Brewing control method and system based on historical data analysis

Abstract

This invention discloses a water boiling control method and system based on historical data analysis. The method includes: acquiring motion sensing data from multiple historical time points using a sensor installed at the bottom of the kettle; training a motion prediction model based on the motion sensing data corresponding to all historical time points; predicting user motion characteristics at the current time point based on the motion prediction model and the motion sensing data from the previous time point; and determining the corresponding water boiling control command for the kettle based on the predicted user motion characteristics in response to the user motion detection signal at the current time point. Therefore, this invention can achieve accurate water boiling command generation based on motion timing analysis, improving the accuracy of intelligent kettle operation and user experience.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a boiling water control method and system based on historical data analysis. Background Technology

[0002] With the rapid growth in demand for smart kettles, businesses and users are increasingly emphasizing the improvement of ease of operation and user experience through intelligent control. Existing technologies typically collect real-time motion data from sensors at the bottom of the kettle, using simple signal triggering or fixed threshold analysis methods to determine user intent and generate boiling commands based on standard rules to control the device. However, existing solutions lack time-series analysis of historical motion data and dynamic prediction of user action characteristics, making it difficult to accurately identify complex operational intentions. Commonly used static detection mechanisms cannot adapt to diverse usage scenarios, resulting in insufficient accuracy in command generation, easily leading to misoperation or device response delays, thus limiting the level of intelligence and operational safety of the kettle. Therefore, existing technologies have shortcomings that urgently need to be addressed. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a water boiling control method and system based on historical data analysis, which can realize the generation of accurate water boiling instructions based on action timing analysis, thereby improving the accuracy of intelligent operation of the kettle and the user experience.

[0004] To address the aforementioned technical problems, the first aspect of this invention discloses a water boiling control method based on historical data analysis, the method comprising:

[0005] Motion sensing data from multiple historical time points is acquired by a sensor placed at the bottom of the kettle.

[0006] Based on the motion sensing data corresponding to all the historical time points, a motion prediction model is trained.

[0007] Based on the action prediction model, the predicted user action features at the current time point are predicted according to the action sensing data at the previous time point.

[0008] In response to the user action detection signal at the current time point, the boiling control command corresponding to the kettle is determined based on the predicted user action characteristics.

[0009] As an optional implementation, in the first aspect of the present invention, training the action prediction model based on the action sensing data corresponding to all the historical time points includes:

[0010] Based on action recognition rules, determine the action feature parameters corresponding to each action sensing data;

[0011] Based on the historical time points, the motion feature parameters corresponding to all the motion sensing data are sorted from morning to night to obtain the parameter time series;

[0012] Using the time series of the parameters as the training dataset, a preset LSTM neural network is trained to obtain an action prediction model.

[0013] As an optional implementation, in the first aspect of the present invention, the action feature parameters include action type, action height, action angle, action speed, and action frequency; the action type is pouring water out of the kettle, adding water to the kettle, adding tea leaves to the kettle, starting to heat the kettle, or stopping to heat the kettle.

[0014] As an optional implementation, in the first aspect of the present invention, the step of training a preset LSTM neural network using the parameter time series as a training dataset to obtain an action prediction model includes:

[0015] For any two adjacent motion feature parameters in the parameter time series, calculate the difference between the motion heights of the two motion feature parameters to obtain the height difference.

[0016] Calculate the difference between the motion angles of the two motion feature parameters to obtain the angle difference;

[0017] Calculate the difference between the movement frequencies of the two aforementioned movement feature parameters to obtain the frequency difference;

[0018] Calculate the difference between the motion velocities of the two motion characteristic parameters to obtain the velocity difference;

[0019] The weighted sum of the height difference, angle difference, speed difference, and frequency difference is calculated to obtain the change characteristic parameters;

[0020] Calculate the ratio of the time difference between the change feature parameter and the historical time point corresponding to the two action feature parameters to obtain the correlation feature between the two action feature parameters;

[0021] The time series of the parameters and all the corresponding associated features are concatenated into a fusion feature, which is then input into a preset LSTM neural network for training to obtain an action prediction model.

[0022] As an optional implementation, in a first aspect of the present invention, the loss function of the LSTM neural network is designed as the product of a first cross-entropy and a second cross-entropy; the first cross-entropy is the difference between the predicted output and the label corresponding to the parameter time series; the second cross-entropy is the difference between the output association feature and the actual association feature; the output association feature is the association feature between the predicted output and the predicted output of the previous time step; the actual association feature is the actual association feature between the corresponding parameters of the parameter time series.

[0023] As an optional implementation, in a first aspect of the invention, the sensor includes a controller and a gyroscope sensor, an infrared altitude sensor, and a weight sensor respectively connected to the controller; the gyroscope sensor is used to detect the motion angle and the motion speed; the infrared altitude sensor is used to detect the motion height; the weight sensor is used to detect the weight parameter of the kettle; and the controller is used to identify the motion type and the motion frequency based on the motion height, the motion angle, and the motion speed.

[0024] As an optional implementation, in the first aspect of the present invention, the step of predicting the predicted user action features at the current time point based on the action prediction model and the action sensing data at the previous time point includes:

[0025] Acquire motion sensor data from at least the two time points preceding the current time point and identify them as historical sensor data;

[0026] Calculate the correlation features between the two historical sensor data sets;

[0027] The historical sensor data and corresponding associated features are input into the action prediction model to obtain the predicted user action features at the current time point.

[0028] As an optional implementation, in the first aspect of the present invention, determining the water boiling control command corresponding to the kettle based on the predicted user action characteristics includes:

[0029] Obtain the current sensor data and corresponding detection device parameters corresponding to the user action detection signal;

[0030] Determine the reference sensing features corresponding to the predicted user action features in the preset database corresponding to the detection device parameters;

[0031] Calculate the feature similarity between the current sensing data and the reference sensing features;

[0032] Determine whether the feature similarity is greater than a preset similarity threshold;

[0033] If not, wait for the user to actually press the button to determine the corresponding boiling control command for the kettle;

[0034] If so, based on the predicted user action characteristics and the preset correspondence between actions and boiling water instructions, the corresponding boiling water control instruction for the kettle is determined.

[0035] A second aspect of this invention discloses a water boiling control system based on historical data analysis, the system comprising:

[0036] The acquisition module is used to acquire motion sensing data at multiple historical time points through a sensor located at the bottom of the kettle;

[0037] The training module is used to train an action prediction model based on the action sensing data corresponding to all the historical time points.

[0038] The prediction module is used to predict the user's action characteristics at the current time point based on the action prediction model and the action sensing data at the previous time point.

[0039] The determination module is used to respond to the user action detection signal at the current time point and determine the water boiling control command corresponding to the kettle based on the predicted user action characteristics.

[0040] As an optional implementation, in the second aspect of the present invention, the specific method by which the training module trains the action prediction model based on the action sensing data corresponding to all the historical time points includes:

[0041] Based on action recognition rules, determine the action feature parameters corresponding to each action sensing data;

[0042] Based on the historical time points, the motion feature parameters corresponding to all the motion sensing data are sorted from morning to night to obtain the parameter time series;

[0043] Using the time series of the parameters as the training dataset, a preset LSTM neural network is trained to obtain an action prediction model.

[0044] As an optional implementation, in the second aspect of the present invention, the motion characteristic parameters include motion type, motion height, motion angle, motion speed, and motion frequency; the motion type is pouring water out of the kettle, adding water to the kettle, adding tea leaves to the kettle, starting to heat the kettle, or stopping to heat the kettle.

[0045] As an optional implementation, in a second aspect of the present invention, the training module uses the parameter time series as a training dataset to train a preset LSTM neural network to obtain an action prediction model, specifically in the following manner:

[0046] For any two adjacent motion feature parameters in the parameter time series, calculate the difference between the motion heights of the two motion feature parameters to obtain the height difference.

[0047] Calculate the difference between the motion angles of the two motion feature parameters to obtain the angle difference;

[0048] Calculate the difference between the movement frequencies of the two aforementioned movement feature parameters to obtain the frequency difference;

[0049] Calculate the difference between the motion velocities of the two motion characteristic parameters to obtain the velocity difference;

[0050] The weighted sum of the height difference, angle difference, speed difference, and frequency difference is calculated to obtain the change characteristic parameters;

[0051] Calculate the ratio of the time difference between the change feature parameter and the historical time point corresponding to the two action feature parameters to obtain the correlation feature between the two action feature parameters;

[0052] The time series of the parameters and all the corresponding associated features are concatenated into a fusion feature, which is then input into a preset LSTM neural network for training to obtain an action prediction model.

[0053] As an optional implementation, in a second aspect of the invention, the loss function of the LSTM neural network is designed as the product of a first cross-entropy and a second cross-entropy; the first cross-entropy is the difference between the predicted output and the label corresponding to the parameter time series; the second cross-entropy is the difference between the output association feature and the actual association feature; the output association feature is the association feature between the predicted output and the predicted output of the previous time step; the actual association feature is the actual association feature between the corresponding parameters of the parameter time series.

[0054] As an optional implementation, in a second aspect of the invention, the sensor includes a controller and a gyroscope sensor, an infrared altitude sensor, and a weight sensor respectively connected to the controller; the gyroscope sensor is used to detect the motion angle and the motion speed; the infrared altitude sensor is used to detect the motion height; the weight sensor is used to detect the weight parameter of the kettle; and the controller is used to identify the motion type and the motion frequency based on the motion height, the motion angle, and the motion speed.

[0055] As an optional implementation, in a second aspect of the invention, the specific method by which the prediction module predicts the user's action characteristics at the current time point based on the action prediction model and the action sensing data from the previous time point includes:

[0056] Acquire motion sensor data from at least the two time points preceding the current time point and identify them as historical sensor data;

[0057] Calculate the correlation features between the two historical sensor data sets;

[0058] The historical sensor data and corresponding associated features are input into the action prediction model to obtain the predicted user action features at the current time point.

[0059] As an optional implementation, in a second aspect of the invention, the determining module determines the specific method of the water boiling control command corresponding to the kettle based on the predicted user action characteristics, including:

[0060] Obtain the current sensor data and corresponding detection device parameters corresponding to the user action detection signal;

[0061] Determine the reference sensing features corresponding to the predicted user action features in the preset database corresponding to the detection device parameters;

[0062] Calculate the feature similarity between the current sensing data and the reference sensing features;

[0063] Determine whether the feature similarity is greater than a preset similarity threshold;

[0064] If not, wait for the user to actually press the button to determine the corresponding boiling control command for the kettle;

[0065] If so, based on the predicted user action characteristics and the preset correspondence between actions and boiling water instructions, the corresponding boiling water control instruction for the kettle is determined.

[0066] A third aspect of this invention discloses another water boiling control system based on historical data analysis, the system comprising:

[0067] Memory containing executable program code;

[0068] A processor coupled to the memory;

[0069] The processor calls the executable program code stored in the memory to execute some or all of the steps in the boiling water control method based on historical data analysis disclosed in the first aspect of the present invention.

[0070] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in the boiling water control method based on historical data analysis disclosed in the first aspect of the present invention.

[0071] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0072] This invention trains a motion prediction model by acquiring motion sensing data from the bottom sensor of the kettle at historical time points. Based on the data from the previous time point, it predicts the current user's motion characteristics and further determines the boiling control command. This enables the generation of accurate boiling commands based on motion timing analysis, improving the accuracy of intelligent operation of the kettle and the user experience. Attached Figure Description

[0073] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0074] Figure 1 This is a schematic flowchart of a water boiling control method based on historical data analysis disclosed in an embodiment of the present invention.

[0075] Figure 2 This is a schematic diagram of a water boiling control system based on historical data analysis disclosed in an embodiment of the present invention.

[0076] Figure 3 This is a schematic diagram of another water boiling control system based on historical data analysis disclosed in an embodiment of the present invention. Detailed Implementation

[0077] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0078] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0079] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0080] This invention discloses a water boiling control method and system based on historical data analysis. By acquiring motion sensing data from sensors at the bottom of the kettle at historical time points to train a motion prediction model, and predicting the current user's motion characteristics based on data from the previous time point, the water boiling control command is further determined. This enables the generation of precise water boiling commands based on motion timing analysis, improving the accuracy of intelligent kettle operation and user experience. Detailed descriptions follow.

[0081] Example 1

[0082] Please see Figure 1 , Figure 1 This is a schematic flowchart of a water boiling control method based on historical data analysis disclosed in an embodiment of the present invention. Figure 1 The described water boiling control method based on historical data analysis can be applied to data processing systems / data processing equipment / data processing servers (including local processing servers or cloud processing servers). For example... Figure 1 As shown, the water boiling control method based on historical data analysis may include the following operations:

[0083] 101. Motion sensing data from multiple historical time points are acquired through a sensor located at the bottom of the kettle.

[0084] Optionally, the sensor can be an infrared sensor, a gyroscope sensor, a weight sensor, or a combination of multiple sensors; the present invention does not limit this.

[0085] Optionally, the acquisition process can be implemented based on real-time acquisition, periodic sampling, triggered acquisition, or low-power acquisition, and this invention does not limit it.

[0086] 102. Based on motion sensing data corresponding to all historical time points, a motion prediction model is trained.

[0087] Optionally, the action prediction model can be an LSTM neural network model, a Transformer model, or a recurrent neural network model; this invention does not impose any limitations.

[0088] Optionally, the training process can be implemented based on supervised learning, semi-supervised learning, or reinforcement learning, and this invention does not limit it.

[0089] Optionally, the action prediction model can be optimized by combining sensor characteristics or user behavior patterns, but this invention does not limit it.

[0090] 103. Based on the action prediction model, predict the user action characteristics at the current time point based on the action sensing data at the previous time point.

[0091] Optionally, the predicted user action characteristics may include action type, action amplitude, action frequency, or action speed, which is not limited in this invention.

[0092] Optionally, the prediction process can be implemented based on real-time inference, batch processing, or incremental calculation, and the present invention does not limit it.

[0093] 104. In response to the user action detection signal at the current time point, determine the corresponding boiling control command for the kettle based on the predicted user action characteristics.

[0094] Optionally, the user action detection signal can be a sensor trigger signal, a user interaction signal, or an external input signal; the present invention does not limit this.

[0095] Optionally, the water boiling control command may include commands to start heating, stop heating, add water, pour out water, or keep warm; this invention does not limit this.

[0096] As can be seen, the above-mentioned embodiments of the invention train an action prediction model by acquiring motion sensing data from the bottom sensor of the kettle at historical time points, predicting the current user's action characteristics based on the data from the previous time point, and further determining the boiling control command. This enables the generation of accurate boiling commands based on action timing analysis, improving the accuracy of the kettle's intelligent operation and the user experience.

[0097] As an optional embodiment, the step above, training the action prediction model based on action sensing data corresponding to all historical time points, includes:

[0098] Based on action recognition rules, determine the action feature parameters corresponding to each action sensing data;

[0099] Based on historical time points, the motion feature parameters corresponding to all motion sensing data are sorted from morning to night to obtain the parameter time series;

[0100] Using the parameter time series as the training dataset, a pre-defined LSTM neural network is trained to obtain an action prediction model.

[0101] Optionally, the action recognition rule can be a feature extraction rule built into the sensor or a corresponding pattern matching rule or machine learning model, and this invention does not limit it.

[0102] Optionally, the LSTM neural network can be a single-layer LSTM, a multi-layer LSTM, a bidirectional LSTM, or an attention mechanism LSTM; this invention does not limit the specific type of LSTM.

[0103] As can be seen, through the above optional embodiments, by extracting action feature parameters based on action recognition rules and generating parameter time series by sorting them by time, and training an LSTM neural network to obtain an action prediction model, the accuracy of action prediction is improved by extracting time series features and training the model on the basis of accurate water boiling command generation. This provides reliable action feature support for water boiling control commands and reduces the risk of command misjudgment caused by deviations in action feature extraction.

[0104] As an optional embodiment, the action characteristic parameters in the above steps include action type, action height, action angle, action speed, and action frequency; the action type is pouring water out of the kettle, adding water to the kettle, adding tea leaves to the kettle, starting to heat the kettle, or stopping to heat the kettle.

[0105] As can be seen, the above optional embodiments define the content of the action feature parameters to comprehensively characterize the user's action features in operating the kettle, assisting in the generation of accurate boiling instructions based on action timing analysis, and improving the accuracy of intelligent operation of the kettle and user experience.

[0106] As an optional embodiment, the step described above, training a pre-defined LSTM neural network using the parameter time series as the training dataset to obtain an action prediction model, includes:

[0107] For any two action feature parameters that are adjacent in the parameter time series, calculate the difference between the action heights of the two action feature parameters to obtain the height difference;

[0108] Calculate the difference between the motion angles of the two motion feature parameters to obtain the angle difference;

[0109] Calculate the difference between the movement frequencies of the two movement feature parameters to obtain the frequency difference.

[0110] Calculate the difference between the motion velocities of the two motion characteristic parameters to obtain the velocity difference;

[0111] The weighted sum of the height difference, angle difference, speed difference, and frequency difference is calculated to obtain the characteristic parameters of change.

[0112] Calculate the ratio of the time difference between the change feature parameter and the historical time point corresponding to the two action feature parameters to obtain the correlation feature between the two action feature parameters;

[0113] The parameter time series and all its corresponding associated features are concatenated into a fusion feature, which is then input into a pre-defined LSTM neural network for training to obtain an action prediction model.

[0114] Optionally, the weighted summation can be calculated using fixed weights, dynamic weights, or adaptive weights; this invention does not impose any limitations on this.

[0115] Optionally, the fused feature can be a multidimensional feature vector, a sequence feature matrix, or a mixed feature set; this invention does not impose any limitations.

[0116] Optionally, this splicing process can be implemented based on feature splicing, data alignment, or dimensional transformation, and the present invention does not limit it.

[0117] Optionally, the training process can be optimized by combining data augmentation, sequence length optimization, or model regularization, and this invention does not limit it.

[0118] As can be seen, through the above optional embodiments, the weighted summation of the differences in height, angle, speed and frequency of adjacent action feature parameters is used to obtain the changing feature parameters. Combined with the time difference, the associated features are generated and spliced ​​into fused features to train the LSTM neural network. Thus, based on the training of the accurate action prediction model, the model's ability to capture action changes is improved through multi-dimensional feature analysis and associated feature fusion, providing more accurate prediction support for the generation of boiling water instructions and reducing the risk of action prediction errors.

[0119] As an optional embodiment, in the above steps, the loss function of the LSTM neural network is designed as the product of the first cross-entropy and the second cross-entropy; the first cross-entropy is the difference between the predicted output and the label corresponding to the parameter time series; the second cross-entropy is the difference between the output association feature and the actual association feature; the output association feature is the association feature between the predicted output and the predicted output of the previous time step; the actual association feature is the actual association feature between the corresponding parameters of the parameter time series.

[0120] Optionally, the loss function can be implemented by combining regularization terms, weight adjustments, or dynamic loss balancing; this invention does not impose any limitations on this.

[0121] Optionally, the first cross-entropy and the second cross-entropy can be categorical cross-entropy, regression cross-entropy, or mixed cross-entropy, and the present invention does not limit them.

[0122] As can be seen, through the above optional embodiments, by designing the loss function of the LSTM neural network as the product of the first cross-entropy and the second cross-entropy, the prediction accuracy of the model for action sequences and related features is improved by optimizing the double cross-entropy on the basis of training the accurate action prediction model, providing more reliable model support for the generation of boiling water instructions and reducing the risk of prediction bias caused by insufficient model training.

[0123] As an optional embodiment, in the above steps, the sensor includes a controller and a gyroscope sensor, an infrared height sensor, and a weight sensor respectively connected to the controller; the gyroscope sensor is used to detect the motion angle and motion speed; the infrared height sensor is used to detect the motion height; the weight sensor is used to detect the weight parameter of the kettle; and the controller is used to identify the motion type and motion frequency based on the motion height, motion angle, and motion speed.

[0124] Optionally, the controller can be an embedded microcontroller, a single-chip microcomputer, or a dedicated chip, which has a built-in recognition algorithm model to identify the action type and frequency based on the action height, action angle, and action speed.

[0125] Optionally, the gyroscope sensor can be a three-axis gyroscope, a six-axis sensor, or a high-precision gyroscope; this invention does not impose any limitations.

[0126] Optionally, the infrared height sensor can be a single-point infrared sensor, a multi-point infrared sensor, or an array infrared sensor; the present invention does not limit this.

[0127] Optionally, the weight sensor can be a piezoelectric sensor, a strain gauge sensor, or a high-precision weight sensor; the present invention does not limit the type of weight sensor.

[0128] As can be seen, through the above optional embodiments, by integrating gyroscope sensors, infrared height sensors and weight sensors to detect motion angle, speed, height and weight parameters, the controller identifies the motion type and frequency. Thus, based on the precise generation of boiling water instructions, the comprehensiveness and accuracy of motion feature extraction are improved through multi-sensor collaboration and motion type analysis, providing high-quality data input for the motion prediction model and reducing the risk of recognition errors caused by insufficient data from a single sensor.

[0129] As an optional embodiment, the step above, predicting the user's action characteristics at the current time point based on the action prediction model and the action sensing data from the previous time point, includes:

[0130] Acquire motion sensor data from at least the two time points preceding the current time point and identify them as historical sensor data;

[0131] Calculate the correlation features between two historical sensor data sets;

[0132] Historical sensor data and corresponding associated features are input into the action prediction model to obtain the predicted user action features at the current time point.

[0133] As can be seen, through the above optional embodiments, by acquiring historical sensor data from at least two time points prior to the current time point and calculating associated features, the action prediction model is input to predict the current user's action features. Thus, based on the accurate generation of water boiling instructions, the temporal accuracy of action feature prediction is improved through historical data and associated feature analysis, providing accurate action basis for water boiling control instructions and reducing the risk of prediction errors caused by insufficient temporal data.

[0134] As an optional embodiment, the step above, determining the water boiling control command corresponding to the kettle based on predicted user action characteristics, includes:

[0135] Acquire the current sensor data and corresponding detection device parameters corresponding to the user action detection signal;

[0136] Determine the reference sensor features corresponding to the predicted user action features in a preset database corresponding to the detection device parameters;

[0137] Calculate the feature similarity between the current sensing data and the reference sensing features;

[0138] Determine whether the feature similarity is greater than a preset similarity threshold;

[0139] If not, wait for the user to actually press the button to determine the corresponding boiling control command for the kettle;

[0140] If so, based on the predicted user action characteristics and the pre-defined correspondence between actions and boiling instructions, determine the corresponding boiling control instruction for the kettle.

[0141] Optionally, the feature similarity can be cosine similarity, Euclidean distance, Jaccard coefficient, or dynamic time warping distance; this invention does not limit the specific features.

[0142] Optionally, the calculation of feature similarity can be based on vector comparison, statistical analysis, or feature matching, and this invention does not limit it.

[0143] Optionally, the similarity threshold can be a fixed threshold, a dynamic threshold, or a threshold adjusted based on the action type; this invention does not impose any limitations.

[0144] Optionally, the current sensing data may include angle data, height data, speed data, or weight data, and the present invention does not limit it.

[0145] Optionally, the parameters of the detection equipment may include sensor sensitivity, sampling frequency, installation location, or equipment calibration parameters, which are not limited in this invention.

[0146] Optionally, the acquisition process of the current sensing data and the corresponding detection device parameters can be achieved through direct sensor acquisition, controller processing, or data interface transmission; this invention does not impose any limitations on this.

[0147] Optionally, the current sensing data is acquired by an infrared sensor located on top of the kettle, which is used only to acquire sensing parameters of the user's proximity action in order to initiate the generation of the corresponding boiling control command.

[0148] Optionally, the button operation can be triggered by a physical button, a touch button, or a voice command; this invention does not impose any limitations.

[0149] Optionally, the waiting process can be implemented based on event listening, user interaction detection, or timeout mechanism, and the present invention does not limit it.

[0150] Optionally, the correspondence between the preset action and the boiling water instruction can be a mapping table, a rule base, a data-driven model, or a conditional logic table; this invention does not impose any limitations.

[0151] Optionally, the water boiling control command can be a single command, a combination of commands, a condition-triggered command, or a priority command; the present invention does not impose any limitations on this.

[0152] As can be seen, through the above optional embodiments, by comparing the similarity between the current sensing data and the reference sensing features and determining the water boiling control command based on the correspondence between predicted user action features and action commands, the accuracy and reliability of command generation are improved through feature similarity verification and action mapping on the basis of accurate water boiling command generation, reducing the risk of command errors caused by misjudgment of action detection signals or user differences.

[0153] Example 2

[0154] Please see Figure 2 , Figure 2 This is a schematic diagram of a water boiling control system based on historical data analysis disclosed in an embodiment of the present invention. Figure 2 The described water boiling control system based on historical data analysis can be applied to data processing systems / data processing equipment / data processing servers (including local processing servers or cloud processing servers). For example... Figure 2 As shown, the water boiling control system based on historical data analysis may include:

[0155] The acquisition module 201 is used to acquire motion sensing data at multiple historical time points through a sensor set at the bottom of the kettle.

[0156] Training module 202 is used to train an action prediction model based on action sensing data corresponding to all historical time points.

[0157] The prediction module 203 is used to predict the user's action characteristics at the current time point based on the action prediction model and the action sensing data at the previous time point.

[0158] The determination module 204 is used to respond to the user action detection signal at the current time point and determine the water boiling control command corresponding to the kettle based on the predicted user action characteristics.

[0159] As can be seen, the above-mentioned embodiments of the invention train an action prediction model by acquiring motion sensing data from the bottom sensor of the kettle at historical time points, predicting the current user's action characteristics based on the data from the previous time point, and further determining the boiling control command. This enables the generation of accurate boiling commands based on action timing analysis, improving the accuracy of the kettle's intelligent operation and the user experience.

[0160] As an optional implementation, the training module trains the action prediction model based on action sensing data corresponding to all historical time points in the following specific ways:

[0161] Based on action recognition rules, determine the action feature parameters corresponding to each action sensing data;

[0162] Based on historical time points, the motion feature parameters corresponding to all motion sensing data are sorted from morning to night to obtain the parameter time series;

[0163] Using the parameter time series as the training dataset, a pre-defined LSTM neural network is trained to obtain an action prediction model.

[0164] As can be seen, through the above optional embodiments, by extracting action feature parameters based on action recognition rules and generating parameter time series by sorting them by time, and training an LSTM neural network to obtain an action prediction model, the accuracy of action prediction is improved by extracting time series features and training the model on the basis of accurate water boiling command generation. This provides reliable action feature support for water boiling control commands and reduces the risk of command misjudgment caused by deviations in action feature extraction.

[0165] As an optional embodiment, the motion characteristic parameters include motion type, motion height, motion angle, motion speed, and motion frequency; the motion type is pouring water out of the kettle, filling the kettle with water, adding tea leaves to the kettle, starting to heat the kettle, or stopping to heat the kettle.

[0166] As can be seen, the above optional embodiments define the content of the action feature parameters to comprehensively characterize the user's action features in operating the kettle, assisting in the generation of accurate boiling instructions based on action timing analysis, and improving the accuracy of intelligent operation of the kettle and user experience.

[0167] As an optional embodiment, the training module uses the parameter time series as the training dataset to train a pre-defined LSTM neural network to obtain the action prediction model. The specific methods include:

[0168] For any two action feature parameters that are adjacent in the parameter time series, calculate the difference between the action heights of the two action feature parameters to obtain the height difference;

[0169] Calculate the difference between the motion angles of the two motion feature parameters to obtain the angle difference;

[0170] Calculate the difference between the movement frequencies of the two movement feature parameters to obtain the frequency difference.

[0171] Calculate the difference between the motion velocities of the two motion characteristic parameters to obtain the velocity difference;

[0172] The weighted sum of the height difference, angle difference, speed difference, and frequency difference is calculated to obtain the characteristic parameters of change.

[0173] Calculate the ratio of the time difference between the change feature parameter and the historical time point corresponding to the two action feature parameters to obtain the correlation feature between the two action feature parameters;

[0174] The parameter time series and all its corresponding associated features are concatenated into a fusion feature, which is then input into a pre-defined LSTM neural network for training to obtain an action prediction model.

[0175] As can be seen, through the above optional embodiments, the weighted summation of the differences in height, angle, speed and frequency of adjacent action feature parameters is used to obtain the changing feature parameters. Combined with the time difference, the associated features are generated and spliced ​​into fused features to train the LSTM neural network. Thus, based on the training of the accurate action prediction model, the model's ability to capture action changes is improved through multi-dimensional feature analysis and associated feature fusion, providing more accurate prediction support for the generation of boiling water instructions and reducing the risk of action prediction errors.

[0176] As an optional embodiment, the loss function of the LSTM neural network is designed as the product of a first cross-entropy and a second cross-entropy; the first cross-entropy is the difference between the predicted output and the label corresponding to the parameter time series; the second cross-entropy is the difference between the output association feature and the actual association feature; the output association feature is the association feature between the predicted output and the predicted output of the previous time step; the actual association feature is the actual association feature between the corresponding parameters of the parameter time series.

[0177] As can be seen, through the above optional embodiments, by designing the loss function of the LSTM neural network as the product of the first cross-entropy and the second cross-entropy, the prediction accuracy of the model for action sequences and related features is improved by optimizing the double cross-entropy on the basis of training the accurate action prediction model, providing more reliable model support for the generation of boiling water instructions and reducing the risk of prediction bias caused by insufficient model training.

[0178] As an optional embodiment, the sensor includes a controller and a gyroscope sensor, an infrared altitude sensor, and a weight sensor respectively connected to the controller; the gyroscope sensor is used to detect the motion angle and motion speed; the infrared altitude sensor is used to detect the motion height; the weight sensor is used to detect the weight parameter of the kettle; and the controller is used to identify the motion type and motion frequency based on the motion height, motion angle, and motion speed.

[0179] As can be seen, through the above optional embodiments, by integrating gyroscope sensors, infrared height sensors and weight sensors to detect motion angle, speed, height and weight parameters, the controller identifies the motion type and frequency. Thus, based on the precise generation of boiling water instructions, the comprehensiveness and accuracy of motion feature extraction are improved through multi-sensor collaboration and motion type analysis, providing high-quality data input for the motion prediction model and reducing the risk of recognition errors caused by insufficient data from a single sensor.

[0180] As an optional embodiment, the prediction module, based on the action prediction model and the action sensing data from the previous time point, predicts the user's action characteristics at the current time point in a specific way, including:

[0181] Acquire motion sensor data from at least the two time points preceding the current time point and identify them as historical sensor data;

[0182] Calculate the correlation features between two historical sensor data sets;

[0183] Historical sensor data and corresponding associated features are input into the action prediction model to obtain the predicted user action features at the current time point.

[0184] As can be seen, through the above optional embodiments, by acquiring historical sensor data from at least two time points prior to the current time point and calculating associated features, the action prediction model is input to predict the current user's action features. Thus, based on the accurate generation of water boiling instructions, the temporal accuracy of action feature prediction is improved through historical data and associated feature analysis, providing accurate action basis for water boiling control instructions and reducing the risk of prediction errors caused by insufficient temporal data.

[0185] As an optional embodiment, the determining module determines the specific method of the boiling control command corresponding to the kettle based on the predicted user action characteristics, including:

[0186] Acquire the current sensor data and corresponding detection device parameters corresponding to the user action detection signal;

[0187] Determine the reference sensor features corresponding to the predicted user action features in a preset database corresponding to the detection device parameters;

[0188] Calculate the feature similarity between the current sensing data and the reference sensing features;

[0189] Determine whether the feature similarity is greater than a preset similarity threshold;

[0190] If not, wait for the user to actually press the button to determine the corresponding boiling control command for the kettle;

[0191] If so, based on the predicted user action characteristics and the pre-defined correspondence between actions and boiling instructions, determine the corresponding boiling control instruction for the kettle.

[0192] As can be seen, through the above optional embodiments, by comparing the similarity between the current sensing data and the reference sensing features and determining the water boiling control command based on the correspondence between predicted user action features and action commands, the accuracy and reliability of command generation are improved through feature similarity verification and action mapping on the basis of accurate water boiling command generation, reducing the risk of command errors caused by misjudgment of action detection signals or user differences.

[0193] Example 3

[0194] Please see Figure 3 , Figure 3 This is another water boiling control system based on historical data analysis disclosed in the embodiments of the present invention. Figure 3 The described water boiling control system based on historical data analysis is applied in a data processing system / data processing equipment / data processing server (wherein, the server includes a local processing server or a cloud processing server). For example... Figure 3 As shown, the water boiling control system based on historical data analysis may include:

[0195] Memory 301 storing executable program code;

[0196] Processor 302 coupled to memory 301;

[0197] The processor 302 calls the executable program code stored in the memory 301 to execute the steps of the boiling water control method based on historical data analysis described in Embodiment 1.

[0198] Example 4

[0199] This invention discloses a computer read storage medium that stores a computer program for electronic data interchange, wherein the computer program causes a computer to execute the steps of the boiling water control method based on historical data analysis described in Embodiment 1.

[0200] Example 5

[0201] This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps of the boiling water control method based on historical data analysis described in Embodiment 1.

[0202] The foregoing has described specific embodiments of this specification; other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than those shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily have to follow the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0203] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0204] For ease of description, the above devices are described in terms of function, divided into various units. Of course, in implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware components.

[0205] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0206] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0207] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0208] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0209] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0210] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0211] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0212] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0213] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0214] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0215] Finally, it should be noted that the boiling water control method and system based on historical data analysis disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method of controlling water boiling based on historical data analysis, characterized by, The method comprises: acquiring action sensing data at a plurality of historical time points through a sensor arranged at the bottom of the water boiler; the sensor comprises a controller and a gyroscope sensor, an infrared height sensor and a weight sensor connected to the controller respectively; training an action prediction model based on the action sensing data corresponding to all the historical time points, comprising: determining an action characteristic parameter corresponding to each of the action sensing data based on an action recognition rule; the action recognition rule is a feature extraction rule, a pattern matching rule or a machine learning model built-in or corresponding to the sensor; the action characteristic parameter comprises an action type, an action height, an action angle, an action speed and an action frequency; sorting the action characteristic parameters corresponding to all the action sensing data from early to late based on the historical time points to obtain a parameter time sequence; calculating a difference value between the action heights of any two action characteristic parameters adjacent in the parameter time sequence to obtain a height difference value; calculating a difference value between the action angles of the two action characteristic parameters to obtain an angle difference value; calculating a difference value between the action frequencies of the two action characteristic parameters to obtain a frequency difference value; calculating a difference value between the action speeds of the two action characteristic parameters to obtain a speed difference value; calculating a weighted sum of the height difference value, the angle difference value, the speed difference value and the frequency difference value to obtain a change characteristic parameter; calculating a ratio of the change characteristic parameter to a time difference value between the historical time points corresponding to the two action characteristic parameters to obtain a correlation characteristic between the two action characteristic parameters; concatenating the parameter time sequence and all the correlation characteristics corresponding thereto into a fusion characteristic and inputting the fusion characteristic into a preset LSTM neural network for training to obtain an action prediction model; based on the action prediction model, predicting a predicted user action characteristic at a current time point according to action sensing data at a previous time point; the predicted user action characteristic comprises an action type, an action amplitude, an action frequency or an action speed; in response to a user action detection signal at the current time point, determining a water boiling control instruction corresponding to the water boiler according to the predicted user action characteristic.

2. The historical data analysis-based water boiling control method according to claim 1, characterized by, The action type is pouring water out of the water boiler, pouring water into the water boiler, adding tea leaves to the water boiler, starting to heat the water boiler or stopping heating the water boiler.

3. The historical data analysis-based water boiling control method according to claim 1, characterized by, The loss function of the LSTM neural network is designed as a product of a first cross entropy and a second cross entropy; the first cross entropy is a difference between a predicted output and a label corresponding to the parameter time sequence; the second cross entropy is a difference between an output correlation characteristic and an actual correlation characteristic; the output correlation characteristic is a correlation characteristic between a predicted output and a predicted output at a previous time step; the actual correlation characteristic is an actual correlation characteristic between corresponding parameters of the parameter time sequence.

4. The historical data analysis-based water boiling control method according to claim 2, characterized by, The gyroscope sensor is used to detect the action angle and the action speed; the infrared height sensor is used to detect the action height; the weight sensor is used to detect the weight parameter of the water boiler; and the controller is used to identify the action type and the action frequency according to the action height, the action angle and the action speed.

5. The historical data analysis based water boiling control method according to claim 1, wherein, The action prediction model is used to predict the predicted user action feature at the current time point according to the action sensing data at the previous time point, including: The action sensing data at at least the previous two time points at the current time point is obtained, and the action sensing data is determined as historical sensing data; The correlation feature between the two historical sensing data is calculated; The historical sensing data and the corresponding correlation feature are input into the action prediction model to obtain the output predicted user action feature at the current time point.

6. The historical data analysis-based water boiling control method according to claim 5, characterized by, The predicted user action feature is used to determine the corresponding water boiling control instruction of the water boiler, including: The current sensing data corresponding to the user action detection signal and the corresponding detection device parameter are obtained; The predicted user action feature is determined to correspond to the reference sensing feature in the preset database corresponding to the detection device parameter; The feature similarity between the current sensing data and the reference sensing feature is calculated; It is judged whether the feature similarity is greater than the preset similarity threshold; If not, the user's actual key operation is waited to determine the water boiling control instruction corresponding to the water boiler; If yes, the predicted user action feature is used to determine the water boiling control instruction corresponding to the water boiler based on the preset corresponding relationship between the action and the water boiling instruction.

7. A history data analysis based water boiling control system, characterized by, The system includes: An acquisition module is used to obtain the action sensing data at a plurality of historical time points through a sensor arranged at the bottom of the water boiler; the sensor includes a controller and a gyroscope sensor, an infrared height sensor and a weight sensor connected to the controller, respectively; A training module is used to train an action prediction model based on the action sensing data corresponding to all the historical time points, including: An action recognition rule is used to determine the action feature parameter corresponding to each action sensing data; the action recognition rule is a feature extraction rule, a pattern matching rule or a machine learning model built-in or corresponding to the sensor; the action feature parameter includes an action type, an action height, an action angle, an action speed and an action frequency; The action feature parameters corresponding to all the action sensing data from early to late at the historical time points are sorted to obtain a parameter time sequence; For any two adjacent action feature parameters in the parameter time sequence, the difference between the action heights of the two action feature parameters is calculated to obtain a height difference value; The difference between the action angles of the two action feature parameters is calculated to obtain an angle difference value; The difference between the action frequencies of the two action feature parameters is calculated to obtain a frequency difference value; The difference between the action speeds of the two action feature parameters is calculated to obtain a speed difference value; The weighted sum of the height difference value, the angle difference value, the speed difference value and the frequency difference value is calculated to obtain a change feature parameter; a ratio of a time difference between the change feature parameter and a historical time point corresponding to the two action feature parameters, to obtain a correlation feature between the two action feature parameters; concatenate the parameter time sequence and all the corresponding correlation features to obtain a fusion feature, and input the fusion feature into a preset LSTM neural network for training to obtain an action prediction model; a prediction module configured to predict a predicted user action feature at a current time point based on the action prediction model and action sensing data at a previous time point, the predicted user action feature including an action type, an action amplitude, an action frequency, or an action speed; a determination module configured to determine a water boiling control instruction corresponding to the water boiler based on the predicted user action feature in response to a user action detection signal at the current time point.

8. A history data analysis based water boiling control system, characterized by, The system comprises: a memory storing executable program codes; a processor coupled to the memory; the processor invokes the executable program codes stored in the memory to execute the water boiling control method based on historical data analysis according to any one of claims 1-6.

Images