Fragrance release control method based on artificial intelligence

Abstract

The invention discloses a fragrance release control method based on artificial intelligence, and relates to the technical field of fragrance release, and the method comprises the steps: carrying out the modeling of environment parameters of a target space, calculating environment impact factors based on a weighted linear combination model, and quantifying the impact degrees of different environments on fragrance diffusion; the weight is solved through a least square method, so that the model can adapt to different target spaces, and release parameters including a release mode, a release amount, a spraying angle and diffusion intensity are calculated, so that the release mode conforms to spatial characteristics; meanwhile, a proper release mode is selected according to environmental parameters, and the release interval and duration are dynamically adjusted, so that excessive accumulation of the fragrance in a small space is avoided, uniform diffusion in a large space is ensured, and the diffusion stability is improved; the air velocity, humidity and fragrance concentration are monitored in real time, release parameters are adjusted based on diffusion state feedback, the diffusion intensity is improved when the air velocity is increased, and the release mode is adjusted when the humidity is reduced so as to prolong the fragrance retention time.

Description

Technical Field

[0001] The present invention relates to the technical field of fragrance release, and particularly to an artificial intelligence-based fragrance release control method. Background Art

[0002] The fragrance release technology is widely applied in home, vehicle and commercial environments. With the change of demands, the control mode has developed from single volatilization to intelligent regulation.

[0003] The current intelligent control methods for fragrance release mainly rely on air quality sensors or timed spraying modes, and cannot perform corresponding regulation for different environments. In a narrow and enclosed space such as a vehicle or a wardrobe, excessive fragrance release may make the smell too strong in a short time, even irritating the eyes and nose and causing discomfort. While in a larger space such as a living room or a meeting room, due to the relatively fast air flow, the fragrance is easily diluted quickly, resulting in the fragrance being difficult to last.

[0004] To address the above problems, some products optimize the experience by adjusting the spraying duration or manually adjusting the release intensity, but this still relies on user initiative and lacks intelligence. For example, in a low-humidity environment, the fragrance volatilizes faster and the fragrance duration is shortened, while in a high-humidity environment, the diffusion speed decreases and it is easy to accumulate in local areas. It can be seen that the existing solutions are difficult to achieve efficient diffusion in different space environments. Therefore, there is an urgent need for an artificial intelligence-based fragrance release control method to solve such problems. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides an artificial intelligence-based fragrance release control method to solve the problem that the existing fragrance release control methods fail to adapt to different spaces, resulting in too strong smell in small spaces and insufficient fragrance diffusion in large spaces, affecting the experience.

[0007] To solve the above technical problems, the present invention provides the following technical solutions:

[0008] An embodiment of the present invention provides an artificial intelligence-based fragrance release control method, which includes:

[0009] Step S1, detecting environmental parameters of a target space, inputting the environmental parameters into a fragrance control model, and calculating release parameters, where the release parameters include a release mode, a release amount, a spraying angle, and a diffusion intensity;

[0010] Step S2, determining a release strategy based on the release parameters calculated in step S1;

[0011] Step S3: According to the release strategy determined in step S2, control the fragrance release device to execute the release, synchronously monitor the environmental status, and provide feedback on the release effect during the fragrance diffusion process to obtain the actual diffusion status;

[0012] Step S4: Based on the actual diffusion status obtained in step S3, adjust the release parameters;

[0013] Step S5: Based on the release parameters adjusted in step S4, optimize the fragrance control model, record the release effects in different environments, and improve future release strategies through artificial intelligence learning of historical data;

[0014] In step S5, combine the user feedback data under different environmental conditions to train a personalized fragrance preference model and optimize future personalized release strategies.

[0015] As a preferred solution of the fragrance release control method based on artificial intelligence according to the present invention, wherein: the environmental parameters include space volume, air flow rate, humidity, temperature, and pollutant concentration.

[0016] As a preferred solution of the fragrance release control method based on artificial intelligence according to the present invention, wherein: in step S1, the fragrance control model calculates the fragrance release parameters based on the input of environmental parameters. Specifically:

[0017] Perform environmental parameter modeling and define the environmental parameter input vector as X:

[0018] X = (x1, x2, x3, x4, x5),

[0019] where x1 represents the volume of the target space, x2 represents the air flow rate, x3 represents the environmental humidity, x4 represents the environmental temperature, and x5 represents the air pollutant concentration.

[0020] Define the environmental impact factor E to quantify the impact of the environment on fragrance diffusion:

[0021] E = f(x1, x2, x3, x4, x5),

[0022] where f(·) is the environmental parameter mapping function that converts the combination of environmental parameters into an impact factor, and the value range of E is normalized to [0, 1] according to historical data.

[0023] Use a weighted linear combination model to calculate E:

[0024] E = w1x1 + w2x2 + w3x3 + w4x4 + w5x5,

[0025] Among them, w1, w2, w3, w4, w5 are weight coefficients used to measure the influence degree of each environmental parameter on fragrance diffusion. The weight coefficients are obtained through training based on historical data, and the calculation method is:

[0026] w = (X T X + λI) -1 X T Y,

[0027] where w is the weight vector (w1, w2, w3, w4, w5), X is the environmental parameter data matrix, (·) T is the transpose operation, Y is the historically measured fragrance diffusion effect, λ is the regularization parameter, I is the identity matrix, and the weight coefficients are solved by the least squares method;

[0028] Calculate the release parameter, which is used to control the way of fragrance diffusion. Define the release parameter vector R:

[0029] R = (r1, r2, r3, r4),

[0030] where r1 represents the release mode, including continuous release and intermittent release, r2 represents the fragrance release amount, which is the amount of fragrance released per unit time, r3 represents the spray angle, and r4 represents the diffusion intensity.

[0031] Select the release mode based on the space volume x1 and the air velocity x2:

[0032] If x1 < x th1 and x2 < x th2 , r1 = 1; otherwise r1 = 0, and the continuous mode is selected;

[0033] where x th1 and x th2 are set thresholds. The intermittent mode is preferentially selected in small spaces and low wind speeds.

[0034] Calculate the release amount r2 based on the environmental impact factor. The calculation formula is:

[0035] r2 = k1E + k2,

[0036] where k1, k2 are the release amount calculation coefficients.

[0037] Calculate the spray angle r3 based on the air velocity x2. The calculation formula is:

[0038] r3 = arctan(k3x2 + k4),

[0039] where k3, k4 are the spray angle calculation coefficients.

[0040] Based on the environmental impact factor E, use the quadratic function model to calculate the diffusion intensity r4:

[0041] r4 = k5E 2 + k6E + k7,

[0042] where k5, k6, and k7 are diffusion intensity calculation coefficients.

[0043] As a preferred solution of the method for controlling fragrance release based on artificial intelligence according to the present invention, wherein: the release strategy includes selecting a release mode adapted to the target space, and setting a release interval and a duration, wherein:

[0044] Reduce the release amount in a small space and optimize the intermittent release mode;

[0045] Adjust the spray angle in a large space to enhance the diffusion intensity.

[0046] As a preferred solution of the method for controlling fragrance release based on artificial intelligence according to the present invention, wherein: the step of determining the release strategy based on the release parameters calculated in step S1 is

[0047] Define the release strategy vector as S:

[0048] S = (s1, s2, s3),

[0049] where s1 represents the release mode, which is adjusted to adapt to the environment, s2 represents the release interval, which controls the time interval, and s3 represents the release duration, which controls the total duration of fragrance diffusion,

[0050] Adjust the release mode based on the space size and air flow rate, and the adjustment method is:

[0051] If x1 < x th3 and x2 < x th4 , then s1 = 1, otherwise s1 = 0 for continuous release;

[0052] where x th3 and x th4 are decision thresholds for the release mode,

[0053] Calculate the release interval, which depends on the environmental impact factor E and the air flow rate x2, and the calculation formula is:

[0054]

[0055] where k8, k9, k 10 are release interval adjustment coefficients, E is the environmental impact factor calculated from environmental parameters, and x2 is the air flow rate,

[0056] The optimization objective of the release interval adjustment coefficient is:

[0057] Improve the uniformity of fragrance diffusion: A too short interval will cause excessive concentration of fragrance, while a too long interval may lead to insufficient fragrance diffusion.

[0058] Adapt to different air flow states: When the air flow rate is large, appropriately reduce the release interval to prevent the fragrance from being diluted too quickly.

[0059] Reduce fragrance consumption: Avoid excessive release and improve the usage efficiency.

[0060] Here, the least squares regression method is used to fit the relationship between the release interval and environmental impact factors and air flow rate based on historical data, construct training data, and collect historical data sets. Among them, E i is the environmental impact factor, x 2i is the air flow rate, s 2i is the best release interval measured historically.

[0061] Based on historical data, establish an objective function:

[0062]

[0063] Among them, L(k8, k9, k 10 ) is the mean square error loss function. By minimizing the loss function, solve for the optimal parameters k8, k9, k 10 ,

[0064] Perform gradient update on the loss function. The formula is:

[0065]

[0066] Use gradient descent to update the parameters:

[0067]

[0068] Among them, α is the learning rate. Iterate until the loss function converges to obtain the optimal k8, k9, k 10 ; After the parameter training is completed, conduct model verification, including cross-validation and error analysis, and finally determine the values of k8, k9, k 10 .

[0069] Optimize the release duration based on the pollutant concentration x5 and humidity x3. The optimization formula is:

[0070] s3 = k 11 ln(x5 + 1) + k 12 x3,

[0071] Among them, k 11 , k 12 are the duration calculation coefficients.

[0072] As a preferred embodiment of the method for controlling fragrance release based on artificial intelligence according to the present invention, wherein: the environmental state includes air velocity, humidity, and fragrance concentration.

[0073] As a preferred embodiment of the method for controlling fragrance release based on artificial intelligence according to the present invention, wherein: the step of providing feedback on the release effect during the fragrance diffusion process to obtain the actual diffusion state is as follows.

[0074] During the fragrance diffusion process, the environmental state is monitored in real time by a sensor, and in combination with the change in fragrance concentration, feedback is provided on the release effect. The environmental state monitoring includes: monitoring the air velocity to determine whether the diffusion range meets the expectation; monitoring the humidity to evaluate the suspension of the fragrance in the air; monitoring the fragrance concentration to obtain real-time diffusion data;

[0075] Calculating the uniformity of fragrance diffusion to determine whether there is over-release or under-diffusion.

[0076] Evaluating the change in fragrance concentration over time to detect whether the diffusion reaches the set threshold.

[0077] Combining the air flow situation to adjust the subsequent release strategy.

[0078] If the fragrance concentration is lower than the expected value, determine whether to increase the release amount.

[0079] If the fragrance concentration is too high, automatically adjust the release interval or reduce the release amount.

[0080] If the air velocity changes drastically, trigger the dynamic adjustment mechanism.

[0081] As a preferred embodiment of the method for controlling fragrance release based on artificial intelligence according to the present invention, wherein: the ways to adjust the release parameters include:

[0082] When the air velocity increases, increase the diffusion intensity.

[0083] When the humidity decreases, adjust the release mode.

[0084] As a preferred embodiment of the method for controlling fragrance release based on artificial intelligence according to the present invention, wherein: the step of adjusting the release parameters based on the actual diffusion state obtained in step S3 is as follows.

[0085] Calculate the release parameter adjustment factor A, and the formula is:

[0086] A = h(y1, y2, y3),

[0087] where y1 is the real-time air velocity, y2 is the real-time humidity, y3 is the real-time fragrance concentration, and h(·) is the parameter adjustment function.

[0088] The calculation formula of the adjustment factor is as follows:

[0089] A = m1y1 + m2y2 + m3y3,

[0090] where m1, m2, and m3 are adjustment coefficients,

[0091] Optimize the release parameters according to the adjustment factor A:

[0092] r′2 = r2 + n1A,

[0093] r′3 = r3 + n2A,

[0094] r′4 = r4 + n3A,

[0095] where r'2, r'3, and r'4 are the adjusted release amount, spray angle, and diffusion intensity respectively, and n1, n2, and n3 are adjustment weight coefficients.

[0096] As a preferred solution of the fragrance release control method based on artificial intelligence described in the present invention, among them: the steps of combining user feedback data under different environmental conditions, training a personalized fragrance preference model, and optimizing future personalized release strategies are as follows,

[0097] Construct a training data set D u :

[0098]

[0099] where E j is the jth environmental impact factor, R j is the jth release parameter, S j is the jth release strategy, F j is the jth user feedback score, in the range [0, 1], and the feedback score F j is composed of subjective evaluation data of the user on the fragrance release effect in different environments, and is obtained by selecting questionnaires or sensor speculation,

[0100] Establish a personalized fragrance preference model, and define a user preference scoring function P(E, R, S):

[0101] P(E, R, S) = v1E + v2R + v3S + v4,

[0102] where P(E, R, S) predicts the score of the user for a specific environment and release strategy, and v1, v2, v3, and v4 are preference weights to be trained,

[0103] Solve v using least squares regression:

[0104]

[0105] Among them, v is the user preference weight vector (v1, v2, v3, v4), and X u is the training data matrix, where each row is (E j , R j , S j , 1), F is the user feedback score vector, and λ u is the regularization parameter.

[0106] Optimize the future release strategy based on the trained model to maximize user satisfaction and predict the personalized release strategy:

[0107]

[0108] Use the gradient ascent method to optimize the strategy:

[0109]

[0110] Among them, S is the optional release strategy, and η is the step size;

[0111] Adopt an online update strategy to continuously optimize the model by combining new user data:

[0112] When new feedback data arrives, update the training dataset D u ,

[0113] Recalculate the weight v to adapt to the new user preferences,

[0114] and monitor the changes in historical scores to dynamically adjust λ u to balance the influence of new and old data.

[0115] The beneficial effects of the present invention are as follows: In the present invention, the environmental parameters of the target space are modeled, and the environmental impact factor is calculated based on the weighted linear combination model to quantify the influence degree of different environments on the fragrance diffusion; the weights are solved by the least squares method, enabling the model to adapt to different target spaces, and the release parameters are calculated, including the release mode, release amount, spray angle, and diffusion intensity, so that the release method conforms to the space characteristics; in terms of release strategy optimization, the appropriate release mode is selected according to the environmental parameters, and by dynamically adjusting the release interval and duration, the fragrance is prevented from accumulating excessively in small spaces and ensured to be evenly diffused in large spaces, thereby improving the diffusion stability.

[0116] The present invention monitors the air flow rate, humidity, and fragrance concentration in real time, adjusts the release parameters based on the feedback of the diffusion state, increases the diffusion intensity when the air flow rate increases, and adjusts the release mode when the humidity decreases to extend the fragrance retention time, so that the diffusion effect is adaptively adjusted according to the environmental changes; combined with artificial intelligence optimization, records the release effects in different environments, and uses machine learning to train a personalized fragrance preference model, and optimizes the release strategy through least squares regression and gradient ascent, so that the fragrance release can not only adapt to the spatial characteristics, but also match the personalized needs of users.

[0117] In summary, the present invention makes the fragrance release more accurate, stable and intelligent, improves the odor experience in different spaces, reduces energy consumption, and enhances user comfort. Brief Description of the Drawings

[0118] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0119] Figure 1 Flow chart of the fragrance release control method based on artificial intelligence of the present invention. Detailed Embodiments

[0120] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the detailed embodiments of the present invention with reference to the drawings of the specification.

[0121] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar promotions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0122] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The "in one embodiment" that appears in different places in this specification does not all refer to the same embodiment, nor is it a separate or alternative embodiment that excludes other embodiments.

[0123] Embodiment 1, referring to Figure 1 , this embodiment provides a fragrance release control method based on artificial intelligence, including the following steps:

[0124] Step S1: Detect the environmental parameters of the target space, input the environmental parameters into the fragrance control model, and calculate the release parameters, where the release parameters include the release mode, release amount, spray angle, and diffusion intensity;

[0125] The environmental parameters include the space volume, air flow rate, humidity, temperature, and pollutant concentration;

[0126] In step S1, based on the input of the environmental parameters, the fragrance control model calculates the fragrance release parameters. Specifically:

[0127] Perform environmental parameter modeling and define the environmental parameter input vector as X:

[0128] X = (x1, x2, x3, x4, x5),

[0129] where x1 represents the volume of the target space, x2 represents the air flow rate, x3 represents the environmental humidity, x4 represents the environmental temperature, and x5 represents the air pollutant concentration.

[0130] Define the environmental impact factor W to quantify the impact of the environment on fragrance diffusion:

[0131] E = f(x1, x2, x3, x4, x5),

[0132] where f(·) is the environmental parameter mapping function that converts the environmental parameter combination into the impact factor, and the value range of E is normalized to [0, 1] based on historical data.

[0133] Use the weighted linear combination model to calculate E:

[0134] E = w1x1 + w2x2 + w3x3 + w4x4 + w5x5,

[0135] where w1, w2, w3, w4, w5 are weight coefficients used to measure the influence degree of each environmental parameter on fragrance diffusion. The weight coefficients are obtained through training based on historical data, and the calculation method is:

[0136] w = (X T X + λI) -1 X T Y,

[0137] where w is the weight vector (w1, w2, w3, w4, w5), X is the environmental parameter data matrix, (·) T is the transpose operation, Y is the historically measured fragrance diffusion effect, λ is the regularization parameter, and I is the identity matrix. The weight coefficients are solved by the least squares method;

[0138] Calculate the release parameters for controlling the way of fragrance diffusion, and define the release parameter vector R:

[0139] R = (r1, r2, r3, r4),

[0140] where r1 represents the release mode, including continuous release and intermittent release; r2 represents the fragrance release amount, which is the amount of fragrance released per unit time; r3 represents the spray angle; r4 represents the diffusion intensity.

[0141] The release mode is selected based on the space volume x1 and the air velocity x2:

[0142] If x1 < x th1 and x2 < x th2 , r1 = 1; otherwise r1 = 0, and the continuous mode is selected.

[0143] where x th1 and x th2 are set thresholds. The intermittent mode is preferentially selected in small spaces and at low wind speeds.

[0144] The release amount r2 is calculated based on the environmental impact factor, and the calculation formula is:

[0145] r2 = k1E + k2,

[0146] where k1 and k2 are release amount calculation coefficients.

[0147] The spray angle r3 is calculated based on the air velocity x2, and the calculation formula is:

[0148] r3 = arctan(k3x2 + k4),

[0149] where k3 and k4 are spray angle calculation coefficients.

[0150] Based on the environmental impact factor E, the quadratic function model is used to calculate the diffusion intensity r4:

[0151] r4 = k5E 2 + k6E + k7,

[0152] where k5, k6, and k7 are diffusion intensity calculation coefficients.

[0153] Specifically, a fragrance control model is constructed here. The environmental parameters are used as the input vector, and the environmental impact factor is calculated through a weighted linear model to measure the influence degree of different environmental conditions on fragrance diffusion. The least squares method is used to solve the weight coefficients so that the model can adapt to different target spaces.

[0154] In terms of release parameter calculation, the release mode is classified based on the space volume and air velocity. The release amount is determined by a linear model, the spray angle is determined by an arctangent mapping, and the diffusion intensity is determined by a quadratic function modeling, so that the fragrance can achieve the best diffusion effect under different environmental conditions.

[0155] Step S2: Determine the release strategy based on the release parameters calculated in Step S1;

[0156] The release strategy includes selecting a release mode adapted to the target space and setting the release interval and duration, where:

[0157] For a small space, reduce the release amount and optimize the intermittent release mode;

[0158] For a large space, adjust the spray angle to enhance the diffusion intensity;

[0159] The steps to determine the release strategy based on the release parameters calculated in Step S1 are as follows:

[0160] Define the release strategy vector as S:

[0161] S = (s1, s2, s3),

[0162] where s1 represents the release mode, which is adjusted to adapt to the environment, s2 represents the release interval, which controls the time interval, and s3 represents the release duration, which controls the total duration of fragrance diffusion;

[0163] Adjust the release mode based on the space size and air flow rate. The adjustment method is as follows:

[0164] If x1 < x th3 and x2 < x th4 , then s1 = 1; otherwise, s1 = 0 for continuous release;

[0165] where x th3 and x th4 are the decision thresholds for the release mode;

[0166] Calculate the release interval. The release interval depends on the environmental impact factor E and the air flow rate x2. The calculation formula is:

[0167]

[0168] where k8, k9, k 10 are the release interval adjustment coefficients, E is the environmental impact factor calculated from environmental parameters, and x2 is the air flow rate;

[0169] The optimization objectives of the release interval adjustment coefficient are:

[0170] Improve the uniformity of fragrance diffusion: Too short an interval will cause excessive concentration of the fragrance, while too long an interval may result in insufficient fragrance diffusion;

[0171] Adapt to different air flow states: When the air flow rate is large, appropriately reduce the release interval to prevent the fragrance from being diluted too quickly;

[0172] Reduce fragrance consumption: Avoid excessive release and improve the usage efficiency;

[0173] Here, the least squares regression method is adopted to fit the relationship between the release interval, environmental impact factor, and air velocity based on historical data, construct the training data, and collect the historical data set. Among them, E i is the environmental impact factor, x 2i is the air velocity, and s 2i is the best release interval measured historically.

[0174] Based on the historical data, establish the objective function:

[0175]

[0176] Among them, L(k8, k9, k 10 ) is the mean square error loss function. By minimizing the loss function, solve for the optimal parameters k8, k9, k 10 .

[0177] Perform gradient update on the loss function. The formula is:

[0178]

[0179] Use gradient descent to update the parameters:

[0180]

[0181] Among them, α is the learning rate. Iterate until the loss function converges to obtain the optimal k8, k9, k 10 ; After the parameter training is completed, perform model verification, including cross-validation and error analysis, and finally determine the values of k8, k9, k 10 ;

[0182] Optimize the release duration based on the pollutant concentration x5 and humidity x3. The optimization formula is:

[0183] s3 = k 11 ln(x5 + 1) + k 12 x3,

[0184] Among them, k 11 , k 12 are the duration calculation coefficients.

[0185] Specifically, based on the release parameters in step S1, the optimal release strategy is calculated. The release mode is selected according to the space volume and air flow rate. When the space is small, intermittent release is preferred, and the release interval is dynamically adjusted by the environmental impact factor E and the air flow rate to make the diffusion of the fragrance adapt to different air flow states. The release duration is determined by comprehensively considering the air pollutant concentration and environmental humidity, so as to ensure that the diffusion time is extended under the conditions of high pollution or low humidity, and improve the coverage effect of the fragrance.

[0186] Step S3: According to the release strategy determined in step S2, control the fragrance release device to execute the release, synchronously monitor the environmental status, and feedback the release effect during the diffusion process of the fragrance to obtain the actual diffusion status.

[0187] The environmental status includes air flow rate, humidity and fragrance concentration.

[0188] The steps for feeding back the release effect during the diffusion process of the fragrance to obtain the actual diffusion status are as follows:

[0189] During the diffusion process of the fragrance, the environmental status is monitored in real time by sensors, and combined with the change of fragrance concentration, the release effect is fed back. The environmental status monitoring includes: monitoring the air flow rate to judge whether the diffusion range meets the expectation; monitoring the humidity to evaluate the suspension situation of the fragrance in the air; monitoring the fragrance concentration to obtain real-time diffusion data.

[0190] Calculate the uniformity of the fragrance diffusion, and judge whether there is over-release or insufficient diffusion.

[0191] Evaluate the change of fragrance concentration over time, and detect whether the diffusion reaches the set threshold.

[0192] Combined with the air flow situation, adjust the subsequent release strategy.

[0193] If the fragrance concentration is lower than the expected value, judge whether it is necessary to increase the release amount.

[0194] If the fragrance concentration is too high, automatically adjust the release interval or reduce the release amount.

[0195] If the air flow rate changes violently, trigger the dynamic adjustment mechanism.

[0196] Step S4: Based on the actual diffusion status obtained in step S3, adjust the release parameters.

[0197] The ways to adjust the release parameters include:

[0198] When the air flow rate increases, increase the diffusion intensity.

[0199] When the humidity decreases, adjust the release mode.

[0200] Based on the actual diffusion state obtained in step S3, the steps to adjust the release parameters are as follows:

[0201] Calculate the release parameter adjustment factor A, and the formula is:

[0202] A = h(y1, y2, y3),

[0203] where y1 is the real-time air velocity, y2 is the real-time humidity, y3 is the real-time fragrance concentration, and h(·) is the parameter adjustment function.

[0204] The calculation formula for the adjustment factor is:

[0205] A = m1y1 + m2y2 + m3y3,

[0206] where m1, m2, and m3 are adjustment coefficients.

[0207] Optimize the release parameters according to the adjustment factor A:

[0208] r′2 = r2 + n1A,

[0209] r′3 = r3 + n2A,

[0210] r′4 = r4 + n3A,

[0211] where r'2, r'3, and r'4 are the adjusted release amount, spray angle, and diffusion intensity respectively, and n1, n2, and n3 are adjustment weight coefficients.

[0212] Specifically, here the release parameters are adjusted based on the feedback of the real-time environmental state to ensure stable fragrance diffusion effect. The release parameter adjustment factor A is calculated by combining the air velocity, humidity, and fragrance concentration, and then affects the release amount, spray angle, and diffusion intensity, so that the release parameters can be adaptively adjusted according to the environmental changes.

[0213] Step S5: Based on the release parameters adjusted in step S4, optimize the fragrance control model, record the release effects in different environments, and improve the future release strategy through artificial intelligence learning of historical data.

[0214] In step S5, combine the user feedback data under different environmental conditions to train a personalized fragrance preference model and optimize the future personalized release strategy.

[0215] The steps to combine the user feedback data under different environmental conditions to train a personalized fragrance preference model and optimize the future personalized release strategy are as follows:

[0216] Construct the training dataset D u :

[0217]

[0218] Among them, E j is the j-th environmental impact factor, R j is the j-th release parameter, S j is the j-th release strategy, F j is the j-th user feedback score, in the range [0, 1]. The feedback score F j is composed of subjective evaluation data of the fragrance release effect by users in different environments, and is obtained by using questionnaires or sensors for speculation.

[0219] Build a personalized fragrance preference model and define the user preference scoring function P(E, R, S):

[0220] P(E, R, S) = v1E + v2R + v3S + v4,

[0221] Among them, P(E, R, S) predicts the score of users for a specific environment and release strategy, and v1, v2, v3, v4 are the preference weights to be trained.

[0222] Use least squares regression to solve for v:

[0223]

[0224] Among them, v is the user preference weight vector (v1, v2, v3, v4), X u is the training data matrix, where each row is (E j , R j , S j , 1), F is the user feedback score vector, and λ u is the regularization parameter.

[0225] Based on the trained model, optimize the future release strategy to maximize user satisfaction and predict the personalized release strategy:

[0226]

[0227] Use the gradient ascent method to optimize the strategy:

[0228]

[0229] Among them, S is the optional release strategy and η is the step size;

[0230] Adopt an online update strategy and continuously optimize the model by combining new user data:

[0231] When new feedback data arrives, update the training dataset D u ,

[0232] Recalculate the weight v to adapt to the new user preferences.

[0233] And monitor the changes in historical scores and dynamically adjust λ u Balance the influence of new and old data;

[0234] Specifically, user feedback data is collected here to train a personalized fragrance preference model, enabling the fragrance release strategy to be optimized according to changes in user needs. The least squares method is used to solve the preference weights, and the gradient ascent method is combined to optimize the release strategy. The prediction accuracy is continuously improved through online updates, making the fragrance diffusion more in line with the preferences of individual users and enhancing the personalized experience.

[0235] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A fragrance release control method based on artificial intelligence, characterized in that: include, Step S1, detecting environmental parameters of the target space, and inputting the environmental parameters into the fragrance control model to calculate release parameters, wherein the release parameters include release mode, release amount, spray angle and diffusion intensity; Step S2, determining a release strategy based on the release parameters calculated in step S1; Step S3, according to the release strategy determined in step S2, control the fragrance release device to execute the release, synchronously monitor the environmental state, and provide feedback on the release effect during the fragrance diffusion process to obtain the actual diffusion state; Step S4, adjusting the release parameters based on the actual diffusion state obtained in step S3; Step S5, based on the release parameters adjusted in step S4, optimizing the fragrance control model, recording the release effects under different environments, and learning historical data through artificial intelligence to improve future release strategies; In step S5, the user feedback data under different environmental conditions is combined to train a personalized fragrance preference model to optimize the future personalized release strategy.

2. The method for controlling fragrance release based on artificial intelligence according to claim 1, characterized in that: The environmental parameters include space volume, air flow rate, humidity, temperature and pollutant concentration.

3. The method for controlling fragrance release based on artificial intelligence as claimed in claim 2, characterized in that: In step S1, the fragrance control model calculates the fragrance release parameters based on the input of the environmental parameters, specifically: Perform environmental parameter modeling and define the environmental parameter input vector as X: X=(x1,x2,x3,x4,x5), Among them, x1 represents the volume of the target space, x2 represents the air flow rate, x3 represents the ambient humidity, x4 represents the ambient temperature, and x5 represents the concentration of air pollutants. Define the environmental impact factor E to quantify the impact of the environment on fragrance diffusion: E=f(x1,x2,x3,x4,x5), Among them, f(·) is the environmental parameter mapping function, which converts the environmental parameter combination into an impact factor. The value range of E is normalized to [0,1] according to historical data. The weighted linear combination model is used to calculate E: E=w1x1+w2x2+w3x3+w4x4+w5x5, Among them, w1, w2, w3, w4, and w5 are weight coefficients, which are used to measure the influence of various environmental parameters on the diffusion of fragrance. The weight coefficients are obtained based on historical data training and are calculated as follows: w=(X T X+λI) -1 X T Y, Where w is the weight vector (w1, w2, w3, w4, w5), X is the environmental parameter data matrix, (·) T is the transposition operation, Y is the historically measured fragrance diffusion effect, λ is the regularization parameter, I is the identity matrix, and the weight coefficient is solved by the least squares method; Calculate the release parameters to control the way the fragrance diffuses and define the release parameter vector R: R＝(r1,r2,r3,r4)， Among them, r1 represents the release mode, including continuous release and intermittent release, r2 represents the fragrance release amount, which is the amount of fragrance released per unit time, r3 represents the spray angle, and r4 represents the diffusion intensity. Select the release mode based on the space volume x1 and air flow rate x2: If x1 <x th1 And x2 <x th2 , r1=1; otherwise r1=0, select continuous mode; Among them, x th1 and x th2 For the set threshold, the intermittent mode is preferred in small spaces and low wind speeds. The release amount r2 is calculated based on the environmental impact factor, and the calculation formula is: r2=k1E+k2, Among them, k1 and k2 are the release calculation coefficients. The spray angle r3 is calculated based on the air flow rate x2, and the calculation formula is: r3=arctan(k3x2+k4), Among them, k3, k4 are spray angle calculation coefficients, Based on the environmental impact factor E, the diffusion intensity r4 is calculated using the quadratic function model: r4=k5E 2 +k6E+k7, Among them, k5, k6, k7 are diffusion intensity calculation coefficients.

4. The method for controlling fragrance release based on artificial intelligence as claimed in claim 3, characterized in that: The release strategy includes selecting a release mode that is adapted to the target space and setting a release interval and duration, wherein: Small space reduces release volume; Large space to adjust the spray angle.

5. The method for controlling fragrance release based on artificial intelligence as claimed in claim 4, characterized in that: The step of determining the release strategy based on the release parameters calculated in step S1 is: Define the release strategy vector as S: S=(s1,s2,s3), Among them, s1 represents the release mode, which is adjusted to adapt to the environment, s2 represents the release interval, which controls the time interval, and s3 represents the release duration, which controls the total duration of fragrance diffusion. Adjust the release mode based on the space size and air flow rate. The adjustment method is: If x1 <x th3 And x2 <x th4 , then s1=1, otherwise s1=0, and continuous release is performed; Among them, x th3 and x th4 is the decision threshold of the release mode, Calculate the release interval, which depends on the environmental impact factor E and the air flow rate x2. The calculation formula is: Among them, k8, k9, k 10 is the release interval adjustment coefficient, E is the environmental impact factor, which is calculated from the environmental parameters, x2 is the air flow rate, The least squares regression method is used here to fit the relationship between the release interval, environmental influencing factors, and air velocity based on historical data, build training data, and collect historical data sets. Among them, E i is the environmental impact factor, x 2i is the air velocity, s 2i is the best release interval measured historically, Based on historical data, establish the objective function: Among them, L(k8,k9,k 10 ) is the mean square error loss function. By minimizing the loss function, the optimal parameters k8, k9, k 10 , The gradient update of the loss function is as follows: Update the parameters using gradient descent: Among them, α is the learning rate, and the iteration is carried out until the loss function converges to obtain the optimal k8, k9, k 10 ; The release duration is optimized based on pollutant concentration x5 and humidity x3. The optimization formula is: <h2 style=";text-align:left;direction:ltr">s3=k<h2 style=";text-align:left;direction:ltr"> 11 <h2 style=";text-align:left;direction:ltr"> ln(x5+1)+k<h2 style=";text-align:left;direction:ltr"> 12 <h2 style=";text-align:left;direction:ltr"> x3, Among them, k 11 ,k 12 Calculates a factor for duration.

6. The method for controlling fragrance release based on artificial intelligence as claimed in claim 5, characterized in that: The environmental conditions include air velocity, humidity and fragrance concentration.

7. The method for controlling fragrance release based on artificial intelligence according to claim 6, characterized in that: The step of providing feedback on the release effect during the fragrance diffusion process to obtain the actual diffusion state is: During the fragrance diffusion process, the sensor monitors the environmental status in real time, and combines the change of fragrance concentration to provide feedback on the release effect and monitor the environmental status, including: monitoring the air flow rate to determine whether the diffusion range meets expectations; monitoring humidity to evaluate the suspension of fragrance in the air; monitoring fragrance concentration to obtain real-time diffusion data; Calculate the uniformity of fragrance diffusion and determine whether there is over-release or under-diffusion. Evaluate the change of fragrance concentration over time and detect whether the diffusion reaches the set threshold. Adjust the subsequent release strategy based on air flow conditions; If the fragrance concentration is lower than the expected value, determine whether the release amount needs to be increased. If the fragrance concentration is too high, the release interval will be automatically adjusted or the release amount will be reduced. If the air flow rate changes dramatically, the dynamic adjustment mechanism is triggered.

8. The method for controlling fragrance release based on artificial intelligence according to claim 7, characterized in that: The method of adjusting the release parameters includes: When the air velocity increases, the diffusion intensity increases; When humidity decreases, adjust the release pattern.

9. The method for controlling fragrance release based on artificial intelligence as claimed in claim 8, characterized in that: The step of adjusting the release parameters based on the actual diffusion state obtained in step S3 is: Calculate the release parameter adjustment factor A, the formula is: A=h(y1,y2,y3), Where y1 is the real-time air velocity, y2 is the real-time humidity, y3 is the real-time fragrance concentration, and h(·) is the parameter adjustment function. The adjustment factor is calculated as: A=m1y1+m2y2+m3y3, Among them, m1, m2, m3 are adjustment coefficients, Optimize release parameters according to adjustment factor A: r'2=r2+n1A, r'3=r3+n2A, r'4=r4+n3A, Among them, r'2, r'3, r'4 are the adjusted release amount, spray angle and diffusion intensity, respectively, and n1, n2, n3 are the adjustment weight coefficients.

10. The method for controlling fragrance release based on artificial intelligence according to claim 9, characterized in that: The steps of combining user feedback data under different environmental conditions, training a personalized fragrance preference model, and optimizing future personalized release strategies are: Construct training dataset D u : Among them, E j is the jth environmental impact factor, R j is the jth release parameter, S j is the jth release strategy, F j Feedback score for the jth user, range [0,1], feedback score F j It consists of users' subjective evaluation data on the fragrance release effect in different environments, obtained through questionnaires or sensors. Establish a personalized fragrance preference model and define the user preference scoring function P(E,R,S): P(E,R,S)=v1E+v2R+v3S+v4, Among them, P(E,R,S) predicts the user's score under a specific environment and release strategy, v1, v2, v3, and v4 are the preference weights to be trained, Solve for v using least squares regression: Among them, v is the user preference weight vector (v1, v2, v3, v4), X u is the training data matrix, where each line is (E j ,R j ,S j ,1), F is the user feedback rating vector, λ u is the regularization parameter, Optimize future release strategies based on trained models To maximize user satisfaction, predict personalized release strategies: Optimize the strategy using gradient ascent: Among them, S is the optional release strategy, η is the step size; Adopt an online update strategy and continuously optimize the model based on new user data: When new feedback data arrives, update the training dataset D u , Recalculate the weight v to adapt to the new user preference, And monitor the changes in historical scores and dynamically adjust λ u Balance the impact of new and old data.

Images