-time data and filtering the control strategy based on the real-time executable probability.

[0011] Optionally, the method further includes:

[0012] A revenue function is constructed based on probability-adjusted labels. Control policies with a real-time executable probability greater than a preset probability threshold are denoted as target policies. The system then determines whether there are conflicting target policies.

[0013] If so, calculate the payoff value of each target strategy according to the payoff function, integrate the payoff values ​​of all target strategies into a payoff matrix, use the Nash equilibrium algorithm to obtain the target strategy combination that achieves the payoff equilibrium state in the payoff matrix, retain the target strategies contained in the target strategy combination, and delete the remaining target strategies.

[0014] If not, no action will be taken.

[0015] The application builds a benefit function based on the probability adjustment label, so that the execution effect of each control strategy can be quantitatively evaluated. The application also records the control strategy with a real-time executable probability greater than a preset probability threshold as a target strategy, and judges whether there is a mutually conflicting target strategy, which helps to discover and handle potential control conflicts in time, and tries to avoid resource waste or poor control effect caused by the execution of conflicting strategies. In the case of conflicting target strategies, the application calculates the benefit value of each target strategy according to the benefit function, integrates the benefit values of all target strategies into a benefit matrix, and uses the Nash equilibrium algorithm to obtain the target strategy combination when the benefit reaches the equilibrium state in the benefit matrix, which retains the target strategy combination that maximizes the overall benefit and deletes the remaining conflicting strategies. By applying the Nash equilibrium algorithm, the application can effectively solve the conflicts between conflicting target strategies, so that the executed control strategy can maximize the overall benefit.

[0016] Optionally, the method further comprises:

[0017] judging whether the number of mutually conflicting target strategies is greater than two,

[0018] if yes, recording the mutually conflicting target strategies as conflicting strategies, mapping all the conflicting strategies into qubits, constructing a superposition state based on the qubits, constructing a quantum state based on the superposition state, calculating the expected benefit of the quantum state by using a quantum algorithm, collapsing the quantum state corresponding to the maximum expected benefit into a conflicting strategy combination, and updating the conflicting strategies contained in the conflicting strategy combination to target strategies;

[0019] if no, no processing is performed.

[0020] If the number of conflicting target strategies is greater than two, it is proved that there are multiple conflicting control strategies for a certain intelligent door and window, at this time the conflicting strategies are mapped to qubits, and a superposition state is constructed based on the qubits, realizing the representation of the conflicting strategies under the quantum computing framework. The processing of conflicting strategies under the quantum computing framework can utilize the parallelism and advantages of quantum computing to improve the processing efficiency, especially when dealing with a large number of conflicting strategies. By adopting the above scheme, the present application can utilize the parallelism and advantages of quantum computing to process complex conflict problems. Subsequently, the present application calculates the expected return of the quantum state by using quantum algorithm, providing a quantitative basis for the combination optimization of conflicting strategies, and then evaluating the execution effect of various conflicting strategy combinations. Subsequently, the present application collapses the quantum state corresponding to the maximum expected return into a conflicting strategy combination, realizing the conversion from the quantum state to the specific conflicting strategy combination, and then updates the conflicting strategies contained in the conflicting strategy combination to the target strategy, which can select the optimal conflicting strategy combination during execution, thereby improving the overall control effect. The present application improves the efficiency and accuracy of processing complex conflicting strategies through the schemes of judging the number of conflicting strategies, quantum mapping, superposition state construction, quantum state expected return calculation, and quantum state collapse and strategy updating.

[0021] Optionally, the method further comprises:

[0022] The target strategy to be deleted and the conflicting strategy not contained in the conflicting strategy combination are recorded as deleted strategies, the average value of the real-time executable probability of the deleted strategies is calculated, and recorded as first data; the target strategy retained and the conflicting strategy contained in the conflicting strategy combination are recorded as retained strategies, and the average value of the real-time executable probability of the retained strategies is calculated, and recorded as second data;

[0023] It is judged whether the first data is greater than the second data, if yes, an alarm signal is sent, and if not, no processing is performed.

[0024] The present application records the target strategy to be deleted and the conflicting strategy not contained in the conflicting strategy combination as deleted strategies, records the target strategy retained and the conflicting strategy contained in the conflicting strategy combination as retained strategies, and calculates the average value of the real-time executable probability of the deleted strategies (first data) and the average value of the real-time executable probability of the retained strategies (second data), and then judges whether the first data is greater than the second data to detect whether there is an abnormal situation. If the first data (the average execution probability of the deleted strategies) is greater than the second data (the average execution probability of the retained strategies), it means that the deleted control strategy has a higher execution value, or the execution effect of the retained control strategy is not good, at this time an alarm signal needs to be sent to the relevant personnel to inform the relevant personnel that there is an abnormality.

[0025] Optionally, after the step of issuing the alarm signal, the method further comprises: gradually reducing the initial executable probability of the deleted strategy by a preset step size until the first data is not greater than the second data.

[0026] Optionally, the method further comprises:

[0027] calculating a partial derivative of the revenue function with respect to each probability adjustment label, inputting real-time environment data and real-time user data into the partial derivative to obtain a partial derivative value, integrating the partial derivative value into a Jacobian matrix;

[0028] based on elements in the Jacobian matrix, obtaining a sensitivity of the revenue function with respect to each probability adjustment label, respectively normalizing each sensitivity to obtain a normalization result, taking the normalization result as a probability compensation value of corresponding real-time environment data or corresponding real-time user data, and taking a sum of the real-time executable probability and the probability compensation value as a new real-time executable probability.

[0029] The present application calculates a partial derivative of the revenue function with respect to each probability adjustment label to obtain the sensitivity of the revenue function to each probability adjustment label. For example, if the partial derivative of a certain environment label (such as temperature) is large, it indicates that the temperature change has a significant impact on the revenue function, and the present application needs to focus on this label. Subsequently, the present application integrates the partial derivative value into a Jacobian matrix to intuitively display the sensitivity of the revenue function to all probability adjustment labels in the form of a matrix. Subsequently, the present application respectively normalizes each sensitivity to take the normalization result as a probability compensation value of corresponding real-time environment data or corresponding real-time user data, so that the present application can adjust the real-time executable probability according to the sensitivity. By dynamically adjusting the real-time executable probability, the present application can better adapt to the actual use habits and needs of the user, and improve the user's satisfaction and comfort. For example, when the user is accustomed to opening the door and window at a certain time period, the present application can automatically increase the real-time executable probability of the opening strategy in that time period. Taking the normalized sensitivity value as the probability compensation value directly reflects the adjustment range of the real-time environment data or the real-time user data to the real-time executable probability, which improves the rationality and effectiveness of the probability compensation value.

[0030] Optionally, the method further comprises:

[0031] determining whether the sensitivity is a non-negative number,

[0032] if yes, no processing is performed;

[0033] if no, the absolute value of the sensitivity less than zero is recorded as a new sensitivity.

[0034] Optionally, if the number of environment class labels and the number of user class labels are different, a digital zero is used to fill in the vacant positions in the Jacobian matrix.

[0035] Optionally, the method further comprises:

[0036] Obtaining historical user data, obtaining adjustment habits of the user for each smart door and window and adjustment time stamps based on the historical user data, and increasing a new real-time executable probability of a control strategy corresponding to the adjustment habits within a preset time period before the adjustment time stamps.

[0037] In a second aspect, the present application provides a smart door and window control system, which adopts the following technical solution:

[0038] A smart door and window control system, comprising a memory and a processor,

[0039] The memory stores a computer readable storage medium and training data;

[0040] The processor executes the steps of the method of the first aspect when calling the program code in the memory.

[0041] In summary, the beneficial technical effects of the present application are as follows:

[0042] The present application dynamically updates the executable probability of each control strategy through real-time data, so that the execution of each control strategy is more consistent with the actual situation, and the environmental adaptability and user satisfaction are improved. BRIEF DESCRIPTION OF DRAWINGS

[0043] Fig. 1 is a flowchart of embodiment 1 of the present application;

[0044] Fig. 2 is a flowchart of embodiment 2 of the present application;

[0045] Fig. 3 is a flowchart of embodiment 3 of the present application. DETAILED DESCRIPTION

[0046] The following will be combined Figs. 1 to 3 to further illustrate the present application.

[0047] Embodiment 1: The present embodiment discloses a smart door and window control method, referring to Fig. 1 , the method comprises: S11 processing strategy, S12 execution strategy, the present embodiment first establishes a database storing each smart door and window control strategy, and adds at least one probability adjustment label (covering environment and user labels) for each control strategy; then, the present embodiment sets an initial executable probability for each control strategy, and collects real-time data including real-time environment data, user data and door and window state, thereby dynamically updating the initial executable probability of the control strategy, obtaining a real-time executable probability, and executing the control strategy whose real-time executable probability exceeds a preset threshold, the present embodiment comprises the following steps:

[0048] S11 processing strategy, establishing a database storing the control strategy of each smart door and window, each smart door and window contains at least one control strategy, and the database records in detail the control strategy of each smart door and window, including the opening mode of each smart door and window (such as the inside / outside opening of casement window, the sliding opening of sliding window, the rotating opening of top-hung window, the rotating opening of bottom-hung window, the folding opening of folding window, the up-down sliding opening of lift and slide window, the center rotating opening of rotary window, etc.), the closing mode (such as the inside / outside closing of casement window, the sliding closing of sliding window, the rotating closing of top-hung window, the rotating closing of bottom-hung window, the folding closing of folding window, the up-down sliding closing of lift and slide window, the center rotating closing of rotary window, etc.) and the opening degree (such as full opening, half opening, micro opening, etc. specific angle or ratio).

[0049] At least one probability adjustment tag is added to each control strategy to dynamically adjust the execution probability of the strategy according to different environments and user needs, and the probability adjustment tag includes environment tags and user tags.

[0050] The environment tags include temperature, humidity, air quality and other environmental factors, which will affect the opening or closing decision of the smart door and window, as well as the opening degree of each smart door and window.

[0051] The user tags include user preferences, daily behavior patterns, family member composition, etc., which reflect the personalized needs of users in the use of smart doors and windows.

[0052] By adding probability adjustment tags to the control strategy of each smart door and window, this embodiment can dynamically adjust the execution probability of each smart door and window control strategy according to real-time environmental data, real-time user data and real-time state of each smart door and window, so as to realize more intelligent and personalized control.

[0053] S12 execution strategy, setting an initial executable probability for the control strategy of each smart door and window, which reflects the execution possibility of the control strategy under the real-time state of the smart door and window, and the initial executable probability can be set according to historical data, expert experience or system default value.

[0054] The scheme for setting the initial executable probability of a control strategy according to historical data is: by analyzing the usage frequency, time distribution of each control strategy in the past period of time (which can be one month or one week), as well as the user's calling frequency and calling time of the control strategy, the initial executable probability of the strategy is calculated.

[0055] For example, the probability of the user calling the control strategy of opening the bedroom window to 40° at 7:00 every morning in the past 30 days is 0.8, and the probability of non-user calling the control strategy is 0, then the initial executable probability of the control strategy can be set as (0.8+0) / 2=0.4.

[0056] The initial executable probability of a control strategy is set according to expert experience, that is, according to the building type, climate condition and user demand, the initial executable probability of the control strategy is set according to the expert knowledge in the field of architecture, environmental science or smart home.

[0057] For example, the expert suggests that the probability of closing the outside opening window of a high-rise building on a typhoon day is 1, and the initial executable probability of the strategy can be set to 1.

[0058] The initial executable probability of a control strategy is set according to system default value, that is, the initial executable probability of the control strategy is preset by the system developer according to general scenarios or industry standards.

[0059] For example, the system defaults the initial executable probability of the control strategy of "closing all windows at night" to 0.7.

[0060] The current environmental data, user data and real-time state of the smart door and window are obtained to dynamically adjust the execution probability of the strategy.

[0061] The real-time environmental data includes the current temperature, humidity, air quality, etc.

[0062] The real-time user data includes the current location and activity state of the user (which can be obtained through sensors in the smart home system or an Internet of Things terminal worn by the user, which can be a smart watch).

[0063] The real-time state of the smart door and window includes whether it is currently opened, the opening size, etc.

[0064] The initial executable probability of each control strategy is dynamically adjusted according to the real-time data to reflect the changes in the current environment and user demand.

[0065] The existing control strategies, the probability adjustment tags and the initial executable probabilities of each control strategy are shown in Table 1, the real-time data is temperature 24℃, humidity 30%, air quality good, the user is in a resting state, smart door and window one is in an opening state of 80°, smart door and window two is in a closed state, smart door and window three is in an open state of 20°, and smart door and window four is in a closed state.

[0066] Table 1: Control strategies and their probability adjustment tags and initial executable probabilities

[0067]

[0068] In this embodiment, the initial executable probability is updated by a voting method, and the update step is set to 0.1, and the method is as follows,

[0069] Taking the control strategy "casement window, open 20°" of the intelligent door and window one as an example, the real-time temperature, the real-time humidity and the air quality all meet the probability adjustment label of the intelligent door and window one, so the initial executable probability of the control strategy is adjusted from 0.2 to 0.5, that is, the real-time executable probability of the control strategy is 0.5.

[0070] Similarly, the real-time executable probability of the control strategy "casement window, close 10°" of the intelligent door and window one is 0.7.

[0071] The real-time executable probability of the control strategy "top-hung window, open 10°" of the intelligent door and window two is 0.5.

[0072] The real-time executable probability of the control strategy "top-hung window, close 5°" of the intelligent door and window two is 0.1.

[0073] The real-time executable probability of the control strategy "casement window, close 5°" of the intelligent door and window two is 0.1.

[0074] The real-time executable probability of the control strategy "casement window, open 20°" of the intelligent door and window two is 0.7.

[0075] The real-time executable probability of the control strategy "top-hung window, open 5°" of the intelligent door and window three is 0.4.

[0076] The real-time executable probability of the control strategy "top-hung window, close 5°" of the intelligent door and window three is 0.

[0077] The real-time executable probability of the control strategy "casement window, open 10°" of the intelligent door and window four is 0.5.

[0078] In other embodiments, the real-time executable probability can also be predicted by a machine learning model, and the process is as follows:

[0079] The control strategy, the initial executable probability and the historical data are taken as training data, and the historical executable probability is taken as a training label to train the machine learning model, and a trained machine learning model is obtained. The historical data includes historical environmental data, historical user data and historical state of each intelligent door and window. The historical executable probability refers to the probability value obtained by updating the initial executable probability using the historical data.

[0080] The control strategy, the initial executable probability and the real-time data are input into the trained machine learning model, and the predicted real-time executable probability is output.

[0081] The control strategy with the real-time executable probability exceeding a preset threshold (such as 0.5) is executed to realize the automatic control of the intelligent door and window.

[0082] By adopting the above scheme, the embodiment realizes fine management, dynamic adjustment and automatic execution of the control strategy, thereby improving the flexibility, adaptability and user experience of the control system.

[0083] Embodiment 2: Reference Fig. 2 The difference between the embodiment and the embodiment 1 is that before executing the control strategy with the real-time executable probability exceeding the preset threshold, the method further comprises:

[0084] S21 strategy judgment, a benefit function is constructed based on the probability adjustment label, the benefit function is used to quantify the execution effect of each control strategy under different environments and user demands, and the benefit function The calculation model of the benefit function is as follows,

[0085] ;

[0086] ;

[0087] Wherein, n is the number of probability adjustment labels of the s-th control strategy; is the satisfaction function of the i-th probability adjustment label .

[0088] In other embodiments, other types of functions can also be selected as the benefit function according to the demand.

[0089] The control strategy with the real-time executable probability greater than the preset probability threshold is marked as a target strategy, and the target strategy is a strategy considered to have a high execution value under the current environment.

[0090] Conflict detection is performed on the above target strategy to determine whether there are mutually conflicting target strategies. The conflict is caused by different operation requirements of different strategies on the same intelligent door and window, for example, one strategy requires opening the door and window for ventilation, and another strategy requires closing the door and window for heat preservation.

[0091] If there is a conflict, S22 quantity judgment is performed to further process the conflicting strategies.

[0092] If there is no conflict, no further processing is performed, and the target strategy meeting the condition is directly executed.

[0093] S22 quantity judgment, whether the number of mutually conflicting target strategies is greater than two,

[0094] If the number of conflicting strategies is greater than two, S23 quantum encoding is performed.

[0095] S23 quantum encoding, conflicting target strategies are marked as conflict strategies, and all conflict strategies are mapped to qubits. Qubits are the basic units of quantum computing, with superposition and entanglement states, and can represent multiple states simultaneously. For example, if there are four conflict strategies, each conflict strategy corresponds to a qubit, and the strategy state is represented in binary. |0> represents that the control strategy cannot be executed, and |1> represents that the control strategy can be executed. Therefore, four qubits are required to represent four conflict strategies, which are represented as |s1s2s3s4>.

[0096] Based on the above qubits, a superposition state is constructed, which can simultaneously consider all possible conflict strategy combinations. Then, a quantum state is constructed based on the superposition state, which is a collection of all possible states in a quantum system.

[0097] The expected return of the quantum state is calculated using quantum algorithms (such as quantum annealing algorithm, variational quantum eigensolver, etc.). The expected return reflects the average return of executing conflict strategy combinations under different quantum states. The expected return is calculated as follows,

[0098] ;

[0099] wherein, is the weight of energy saving return; A is the saved electricity, for example, by closing smart doors and windows when the temperature is low to save electricity when the indoor air conditioner is heating, etc. is the weight of user satisfaction; B is the difference between the user satisfaction value and the satisfaction threshold.

[0100] The quantum state corresponding to the maximum expected return is collapsed into a conflict strategy combination. The collapse process is the transition from a quantum state to a classical state, which determines the optimal conflict strategy combination under the current environment. The conflict strategies included in the conflict strategy combination are updated to target strategies, and then S24 strategy screening is performed.

[0101] If the number of conflict strategies is not greater than two, no further processing is performed.

[0102] S24 strategy screening, the return value of each new target strategy is calculated according to the return function. These return values reflect the execution effect of each new target strategy under the current environment. The return values of all new target strategies are integrated into a return matrix, and an element in the return matrix represents the return value of a new target strategy.

[0103] Nash equilibrium algorithm is used to find a new target strategy combination in the return matrix that makes the return reach an equilibrium state. Nash equilibrium is a concept in game theory, which means that in a game, each participant has chosen the optimal strategy, and no participant has the motivation to change their strategy alone to obtain higher returns.

[0104] Reserve the new target strategy contained in the new target strategy portfolio when the revenue reaches the equilibrium state.

[0105] Delete the remaining new target strategies that are not in the equilibrium strategy portfolio, as these strategies are not the best choice in the current environment.

[0106] S25, alarm, mark the new target strategies deleted in S24 strategy screening and the conflict strategies not contained in the conflict strategy portfolio as deleted strategies. These control strategies are considered not to be the best choice in the current environment, and are therefore excluded. Calculate the average of the real-time executable probabilities of the deleted strategies, denoted as the first data.

[0107] Mark the new target strategies retained in S24 strategy screening and the conflict strategies contained in the conflict strategy portfolio as retained strategies, and calculate the average of the real-time executable probabilities of the retained strategies, denoted as the second data.

[0108] Determine whether the first data is greater than the second data. If the first data is greater than the second data, it indicates that the average execution likelihood of the deleted strategies in the current environment is higher than that of the retained strategies, indicating that the initial executable probability of the deleted strategies is set too high, and the executable probability of the deleted strategies needs to be reduced at this time.

[0109] If the first data is not greater than the second data, it indicates that the average execution likelihood of the deleted strategies is not higher than that of the retained strategies, and the current strategy selection is reasonable, and no alarm processing is needed.

[0110] When the alarm signal is issued, the control system gradually reduces the initial executable probability of the deleted strategies according to a preset step size. By adjusting the initial execution likelihood of the strategy, the likelihood of its being incorrectly selected in the future is reduced. After adjusting the initial executable probability of the deleted strategies, the steps of processing the strategy S11 to S25 alarm are re-executed, and a new round of strategy screening, evaluation and adjustment is performed until the first data is no longer greater than the second data.

[0111] Embodiment 3: Refer to Fig. 3 The difference between this embodiment and embodiment 2 is that the method further comprises:

[0112] S31, calculate the sensitivity, use the partial derivative formula in mathematics to derive the revenue function, obtain the partial derivative expression about each probability adjustment label, input the real-time environmental data and real-time user data into the partial derivative expression, and calculate the partial derivative value of each probability adjustment label in the current environment. By calculating the partial derivative of the revenue function with respect to each probability adjustment label, this embodiment can understand the degree of influence of each label on the revenue function.

[0113] The partial derivative values of the reward function of the same control strategy with respect to each probability adjustment label are arranged as elements in the same row or the same column of the Jacobian matrix, and the complete Jacobian matrix is constructed according to the order of the labels, and the Jacobian matrix can more intuitively show the sensitivity of the reward function to all probability adjustment labels. The sensitivity of the reward function to each probability adjustment label is calculated in this embodiment. The sensitivity reflects the degree of response of the reward function to the change of the probability adjustment label.

[0114] S32 value judgment, judging whether the sensitivity is a non-negative number by traversing each element in the Jacobian matrix,

[0115] If yes, no processing is performed;

[0116] If no, the absolute value of the sensitivity less than zero is recorded as a new sensitivity, and S33 updating real-time executable probability is performed.

[0117] S33 updating real-time executable probability, each sensitivity is normalized by using linear normalization or minimum-maximum normalization, and the sensitivity value is mapped to [0, 1], and the normalization result is used as a probability compensation value of the corresponding real-time environmental data or the corresponding real-time user data to reflect the influence of these data on the real-time executable probability, and then the real-time executable probability of each control strategy is added to the corresponding probability compensation value to obtain a new real-time executable probability to realize dynamic adjustment.

[0118] In other embodiments, the method further comprises,

[0119] The adjustment habits of the user for each intelligent door and window and the adjustment time stamp are obtained from the historical user data. Within a preset time period before the adjustment time stamp, the new real-time executable probability of the control strategy corresponding to the adjustment habit is increased.

[0120] In other embodiments, the S31 calculating sensitivity further comprises: if the number of environmental class labels and user class labels is different, a digital zero is used to fill the missing positions in the Jacobian matrix.

[0121] By using the above scheme, the embodiment can more accurately evaluate the influence degree of each probability adjustment label on the reward function, and dynamically adjust the real-time executable probability of each intelligent door and window control strategy based on the sensitivity value, which helps to improve the intelligent level of the control system and the user experience, and makes it better adapt to the changing environment and user needs.

[0122] Embodiment 4: This embodiment discloses an intelligent door and window control system, the system comprising: a memory and a processor,

[0123] The memory stores a computer readable storage medium;

[0124] The processor processes the computer program stored on the computer readable storage medium to implement the method.

[0125] The above are preferred embodiments of the present application, and are not intended to limit the protection scope of the present application, and thus: any equivalent changes made according to the structure, shape, principle of the present application should be covered within the protection scope of the present application.

Claims

1. A smart door and window control method, characterized in that, The method comprises: establishing a database storing control strategies of each smart door and window, adding at least one probability adjustment label to each control strategy, the control strategy including the opening mode, closing mode and opening degree of each smart door and window, and the probability adjustment label including an environmental label and a user label; setting an initial executable probability for the control strategy of each smart door and window, collecting real-time data including real-time environmental data, real-time user data and real-time state of each smart door and window, updating the initial executable probability of the control strategy of each smart door and window based on the real-time data, taking the updated initial executable probability as a real-time executable probability, and executing the control strategy with a real-time executable probability greater than a preset probability threshold; constructing a benefit function based on the probability adjustment label, taking the control strategy with a real-time executable probability greater than the preset probability threshold as a target strategy, and judging whether there are mutually conflicting target strategies, if yes, calculating the benefit value of each target strategy according to the benefit function, integrating the benefit values of all target strategies into a benefit matrix, obtaining a target strategy combination in which the benefit reaches an equilibrium state in the benefit matrix by using a Nash equilibrium algorithm, retaining the target strategies included in the target strategy combination, and deleting the remaining target strategies; if no, no processing is performed; judging whether the number of mutually conflicting target strategies is greater than two, if yes, taking the mutually conflicting target strategies as conflict strategies, mapping all conflict strategies into qubits, constructing a superposition state based on the qubits, constructing a quantum state based on the superposition state, calculating the expected benefit of the quantum state by using a quantum algorithm, collapsing the quantum state corresponding to the maximum expected benefit into a conflict strategy combination, and updating the conflict strategies included in the conflict strategy combination to target strategies; if no, no processing is performed.

2. The intelligent door and window control method according to claim 1, wherein, The method further comprises: taking the deleted target strategies and the conflict strategies not included in the conflict strategy combination as deleted strategies, calculating the average of the real-time executable probabilities of the deleted strategies, and taking the average as first data; taking the retained target strategies and the conflict strategies included in the conflict strategy combination as retained strategies, calculating the average of the real-time executable probabilities of the retained strategies, and taking the average as second data; judging whether the first data is greater than the second data, if yes, an alarm signal is sent; if no, no processing is performed.

3. The intelligent door and window control method of claim 2, wherein, After the alarm signal is sent, the method further comprises: gradually reducing the initial executable probability of the deleted strategies by a preset step size until the first data is not greater than the second data.

4. The intelligent door and window control method according to any one of claims 1-3, characterized in that, The method further comprises: calculating the partial derivative of the benefit function with respect to each probability adjustment label, inputting the real-time environmental data and the real-time user data into the partial derivative to obtain a partial derivative value, and integrating the partial derivative value into a Jacobian matrix; based on the elements in the Jacobian matrix, obtaining the sensitivity of the benefit function with respect to each probability adjustment label, respectively normalizing each sensitivity to obtain a normalization result, taking the normalization result as a probability compensation value of the corresponding real-time environmental data or the corresponding real-time user data, and taking the sum of the real-time executable probability and the probability compensation value as a new real-time executable probability.

5. The intelligent door and window control method of claim 4, wherein, The method further comprises: judging whether the sensitivity is a non-negative number, if yes, no processing is performed; If not, the absolute value of the sensitivity less than zero is recorded as the new sensitivity.

6. The intelligent door and window control method of claim 5, wherein, If the number of environment class labels and user class labels is different, the number zero is used to fill the vacancy position in the Jacobian matrix.

7. The intelligent door and window control method of claim 4, wherein, The method further comprises: Obtaining historical user data, obtaining the adjustment habit of the user to each smart door and window and the adjustment timestamp based on the historical user data, and increasing the new real-time executable probability of the control strategy corresponding to the adjustment habit within a preset time period before the adjustment timestamp.

8. An intelligent door and window control system characterized in that, Comprise: A memory and a processor, The memory stores a computer readable storage medium; The processor processes the computer program stored on the computer readable storage medium to implement the method of any one of claims 1-7.

Images