Parameter adaptive adjustment method and system of intelligent coffee machine fused with environment perception

Abstract

The application discloses a fusion environment perception intelligent coffee machine parameter adaptive adjustment method and system, relates to the technical field of parameter adjustment, and comprises the following steps: collecting brewing environment data through a machine body sensor array, matching coffee bean material labels, and constructing multi-dimensional environment disturbance characteristics; importing a Gaussian mixture distribution model, outputting a probability distribution interval of a powder bed physical state parameter, obtaining a control parameter physical safety boundary; relying on a virtual control sandbox before water injection, carrying out multi-round brewing simulation, and obtaining a solid solubility prediction value; and constructing a target brewing control scheme to drive a coffee machine execution unit to perform parameter adaptive adjustment. The application solves the technical problems that the prior art cannot quantize the uncertainty of the powder bed state in real time under dynamic environment disturbance and dynamically adjust the brewing parameters, and the consistency of extraction flavor and individual taste are insufficient, and achieves the technical effect of improving the control precision of the adaptability of the brewing parameters of the coffee machine under the multi-variable environment and the stability of the flavor.

Description

Technical Field

[0001] This invention relates to the field of parameter adjustment technology, specifically to a method and system for adaptive parameter adjustment of an intelligent coffee machine that integrates environmental perception. Background Technology

[0002] Traditional coffee machines use fixed brewing parameters, such as water temperature, grind size, extraction pressure, and time, designed based on stable, ideal environmental conditions. However, in real-world use, the coffee brewing process is highly susceptible to significant environmental factors. Changes in ambient temperature, relative humidity, atmospheric pressure, and light conditions directly alter the physicochemical properties of coffee grounds. For example, high humidity causes coffee grounds to absorb moisture and clump, increasing bed resistance and triggering channeling effects; low temperature or low pressure lowers the boiling point of water and extraction efficiency, resulting in insufficient dissolution of flavor compounds. Environmental disturbances are coupled with the coffee bean variety, roasting level, and material state, causing the optimal extraction parameters for the same coffee bean to vary significantly depending on the season, altitude, or weather. Existing coffee machines lack the ability to perceive and utilize environmental information in real time and still rely on manual adjustments based on user experience. This makes it difficult to guarantee consistent taste and achieve adaptive optimization across different scenarios. Threshold-based on-off control or simple linear compensation ignores the nonlinear, time-varying, and uncertain characteristics of environmental variables and extraction kinetics, making it difficult to stably reproduce the target flavor under dynamic environmental disturbances. Furthermore, it cannot achieve continuous evolution driven by personalized user taste feedback and sensor data fusion.

[0003] Therefore, current technologies suffer from the inability to quantify the uncertainty of the powder bed state in real time under dynamic environmental disturbances and to dynamically adjust brewing parameters, resulting in insufficient consistency of extracted flavor and personalized taste. Summary of the Invention

[0004] This application provides a method and system for adaptive adjustment of parameters of an intelligent coffee machine that integrates environmental perception. This solves the technical problem in the prior art that it is impossible to quantify the uncertainty of the coffee bed state in real time under dynamic environmental disturbances and dynamically adjust the brewing parameters, resulting in insufficient consistency of extraction flavor and personalized taste. It achieves the technical effect of improving the adaptability of brewing parameters and the control precision of flavor stability of coffee machines under changing environments.

[0005] This application provides a method for adaptive parameter adjustment of an intelligent coffee machine that integrates environmental perception. The method includes: collecting brewing environment data in real time through a sensor array integrated on the coffee machine, and combining it with the input coffee bean material label to generate a multi-dimensional environmental disturbance feature characterizing the evolution of the current brewing environment; inputting the multi-dimensional environmental disturbance feature into a Gaussian mixture distribution model, performing joint probability density mapping using the Gaussian mixture distribution model, and outputting the uncertainty continuous probability distribution interval of the physical state parameters of the coffee bed under the current environmental disturbance; using the uncertainty continuous probability distribution interval as a boundary condition to perform constraint solving to obtain the physical safety boundary of the control parameters; before water injection, performing multiple rounds of brewing simulation in a virtual control sandbox for the physical safety boundary of the control parameters according to a preset time step to obtain a predicted value of solid solubility; constructing a target brewing control scheme based on the predicted value of solid solubility, and driving the execution unit of the coffee machine to execute the target brewing control scheme to perform parameter adaptive adjustment.

[0006] In a possible implementation, the physical safety boundary of the control parameter is a range of control parameters that keeps the total solids solubility within a preset refined extraction range.

[0007] In a possible implementation, a sensor array integrated into the coffee machine collects brewing environment data in real time. Combined with the input coffee bean material label, this generates a multidimensional environmental disturbance feature characterizing the evolution of the current brewing environment. This includes: extracting instantaneous ambient temperature and relative humidity at the same time based on the brewing environment data, calculating the moisture absorption integral of relative humidity over time, and generating a dynamic moisture absorption accumulation effect term; using the derivative of the instantaneous ambient temperature with respect to time as the instantaneous ambient temperature gradient; constructing a temperature and humidity transient coupling matrix using the instantaneous ambient temperature gradient as the main diagonal element and the dynamic moisture absorption accumulation effect term as the off-diagonal interleaving element; performing interference analysis on the material corresponding to the coffee bean material label based on the temperature and humidity transient coupling matrix to obtain a nonlinear interference coefficient; and extracting the real-time ambient atmospheric pressure time-series fluctuation rate from the brewing environment data and cascading it with the nonlinear interference coefficient to obtain the multidimensional environmental disturbance feature.

[0008] In a possible implementation, interference analysis is performed on the material corresponding to the coffee bean material label based on the transient coupling matrix of temperature and humidity to obtain a nonlinear interference coefficient. This includes: obtaining a preset material moisture absorption sensitivity feature vector corresponding to the material based on the coffee bean material label; multiplying the transient coupling matrix of temperature and humidity by the preset material moisture absorption sensitivity feature vector, and performing linear decoupling operation of the temperature and humidity coupling features through matrix multiplication to obtain the nonlinear interference coefficient.

[0009] In possible implementations, the physical state parameters of the powder bed include the initial moisture content range of the powder bed, the dynamic exhaust resistance coefficient of carbon dioxide in the powder bed, and the dynamic heat loss rate of the cooking head.

[0010] In a possible implementation, the multidimensional environmental disturbance features are input into a Gaussian mixture distribution model (GMM), and a joint probability density mapping is performed using the GMM to output the uncertainty continuous probability distribution interval of the coffee bean bed physical state parameters under the current environmental disturbance. This includes: inputting the multidimensional environmental disturbance features as an input vector into the GMM; using multiple pre-set sub-Gaussian probability density functions within the GMM to calculate the conditional probability distribution of the input vector under each sub-Gaussian component; using pre-set component weight coefficients to linearly weight and superimpose the conditional probability distributions under each sub-Gaussian component to obtain a global joint probability density function, wherein the expectation vector and covariance matrix of each sub-Gaussian component are pre-trained and determined by the coffee bean material label; setting a target confidence threshold, integrating the global joint probability density function, and extracting the continuous integral boundary that meets the target confidence threshold as the uncertainty continuous probability distribution interval.

[0011] In one possible implementation, before water injection is initiated, multiple rounds of brewing simulation are performed in a virtual control sandbox at a preset time step, targeting the physical safety boundary of the control parameters, to obtain a predicted value for solid solubility. This includes: loading a kinetic virtual operator in the virtual control sandbox that covers the relationship between the mass change rate of the material corresponding to the coffee bean material label and the total solid solubility change rate; and, within the preset time step and within the physical safety boundary of the control parameters, traversing different virtual execution power combinations in the historical virtual power combination library to perform multiple rounds of brewing simulation, running the kinetic virtual operator, and obtaining the predicted value for solid solubility.

[0012] In a possible implementation, a target brewing control scheme is constructed based on the predicted solid solubility value, including: retrieving the shear force distribution gradient of fluid penetrating the virtual porous powder bed when predicting solubility in each brewing simulation using the kinetic virtual operator; calculating the probability value of channel effect caused by sudden flow velocity changes based on the shear force distribution gradient corresponding to each brewing simulation; when the probability value exceeds a preset safety threshold, introducing a numerical penalty term into the control parameters of the simulation round to obtain a physical compliance score; traversing the predicted solid solubility value and the physical compliance score, extracting the virtual execution power combination parameter corresponding to the brewing simulation round where the predicted solid solubility value converges within a preset premium extraction range and the physical compliance score is the highest, and using this as the target brewing control scheme.

[0013] In a possible implementation, retrieving the shear force distribution gradient of fluid penetrating the virtual porous powder bed when predicting solubility in each round of brewing simulation using the kinetic virtual operator includes: discretizing the virtual porous powder bed along the fluid flow direction and radial cross section to obtain grid cells; extracting the fluid shear force values ​​of each grid cell within the preset time step to construct the shear force distribution gradient corresponding to each round of brewing simulation.

[0014] This application also provides an intelligent coffee machine parameter adaptive adjustment system integrating environmental perception. The system includes: an environmental data acquisition module, used to collect brewing environment data in real time through a sensor array integrated on the coffee machine, and combine it with the input coffee bean material label to generate a multi-dimensional environmental disturbance feature characterizing the evolution of the current brewing environment; a probability distribution output module, used to input the multi-dimensional environmental disturbance feature into a Gaussian mixture distribution model, and use the Gaussian mixture distribution model to perform joint probability density mapping to output the uncertainty continuous probability distribution interval of the physical state parameters of the coffee bed under the current environmental disturbance; a parameter boundary acquisition module, used to use the uncertainty continuous probability distribution interval as boundary conditions to perform constraint solving to obtain the physical safety boundary of the control parameters; a brewing simulation module, used to perform multiple rounds of brewing simulation in a virtual control sandbox according to a preset time step for the physical safety boundary of the control parameters before water injection, to obtain the predicted value of solid solubility; and a brewing control scheme construction module, used to construct a target brewing control scheme based on the predicted value of solid solubility, and drive the execution unit of the coffee machine to execute the target brewing control scheme for parameter adaptive adjustment.

[0015] This application proposes a method and system for adaptive parameter adjustment of an intelligent coffee machine that integrates environmental perception. This involves collecting brewing environment data through a sensor array on the machine body, combining it with coffee bean material tags to construct multi-dimensional environmental disturbance characteristics; importing a Gaussian mixture distribution model to output the probability distribution range of the physical state parameters of the coffee bed, thus obtaining the physical safety boundary of the control parameters; conducting multiple rounds of brewing simulations using a virtual control sandbox before water injection to obtain predicted values ​​of solid solubility; and constructing a target brewing control scheme to drive the coffee machine's execution unit to adaptively adjust parameters. This solves the technical problem in existing technologies where the uncertainty of the coffee bed state cannot be quantified in real time under dynamic environmental disturbances and brewing parameters cannot be dynamically adjusted, leading to insufficient consistency in extracted flavor and personalized taste. It achieves the technical effect of improving the adaptability of brewing parameters and the control precision of flavor stability in coffee machines under changing environments. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0017] Figure 1 A schematic flowchart of the intelligent coffee machine parameter adaptive adjustment method integrating environmental perception provided in the embodiments of this application.

[0018] Figure 2 A schematic diagram of the structure of the intelligent coffee machine parameter adaptive adjustment system that integrates environmental perception, provided in an embodiment of this application.

[0019] Figure labeling: Environmental data acquisition module 10, probability distribution output module 20, parameter boundary acquisition module 30, brewing simulation module 40, brewing control scheme construction module 50. Detailed Implementation

[0020] To further illustrate the technical means and effects adopted by the present invention in order to achieve the intended purpose, the following detailed description is provided in conjunction with the accompanying drawings and preferred embodiments, based on the specific implementation methods, structures, features and effects of the present invention.

[0021] This application provides an embodiment of a smart coffee machine parameter adaptive adjustment method that integrates environmental perception, such as... Figure 1 As shown, the method includes: In step S100, the brewing environment data is collected in real time by the sensor array integrated on the coffee machine, and combined with the input coffee bean material label, a multi-dimensional environmental disturbance feature characterizing the evolution of the current brewing environment is generated.

[0022] Preferably, the coffee machine integrates a temperature sensor, a relative humidity sensor, an atmospheric pressure sensor, and a light sensor to form a sensor array, which collects brewing environment data in real time before and during brewing, including at least instantaneous ambient temperature, relative humidity, real-time ambient atmospheric pressure, and ambient light intensity.

[0023] Furthermore, step S100 also includes: extracting the instantaneous ambient temperature and relative humidity at the same time based on the brewing environment data, and calculating the moisture absorption integral of the relative humidity on the time axis to generate a dynamic moisture absorption accumulation effect term; using the derivative of the instantaneous ambient temperature with respect to time as the instantaneous ambient temperature gradient; using the instantaneous ambient temperature gradient as the main diagonal element and the dynamic moisture absorption accumulation effect term as the off-diagonal interleaving element to construct a temperature and humidity transient coupling matrix; performing interference analysis on the material corresponding to the coffee bean material label based on the temperature and humidity transient coupling matrix to obtain a nonlinear interference coefficient; extracting the real-time ambient atmospheric pressure time-series fluctuation rate from the brewing environment data, and cascading and splicing it with the nonlinear interference coefficient to obtain multidimensional environmental disturbance characteristics.

[0024] Preferably, the instantaneous ambient temperature and relative humidity are extracted from the brewing environment data at the same time. The curve of relative humidity changing with time is integrated, with the integration interval from a preset moment before brewing to the current moment, generating a dynamic moisture absorption accumulation effect term to characterize the degree of influence of air humidity on the cumulative moisture absorption of coffee powder. The first derivative of the instantaneous ambient temperature with respect to time is calculated to obtain the instantaneous ambient temperature gradient, characterizing the rate of temperature change with time. Then, the instantaneous ambient temperature gradient is used as the main diagonal element, and the dynamic moisture absorption accumulation effect term is used as the off-diagonal interleaving element to construct a transient temperature and humidity coupling matrix, which is used to characterize the temperature change and humidity accumulation effect. The interaction between temperature and humidity is analyzed to determine the nonlinear interaction between temperature and humidity, which physically interferes with the saturated water vapor pressure difference on the material surface. Next, the transient coupling matrix of temperature and humidity is analyzed against the material corresponding to the coffee bean material label to determine the nonlinear interference coefficient, characterizing the degree of interference of temperature and humidity coupling on the current coffee bean material. The volatility of atmospheric pressure over time is calculated based on brewing environment data, using statistical measures such as standard deviation, variance, or the sum of the absolute values ​​of differences between adjacent time points to characterize the severity of pressure fluctuations. Finally, the real-time atmospheric pressure temporal volatility and the nonlinear interference coefficient are concatenated in the vector dimension to generate multidimensional environmental disturbance characteristics.

[0025] Furthermore, step S100 also includes obtaining a preset material moisture absorption sensitivity feature vector based on the coffee bean material label; multiplying the temperature and humidity transient coupling matrix by the preset material moisture absorption sensitivity feature vector, and performing linear decoupling operation of the temperature and humidity coupling features through matrix multiplication to obtain the nonlinear interference coefficient.

[0026] Preferably, the coffee bean material label refers to information input by the user or pre-stored by the system to identify attributes such as coffee bean variety, origin, roasting degree, and processing method. Based on the coffee bean material label, a preset material moisture absorption sensitivity feature vector corresponding to the material label is retrieved from a pre-built database. The values ​​of each dimension of the vector represent the sensitivity of the coffee bean material to changes in environmental factors such as humidity and temperature. This vector is pre-trained and stored in the internal memory of the coffee machine through offline experiments or historical data. Then, the transient coupling matrix of temperature and humidity is multiplied by the preset material moisture absorption sensitivity feature vector, that is, each row of the matrix is ​​multiplied by the vector. The coupling information in the transient coupling matrix of temperature and humidity (the intertwined influence of temperature gradient and humidity cumulative effect) is linearly combined with the moisture absorption sensitivity characteristics of the material itself. This decomposes the temperature and humidity influencing factors into independent contribution components aligned with the material characteristics. The vector obtained by the matrix multiplication operation is the nonlinear interference coefficient, which represents the comprehensive nonlinear interference degree generated on a specific coffee bean material under the current temperature and humidity coupling change conditions.

[0027] Step S200: Input the multidimensional environmental disturbance features into the Gaussian mixture distribution model, use the Gaussian mixture distribution model to perform joint probability density mapping, and output the uncertainty continuous probability distribution interval of the physical state parameters of the powder bed under the current environmental disturbance.

[0028] Step S200 further includes: inputting the multidimensional environmental disturbance features as input vectors into the Gaussian mixture distribution model; calculating the conditional probability distribution of the input vector under each sub-Gaussian component using multiple preset sub-Gaussian probability density functions within the Gaussian mixture distribution model; linearly weighting and superimposing the conditional probability distributions under each sub-Gaussian component using preset component weight coefficients to obtain a global joint probability density function, wherein the expectation vector and covariance matrix of each sub-Gaussian component are pre-trained and determined by the coffee bean material label; setting a target confidence threshold, integrating the global joint probability density function, and extracting the continuous integral boundary that meets the target confidence threshold as the uncertainty continuous probability distribution interval.

[0029] Preferably, multidimensional environmental perturbation features are used as input and fed into a pre-constructed and stored Gaussian mixture distribution model. This model contains multiple sub-Gaussian probability density functions, each corresponding to a Gaussian component. The conditional probability density values ​​of the input vector under each sub-Gaussian component are calculated using these sub-Gaussian probability density functions to obtain the conditional probability distribution. Using preset component weight coefficients, the conditional probability distributions under each sub-Gaussian component are linearly weighted and superimposed to obtain a global joint probability density function, representing the joint probability distribution relationship between the input vector and the output variable. These preset component weight coefficients are pre-trained and determined offline based on experimental data corresponding to coffee bean material labels, including combinations of specific origins, varieties, roasting levels, and processing methods for each coffee bean material label. A large amount of sample data was collected, including multidimensional environmental disturbance feature vectors and corresponding measured values ​​of the physical state parameters of the coal bed. Using the expectation-maximization algorithm or variational inference, a Gaussian mixture model was fitted to the sample data, simultaneously estimating the weight coefficients, expectation vectors, and covariance matrices of each sub-Gaussian component. A target confidence threshold, such as 95% or 99%, was set, representing the probability that the output interval would contain the true value. The global joint probability density function was integrally solved over the range of output variable values ​​to determine a continuous integration interval. The integral value in this interval was equal to or greater than the target confidence threshold, i.e., the continuous region with the highest integration density. Finally, its boundary was taken as the uncertainty continuous probability distribution interval, including lower and upper bounds, indicating that under the current environmental disturbance, the physical state parameters of the coal bed fall within this interval with the target confidence level.

[0030] Preferably, for the same coffee bean material label, brewing experiments are conducted under various environmental conditions, collecting N sets of training data samples. Each set of samples contains a multi-dimensional environmental disturbance feature vector and measured values ​​of the physical state parameters of the coffee bed. The K-means clustering algorithm is used to divide the N samples into K clusters. For the k-th cluster, the weight coefficients are initialized to the number of samples in the cluster divided by the total number of samples, the expectation vector is initialized to the mean vector of the samples in the cluster, and the covariance matrix is ​​initialized to the covariance matrix of the samples in the cluster. Then, the expectation-maximization algorithm is executed iteratively for training, including repeatedly calculating the posterior probability of each sample belonging to the k-th sub-Gaussian component and updating the parameters, calculating the log-likelihood function under the current parameters, and if the absolute value of the change in the log-likelihood function between two consecutive iterations is less than a preset threshold (e.g., 10), the training is performed. −5 If the iteration stops, the output parameters are stored in the internal memory of the coffee machine and associated with the corresponding coffee bean material label, thus obtaining the Gaussian mixture distribution model.

[0031] Furthermore, step S200 also includes the following physical state parameters of the powder bed: the initial moisture content range of the powder bed, the dynamic exhaust resistance coefficient of carbon dioxide in the powder bed, and the dynamic heat loss rate of the brewing head.

[0032] Preferably, the initial moisture content range of the coffee grounds refers to the possible range of moisture content inside the coffee puck before brewing under current environmental disturbance conditions. That is, it represents the percentage of water mass in the total mass of the coffee grounds. Due to the uncertainty of environmental disturbances, the initial moisture content of the coffee grounds is a continuous range including a lower and upper limit. Influencing factors include relative humidity, ambient temperature, hygroscopic characteristics of coffee beans, initial moisture content of coffee beans, and exposure time. The initial moisture content of the coffee grounds determines the degree of clumping of the coffee grounds, the flow resistance characteristics of the coffee grounds, and the water absorption rate during water injection. The dynamic carbon dioxide exhaust resistance coefficient of the coffee grounds is a quantitative indicator of the resistance encountered by freshly roasted coffee beans when releasing carbon dioxide gas during brewing, as the gas penetrates the coffee grounds. The carbon dioxide gas inside freshly roasted coffee beans expands and is released rapidly upon contact with hot water, dynamically changing with brewing time, temperature, and pressure. It describes the degree to which the coffee grounds impede gas flow. The carbon dioxide exhaust resistance coefficient determines whether bubbles obstruct water flow in the initial brewing stage, whether the coffee grounds form a channel effect due to gas accumulation, and whether a pre-soaking stage is needed to remove gas. The dynamic heat loss rate of the brew head refers to the rate at which heat is lost during the brewing process as hot water flows from the heating system through the brew head to the coffee grounds. The brew head is the metal part in the coffee machine that comes into contact with the coffee grounds via the hot water outlet. The heat loss rate changes dynamically with ambient temperature, the temperature of the brew head itself, water flow rate, and brewing time. The dynamic heat loss rate of the brew head determines the deviation between the actual water temperature in contact with the coffee grounds and the set water temperature, the true temperature curve during the brewing process, and whether preheating of the brew head or temperature compensation is required.

[0033] Step S300: Use the interval of continuous probability distribution of uncertainty as the boundary condition to perform constraint solution and obtain the physical safety boundary of control parameters.

[0034] Preferably, the uncertainty continuous probability distribution interval refers to the range of values ​​for physical state parameters of the powder bed, such as initial moisture content of the powder bed, dynamic exhaust resistance coefficient of carbon dioxide in the powder bed, and dynamic heat loss rate of the brewing head, output by the Gaussian mixture distribution model, under the target confidence threshold. This range is used as boundary conditions for constraint solving, establishing a physical model of the control parameters and the powder bed state. The range of the powder bed physical state parameters is input into the physical model, and the inequality constraints that the control parameters need to satisfy are calculated. For each control parameter, the intersection of feasible ranges under all possible powder bed states is taken to determine the value range of control parameters such as grinding degree, water temperature, extraction pressure, and extraction time. This determines the physical safety boundary of the control parameters, ensuring that any value within this range... All of these measures ensure that the extraction process does not violate physical safety constraints. Specifically, for the flow resistance constraint of the coffee bed, based on Darcy's law or the Ogun equation, the resistance of the coffee bed to the water flow is related to the initial moisture content of the coffee bed and the carbon dioxide exhaust resistance coefficient. The constraint is that the pressure drop of the water flowing through the coffee bed is less than or equal to the maximum pressure that the coffee machine pump system can provide. For the extraction temperature constraint, the actual water temperature in contact with the coffee bed needs to be maintained within the effective extraction temperature range, that is, the difference between the set water temperature and the heat loss calculated from the dynamic heat loss rate range of the brewing head is within the preset effective extraction temperature range, such as 90℃ to 96℃. For the total solids solubility constraint, the total solids solubility of the extracted coffee should be maintained within the preset premium extraction range, such as 18% to 22%.

[0035] Step S400: Before starting the water injection, multiple rounds of boiling simulation are performed in the virtual control sandbox according to the physical safety boundary of the control parameters and a preset time step to obtain the predicted value of solid solubility.

[0036] Step S400 further includes loading a kinetic virtual operator in the virtual control sandbox that covers the relationship between the mass change rate and the total solids solubility change rate of the material corresponding to the coffee bean material label; according to a preset time step, within the physical safety boundary of the control parameters, traversing different virtual execution power combinations in the historical virtual power combination library to perform multiple rounds of brewing simulation, running the kinetic virtual operator, and obtaining the solids solubility prediction value.

[0037] Preferably, the virtual control sandbox is a software simulation environment built in the internal memory of the coffee machine. It is used to simulate the brewing process and extraction results under different control parameters without consuming coffee materials or actually running the pump and heater. The kinetic virtual operator is a mathematical model loaded in the virtual control sandbox, which covers the relationship between the mass change rate and the total solids solubility change rate of the materials corresponding to the coffee bean material label. Here, the mass change rate refers to the rate of change in the mass of solid substances in the coffee grounds that dissolve and migrate into the coffee liquid per unit time during the brewing process, and the total solids solubility change rate refers to the rate of change in the total solids solubility in the coffee liquid per unit time during the brewing process. The functional relationship expressed by the kinetic virtual operator is obtained by fitting offline experimental data and associated with and stored with specific coffee bean material labels. The preset time step refers to the time discretization interval of the brewing simulation, such as 0.1 seconds, 0.5 seconds, or 1.0 seconds, which is used to discretize the continuous brewing process into multiple time steps. In each time step, the mass change rate, the total solids solubility change rate, and other state variables are updated sequentially.

[0038] Preferably, a historical virtual power combination library is pre-stored inside the coffee machine. This library contains multiple different virtual execution power combinations, each of which is a vector containing specific values ​​for multiple control parameters. The historical virtual execution power combination library may include control parameter grid points generated by offline optimization algorithms and control parameter combinations that have performed well in historical actual brewing. Then, within the physical safety boundary of the control parameters, multiple rounds of brewing simulation are performed for each virtual execution power combination. Specifically, it is determined whether each control parameter in the virtual execution power combination falls within the corresponding physical safety boundary. If all parameters are within the boundary, the calculation is iteratively performed step by step according to a preset time step. That is, the coffee bed state is initialized at the initial moment. At each time step, the mass change rate is calculated based on the current control parameters and the current coffee bed state. Then, the kinetic virtual operator is input, and the total solids solubility change rate is calculated in combination with the current control parameters and the coffee bed state. Then, the cumulative dissolved mass, total solids solubility, and coffee bed state are updated. After the brewing time is reached, the final total solids solubility prediction value is recorded. The process is iteratively executed to traverse all virtual execution power combinations to obtain the solids solubility prediction value. Each value corresponds to the simulated extraction result of a set of virtual execution power combinations within the physical safety boundary.

[0039] Step S500: Based on the predicted solid solubility value, a target brewing control scheme is constructed, and the execution unit of the coffee machine is driven to execute the target brewing control scheme to perform parameter adaptive adjustment.

[0040] Step S500 further includes: retrieving the shear force distribution gradient of fluid penetrating the virtual porous powder bed when predicting solubility in each round of brewing simulation using the kinetic virtual operator; calculating the probability value of channel effect caused by sudden flow velocity changes based on the shear force distribution gradient corresponding to each round of brewing simulation; when the probability value exceeds a preset safety threshold, introducing a numerical penalty term into the control parameters of the simulation round to obtain a physical compliance score; traversing the solid solubility prediction value and the physical compliance score, extracting the virtual execution power combination parameter corresponding to the brewing simulation round where the solid solubility prediction value converges within a preset premium extraction range and the physical compliance score is the highest, and using it as the target brewing control scheme.

[0041] Preferably, the shear force distribution gradient is retrieved when predicting solubility in each brewing round using the kinetic virtual operator. The shear force distribution gradient refers to the spatial distribution difference of the shear force exerted by hot water on the surface of coffee particles at different locations as it penetrates the virtual porous coffee bed. Channel effect refers to the phenomenon where fluid preferentially flows along one or a few low-resistance paths when penetrating the coffee bed, causing most of the fluid to bypass other coffee bed areas, resulting in uneven extraction. When the shear force distribution gradient shows a localized area with significantly higher shear force than adjacent areas, it indicates a sudden increase in flow velocity in that area. The greater the magnitude of the velocity change, the higher the probability of channel effect. Based on the shear force distribution gradient corresponding to each brewing round, the probability value of channel effect caused by the velocity change is calculated. ,in, It is the standard deviation of the shear force distribution gradient in the j-th simulation, characterizing the degree of non-uniformity. It is the mean of the shear force distribution gradient in the j-th simulation. It is a predefined mapping function, such as the Sigmoid function or a piecewise linear function, that maps the coefficient of variation to a probability value between 0 and 1.

[0042] Preferably, a preset safety threshold, such as 0.3 or 0.5, represents the upper limit of the acceptable channel effect probability. When the probability value exceeds the preset safety threshold, a numerical penalty term, i.e., a non-negative number, is introduced into the control parameters of the simulation round. The difference between the initial score and the introduced numerical penalty term is calculated to determine the physical compliance score. If the penalty term is +∞, the physical compliance score of the simulation round is directly set to negative infinity or marked as invalid, thus obtaining the physical compliance score. Then, the predicted solid solubility value and physical compliance score of each round of brewing simulation are obtained. The virtual execution power combination parameters corresponding to the brewing simulation round with the highest physical compliance score and the predicted solid solubility value converging within the preset premium extraction range are retained as the target brewing control scheme, which includes grinding degree, water temperature, extraction pressure, extraction time, etc.

[0043] Furthermore, step S500 also includes discretizing the virtual porous powder bed along the fluid flow direction and radial cross section to obtain grid cells; extracting the fluid shear force values ​​of each grid cell within the preset time step to construct the shear force distribution gradient corresponding to each round of brewing simulation.

[0044] Preferably, the virtual porous powder bed is divided into grids along the fluid flow direction (axial) and radial cross-section to obtain grid cells. Within each preset time step, the shear force value exerted by the fluid on the powder bed in each grid cell is calculated. Then, the shear force values ​​of all grid cells are organized according to spatial location to form the shear force distribution gradient corresponding to each round of boiling simulation.

[0045] In the above text, refer to Figure 1 This paper describes in detail a method for adaptive adjustment of parameters of an intelligent coffee machine based on environmental perception, according to an embodiment of the present invention. Next, reference will be made to... Figure 2 This invention describes an intelligent coffee machine parameter adaptive adjustment system that integrates environmental perception according to an embodiment of the present invention.

[0046] The intelligent coffee machine parameter adaptive adjustment system integrating environmental perception according to embodiments of the present invention is used to solve the technical problem in the prior art that it is impossible to quantify the uncertainty of the coffee bed state in real time under dynamic environmental disturbances and dynamically adjust the brewing parameters, resulting in insufficient consistency of extraction flavor and personalized taste. It achieves the technical effect of improving the adaptability of brewing parameters and the control precision of flavor stability in coffee machines under changing environments. Figure 2 As shown, the intelligent coffee machine parameter adaptive adjustment system integrating environmental perception includes: an environmental data acquisition module 10, a probability distribution output module 20, a parameter boundary acquisition module 30, a brewing simulation module 40, and a brewing control scheme construction module 50.

[0047] The environmental data acquisition module 10 is used to collect brewing environment data in real time through a sensor array integrated on the coffee machine, and combine it with the input coffee bean material label to generate multi-dimensional environmental disturbance features characterizing the evolution of the current brewing environment; the probability distribution output module 20 is used to input the multi-dimensional environmental disturbance features into a Gaussian mixture distribution model, and use the Gaussian mixture distribution model to perform joint probability density mapping to output the uncertainty continuous probability distribution interval of the physical state parameters of the coffee bed under the current environmental disturbance; the parameter boundary acquisition module 30 is used to use the uncertainty continuous probability distribution interval as boundary conditions to perform constraint solving to obtain the physical safety boundary of the control parameters; the brewing simulation module 40 is used to perform multiple rounds of brewing simulation in a virtual control sandbox before water injection, based on the physical safety boundary of the control parameters according to a preset time step, to obtain the predicted value of solid solubility; the brewing control scheme construction module 50 is used to construct a target brewing control scheme based on the predicted value of solid solubility, and drive the execution unit of the coffee machine to execute the target brewing control scheme for parameter adaptive adjustment.

[0048] The specific configuration of the parameter boundary acquisition module 30 will be described in detail below. The parameter boundary acquisition module 30 further includes: the physical safety boundary of the control parameter is a control parameter range that keeps the total solids solubility within a preset refined extraction range.

[0049] The specific configuration of the environmental data acquisition module 10 will be described in detail below. The environmental data acquisition module 10 further includes: extracting the instantaneous ambient temperature and relative humidity at the same time based on the brewing environment data, and calculating the moisture absorption integral of the relative humidity on the time axis to generate a dynamic moisture absorption cumulative effect term; using the derivative of the instantaneous ambient temperature with respect to time as the instantaneous ambient temperature gradient; using the instantaneous ambient temperature gradient as the main diagonal element and the dynamic moisture absorption cumulative effect term as the off-diagonal interleaving element to construct a temperature and humidity transient coupling matrix; performing interference analysis on the material corresponding to the coffee bean material label based on the temperature and humidity transient coupling matrix to obtain a nonlinear interference coefficient; extracting the real-time ambient atmospheric pressure time-series fluctuation rate from the brewing environment data, and cascading and splicing it with the nonlinear interference coefficient to obtain multidimensional environmental disturbance characteristics.

[0050] The specific configuration of the environmental data acquisition module 10 will be described in detail below. The environmental data acquisition module 10 further includes: obtaining a preset material moisture absorption sensitivity feature vector based on the coffee bean material label; multiplying the temperature and humidity transient coupling matrix by the preset material moisture absorption sensitivity feature vector, and performing linear decoupling operation of the temperature and humidity coupling features through matrix multiplication to obtain the nonlinear interference coefficient.

[0051] The specific configuration of the probability distribution output module 20 will be described in detail below. The probability distribution output module 20 further includes: the physical state parameters of the powder bed including the initial moisture content range of the powder bed, the dynamic exhaust resistance coefficient of carbon dioxide in the powder bed, and the dynamic heat loss rate of the cooking head.

[0052] The specific configuration of the probability distribution output module 20 will be described in detail below. The probability distribution output module 20 further includes: inputting the multidimensional environmental disturbance features as an input vector into the Gaussian mixture distribution model; calculating the conditional probability distribution of the input vector under each sub-Gaussian component using multiple preset sub-Gaussian probability density functions within the Gaussian mixture distribution model; linearly weighting and superimposing the conditional probability distributions under each sub-Gaussian component using preset component weight coefficients to obtain a global joint probability density function, wherein the expectation vector and covariance matrix of each sub-Gaussian component are pre-trained and determined by the coffee bean material label; setting a target confidence threshold, integrating the global joint probability density function, and extracting the continuous integral boundary that meets the target confidence threshold as the uncertainty continuous probability distribution interval.

[0053] The specific configuration of the brewing simulation module 40 will be described in detail below. The brewing simulation module 40 further includes: loading a kinetic virtual operator in the virtual control sandbox that covers the relationship between the mass change rate and the total solids solubility change rate of the material corresponding to the coffee bean material label; performing multiple rounds of brewing simulation by traversing different virtual execution power combinations in the historical virtual power combination library within the physical safety boundary of the control parameters according to a preset time step, running the kinetic virtual operator, and obtaining the predicted value of solids solubility.

[0054] The specific configuration of the brewing control scheme construction module 50 will be described in detail below. The brewing control scheme construction module 50 further includes: retrieving the shear force distribution gradient of the fluid penetrating the virtual porous powder bed during solubility prediction in each round of brewing simulation using the kinetic virtual operator; calculating the probability value of channel effects caused by sudden changes in flow rate based on the shear force distribution gradient corresponding to each round of brewing simulation; when the probability value exceeds a preset safety threshold, introducing a numerical penalty term into the control parameters of that simulation round to obtain a physical compliance score; traversing the solid solubility prediction value and the physical compliance score, extracting the virtual execution power combination parameter corresponding to the brewing simulation round where the solid solubility prediction value converges within a preset premium extraction range and has the highest physical compliance score, and using this as the target brewing control scheme.

[0055] The specific configuration of the brewing control scheme construction module 50 will be described in detail below. The brewing control scheme construction module 50 further includes: discretizing the virtual porous powder bed along the fluid flow direction and radial cross section to obtain grid cells; extracting the fluid shear force values ​​of each grid cell within the preset time step to construct the shear force distribution gradient corresponding to each brewing simulation.

[0056] The intelligent coffee machine parameter adaptive adjustment system with integrated environmental perception provided in this embodiment of the invention can execute the intelligent coffee machine parameter adaptive adjustment method with integrated environmental perception provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0057] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for adaptive parameter adjustment of an intelligent coffee machine integrating environmental perception, characterized in that, The method includes: By collecting real-time brewing environment data through a sensor array integrated on the coffee machine, and combining it with the input coffee bean material label, a multi-dimensional environmental disturbance feature characterizing the evolution of the current brewing environment is generated. The multidimensional environmental disturbance features are input into the Gaussian mixture distribution model, and the joint probability density mapping is performed using the Gaussian mixture distribution model to output the uncertainty continuous probability distribution interval of the physical state parameters of the powder bed under the current environmental disturbance. The uncertainty continuous probability distribution interval is used as the boundary condition to perform constraint solving to obtain the physical safety boundary of the control parameters; Before water injection is started, multiple rounds of boiling simulation are performed in a virtual control sandbox according to a preset time step, targeting the physical safety boundary of the control parameters, to obtain the predicted value of solid solubility. Based on the predicted solid solubility value, a target brewing control scheme is constructed, and the execution unit of the coffee machine is driven to execute the target brewing control scheme to perform adaptive parameter adjustment; Specifically, a sensor array integrated into the coffee machine collects real-time brewing environment data and, combined with the input coffee bean material tags, generates multi-dimensional environmental disturbance characteristics representing the evolution of the current brewing environment, including: Based on the brewing environment data, the instantaneous ambient temperature and relative humidity at the same time are extracted, and the moisture absorption integral of the relative humidity on the time axis is calculated to generate a dynamic moisture absorption cumulative effect term. The derivative of the instantaneous ambient temperature with respect to time is taken as the instantaneous ambient temperature gradient; The instantaneous ambient temperature gradient is used as the main diagonal element, and the dynamic moisture absorption accumulation effect term is used as the off-diagonal interleaved element to construct a transient coupling matrix of temperature and humidity. Based on the transient coupling matrix of temperature and humidity, interference analysis is performed on the material corresponding to the coffee bean material label to obtain the nonlinear interference coefficient. The real-time atmospheric pressure fluctuation rate is extracted from the brewing environment data and concatenated with the nonlinear interference coefficient to obtain multi-dimensional environmental disturbance characteristics. The physical state parameters of the powder bed include the initial moisture content range of the powder bed, the dynamic exhaust resistance coefficient of carbon dioxide in the powder bed, and the dynamic heat loss rate of the cooking head.

2. The intelligent coffee machine parameter adaptive adjustment method integrating environmental perception as described in claim 1, characterized in that, The physical safety boundary of the control parameters is the range of control parameters that keeps the total solids solubility within the preset refined extraction range.

3. The intelligent coffee machine parameter adaptive adjustment method integrating environmental perception as described in claim 1, characterized in that, Based on the transient coupling matrix of temperature and humidity, interference analysis is performed on the material corresponding to the coffee bean material label to obtain the nonlinear interference coefficient, including: Based on the coffee bean material label, obtain the preset material moisture absorption sensitivity feature vector corresponding to the material; The transient coupling matrix of temperature and humidity is multiplied by a preset material moisture absorption sensitivity feature vector. The linear decoupling operation of the temperature and humidity coupling feature is completed by matrix multiplication to obtain the nonlinear interference coefficient.

4. The intelligent coffee machine parameter adaptive adjustment method integrating environmental perception as described in claim 1, characterized in that, The multidimensional environmental disturbance features are input into a Gaussian mixture model. A joint probability density mapping is performed using the Gaussian mixture model, outputting the continuous probability distribution interval of the uncertainty of the physical state parameters of the powder bed under the current environmental disturbance, including: The multidimensional environmental disturbance features are input as input vectors into the Gaussian mixture distribution model. Multiple sub-Gaussian probability density functions preset within the Gaussian mixture distribution model are used to calculate the conditional probability distribution of the input vector under each sub-Gaussian component. Using preset component weighting coefficients, the conditional probability distributions under each sub-Gaussian component are linearly weighted and superimposed to obtain the global joint probability density function. The expectation vector and covariance matrix of each sub-Gaussian component are pre-trained and determined by the coffee bean material label. Set a target confidence threshold, integrate the global joint probability density function, and extract the continuous integral boundary that meets the target confidence threshold as the uncertainty continuous probability distribution interval.

5. The intelligent coffee machine parameter adaptive adjustment method integrating environmental perception as described in claim 1, characterized in that, Before water injection is initiated, multiple rounds of boiling simulations are performed in a virtual control sandbox according to a preset time step, targeting the physical safety boundaries of the control parameters, to obtain predicted values ​​for solid solubility, including: A kinetic virtual operator covering the relationship between the mass change rate and the total solids solubility change rate of the materials corresponding to the coffee bean material labels is loaded into the virtual control sandbox; According to the preset time step, within the physical safety boundary of the control parameters, multiple rounds of brewing simulation are performed by traversing different virtual execution power combinations in the historical virtual power combination library, and the kinetic virtual operator is run to obtain the predicted value of solid solubility.

6. The intelligent coffee machine parameter adaptive adjustment method integrating environmental perception as described in claim 5, characterized in that, Based on the predicted solid solubility values, a target brewing control scheme is constructed, including: The shear force distribution gradient of fluid penetrating the virtual porous powder bed is obtained when the solubility is predicted in each round of boiling simulation by retrieving the aforementioned kinetic virtual operator; The probability value of channel effect caused by sudden flow velocity change is calculated based on the shear force distribution gradient corresponding to each round of boiling simulation. When the probability value exceeds a preset safety threshold, a numerical penalty term is introduced into the control parameters of the simulation round to obtain a physical compliance score. By iterating through the predicted solid solubility values ​​and physical compliance scores, the virtual execution power combination parameters corresponding to the brewing simulation rounds with the highest physical compliance scores and whose predicted solid solubility values ​​converge within the preset premium extraction range are extracted and used as the target brewing control scheme.

7. The intelligent coffee machine parameter adaptive adjustment method integrating environmental perception as described in claim 6, characterized in that, The shear force distribution gradient of fluid penetrating the virtual porous powder bed when retrieving the solubility prediction of each round of boiling simulation using the aforementioned kinetic virtual operator includes: The virtual porous powder bed is discretized into a mesh space along the fluid flow direction and radial section to obtain multiple mesh elements; Extract the fluid shear force values ​​of each grid cell within the preset time step from multiple grid cells, and construct the shear force distribution gradient corresponding to each round of boiling simulation.

8. An intelligent coffee machine parameter adaptive adjustment system integrating environmental perception, characterized in that, The system is used to implement the intelligent coffee machine parameter adaptive adjustment method integrating environmental perception as described in any one of claims 1 to 7, the system comprising: The environmental data acquisition module is used to collect brewing environment data in real time through a sensor array integrated on the coffee machine, and combine it with the input coffee bean material label to generate multi-dimensional environmental disturbance characteristics that characterize the evolution of the current brewing environment. The probability distribution output module is used to input the multidimensional environmental disturbance features into the Gaussian mixture distribution model, perform joint probability density mapping using the Gaussian mixture distribution model, and output the uncertainty continuous probability distribution interval of the physical state parameters of the powder bed under the current environmental disturbance. The parameter boundary acquisition module is used to perform constraint solving by taking the continuous probability distribution interval of the uncertainty as the boundary condition to obtain the physical safety boundary of the control parameters. The brewing simulation module is used to perform multiple rounds of brewing simulation in a virtual control sandbox before water injection starts, based on the physical safety boundary of the control parameters and according to a preset time step, to obtain the predicted value of solid solubility. The brewing control scheme construction module is used to construct a target brewing control scheme based on the predicted solid solubility value, and drive the coffee machine's execution unit to execute the target brewing control scheme for adaptive parameter adjustment.

Images