Motor control method for food processor

Abstract

A motor control method for a food processor relates to the field of motor control technology. The method comprises: obtaining motor load data and vibration data during the dough mixing process; based on the motor load data and vibration data, identifying the state of ice cubes during the dough mixing process, including the ice cube sinking stage, the ice cube throwing stage and the ice crushing stage; according to the identified ice cube state, implementing different motor control strategies: in the ice cube sinking stage, adopting a first motor control strategy to prevent the ice cubes from continuously sinking to the bottom to promote uniform hydration of the bottom flour; in the ice cube throwing stage, adopting a second motor control strategy to control the ice crushing process and reduce motor load fluctuations; in the ice crushing stage, adopting a third motor control strategy to control the uniform distribution of crushed ice and avoid over-mixing. The motor control method for a food processor provided in this application can better mix the dough.

Description

Technical Field

[0001] The present application relates to the field of motor control technology, and in particular to a motor control method for a food processor. Background Art

[0002] An important function of a home kitchen machine is to mix dough, and the quality of this function directly affects the quality of the product. The motors used in existing kitchen machines are generally divided into multiple mixing speed gears, and each gear can drive the dough to rotate and mix at a fixed speed.

[0003] In order to control the temperature of the dough and prevent it from fermenting prematurely during mixing, users usually add ice water containing ice cubes when mixing water and flour.

[0004] However, since the ice cubes may change during the mixing process, the prior art still has some problems, resulting in uneven mixing of the dough. Summary of the Invention

[0005] In order to solve the above technical problems or at least partially solve the above technical problems, the present application provides a motor control method for a food processor, which can make the dough mixing effect of the household food processor better.

[0006] The present application provides a motor control method for a food processor, the motor control method for a food processor comprising the following steps:

[0007] Obtaining motor load data and vibration data during the dough mixing process; based on the motor load data and vibration data, identifying the state of the ice cubes during the dough mixing process, including the ice cube sinking stage, the ice cube throwing stage, and the ice crushing stage;

[0008] Depending on the identified ice state, different motor control strategies are implemented accordingly:

[0009] During the ice cube sinking stage, the first motor control strategy is adopted to prevent the ice cube from continuously sinking to the bottom to promote uniform hydration of the flour at the bottom;

[0010] During the ice-throwing stage, the second motor control strategy is adopted to control the ice crushing process and reduce motor load fluctuations;

[0011] During the crushed ice entrainment stage, the third motor control strategy is adopted to control the uniform distribution of crushed ice and avoid excessive stirring.

[0012] Optionally, the motor control method for a food processor further includes:

[0013] During the ice wall throwing stage, predict ice breakage data;

[0014] According to the predicted ice crushing data, the signal recognition threshold of the ice crushing stage is adjusted;

[0015] Based on the adjusted signal recognition threshold, the current crushed ice entrainment state is detected to adjust the third motor control strategy, thereby avoiding excessive stirring while ensuring uniform distribution of the crushed ice.

[0016] Optionally, the first motor control strategy, the second motor control strategy and the third motor control strategy are output through a preset food processor motor control model;

[0017] The food processor motor control model includes:

[0018] Input layer, used to receive motor load data and vibration data;

[0019] The feature extraction layer is used to receive data from the input layer and process the data;

[0020] The stage discrimination layer receives the output of the feature extraction layer to identify the stage of the ice cube in the dough mixing process;

[0021] The early information memory layer is used to predict and record ice breakage prediction data based on the data from the feature extraction layer during the ice throwing stage;

[0022] A threshold adjustment layer, configured to dynamically adjust the signal recognition threshold of the ice entrainment phase according to the ice breakage prediction data stored in the previous information memory layer;

[0023] The control strategy layer is used to obtain the corresponding motor control parameters based on the output of the feature extraction layer using the corresponding expert neural network according to the identified ice stage;

[0024] In the crushed ice entrainment stage, the crushed ice entrainment intensity is identified based on the adjusted signal recognition threshold, and the motor control parameters are adjusted accordingly;

[0025] The output layer is used to convert the motor control parameters of different stages into motor drive instructions, thereby obtaining the first motor control strategy, the second motor control strategy and the third motor control strategy respectively.

[0026] Optionally, the feature extraction layer includes:

[0027] a time-frequency preprocessing unit, configured to perform a short-time Fourier transform on the motor load data and the vibration data to generate a time-frequency spectrum;

[0028] The shared feature extraction unit adopts a one-dimensional residual convolutional network structure to extract cross-channel temporal features from the time-frequency spectrum to form a shared feature vector.

[0029] Optionally, the stage discrimination layer adopts a Transformer structure, which is used to receive the shared feature vector and output a stage probability vector representing the ice sinking stage, the ice throwing wall stage and the broken ice entrainment stage.

[0030] Optionally, the early information memory layer adopts a fully connected neural network + LSTM structure, which is activated when the ice cube wall-throwing stage is determined, and predicts the following indicators based on the shared feature vector of the ice cube wall-throwing stage:

[0031] The size of the ice when it breaks;

[0032] The degree of ice fragmentation.

[0033] Optionally, the threshold adjustment layer adopts a fully connected neural network structure, which receives the size of the ice cubes when broken and the degree of ice breakage predicted by the previous information memory layer, and outputs the motor load recognition threshold and vibration recognition threshold required for the ice crushing stage.

[0034] Optionally, the control strategy layer includes:

[0035] The stage selection unit adopts a gated logic structure. Based on the stage probability vector output by the stage discrimination layer, it activates the expert neural network corresponding to the ice stage with the highest probability in the current stage from the expert network unit in a hard-gated manner.

[0036] The expert network unit includes three groups of expert neural networks that independently correspond to the ice sinking stage, ice throwing stage, and ice crushing stage, among which:

[0037] The expert neural network for the ice sinking stage and the ice throwing stage both adopt a fully connected neural network structure, taking the shared feature vector output by the feature extraction layer as input and outputting the motor control parameters of the corresponding stage to the output layer;

[0038] The expert neural network in the ice crushing stage includes a shallow convolutional autoencoder and a fully connected neural network structure. The shallow convolutional autoencoder is used to compress and reconstruct the input shared feature vector to obtain abnormal residual features that characterize the ice crushing signal.

[0039] The abnormal residual features are compared with the motor load identification threshold and vibration identification threshold provided in real time by the threshold adjustment layer, so as to quantify the current crushed ice entrainment intensity. The quantified crushed ice entrainment intensity and the shared feature vector are used as the input of the expert neural network in the crushed ice entrainment stage, and the motor control parameters of the crushed ice entrainment stage are output to the output layer.

[0040] Optionally, the output layer includes:

[0041] A parameter safety domain clipping unit, configured to receive the control parameters output by the control strategy layer and clip the control parameters to a safety domain range to ensure that the output control instructions are within a safe operating range of the chef machine motor;

[0042] The control instruction generating unit is used to convert the control parameters after being clipped by the safety domain into a food processor motor driving instruction, wherein the food processor motor driving instruction at least includes a target motor rotation time, a target motor speed instruction, and a target motor torque instruction.

[0043] The technical solution provided by this application has the following advantages compared with the existing technology:

[0044] One of its working principles and beneficial effects is that when mixing dough in a food processor, users often add water containing ice cubes to reduce the temperature of the dough and prevent it from fermenting prematurely during mixing. Due to the addition of ice cubes, the state of the ice cubes during mixing will go through three stages: sinking to the bottom, being thrown off the wall, and being entrained by crushed ice. Different stages have different effects on dough mixing, for example:

[0045] When mixing flour and water at a low speed, due to the weight of the ice cubes, they will gather at the bottom due to the stirring effect to form a local low-temperature area, forming a sinking stage.

[0046] Low temperature will inhibit the hydration of flour, and the ice cubes are large and smooth at this time, making them difficult to flip at a slow speed, thus preventing the flour pressed under the ice cubes from flipping, resulting in insufficient hydration of the flour at the bottom and forming hard lumps at the bottom. These lumps will also cause uneven dough after mixing later.

[0047] After the flour and water are initially mixed, the speed of the paddle increases, and the ice cubes are thrown toward the inner wall of the container due to centrifugal force, causing the outer edge of the paddle to erratically strike the ice cubes. If the motor uses dynamic torque output, the motor torque will frequently fluctuate between "high torque" and "low torque" as the paddle strikes and pushes away the ice cubes, causing large fluctuations in the motor load, fatigue of the mechanical components, and affecting the stability of the equipment.

[0048] The ice cubes will be broken by the paddles during the wall-throwing stage, and the resulting crushed ice will be wrapped by the formed dough, entering the crushed ice entrainment stage.

[0049] If the dough is not stirred sufficiently, localized water spots will appear after the enclosed crushed ice melts, affecting the uniformity of the dough. Therefore, it is generally necessary to extend the stirring time, but this may also lead to over-mixing of the dough, resulting in an undesirable gluten structure or requiring more energy.

[0050] This application proposes a motor control method for a food processor that can identify the different stages of ice cubes during the dough mixing process and adopt different motor control strategies for each stage. During the ice sinking stage, the method adopts a first motor control strategy, which prevents the ice cubes from sinking to the bottom for a long time by reasonably adjusting the motor parameters, promotes full contact between the bottom flour and water, and achieves uniform hydration; when the ice cubes enter the wall-throwing stage, the system switches to a second motor control strategy, which alleviates the violent fluctuations in the motor torque output caused by the impact of the ice cubes; finally, during the crushed ice entrainment stage, the third motor control strategy is activated, and an intelligent algorithm is used to control the uniform distribution of crushed ice in the dough, while avoiding excessive mixing that may damage the gluten structure.

[0051] Therefore, the motor control method for a food processor provided in this application can better stir the dough.

[0052] The second working principle and its beneficial effect is that during the crushed ice entrainment stage, since the crushed ice is wrapped in the dough and highly overlaps with the dough's stirring resistance signal, the traditional motor control method is difficult to accurately identify the current degree of entrainment of the crushed ice at this stage, resulting in the inability to adjust the motor's operation according to the current degree of ice breakage and the current state of the dough, thereby controlling the dough's stirring time.

[0053] This application introduces a preliminary information memory layer and a dynamic threshold adjustment mechanism. When the ice cubes enter the wall-throwing stage, the motor control system will predict and record the data of ice crushing through the preliminary information memory layer. By being able to better capture the characteristic signals of ice cubes in the early stage, the predicted and recorded data of ice crushing can be used as a reference for determining the strength of ice crushing separation in the subsequent ice crushing and entrainment stage. By making corresponding judgments in the early stages of ice crushing, a more accurate basis is provided for signal recognition in the later ice crushing and entrainment stage, thereby avoiding the problem in the traditional method that the strength of ice crushing is difficult to determine due to inaccurate separation threshold setting, resulting in insufficient or excessive dough mixing.

[0054] Therefore, the motor control method for a food processor provided in this application can better stir the dough. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 A flow chart of a motor control method for a food processor provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] The technical solution in this application will be described below with reference to the accompanying drawings.

[0057] The following description sets forth many specific details to facilitate a thorough understanding of the present application. However, the present application may also be implemented in other ways than those described herein. It is apparent that the embodiments described in the specification are only some of the embodiments of the present application, not all of them. It should be noted that the embodiments of the present application and the features therein may be combined with each other unless there is a conflict.

[0058] An embodiment of the present application provides a motor control method for a food processor, the motor control method for a food processor comprising the following steps:

[0059] S101: Acquire motor load data and vibration data during dough mixing.

[0060] Specifically, in the embodiment of the present application, the motor load data can be obtained by monitoring the load condition of the motor in real time through a current sensor installed in the motor control system.

[0061] Vibration data can be obtained by acquiring vibration acceleration in real time through a three-axis accelerometer installed coaxially with the motor.

[0062] S102: Based on the motor load data and vibration data, identifying the state of the ice cubes during the dough mixing process, including the ice cube sinking stage, the ice cube throwing stage, and the ice crushing stage.

[0063] S103: Implement different motor control strategies based on the identified ice state:

[0064] During the ice cube sinking stage, the first motor control strategy is adopted to prevent the ice cube from continuously sinking to the bottom to promote uniform hydration of the flour at the bottom;

[0065] During the ice-throwing stage, the second motor control strategy is adopted to control the ice crushing process and reduce motor load fluctuations;

[0066] During the crushed ice entrainment stage, the third motor control strategy is adopted to control the uniform distribution of crushed ice and avoid excessive stirring.

[0067] Specifically, the motor control method for a food processor further includes:

[0068] During the ice wall throwing stage, predict ice breakage data;

[0069] According to the predicted ice crushing data, the signal recognition threshold of the ice crushing stage is adjusted;

[0070] Based on the adjusted signal recognition threshold, the current crushed ice entrainment state is detected to adjust the third motor control strategy, thereby avoiding excessive stirring while ensuring uniform distribution of the crushed ice.

[0071] In the embodiment of the present application, the steps S102-S103 are specifically implemented by the following constructed food processor motor control model:

[0072] The food processor motor control model includes:

[0073] Input layer, used to receive motor load data and vibration data;

[0074] The feature extraction layer is used to receive data from the input layer and process the data;

[0075] Specifically, the feature extraction layer includes:

[0076] The time-frequency preprocessing unit is used to perform short-time Fourier transform on the motor load data and vibration data to generate a time-frequency spectrum.

[0077] Specifically, the processing flow of the time-frequency preprocessing unit includes:

[0078] Data segmentation and preprocessing: the continuously collected motor load data and vibration data are segmented according to fixed time windows (window size is 256ms, step size is 64ms);

[0079] Multi-channel synchronous processing, performing short-time Fourier transform on motor load data (single channel) and vibration data (three-axis acceleration, three channels in total) to form a four-channel time-frequency representation;

[0080] Noise suppression: Apply adaptive threshold filtering to the obtained time-frequency spectrum to remove environmental interference and inherent noise components generated by the motor operation itself;

[0081] Frequency band selective enhancement: It selectively enhances the characteristic frequency bands generated by ice in different states (ice sinking to the bottom: 5-30Hz, ice throwing off the wall: 50-200Hz, ice crushing: 20-80Hz) to improve the significance of the features;

[0082] Normalization of time-frequency features: amplitude normalization is performed on the time-frequency spectrograms of different channels to make the amplitude range of each channel data consistent.

[0083] The shared feature extraction unit adopts a one-dimensional residual convolutional network structure to extract cross-channel temporal features from the time-frequency spectrum to form a shared feature vector.

[0084] Specifically, the one-dimensional residual convolutional network structure includes:

[0085] The input layer is used to receive the normalized multi-channel time-frequency spectrum;

[0086] The channel fusion layer receives the output of the input layer and uses a 1×1 convolution kernel to perform preliminary fusion of the time-frequency information of the four channels. The number of output feature channels is 32.

[0087] The residual block stacking layer receives the output of the channel fusion layer and contains 5 consecutive residual blocks. Each residual block consists of two one-dimensional convolutional layers (with kernel sizes of 3 and 5 respectively), a batch normalization layer, and a ReLU activation function. It also has a skip connection structure to alleviate the gradient vanishing problem.

[0088] The global pooling layer receives the output of the residual stacking layer and uses adaptive average pooling to extract global timing information;

[0089] The feature dimension unification layer receives the output of the global pooling layer and maps the extracted features to a 128-dimensional feature vector space through a fully connected layer to form the final shared feature vector.

[0090] The stage discrimination layer receives the output of the feature extraction layer to identify the stage of the ice cube in the dough mixing process;

[0091] Specifically, the stage discrimination layer adopts a Transformer structure, which is used to receive the shared feature vector and output a stage probability vector representing the ice sinking stage, the ice throwing wall stage, and the broken ice entrainment stage.

[0092] In a specific implementation, the Transformer structure includes:

[0093] A multi-head self-attention mechanism unit is used to capture long-term dependencies in shared feature vectors. It uses four attention heads, each with a dimension of 64. By calculating the correlation between different time steps in the feature sequence, it enhances the ability to perceive the continuous changes in the state of the ice cube;

[0094] A feedforward neural network unit, consisting of two fully connected layers. The first layer uses 256 neurons and is equipped with a ReLU activation function. The second layer uses 128 neurons to perform nonlinear transformations on the features extracted by the multi-head self-attention mechanism.

[0095] Layer normalization units, applied to the outputs of multi-head self-attention units and feedforward neural network units;

[0096] Residual connection units, used to preserve original feature information and prevent the gradient vanishing problem in deep networks;

[0097] The temporal encoding unit, by adding position encoding, enables the model to distinguish features at different time steps and enhances the ability to recognize changes in ice state during the dough mixing process;

[0098] The classification head unit consists of a fully connected layer with 3 neurons, corresponding to the three ice states, and outputs a vector representing the probabilities of the three states through the softmax function.

[0099] The training process of the discriminant layer in this stage includes:

[0100] Collect an annotated dataset containing motor load and vibration data from different users during dough mixing under various conditions, as well as corresponding ice state annotations (sinking, wall-throwing, or ice-entrained).

[0101] The cross entropy loss function is used as the optimization target, and the Adam optimizer is used for training. The initial learning rate is set to 0.001, and the learning rate decay strategy is adopted to train the Transformer structure of the stage discriminant layer.

[0102] The early information memory layer is used to predict and record ice breakage prediction data based on the data from the feature extraction layer during the ice throwing stage;

[0103] Specifically, the early information memory layer adopts a fully connected neural network + LSTM structure, which is activated when the ice cube is judged to be in the wall-throwing stage, and predicts the following indicators based on the shared feature vector of the ice cube wall-throwing stage:

[0104] The size of the ice when it breaks;

[0105] The degree of ice fragmentation.

[0106] Specifically, the structure of the early information memory layer is as follows:

[0107] Selective activation layer: When the probability of the ice cube falling wall stage in the output of the stage discrimination layer exceeds the preset activation threshold, the previous information memory layer is activated, starts to receive shared feature vectors and perform prediction tasks;

[0108] A multi-layer perceptron structure, consisting of multiple fully connected layers, with a hierarchical structure of 128-256-128-64 units, and a LeakyReLU activation function between each layer, is used to receive the shared feature vector, process the data and output it;

[0109] Time series feature aggregation uses a long short-term memory (LSTM) network with a capacity of 64 units to aggregate feature information from multiple consecutive time windows during the wall-throwing phase to capture the dynamic changes in the ice's characteristics when it is impacted.

[0110] The dual-task output head is divided into two parallel branches:

[0111] Ice size prediction branch: used to output the estimated average size of ice cubes when they are broken;

[0112] Crushing degree prediction branch: used to classify the degree of ice crushing into three levels (sufficient, moderate, and insufficient).

[0113] A threshold adjustment layer, configured to dynamically adjust the signal recognition threshold of the ice entrainment phase according to the ice breakage prediction data stored in the previous information memory layer;

[0114] Specifically, the threshold adjustment layer adopts a fully connected neural network structure, which receives the size of the ice cubes when broken and the degree of ice breakage predicted by the previous information memory layer, and outputs the motor load recognition threshold and vibration recognition threshold required for the ice crushing stage.

[0115] In the embodiment of the present application, the threshold adjustment layer adopts a dual-path fully connected neural network structure, including:

[0116] Input the normalization unit to perform Min-Max normalization on the ice cube size and crushing degree data, respectively, and map them to the range of 0-1;

[0117] The feature extraction dual-pathway structure, where the size feature path contains two layers of 32-node fully connected networks, and the fragmentation feature path contains two layers of 16 to 32-node fully connected networks, and the two paths extract relevant feature information respectively;

[0118] The threshold generation unit, through a fully connected network layer, fuses the two features and generates the motor load recognition threshold and vibration recognition threshold.

[0119] The control strategy layer is used to obtain the corresponding motor control parameters based on the output of the feature extraction layer using the corresponding expert neural network according to the identified ice stage;

[0120] In the crushed ice entrainment stage, the crushed ice entrainment intensity is identified based on the adjusted signal recognition threshold, and the motor control parameters are adjusted accordingly;

[0121] Specifically, the control strategy layer includes:

[0122] The stage selection unit adopts a gated logic structure. Based on the stage probability vector output by the stage discrimination layer, it activates the expert neural network corresponding to the ice stage with the highest probability in the current stage from the expert network unit in a hard-gated manner.

[0123] The expert network unit includes three groups of expert neural networks that independently correspond to the ice sinking stage, ice throwing stage, and ice crushing stage, among which:

[0124] The expert neural network for the ice sinking stage and the ice throwing stage both adopt a fully connected neural network structure, taking the shared feature vector output by the feature extraction layer as input and outputting the motor control parameters of the corresponding stage to the output layer;

[0125] Among them, the expert neural network is specially trained to generate the first motor control instruction during the ice sinking stage. The first motor control strategy is to adopt a variable speed stirring mode during the ice sinking stage and periodically adjust the motor speed and direction switching.

[0126] During the ice sinking phase, the expert neural network determines the specific values ​​of the following motor parameters within the specified range (i.e., the range in brackets) based on the actual detected motor load and vibration data:

[0127] Base speed (50-80) RPM;

[0128] The speed is increased by (0.5-1) seconds every (5-8) seconds (the increase is 30%-50% of the basic speed);

[0129] After rotating in one direction for (3-5) seconds, reverse;

[0130] The value of torque output (70%-85% of rated torque).

[0131] Among them, the expert neural network in the ice-throwing wall stage is specially trained to generate the second motor control instruction. The second motor control strategy is to adopt an adaptive torque control mode with smooth torque changes in the ice-throwing wall stage, that is, when the motor load increases and the torque needs to be increased, the torque is slowly increased.

[0132] During the ice-throwing phase, the expert neural network determines the specific values ​​of the following motor parameters within the specified range based on the actual detected motor load and vibration data:

[0133] Base speed (100-150) RPM;

[0134] Torque variation limit (20%-40% of rated torque);

[0135] Torque smoothing coefficient, that is, the rate of change of torque (0.6-0.8).

[0136] The expert neural network in the ice crushing stage includes a shallow convolutional autoencoder and a fully connected neural network structure. The shallow convolutional autoencoder is used to compress and reconstruct the input shared feature vector to obtain abnormal residual features that characterize the ice crushing signal.

[0137] The abnormal residual feature is compared with the motor load identification threshold and the vibration identification threshold provided in real time by the threshold adjustment layer, so as to quantify the current crushed ice entrainment intensity.

[0138] Specifically, the shallow convolutional autoencoder includes an encoder part and a decoder part:

[0139] The encoder part consists of two one-dimensional convolutional layers. The first layer uses 16 convolution kernels, the convolution kernel size is 3, and the stride is 1. The second layer uses 8 convolution kernels, the convolution kernel size is 3, and the stride is 2. Each layer is followed by a batch normalization layer and a ReLU activation function to compress the input 128-dimensional shared feature vector into a 32-dimensional potential representation.

[0140] The decoder part consists of two one-dimensional transposed convolutional layers. The first layer uses 16 convolution kernels with a kernel size of 3 and a stride of 2. The second layer uses a convolution kernel with the same number of channels as the input features, a kernel size of 3 and a stride of 1. Each layer is followed by a batch normalization layer and a ReLU activation function to reconstruct the 32-dimensional potential representation into a feature vector with the same dimension as the original input.

[0141] During the training phase, the autoencoder uses normal dough mixing data (without crushed ice entrainment) for unsupervised learning. The mean squared error loss function is used to optimize the reconstruction error, enabling the network to learn the characteristic distribution of normal dough signals.

[0142] When the input contains a feature vector containing a crushed ice entrainment signal, the reconstruction process will produce an error because the network only learns the feature distribution of normal dough. This error is the abnormal residual feature RE(t).

[0143] Then, the subspace division of the shared feature vector is obtained based on prior knowledge. For example, in the embodiment of the present application, the first 32 dimensions of the shared feature vector correspond to the motor load characteristics, and the last 96 dimensions are vibration characteristics.

[0144] The division of the subspace needs to be based on the characteristics of the one-dimensional residual convolutional network structure in the shared feature extraction unit, because the network retains the basic correspondence between the original signal channels when extracting features.

[0145] The reconstruction error is divided into subspaces to process abnormal residual features RE(t):

[0146] Motor load anomaly residual: RL(t) = ||RE(t)[0:32]||², which is the sum of the squares of the reconstruction errors of the first 32 dimensions;

[0147] Vibration anomaly residual: RV(t) = ||RE(t)[32:128]||², which is the sum of squares of the reconstruction errors of the last 64 dimensions.

[0148] After obtaining the motor load abnormality residual and the vibration abnormality residual, the crushed ice entrainment strength can be calculated using the following formula. The quantitative calculation method of the crushed ice entrainment strength S(t) is:

[0149] S(t) = RL(t) / TL+ RV(t) / TV

[0150] Among them, TL and TV are the motor load recognition threshold and vibration recognition threshold provided by the threshold adjustment layer respectively;

[0151] The quantized crushed ice entrainment intensity and the shared feature vector are used as the input of the expert neural network in the crushed ice entrainment stage, and the motor control parameters of the crushed ice entrainment stage are output to the output layer.

[0152] Among them, the expert neural network in the crushed ice entrainment stage is specially trained to generate a third motor control instruction, and the third motor control strategy is to dynamically adjust the stirring time in the crushed ice entrainment stage.

[0153] During the ice entrainment phase, the expert neural network determines the specific values ​​of the following motor parameters within a specified range based on the actual detected motor load data, vibration data, and ice entrainment intensity:

[0154] Base speed (120-180) RPM;

[0155] Stirring time is (60-360) seconds.

[0156] The output layer is used to convert the motor control parameters of different stages into motor drive instructions, thereby obtaining the first motor control strategy, the second motor control strategy and the third motor control strategy respectively.

[0157] Specifically, the output layer includes:

[0158] A parameter safety domain clipping unit, configured to receive the control parameters output by the control strategy layer and clip the control parameters to a safety domain range to ensure that the output control instructions are within a safe operating range of the chef machine motor;

[0159] The control instruction generating unit is used to convert the control parameters after being clipped by the safety domain into a food processor motor driving instruction, wherein the food processor motor driving instruction at least includes a target motor rotation time, a target motor speed instruction, and a target motor torque instruction.

[0160] The technical solution provided by the embodiments of the present application has the following advantages compared with the prior art:

[0161] One of its working principles and beneficial effects is that when mixing dough in a food processor, users often add water containing ice cubes to reduce the temperature of the dough and prevent it from fermenting prematurely during mixing. Due to the addition of ice cubes, the state of the ice cubes during mixing will go through three stages: sinking to the bottom, being thrown off the wall, and being entrained by crushed ice. Different stages have different effects on dough mixing, for example:

[0162] When mixing flour and water at a low speed, due to the weight of the ice cubes, they will gather at the bottom due to the stirring effect to form a local low-temperature area, forming a sinking stage.

[0163] Low temperature will inhibit the hydration of flour, and the ice cubes are large and smooth at this time, making it difficult to flip at a slow speed, preventing the flour pressed under the ice cubes from flipping, resulting in insufficient hydration of the flour at the bottom and forming hard lumps at the bottom. These lumps will also cause uneven dough after mixing later;

[0164] After the flour and water are initially mixed, the speed of the paddle increases, and the ice cubes are thrown toward the inner wall of the container due to centrifugal force, causing the outer edge of the paddle to erratically strike the ice cubes. If the motor uses dynamic torque output, the motor torque will frequently fluctuate between "high torque" and "low torque" as the paddle strikes and pushes away the ice cubes, causing large fluctuations in the motor load, fatigue of the mechanical components, and affecting the stability of the equipment.

[0165] The ice cubes will be broken by the paddles during the wall-throwing stage, and the resulting crushed ice will be wrapped by the formed dough, entering the crushed ice entrainment stage.

[0166] If the dough is not stirred sufficiently, localized water spots will appear after the enclosed crushed ice melts, affecting the uniformity of the dough. Therefore, it is generally necessary to extend the stirring time, but this may also lead to over-mixing of the dough, resulting in an undesirable gluten structure or requiring more energy.

[0167] The embodiment of the present application proposes a motor control method for a food processor, which can identify the different stages of ice cubes during the dough mixing process and adopt different motor control strategies for each stage. During the ice sinking stage, the method adopts a first motor control strategy, which prevents the ice cubes from sinking to the bottom for a long time by reasonably adjusting the motor parameters, promotes full contact between the bottom flour and water, and achieves uniform hydration; when the ice cubes enter the wall throwing stage, the system switches to a second motor control strategy, which alleviates the violent fluctuations in the motor torque output caused by the impact of the ice cubes; finally, during the crushed ice entrainment stage, the third motor control strategy is activated, and an intelligent algorithm is used to control the uniform distribution of crushed ice in the dough, while avoiding excessive mixing that may damage the gluten structure.

[0168] Therefore, the motor control method for a food processor provided in the embodiment of the present application can better stir the dough.

[0169] The second working principle and its beneficial effect is that during the crushed ice entrainment stage, since the crushed ice is wrapped in the dough and highly overlaps with the dough's stirring resistance signal, the traditional motor control method is difficult to accurately identify the current degree of entrainment of the crushed ice at this stage, resulting in the inability to adjust the motor's operation according to the current degree of ice breakage and the current state of the dough, thereby controlling the dough's stirring time.

[0170] The embodiment of the present application introduces an early information memory layer and a dynamic threshold adjustment mechanism. When the ice cubes enter the wall-throwing stage, the motor control system predicts and records the data of ice crushing through the early information memory layer. By being able to better capture the characteristic signals of the ice cubes in the early stage, the predicted and recorded data of the ice crushing are used as a reference for determining the separation strength of the crushed ice in the subsequent crushing and entrainment stage. By making corresponding judgments in the early stages of ice crushing, a more accurate basis is provided for signal identification in the later crushing and entrainment stage, thereby avoiding the problem in the traditional method that the crushed ice entrainment strength is difficult to determine due to inaccurate separation threshold setting, resulting in insufficient or excessive dough mixing.

[0171] Therefore, the motor control method for a food processor provided in the embodiment of the present application can better stir the dough.

[0172] It should be noted that, in this document, relational terms such as "first" and "second" are used solely to distinguish one entity or operation from another, and do not necessarily require or imply any actual relationship or order between these entities or operations. Furthermore, the terms "include," "comprises," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, article, or device. Without further limitation, an element defined by the phrase "comprising a..." does not preclude the presence of additional identical elements in the process, method, article, or device comprising the element. Furthermore, in the description of the embodiments of this application, unless otherwise specified, " / " represents or. For example, A / B can represent either A or B. "And / or" herein is merely a description of an associative relationship between associated objects, indicating that three relationships can exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, in the description of the embodiments of the present application, “plurality” refers to two or more than two.

[0173] The foregoing description is intended only to provide specific embodiments of the present application, which will enable those skilled in the art to understand and implement the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments described herein, but is intended to be construed in the broadest manner consistent with the principles and novel features disclosed herein.

Claims

1. A motor control method for a food processor, characterized in that: The motor control method for a food processor comprises the following steps: Obtaining motor load data and vibration data during the dough mixing process; based on the motor load data and vibration data, identifying the state of the ice cubes during the dough mixing process, including the ice cube sinking stage, the ice cube throwing stage, and the ice crushing stage; Depending on the identified ice state, different motor control strategies are implemented accordingly: During the ice sinking stage, a first motor control strategy is adopted, wherein the first motor control strategy is to adopt a variable speed stirring mode during the ice sinking stage and periodically adjust the motor speed and direction switching to prevent the ice from continuously sinking to the bottom and promote uniform hydration of the flour at the bottom; During the ice-wall throwing stage, a second motor control strategy is adopted, which is an adaptive torque control mode with smooth torque changes during the ice-wall throwing stage, thereby controlling the ice crushing process and reducing motor load fluctuations; During the crushed ice entrainment stage, a third motor control strategy is adopted, which dynamically adjusts the stirring time during the crushed ice entrainment stage, thereby controlling the uniform distribution of the crushed ice and avoiding excessive stirring.

2. The motor control method for a food processor according to claim 1, characterized in that: The motor control method for a food processor further includes: During the ice wall throwing stage, predict ice breakage data; According to the predicted ice crushing data, the signal recognition threshold of the ice crushing stage is adjusted; Based on the adjusted signal recognition threshold, the current crushed ice entrainment state is detected to adjust the third motor control strategy, thereby avoiding excessive stirring while ensuring uniform distribution of the crushed ice.

3. The motor control method for a food processor according to claim 2, characterized in that: The first motor control strategy, the second motor control strategy and the third motor control strategy are outputted through a preset food processor motor control model; The food processor motor control model includes: Input layer, used to receive motor load data and vibration data; The feature extraction layer is used to receive data from the input layer and process the data; The stage discrimination layer receives the output of the feature extraction layer to identify the stage of the ice cube in the dough mixing process; The early information memory layer is used to predict and record ice breakage prediction data based on the data from the feature extraction layer during the ice throwing stage; A threshold adjustment layer, configured to dynamically adjust the signal recognition threshold of the ice entrainment phase according to the ice breakage prediction data stored in the previous information memory layer; The control strategy layer is used to obtain the corresponding motor control parameters based on the output of the feature extraction layer using the corresponding expert neural network according to the identified ice stage; In the crushed ice entrainment stage, the crushed ice entrainment intensity is identified based on the adjusted signal recognition threshold, and the motor control parameters are adjusted accordingly; The output layer is used to convert the motor control parameters of different stages into motor drive instructions, thereby obtaining the first motor control strategy, the second motor control strategy and the third motor control strategy respectively.

4. The motor control method for a food processor according to claim 3, characterized in that: The feature extraction layer includes: a time-frequency preprocessing unit, configured to perform a short-time Fourier transform on the motor load data and the vibration data to generate a time-frequency spectrum; The shared feature extraction unit adopts a one-dimensional residual convolutional network structure to extract cross-channel temporal features from the time-frequency spectrum to form a shared feature vector.

5. The motor control method for a food processor according to claim 4, characterized in that: The stage discrimination layer adopts a Transformer structure, which is used to receive the shared feature vector and output a stage probability vector representing the ice sinking stage, the ice throwing wall stage and the broken ice entrainment stage.

6. The motor control method for a food processor according to claim 5, characterized in that: The early information memory layer uses a fully connected neural network + LSTM structure, which is activated when the ice cube is judged to be in the wall-throwing stage, and predicts the following indicators based on the shared feature vector of the ice cube wall-throwing stage: The size of the ice when it breaks; The degree of ice fragmentation.

7. The motor control method for a food processor according to claim 6, characterized in that: The threshold adjustment layer adopts a fully connected neural network structure, which receives the size of ice cubes and the degree of ice breakage predicted by the previous information memory layer, and outputs the motor load recognition threshold and vibration recognition threshold required for the ice crushing stage.

8. The motor control method for a food processor according to claim 7, characterized in that: The control strategy layer includes: The stage selection unit adopts a gated logic structure. Based on the stage probability vector output by the stage discrimination layer, it activates the expert neural network corresponding to the ice stage with the highest probability in the current stage from the expert network unit in a hard-gated manner. The expert network unit includes three groups of expert neural networks that independently correspond to the ice sinking stage, ice throwing stage, and ice crushing stage, among which: The expert neural network for the ice sinking stage and the ice throwing stage both adopt a fully connected neural network structure, taking the shared feature vector output by the feature extraction layer as input and outputting the motor control parameters of the corresponding stage to the output layer; The expert neural network in the ice crushing stage includes a shallow convolutional autoencoder and a fully connected neural network structure. The shallow convolutional autoencoder is used to compress and reconstruct the input shared feature vector to obtain abnormal residual features that characterize the ice crushing signal. The abnormal residual features are compared with the motor load identification threshold and vibration identification threshold provided in real time by the threshold adjustment layer, so as to quantify the current crushed ice entrainment intensity. The quantified crushed ice entrainment intensity and the shared feature vector are used as the input of the expert neural network in the crushed ice entrainment stage, and the motor control parameters of the crushed ice entrainment stage are output to the output layer.

9. The motor control method for a food processor according to claim 8, characterized in that: The output layer includes: A parameter safety domain clipping unit, configured to receive the control parameters output by the control strategy layer and clip the control parameters to a safety domain range to ensure that the output control instructions are within a safe operating range of the chef machine motor; The control instruction generating unit is used to convert the control parameters after being clipped by the safety domain into a food processor motor driving instruction, wherein the food processor motor driving instruction at least includes a target motor rotation time, a target motor speed instruction, and a target motor torque instruction.

Images