Aromatherapy machine follow-up mediation method and system for user habit prediction

Abstract

The invention discloses an aromatherapy diffuser follow-up mediation method and system for user habit prediction, and relates to the technical field of user habit modeling, and the method comprises the steps: collecting historical operation data of a user using an aromatherapy diffuser, and constructing a scene habit model; acquiring a real-time sign and a space-time state of the user through a preset permission; matching the scene model according to a real-time state, determining a current scene, and extracting a basic fragrance and an output parameter space; extracting mental and physical sign characteristics, and deciding aromatherapy output parameters in a parameter space in combination with the model; therefore, follow-up adjustment of the aromatherapy machine is achieved. Therefore, the technical effects of improving the adjustment accuracy and the intelligent level of the aromatherapy machine and improving the user experience are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of user habit modeling, in particular to a user habit prediction aromatherapy machine follow-up mediation method and system. BACKGROUND

[0002] Traditional aromatherapy machines mainly rely on preset aroma release programs, or simple timing, fixed concentration and other single modes to operate, and cannot be flexibly adjusted according to the personalized needs of users at different times and in different scenes. There is a disconnection between the control strategy of the aromatherapy machine and the real-time needs of the user, so that the user cannot obtain the best fragrance experience in the use process, and the aroma cannot meet the user's expectations for intelligent and personalized control of the aromatherapy machine. SUMMARY

[0003] The present application provides a user habit prediction aromatherapy machine follow-up mediation method and system to solve the technical problems of low mediation adaptability and intelligence level and affecting user experience in the prior art, and to achieve the technical effects of improving the mediation accuracy and intelligence level of the aromatherapy machine and improving the user experience.

[0004] In a first aspect, the present application provides a user habit prediction aromatherapy machine follow-up mediation method, wherein the user habit prediction aromatherapy machine follow-up mediation method comprises: Collecting historical operation data of the user using the aromatherapy machine, and establishing a user scene habit model based on the historical operation data.

[0005] Obtaining user real-time state data in combination with preset permissions, wherein the user real-time state information includes real-time physical sign information and real-time space-time information.

[0006] Taking the user real-time state data as an index, traversing the user scene habit model for scene matching to determine the real-time user scene, and corresponding extraction of basic fragrance type information and basic output parameter space.

[0007] Performing feature extraction on the user real-time state data to obtain user mental feature items and user physical feature items, and performing output parameter decision in the basic output parameter space in combination with the user scene habit model to obtain aromatherapy output parameters.

[0008] Based on the basic fragrance type information and the aromatherapy output parameters, performing follow-up mediation of the aromatherapy machine.

[0009] In a feasible implementation manner, collecting historical operation data of the user using the aromatherapy machine, and establishing a user scene habit model based on the historical operation data, comprises: Performing one clustering on the historical operation data with fragrance preference as the target, and performing two clustering on the one clustering result with space-time features as the target.

[0010] According to the spatio-temporal characteristics, the secondary clustering results are classified and reconstructed to obtain a plurality of reconstructed operation data clusters.

[0011] Through correlation analysis, a set of typical influence factors of the plurality of reconstructed operation data clusters is screened and determined, and a plurality of user scene habit base models based on state transition probability are constructed correspondingly, wherein the typical influence factors are marked with influence severity coefficients.

[0012] In combination with preset selection and cutting constraints, statistical analysis is performed on the plurality of user scene habit base models to determine a plurality of scene output parameter spaces and associate storage.

[0013] In a feasible implementation manner, user real-time state data is obtained in combination with preset permissions, wherein the user real-time state information includes real-time physical sign information and real-time spatio-temporal information, and includes: The real-time physical sign information of the user is collected through a wearable device, wherein the real-time physical sign information at least includes heart rate, skin temperature and blood oxygen saturation.

[0014] The real-time spatio-temporal information of the user is collected through a positioning device and a people-in-sensor device, wherein the real-time spatio-temporal information at least includes geographic location information and indoor area location information.

[0015] In a feasible implementation manner, the user scene habit model is traversed for scene matching with the user real-time state data as an index to determine a real-time user scene, and corresponding basic fragrance type information and basic output parameter space are extracted, including: The user real-time state data is vectorized to obtain a comparison state vector.

[0016] In combination with the set of typical influence factors, a set of reference state vectors of the user scene habit model is generated.

[0017] Based on the comparison state vector, similarity calculation is performed on the set of reference state vectors, and the user scene habit base model corresponding to the reference state vector with the highest similarity is selected as the real-time user scene.

[0018] According to the real-time user scene, the historical fragrance type probability distribution is extracted as the basic fragrance type information, and the scene output parameter space is extracted as the basic output parameter space.

[0019] In a feasible implementation manner, user mental feature items and user physical sign feature items are obtained through feature extraction on the user real-time state data, and output parameter decision is performed on the basic output parameter space in combination with the user scene habit model to obtain fragrance output parameters, including: Take the user mental feature item and the user sign feature item as the attention factor, and perform scene factor dimension reduction on the user scene habit model.

[0020] Based on the scene factor dimension reduction result, calculate the relative deviation index of the output parameter in the real-time user scene, and obtain a relative deviation index set, wherein the relative deviation index is expressed in polar coordinate form.

[0021] According to the relative deviation index set, establish a user parameter deviation habit, and select fluctuations in the basic output parameter space according to the user parameter deviation habit, and obtain the aromatherapy output parameter.

[0022] In a feasible implementation manner, the aromatherapy machine follow-up mediation method for user habit prediction further includes: Collect the current physical environment information, including temperature, humidity, air quality and light intensity.

[0023] Based on the physical environment information and the preset environment correction model, the environmental influence correction of the aromatherapy output parameter is performed, and the environmental influence correction result is applied to the follow-up mediation of the aromatherapy machine.

[0024] In a feasible implementation manner, the aromatherapy machine follow-up mediation method for user habit prediction further includes: Record the user intervention log, and combine the accumulator to perform scene-based intervention accumulation.

[0025] According to the preset intervention correction constraint, the scene-based intervention accumulation result is combined to perform user habit correction discrimination, wherein the intervention correction constraint at least includes period constraint, accumulation constraint and frequency constraint.

[0026] If the user habit correction discrimination result is intervention, the habit correction of the user scene habit model is performed by the accumulator corresponding to the accumulation source data.

[0027] Secondly, the present application also provides an aromatherapy machine follow-up mediation system for user habit prediction, wherein the aromatherapy machine follow-up mediation system for user habit prediction includes: The habit model establishment module is used to collect the historical operation data of the user using the aromatherapy machine, and establish a user scene habit model based on the historical operation data.

[0028] The real-time state acquisition module is used to combine the preset permission to acquire user real-time state data, wherein the user real-time state information includes real-time sign information and real-time space-time information.

[0029] The scene matching and information extraction module is used for indexing the real-time state data of the user, traversing the user scene habit model to perform scene matching, determining a real-time user scene, and corresponding extraction of basic fragrance type information and a basic output parameter space.

[0030] The output parameter decision module is used for extracting features of the real-time state data of the user, obtaining user mental feature items and user physical feature items, and performing output parameter decision in the basic output parameter space in combination with the user scene habit model, to obtain aromatherapy output parameters.

[0031] The mediation execution module is used for performing follow-up mediation of the aromatherapy machine based on the basic fragrance type information and the aromatherapy output parameters.

[0032] The application discloses an aromatherapy machine follow-up mediation method and system based on user habit prediction, which comprises the following steps: collecting historical operation records of a user in the process of using an aromatherapy machine, and constructing a scene preference model of the user according to the historical operation records; after obtaining authorization of the user, collecting physiological feature data and current location information of the user in real time; taking the real-time state data as a retrieval index, performing matching in the scene preference model, identifying a current user scene, and extracting corresponding basic fragrance types and optional output parameter ranges; performing feature analysis on the real-time state data, obtaining user mental state features and physiological state features, and combining the scene preference model to make decisions on aromatherapy output parameters in the basic output parameter ranges; and driving the aromatherapy machine to perform dynamic mediation based on the selected basic fragrance types and the decided output parameters. The aromatherapy machine follow-up mediation method and system based on user habit prediction disclosed by the application solve the technical problems of low mediation adaptability and intelligent level and influence on user experience, realize the technical effect of improving the mediation accuracy and intelligent level of the aromatherapy machine, and improve the user experience. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 Fig. 1 is a flowchart of the aromatherapy machine follow-up mediation method based on user habit prediction.

[0034] Figure 2 Fig. 2 is a structural schematic diagram of the aromatherapy machine follow-up mediation system based on user habit prediction.

[0035] Reference signs in the drawings: The habit model establishment module 11, the real-time state acquisition module 12, the scene matching and information extraction module 13, the output parameter decision module 14, and the mediation execution module 15. DETAILED DESCRIPTION

[0036] The above technical solutions will be described in detail below in combination with the accompanying drawings and specific embodiments, so that the above technical solutions can be better understood. Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments for explaining the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application. In addition, it should be noted that, for the convenience of description, only the parts related to the present application are shown in the drawings, not all.

[0037] In some embodiments, the method comprises: Figure 1 The flowchart of a user habit prediction aromatherapy machine follow-up mediation method of the present application, wherein the user habit prediction aromatherapy machine follow-up mediation method comprises: S100: Collecting historical operation data of the user using the aromatherapy machine, and establishing a user scene habit model based on the historical operation data.

[0038] Specifically, the historical operation data refers to various operation records generated by the user during the use of the aromatherapy machine, including but not limited to the selected fragrance type, the aromatherapy concentration, the use time, the use location, etc. The user scene habit model is a pattern model (such as a mathematical or neural network model) that can reflect the user's habit of using the aromatherapy machine in different scenes, which is constructed by analyzing the historical operation data. The model can reflect the user's use habits and preferences, and then predict the user's possible needs in a specific scene according to the input data.

[0039] Specifically, first, the historical operation data of the user using the aromatherapy machine is collected, for example, the user usually selects lavender fragrance type and sets the concentration to medium when going to bed at night; while at work, the user selects lemon fragrance type and sets the concentration to low. Then, based on the obtained data, the user's behavior pattern is analyzed by clustering algorithm (such as K-means clustering) or graph neural network, and the user's behavior is classified according to time and space characteristics, such as clustering the user's behavior in the work scene (such as office) and the rest scene (such as bedroom), and further analyzing the typical characteristics in each cluster, such as fragrance preference, concentration preference, etc., to construct the user scene habit model.

[0040] Through the above-mentioned way, the user's habit preference in different scenes can be identified, and the user's use habit can be accurately grasped, which provides a basis for subsequent personalized aromatherapy control.

[0041] In some embodiments, collecting historical operation data of the user using the aromatherapy machine, and establishing a user scene habit model based on the historical operation data, comprises: The historical operation data is clustered once with the fragrance type preference as the target, and the once clustering result is clustered twice with the space-time feature as the target; the twice clustering result is classified and reconstructed according to the space-time feature, and a plurality of reconstructed operation data clusters are obtained; a plurality of typical influence factor sets of the reconstructed operation data clusters are determined through correlation analysis, and a plurality of user scene habit base models based on state transition probability are constructed correspondingly, wherein the typical influence factors are marked with influence severity coefficients; a plurality of the user scene habit base models are statistically analyzed in combination with a preset selection cutting constraint, a plurality of scene output parameter spaces are determined and are stored in association.

[0042] Specifically, the once clustering is a preliminary clustering of the historical operation data with the fragrance type preference as the target, for classifying the user's behavior according to the fragrance type preference; the twice clustering is a further clustering of the once clustering result with the space-time feature as the target, for subdividing the user's behavior according to the time and space features. The reconstructed operation data cluster is obtained by classifying and reconstructing the twice clustering result according to the space-time feature, and a plurality of more specific and subdivided user behavior patterns are obtained.

[0043] Specifically, the typical influence factor set is a factor set that has a greater influence on the user behavior pattern, which is screened out through correlation analysis, such as time, place, environmental state, mood, body state, etc. The user scene habit base model is a model constructed based on state transition probability, for describing the user's behavior habit and preference in different scenes.

[0044] Specifically, first, the historical operation data of the user using the aromatherapy machine is clustered once with the fragrance type preference as the target, i.e., the user's behavior is classified according to the fragrance type preference. For example, the user may prefer lemon fragrance when working, and prefer lavender fragrance when resting. Then, the once clustering result is clustered twice with the space-time feature as the target, so as to further subdivide the user's behavior pattern. For example, the user's behavior in the working scene (including multiple time points such as day shift, lunch break, night shift, and overtime work) and the resting scene (such as the evening of weekdays and the morning, noon, and evening of weekends in the living room) are clustered respectively. Then, the twice clustering result is classified and reconstructed according to the space-time feature, and a plurality of reconstructed operation data clusters are obtained. For example, the user's behavior in the working scene (including working day and lunch break) and the resting scene (including weekend morning, noon, and evening and weekday evening) are classified respectively.

[0045] Further, through correlation analysis, a set of typical influence factors of the multiple reconstruction operation data clusters is screened and determined, such as the time, place, environmental state, user mood, and user body state of the aromatherapy machine working, and multiple user scene habit base models based on state transition probability are constructed correspondingly. The model is used to represent the transition probability of the working strategy of the aromatherapy machine corresponding to different combinations of influence factors, which describes the habits of users. Different users include different combinations of influence factors and different transition probability distributions. Among them, the typical influence factors are marked with an influence severity coefficient. For example, the influence severity coefficient of fragrance preference on user behavior is 0.8, while the influence severity coefficient of time is 0.6.

[0046] Finally, the multiple user scene habit base models are statistically analyzed in combination with preset selection and cutting constraints to determine multiple scene output parameter spaces and associate storage. The selection and cutting constraints refer to a set of data preprocessing rules introduced to improve data quality and model stability before user behavior modeling or output parameter statistical analysis. The constraints are used to filter, standardize, or limit the data samples participating in the analysis. Typical operations include but are not limited to: Boundary cutting of serialized data, for example: removing 5% of the data segments at the beginning and end of each data sequence to avoid abnormal fluctuations during device startup and stop; setting a time window limit to only retain data within the last 7 days or 30 days; setting a user activity threshold to only select user samples with a daily average usage frequency greater than 2 times. This constraint mechanism can effectively improve the representativeness and consistency of data in the subsequent modeling process, avoiding extreme values or boundary effects that interfere with the model.

[0047] S200: Obtain user real-time state data in combination with preset permissions, wherein the user real-time state information includes real-time physical sign information and real-time space-time information.

[0048] Specifically, the preset permission refers to the permission granted by the user to the aromatherapy machine system to access their personal devices (such as wearable devices, positioning devices, etc.), which is used to obtain the user's real-time state data. Real-time physical sign information refers to the user's current physical physiological data, such as heart rate, skin temperature, and blood oxygen saturation. These data are usually collected through wearable devices (such as smart watches, fitness bands, etc.). Real-time space-time information refers to the user's current time and spatial location information, such as geographic location (through GPS positioning) and indoor area location (through indoor positioning devices or sensors).

[0049] Specifically, the user's explicit authorization is first required to ensure the legality and security of data collection. For example, the user can authorize the aromatherapy machine system to access their smart watch health data (real-time physical sign information) and phone positioning function (real-time space-time information) through the phone application. After the user's authorization, the user's real-time physical sign information, such as heart rate, skin temperature, and blood oxygen saturation, can be collected through wearable devices.

[0050] At the same time, by positioning devices and indoor sensors, real-time spatio-temporal information of the user is collected, such as the current geographical location (e.g. in the office or at home) and the indoor area location (e.g. in the living room or bedroom) of the user, thereby providing an important basis for subsequent scene matching and aromatherapy parameter adjustment.

[0051] For example, when the user's heart rate is high (e.g. in a stressful work state), the aromatherapy machine can adjust the fragrance type to have a soothing effect (e.g. lavender fragrance type) and appropriately increase the fragrance concentration according to the user's real-time state, helping the user to relax. At the same time, according to the environment in which the user is located (e.g. office or bedroom), the aromatherapy machine can adjust the fragrance concentration and fragrance type to adapt to different scene requirements.

[0052] Through the above process, combined with the pre-set permission to obtain the real-time state data of the user, the aromatherapy machine can more comprehensively understand the current physical condition of the user and the environment in which the user is located. The acquisition of such real-time data enables the aromatherapy machine to dynamically adjust according to the real-time needs of the user, thereby providing more accurate and personalized aromatherapy services.

[0053] In some embodiments, the real-time state data of the user is obtained in combination with the pre-set permission, wherein the real-time state information of the user includes real-time physiological information and real-time spatio-temporal information, including: The real-time physiological information of the user is collected by a wearable device, wherein the real-time physiological information at least includes heart rate, skin temperature and blood oxygen saturation; the real-time spatio-temporal information of the user is collected by a positioning device and a human body sensing device, wherein the real-time spatio-temporal information at least includes geographical location information and indoor area location information.

[0054] Specifically, the real-time state information of the user refers to a data set that can reflect the current physiological condition of the user and the state of the environment in which the user is located, which is dynamically collected from external perception devices based on user authorization or pre-set permission strategies during the operation of the device. The state information mainly includes two types: Real-time physiological information, i.e. the current physical state parameters of the user obtained by a wearable device, such as heart rate, skin temperature and blood oxygen saturation, used to reflect the physiological activity or stress level of the user; real-time spatio-temporal information, i.e. the spatial location information of the user obtained by a positioning device and an indoor human body sensing device, including macro geographical location (e.g. city, building) and micro indoor area location (e.g. bedroom, living room, study room, etc.), used to identify the life or work scene in which the user is located.

[0055] Specifically, first, the physiological parameters of the user are collected in real time by a wearable device (such as a smart bracelet, a smart watch or a health monitoring patch) bound to the user. For example, the built-in photoplethysmogram (PPG) sensor in the smart bracelet can be used to monitor heart rate and blood oxygen saturation, and a thermistor or thermocouple can be used to detect skin temperature. The obtained data can be automatically uploaded to the aromatherapy control system through Bluetooth or Wi-Fi.

[0056] At the same time, the geographical location information of the user is obtained through a GPS module or a Wi-Fi positioning module, which is used to determine whether the user is in a home, office or outdoor environment; and in combination with indoor installed human sensing devices (such as infrared sensors, millimeter wave radars, ultrasonic sensors, etc.), the specific location area of the user in the indoor environment is further identified, such as whether the user is resting in the bedroom, working in the study or moving in the living room.

[0057] For example, when it is detected that the user's heart rate is low, the skin temperature is moderate, the blood oxygen is normal, and the positioning information shows that the user is in the bedroom area, it can be determined that the user is in a "night rest" state, and thus an appropriate soothing fragrance type is selected and output at a low intensity.

[0058] Through the above process, multi-dimensional perception of the current physiological state of the user and the environmental scenario can be achieved, thereby establishing an intelligent linkage mechanism between the output control of the aromatherapy and the state of the user. Compared with the traditional aromatherapy control mode based on time scheduling or manual setting by the user, by collecting and fusing the sign information and the spatial information in real time, the fragrance type selection and the output parameters are more personalized and context-adaptive, which helps to improve the comfort and satisfaction of the user experience, enhance the intelligent level and response ability of the device, and provide a more forward-looking control strategy basis for the aromatherapy intelligent device.

[0059] S300: indexing with the real-time state data of the user, traversing the user scenario habit model to perform scenario matching, determining the real-time user scenario, and correspondingly extracting the basic fragrance type information and the basic output parameter space.

[0060] Specifically, the real-time state data of the user will be used as the key information for querying, to find a matching scenario in the user scenario habit model. By comparing the real-time state data of the user with the historical data in the user scenario habit model, the most similar scenario mode that best reflects the user's demand can be found.

[0061] Specifically, the basic fragrance type information refers to the fragrance preference information associated with the matched real-time user scenario, and preferably, the fragrance information is obtained according to the historical selection of the user in a similar scenario. The basic output parameter space refers to the range of the output parameters of the aromatherapy machine associated with the matched real-time user scenario, such as the concentration of the aromatherapy and the release frequency, which can also be obtained according to the historical selection of the user in a similar scenario.

[0062] By matching the user real-time state data as an index through the above process, the current scene in which the user is located can be quickly and accurately determined, and the basic fragrance type information and the basic output parameter space associated with the scene are extracted, which helps the subsequent aromatherapy machine to provide more accurate and personalized aromatherapy services according to the current state and historical habits of the user.

[0063] In some embodiments, the user real-time state data is indexed to traverse the user scene habit model for scene matching, determine the real-time user scene, and correspondingly extract the basic fragrance type information and the basic output parameter space, including: The user real-time state data is vectorized to obtain a comparison state vector; a reference state vector set of the user scene habit model is generated in combination with the set of typical influence factors; based on the comparison state vector, similarity calculation is performed by traversing the reference state vector set, and the user scene habit base model corresponding to the reference state vector with the highest similarity is selected as the real-time user scene; the historical fragrance type probability distribution is extracted as the basic fragrance type information according to the real-time user scene, and the scene output parameter space is extracted as the basic output parameter space.

[0064] Specifically, the set of typical influence factors refers to a set of key variables used to represent the mapping relationship between the user state and the scene during the construction of the user scene habit model. For example, at least the user physiological state parameters (such as heart rate, skin temperature, and blood oxygen saturation), spatial position information (such as geographic location and indoor area), and time information (such as current time period and day of the week) are included. In the model training stage, a plurality of reference state vectors can be constructed based on the set of influence factors, wherein each vector represents the state characteristics of a typical user scene, forming a reference state vector set. The comparison state vector refers to the expression form of the user real-time state data collected at the current moment after being vectorized in a unified format, which is used for similarity comparison with the reference state vector.

[0065] Specifically, in actual operation, first, the user real-time state data collected (such as heart rate 72bpm, skin temperature 36.2℃, blood oxygen 97%, current location as a study, and time as afternoon) is normalized and encoded to form a unified dimension comparison state vector, and each dimension in the vector represents a normalized influence factor. Subsequently, the pre-trained user scene habit model is called to extract the reference state vector set corresponding to each typical scene, such as morning wake-up, night relaxation, and work concentration, wherein each reference vector is also a normalized multi-dimensional vector.

[0066] Further, similarity calculation is performed on the contrast state vector and each reference state vector, such as using cosine similarity as a similarity measure, and the user scene habit base model corresponding to the reference state vector with the highest similarity is selected as the current real-time user scene.

[0067] Next, the historical fragrance type probability distribution of the user scene habit base model (such as 60% mint, 30% citrus, and 10% other) is extracted as the basic fragrance type information, and the corresponding output parameter space (such as the output intensity range of 40%-60% and the intermittent period of 3-5 minutes) is extracted as the basic output parameter space.

[0068] Through the above process, accurate mapping from real-time user state data to personalized fragrance type control parameters can be achieved, ensuring that the fragrance type selection and output strategy are highly matched with the current state and behavior scene of the user. By introducing vector modeling and similarity matching mechanism, the accuracy of user state recognition and the intelligence level of fragrance type control can be significantly improved. Compared with traditional fixed rule-based or manually set aromatherapy device control methods, this method has stronger adaptability, scalability, and personalized service capability, facilitating continuous learning of user preferences and dynamic optimization of control strategy, improving user experience and device intelligence level.

[0069] S400: Feature extraction is performed on the real-time user state data to obtain user mental feature items and user physical feature items, and output parameter decision is made in the basic output parameter space combined with the user scene habit model to obtain the aromatherapy output parameters.

[0070] Specifically, the user mental feature items and the user physical feature items are two types of key feature parameters extracted from the real-time user state data for assisting output parameter decision, wherein the mental feature items are used to reflect the current emotional or cognitive state of the user, such as stress level, attention level, emotional trend, etc., which can be inferred by electroencephalogram (EEG), facial expression recognition, voice tone analysis, or heart rate variability (HRV) etc.; the physical feature items refer to the physiological state parameters of the user, such as heart rate, skin temperature, blood oxygen saturation, respiratory rate, etc., which are usually obtained by wearable devices or environmental sensors, and are used to reflect the user's body movement state, such as resting, intense exercise, etc.

[0071] By extracting these two types of features and combining the current determined user scene and its basic output parameter space, output parameter optimization decision can be performed to generate aromatherapy output parameters that are more consistent with the current physical and mental state of the user.

[0072] In some embodiments, the user real-time state data is feature extracted to obtain user mental feature items and user physical feature items, and the user scene habit model is combined with the basic output parameter space to make output parameter decisions to obtain the aromatherapy output parameters, including: The user scene habit model is subjected to scene factor dimension reduction with the user mental feature items and the user physical feature items as the focus factors; based on the scene factor dimension reduction result, a relative deviation index of the output parameter in the real-time user scene is calculated to obtain a relative deviation index set, wherein the relative deviation index is in polar coordinate form; and based on the relative deviation index set, a user parameter deviation habit is established, and fluctuation selection is made in the basic output parameter space according to the user parameter deviation habit to obtain the aromatherapy output parameters.

[0073] Optionally, the user mental feature items generally include user current emotional state, attention level, stress index and the like obtained by analyzing multi-modal sensing data; and the user physical feature items include physiological parameters of the user, such as heart rate, skin electric response, body temperature, blood oxygen saturation and the like. The above-mentioned feature items are high-dimensional representations of the current user state, and can reflect the immediate reaction and preference trend of the user in a specific scene.

[0074] Specifically, the relative deviation index in polar coordinate form refers to the trend of the current user features deviating from the basic output parameter space in terms of angle (direction) and amplitude (degree), and is used to describe the adjustment direction and intensity of the aromatherapy output in the current state. For example, the angle can represent the preference change direction (such as being inclined to fresh, warm, and soothing fragrance dimensions), and the amplitude represents the deviation degree from the basic output (such as increasing the output intensity by 20%, prolonging the cycle by 30%, etc.).

[0075] Specifically, first, the current user mental feature items and physical feature items are used as focus factors to perform dimension reduction processing on various scene factors in the user scene habit model, so as to focus on the most influential dimensions in the current state, for example, from the original 12-dimensional scene factors to 3-5-dimensional core factor groups. Illustratively, this process can use principal component analysis (PCA), linear discriminant analysis (LDA), or self-attention mechanism for feature weight evaluation and screening.

[0076] Specifically, then, based on the dimension-reduced factor groups, the relative deviation values between each output parameter (such as aromatherapy intensity, output cycle, and fragrance switching frequency) in the current state and the basic output parameter are calculated, and are constructed into a deviation index set in polar coordinate form. For example, the fragrance deviation angle is 45° (towards soothing type), the output intensity deviation amplitude is +20%, and the cycle deviation is -10%.

[0077] Further, based on the above-mentioned deviation index set, the user parameter deviation habit is established, that is, the adjustment tendency model of the user to the aromatherapy output parameter in a specific mental state or physical state, and the corresponding selection basis (i.e., the habit of parameter setting) of the aromatherapy output parameter is selected.

[0078] For example, the user tends to increase the output intensity by 15% and shorten the cycle by 10% in a high-pressure state, and the corresponding deviation habit is used as a dynamic adjustment weight to superimpose on the basic output parameter space for fluctuation selection, and finally generate the aromatherapy output parameter in the current scene. For example, the fluctuation selection can take the spatial center of the basic output parameter space as the reference point, and define the fluctuation probability and range in different directions based on the polar coordinate parameters of the user parameter deviation habit.

[0079] Through the above process, the current mental and physical state of the user can be integrated on the basis of the static basic parameter space to realize dynamic personalized adjustment of the aromatherapy output parameter. This method not only improves the personalization of user experience, but also enhances the response ability of the system to the change of the user state, and reflects the intelligent regulation and control characteristics of the aromatherapy according to the user state. In other words, by introducing the deviation index and the user deviation habit model, soft compensation type output adjustment can be realized.

[0080] S500: Based on the basic fragrance information and the aromatherapy output parameter, the follow-up adjustment of the aromatherapy machine is performed.

[0081] Specifically, based on the basic fragrance information and the generated aromatherapy output parameter, the follow-up adjustment operation of the aromatherapy machine is performed to realize dynamic control of the device working state. The working state of the aromatherapy machine includes but is not limited to: Fragrance switching, that is, according to the target fragrance information matched according to the current user state, the fragrance channel or fragrance module inside the aromatherapy machine is controlled to switch, realizing dynamic adjustment of the fragrance type; Aromatherapy output intensity adjustment, that is, according to the intensity index in the output parameter, the concentration or spraying rate of the aromatherapy release is adjusted to meet the user's current olfactory comfort or adjustment requirement; Running time control, that is, according to the user scene or physiological rhythm, the duration or intermittent cycle of the aromatherapy output is set to realize rhythmic or phased aromatherapy release; Aromatherapy module start-stop, that is, in a specific state (such as user leaving, sleep wake-up, etc.), a module start-stop control instruction is issued to control the start or stop of the aromatherapy module, to protect the user safety and device efficiency.

[0082] Through the above follow-up adjustment mechanism, real-time response and personalized adjustment of the aromatherapy device to the user state can be realized, thereby improving the intelligent degree of aromatherapy use and user experience.

[0083] In some embodiments, the user habit prediction-based aromatherapy machine follow-up adjustment method further comprises: collecting current physical environment information, the physical environment information including temperature, humidity, air quality and light intensity; based on the physical environment information and a preset environment correction model, performing environment influence correction on the aromatherapy output parameter, and applying the environment influence correction result to perform follow-up adjustment of the aromatherapy machine.

[0084] Specifically, the physical environment information refers to the environmental parameter data collected by the aromatherapy machine body or its matching sensing module in real time. For example, the physical environment information includes temperature (unit: ℃), humidity (unit: %RH), PM2.5, TVOC, C density, light intensity (unit: Lux), etc.

[0085] Specifically, the environment correction model is a calculation model for evaluating the influence degree of physical environmental factors on the diffusion effect of aromatherapy and user perception, and correcting the aromatherapy output parameter accordingly. The model can be constructed based on empirical rules, experimental data fitting or machine learning algorithms. At the same time, the output of the model can be one or more correction factors for adjusting the strength, duration or fragrance selection of the aromatherapy output, etc.

[0086] Specifically, during the operation of the aromatherapy machine, the current environmental data is collected in real time by the built-in or external sensor. For example, the current environmental temperature is 30℃, the humidity is 85%, the PM2.5 concentration is 80μg / m³, and the light intensity is 100Lux. Then, the set of physical environment data is input into the preset environment correction model, and the corresponding aromatherapy diffusion correction factor is output by the model.

[0087] For example, the environmental temperature is 30℃ which is too high → the fragrance volatilization is accelerated → the output strength is adjusted downward; the humidity is 85% which is too high → the fragrance diffusion is blocked → the output time is extended; the air quality is poor → the fresh or purification type of fragrance is recommended; the light intensity is low → it is presumed to be night or rest scene → the output strength is appropriately reduced, and the following correction suggestions can be output accordingly: The fragrance type is switched to "mint purification"; the output strength is adjusted from 60% to 45%; the running time is extended by 2 minutes.

[0088] Through the above process, the physical state of the current environment can be dynamically perceived, and the aromatherapy output parameter can be adjusted in real time based on the environment correction model, so as to improve the effectiveness of fragrance release and the consistency of user perception. Compared with the traditional static output strategy, the above environment influence correction mechanism provided by the present application can significantly enhance the adaptability of the aromatherapy machine under different environmental conditions, reduce the problems of too strong, too weak or uneven diffusion of fragrance, and improve the comfort and intelligent level of the use of aromatherapy.

[0089] In some embodiments, the user habit prediction and follow-up adjustment method of the aromatherapy machine further comprises: record the user intervention log, and combine the accumulators to perform scenario-based intervention accumulation; according to a preset intervention correction constraint, combine the scenario-based intervention accumulation result to perform user habit correction discrimination, wherein the intervention correction constraint at least includes a period constraint, an accumulation constraint and a frequency constraint; if the user habit correction discrimination result is intervention required, then call the accumulation source data corresponding to the accumulator to correct the user scene habit model.

[0090] Specifically, the user intervention log refers to the operation record generated when the user actively adjusts the aromatherapy output parameters (including fragrance selection, output intensity, running time, etc.) during the operation of the device. The log usually includes operation time, operation type, parameter change value before and after operation, current environment state and user state, etc.

[0091] Specifically, scenario-based intervention accumulation refers to the process of classifying and accumulating similar intervention behaviors based on the context scenarios (such as time period, environmental parameters, user state, etc.) in which the user intervention behaviors occur. This process is implemented through accumulators. Optionally, the accumulators can maintain corresponding intervention behavior counters, time window sliding statistics or trend trackers for each typical scenario.

[0092] Specifically, the intervention correction constraint is used to determine the judgment condition for triggering user habit model correction, and at least includes the following three types: period constraint, used to determine whether the intervention behavior presents a periodic repetition feature; accumulation constraint, used to determine whether the cumulative frequency or intensity of intervention behavior within a set time window exceeds a threshold; frequency constraint, used to determine whether the intervention frequency of the user in similar scenarios is higher than a set frequency threshold.

[0093] For example, according to the frequency constraint, similar intervention occurs in a certain scenario for three consecutive days, and the intervention direction is consistent (for example, all reducing the output intensity or switching to a soothing fragrance), which will trigger intervention correction discrimination, in other words, the intervention correction constraint corresponding to the current user habit model is considered to be unable to accurately reflect the user's real preferences, and needs to be corrected.

[0094] Further, the intervention source data in this scenario is called to adjust the fragrance priority and output intensity parameters in the original user habit model for the corresponding scenario, so that they are closer to the user's actual preferences.

[0095] Through the above process, a behavior feedback closed loop can be constructed based on the user's actual intervention behavior, and dynamic self-adaptive correction of the user habit model can be realized in the long-term use process. Further, the user preference change trend can be more accurately captured in the long-term use, and the individualization matching degree and user satisfaction of the aromatherapy output can be improved.

[0096] In summary, the user habit prediction aromatherapy machine follow-up mediation method provided by the present application has the following technical effects: By collecting the historical operation records of the user during the use of the aromatherapy machine, and constructing the scene preference model of the user accordingly; after obtaining the authorization of the user, the physiological feature data and the current location information of the user are collected in real time; the real-time state data is taken as a retrieval index, and matching is performed in the scene preference model to identify the current user scene, and the corresponding basic fragrance type and optional output parameter range are extracted; the real-time state data is analyzed for features to obtain the mental state features and the physiological state features of the user, and the scene preference model is combined to make decisions on the output parameters of the aromatherapy machine within the basic output parameter range; based on the selected basic fragrance type and the decided output parameters, the aromatherapy machine is driven for dynamic adjustment, thereby achieving the technical effects of improving the adjustment accuracy and the intelligent level of the aromatherapy machine and improving the user experience.

[0097] In some embodiments, the method comprises: Figure 2 is a structural schematic diagram of a user habit prediction aromatherapy machine follow-up adjustment system of the present application. For example, Figure 1 The flowchart of the method can be implemented by the structure as shown in Figure 2 .

[0098] Based on the same idea as the method for predicting the habits of a user and adjusting the aromatherapy machine, the present application also provides a system for predicting the habits of a user and adjusting the aromatherapy machine, which comprises: A habit model establishing module 11 is configured to collect historical operation data of the user using the aromatherapy machine, and establish a user scene habit model based on the historical operation data.

[0099] A real-time state obtaining module 12 is configured to obtain real-time state data of the user in combination with a preset permission, wherein the real-time state information of the user includes real-time physical sign information and real-time space-time information.

[0100] A scene matching and information extracting module 13 is configured to take the real-time state data of the user as an index, traverse the user scene habit model for scene matching, determine the real-time user scene, and correspondingly extract basic fragrance type information and a basic output parameter space.

[0101] An output parameter decision module 14 is configured to extract features from the real-time state data of the user to obtain user mental feature items and user physical feature items, and make decisions on the output parameters of the aromatherapy machine in the basic output parameter space in combination with the user scene habit model, to obtain the output parameters of the aromatherapy machine.

[0102] An adjustment executing module 15 is configured to perform follow-up adjustment of the aromatherapy machine based on the basic fragrance type information and the output parameters of the aromatherapy machine.

[0103] In some embodiments, the habit model establishing module 11 comprises: The clustering unit is configured to cluster the historical operation data once with fragrance preference as a target, and to cluster the once clustering result twice with space-time characteristics as a target.

[0104] The reconstructed operation data cluster acquisition unit is configured to classify and reconstruct the twice clustering result according to the space-time characteristics, and to acquire a plurality of reconstructed operation data clusters.

[0105] The typical influence factor set screening and scenario habit base model construction unit is configured to screen and determine a plurality of typical influence factor sets of the reconstructed operation data clusters through correlation analysis, and to construct a plurality of user scenario habit base models based on state transition probability correspondingly, wherein the typical influence factors are marked with influence severity coefficients.

[0106] The scenario output parameter space determination and association storage unit is configured to statistically analyze a plurality of the user scenario habit base models in combination with a preset selection and cutting constraint, to determine a plurality of scenario output parameter spaces and to store in association.

[0107] In some embodiments, the real-time state acquisition module 12 comprises: The real-time physical sign information acquisition unit is configured to acquire the real-time physical sign information of the user through a wearable device, wherein the real-time physical sign information at least includes heart rate, skin temperature and blood oxygen saturation.

[0108] The real-time space-time information acquisition unit is configured to acquire the real-time space-time information of the user through a positioning device and a human-in-sensor device, wherein the real-time space-time information at least includes geographic location information and indoor area location information.

[0109] In some embodiments, the scenario matching and information extraction module 13 comprises: The user real-time state data vectorization unit is configured to vectorize the user real-time state data, and to acquire a comparison state vector.

[0110] The reference state vector set generation unit is configured to generate a reference state vector set of the user scenario habit model in combination with the typical influence factor set.

[0111] The real-time user scenario determination unit is configured to perform similarity calculation by traversing the reference state vector set based on the comparison state vector, and to select the user scenario habit base model corresponding to the reference state vector with the highest similarity as a real-time user scenario.

[0112] The basic fragrance type information and basic output parameter space extraction unit is configured to extract the historical fragrance probability distribution as the basic fragrance type information and the scenario output parameter space as the basic output parameter space correspondingly according to the real-time user scenario.

[0113] In some embodiments, the output parameter decision module 14 comprises: a scene factor dimension reduction unit, configured to perform scene factor dimension reduction on the user scene habit model by taking the user mental feature item and the user physical feature item as the concerned factors.

[0114] a relative deviation index calculation and acquisition unit, configured to calculate a relative deviation index of the output parameter in a real-time user scene based on the scene factor dimension reduction result, and acquire a set of relative deviation indexes, wherein the relative deviation index is in polar coordinate form.

[0115] a fragrance output parameter acquisition unit, configured to establish a user parameter deviation habit according to the set of relative deviation indexes, and select fluctuations in the base output parameter space according to the user parameter deviation habit to acquire the fragrance output parameter.

[0116] Further, the user habit prediction-based aromatherapy machine follow-up adjustment system further comprises: a physical environment information acquisition unit, configured to acquire current physical environment information, wherein the physical environment information includes temperature, humidity, air quality, and light intensity.

[0117] an environment influence correction unit, configured to perform environment influence correction on the fragrance output parameter based on the physical environment information and a preset environment correction model, and apply the environment influence correction result to perform follow-up adjustment of the aromatherapy machine.

[0118] Further, the user habit prediction-based aromatherapy machine follow-up adjustment system further comprises: a user intervention log recording and accumulation unit, configured to record a user intervention log and perform scenario-based intervention accumulation in combination with an accumulator.

[0119] a user habit correction discrimination unit, configured to perform user habit correction discrimination in combination with a scenario-based intervention accumulation result according to a preset intervention correction constraint, wherein the intervention correction constraint at least includes a periodic constraint, an accumulation constraint, and a frequency constraint.

[0120] a user scene habit model correction unit, configured to perform habit correction on the user scene habit model by using accumulator corresponding accumulation source data if the user habit correction discrimination result indicates that intervention is needed.

[0121] It should be understood that the embodiments mentioned in the specification focus on their differences from other embodiments, and the specific embodiments in the foregoing embodiment one are also applicable to the user habit prediction-based aromatherapy machine follow-up adjustment system in embodiment two. For the sake of brevity of the specification, no further expansion is made here.

[0122] It should be understood that the embodiments disclosed herein and the foregoing description thereof are merely exemplary in nature and, thus, that various changes in the details thereof can be implemented by those skilled in the art without departing from the spirit and scope of the present application. Such changes are intended to fall within the scope of the present application as defined by the appended claims.

Claims

1. A method for adjusting an aromatherapy machine based on user habit prediction, characterized in that: include: Collect historical operation data of the user using the aromatherapy machine, and establish a user scenario habit model based on the historical operation data; Acquire the user's real-time status data in combination with the preset permissions, wherein the user's real-time status information includes real-time vital sign information and real-time spatiotemporal information; Using the user's real-time status data as an index, traverse the user's scene habit model to perform scene matching, determine the real-time user scene, and correspondingly extract basic fragrance information and basic output parameter space; Extracting features from the user's real-time status data to obtain user mental feature items and user physical feature items, and combining the user scenario habit model with the basic output parameter space to make output parameter decisions and obtain aromatherapy output parameters; Based on the basic fragrance information and the fragrance output parameters, the aroma diffuser is adjusted automatically.

2. The aromatherapy machine follow-up adjustment method based on user habit prediction according to claim 1, characterized in that: Collect historical operation data of users using the aromatherapy machine and establish a user scenario habit model based on the historical operation data, including: Performing a primary clustering on the historical operation data based on fragrance preference, and performing a secondary clustering on the primary clustering result based on spatiotemporal characteristics; Classify and reconstruct the secondary clustering results according to the spatiotemporal features to obtain multiple reconstruction operation data clusters; By means of correlation analysis, a set of typical influencing factors of the plurality of reconstruction operation data clusters is screened and determined, and a plurality of user scenario habit base models based on state transition probability are correspondingly constructed, wherein the typical influencing factors are marked with an influence severity coefficient; Statistical analysis is performed on the plurality of user scenario habit base models in combination with preset selection and cropping constraints to determine a plurality of scenario output parameter spaces and store them in association.

3. The method for adjusting the aromatherapy machine based on user habit prediction according to claim 2, characterized in that: Acquire the user's real-time status data in combination with the preset permissions, wherein the user's real-time status information includes real-time vital sign information and real-time spatiotemporal information, including: Collecting the user's real-time vital sign information through a wearable device, wherein the real-time vital sign information includes at least heart rate, skin temperature, and blood oxygen saturation; The real-time spatiotemporal information of the user is collected through the positioning device and the person's sensing device, wherein the real-time spatiotemporal information at least includes geographic location information and indoor area location information.

4. The method for adjusting the aromatherapy machine based on user habit prediction according to claim 3, characterized in that: Using the user's real-time status data as an index, traverse the user's scene habit model to perform scene matching, determine the real-time user scene, and correspondingly extract basic fragrance information and basic output parameter space, including: Vectorizing the user's real-time status data to obtain a comparison state vector; generating a reference state vector set of the user scenario habit model in combination with the typical influencing factor set; Based on the comparison state vector, traverse the reference state vector set to perform similarity calculation, and select the user scenario habit base model corresponding to the reference state vector with the highest similarity as the real-time user scenario; According to the real-time user scenario, the historical aroma probability distribution is correspondingly extracted as the basic aroma information, and the scenario output parameter space is correspondingly extracted as the basic output parameter space.

5. The aromatherapy machine follow-up adjustment method based on user habit prediction according to claim 4, characterized in that: Extracting features from the user's real-time status data to obtain user mental feature items and user physical feature items, and combining the user scenario habit model with the basic output parameter space to make output parameter decisions and obtain aromatherapy output parameters, including: Taking the user mental feature item and the user physical feature item as focus factors, performing scenario factor dimensionality reduction on the user scenario habit model; Based on the result of the scenario factor dimensionality reduction, a relative deviation index of the output parameter in the real-time user scenario is calculated to obtain a relative deviation index set, wherein the relative deviation index is expressed in polar coordinate form; According to the relative deviation index set, a user parameter deviation habit is established, and according to the user parameter deviation habit, fluctuation selection is performed in the basic output parameter space to obtain the aromatherapy output parameter.

6. The aromatherapy machine follow-up adjustment method based on user habit prediction according to claim 1, characterized in that: Also includes: Collecting current physical environment information, including temperature, humidity, air quality, and light intensity; Based on the physical environment information and a preset environment correction model, the environment impact correction of the aromatherapy output parameter is performed, and the environment impact correction result is used to perform follow-up adjustment of the aromatherapy machine.

7. The aromatherapy machine follow-up adjustment method based on user habit prediction according to claim 1, characterized in that: Also includes: Record user intervention logs and combine them with accumulators to perform scenario-based intervention accumulation; According to the preset intervention correction constraints, combined with the scenario-based intervention accumulation results, the user habit correction judgment is performed, wherein the intervention correction constraints include at least period constraints, accumulation constraints and frequency constraints; If the user habit correction judgment result is that intervention is required, the accumulated source data corresponding to the accumulator is called to perform habit correction on the user scenario habit model.

8. A user habit prediction aromatherapy machine follow-up adjustment system, characterized in that: The aromatherapy machine follow-up adjustment method for realizing the user habit prediction according to any one of claims 1 to 7 comprises: A habit model building module is used to collect historical operation data of the user using the aromatherapy machine and build a user scenario habit model based on the historical operation data; A real-time status acquisition module is used to obtain the user's real-time status data in combination with preset permissions, wherein the user's real-time status information includes real-time vital sign information and real-time spatiotemporal information; A scene matching and information extraction module is used to use the user's real-time status data as an index, traverse the user scene habit model to perform scene matching, determine the real-time user scene, and correspondingly extract basic fragrance information and basic output parameter space; An output parameter decision module is used to extract features from the user's real-time status data to obtain user mental feature items and user physical feature items, and combine the user scenario habit model with the basic output parameter space to make output parameter decisions and obtain aromatherapy output parameters; The mediation execution module is used to perform follow-up mediation of the aromatherapy machine based on the basic fragrance information and the aromatherapy output parameters.

Images