Intelligent cooking quality evaluation and feedback optimization system and method

Abstract

The invention belongs to the technical field of cooking quality evaluation, and discloses an intelligent cooking quality evaluation and feedback optimization system and method, and the method comprises the steps: carrying out the signal collection and image collection of cooking parameters and dish appearance in the cooking process of a target dish, and obtaining a corresponding cooking sensing signal and a dish cooking image; respectively performing enhancement processing on the obtained cooking sensing signal and the dish cooking image to obtain a corresponding enhanced cooking signal and an enhanced cooking image; performing quality evaluation on cooking parameters and dish appearance in the cooking process of the target dish based on the obtained enhanced cooking signal and enhanced cooking image to obtain a corresponding quality evaluation result; and performing parameter optimization on the cooking parameters in the cooking process of the target dish based on the obtained quality evaluation result. According to the invention, the scientificity, accuracy and user experience of cooking are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of cooking quality assessment, and more specifically, to an intelligent cooking quality assessment and feedback optimization system and method. Background Art

[0002] In the current cooking field, traditional cooking methods have many limitations. With the development of science and technology and the improvement of people's living standards, the demand for intelligent and precise cooking is becoming increasingly urgent.

[0003] Compared with the existing technology, the traditional evaluation of the cooking quality of dishes mainly relies on the experience of chefs. It is impossible to conduct comprehensive and accurate quantitative monitoring of key parameters in the cooking process, such as temperature, weight, humidity, and changes in gas composition. Chefs can only rely on their senses and experience to judge, which makes it difficult to ensure the stability of the cooking results. Different chefs or the same chef at different times may make the same dish, and there may be large differences in taste, color, doneness, etc., which cannot meet the needs of large-scale standardized catering production and consumers for a stable food experience; and it is highly subjective; even if problems are found in the quality of dishes, it is difficult to quickly and accurately find the corresponding cooking parameter problems and give optimization measures. Users can only rely on limited experience to explore and improve on their own, lacking a systematic feedback mechanism and precise optimization guidance; this leads to a slow improvement in cooking level, and it is impossible to fully utilize modern technology to improve cooking efficiency and quality, and it is difficult to adapt to people's demand for convenient and high-quality cooking in fast-paced life.

[0004] In view of this, the present invention proposes an intelligent cooking quality assessment and feedback optimization system and method to solve the above problems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solutions:

[0006] An intelligent cooking quality evaluation and feedback optimization system, comprising:

[0007] The data acquisition module, based on the pre-built IoT terminal array and terminal, respectively performs signal acquisition and image acquisition on the cooking parameters and the appearance of the target dish during the cooking process to obtain the corresponding dish cooking information; the dish cooking information includes the cooking sensor signal and the dish cooking image;

[0008] Data processing module, used to decompose the cooking sensor signal into Components and perform delayed reconstruction, and obtain the delayed reconstruction The self-clustering coefficient corresponding to the component; and based on it, judge the corresponding Whether the random error of the component meets the requirements, if not, the corresponding The components are adaptively enhanced and reconstructed to obtain corresponding enhanced cooking signals, and the corresponding cooking images of the dishes are simultaneously enhanced to obtain corresponding enhanced cooking images;

[0009] The dish evaluation module evaluates the cooking parameters and the appearance of the target dish during the cooking process based on the preset cooking standards and in combination with the obtained enhanced cooking signals and enhanced cooking images to obtain the corresponding quality evaluation results;

[0010] The feedback optimization module optimizes the cooking parameters of the target dish during the cooking process based on the quality evaluation results obtained, and gives cooking optimization suggestions.

[0011] Furthermore, the process of obtaining the corresponding dish cooking information includes:

[0012] The data acquisition module is provided with an acquisition node and an image terminal; the acquisition node is composed of a plurality of IoT terminal arrays with specific different functions, and is used to collect data on the cooking process of the target dish and obtain corresponding cooking sensor signals; the cooking sensor signals include sensor signals corresponding to temperature, weight, humidity and other cooking parameters;

[0013] The image terminal is used to collect images of the appearance of the dish during the cooking process to obtain corresponding cooking images of the dish;

[0014] The collected dish cooking images and cooking sensor signals are summarized to obtain corresponding dish cooking information.

[0015] Furthermore, the process of acquiring the corresponding enhanced cooking signal and enhanced cooking image includes:

[0016] Perform empirical mode decomposition on the corresponding cooking sensor signal to obtain the corresponding residual signal and several Quantity;

[0017] For the obtained The components are reconstructed with delay to obtain the corresponding delay components, and the corresponding regression coefficients are constructed based on them; the regression coefficients between all data points in the corresponding delay components are obtained, and they are mapped into a pre-constructed blank matrix to obtain the corresponding regression matrix; the matrix elements in the corresponding regression matrix are mapped into the pre-constructed complex network to obtain the corresponding network subgraph;

[0018] Perform image analysis on the obtained network subgraph to obtain the self-clustering coefficient corresponding to the corresponding network subgraph;

[0019] Set the clustering threshold, compare the obtained self-clustering coefficient with the clustering threshold, if the self-clustering coefficient is not less than the clustering threshold, The components are decomposed by wavelet transform to obtain high-frequency coefficients and low-frequency coefficients at different decomposition layers;

[0020] Adaptive threshold processing is performed on the high-frequency coefficients in each decomposition layer, and the high-frequency coefficients and low-frequency coefficients after threshold processing are reconstructed to obtain the corresponding enhancement Quantity;

[0021] The enhanced Quantity and random error meet the requirements The components and the corresponding residual components are recombined to obtain the corresponding enhanced cooking signal;

[0022] The dish cooking image in the corresponding dish cooking information is processed to obtain a corresponding enhanced cooking image, which is then combined with the corresponding enhanced cooking signal to obtain corresponding enhanced cooking information.

[0023] Furthermore, the self-clustering coefficient ; In the formula, is the total number of nodes in the network subgraph; Indicates The local clustering coefficient corresponding to each node; ; In the formula, Representation and network nodes The total number of network nodes with visible edges between them; , and Represents network nodes , and The regression coefficient between For and point Index of the number of network nodes with visible edges between them;

[0024] The formula for adaptive threshold processing is:

[0025] ; Indicates The high frequency coefficients within the layer decomposition layer, represents symbolic function operation, Indicates a preset high frequency coefficient threshold; represents the high frequency coefficient after threshold processing, represents the adjustment coefficient; Represented by natural numbers The exponential function operation with base is.

[0026] Furthermore, the process of obtaining the corresponding quality assessment results includes:

[0027] Inputting the obtained enhanced cooking information into a pre-built dish quality assessment model to obtain a corresponding model output result, and based on the model output result, obtaining a dish cooking result corresponding to the corresponding dish cooking process;

[0028] Comparing the obtained dish cooking results with the pre-set cooking standards, and marking the cooking standards in the corresponding dish cooking results as normal standards and low-quality standards according to the comparison results;

[0029] Based on the low quality standards and normal standards marked by the corresponding dish cooking results, scores are assigned to the corresponding dish cooking results to generate corresponding quality assessment results.

[0030] Furthermore, the construction process of the dish quality evaluation model includes:

[0031] The backbone network of the dish quality assessment model is an improved convolutional neural network, and the basic framework of the improved convolutional neural network is an input layer, a feature layer, a fusion layer and an output layer;

[0032] The input layer is used to receive input training samples and preprocess them to obtain corresponding input data;

[0033] The feature layer includes improved Coding modules and improvements Modules;

[0034] The improvements The encoding module is composed of a convolution unit and a forward propagation unit. The convolution unit is used to perform an encoding operation on the cooking sensor signal in the input data to obtain a corresponding encoding sequence; perform feature extraction on the corresponding encoding sequence to obtain corresponding local features and mark them, and input the marked local features into the forward propagation unit to obtain a corresponding output feature vector;

[0035] The improvements The module is implemented through a The convolution kernel performs convolution processing on the input cooking image of the dish to obtain the corresponding convolution feature map; the convolution feature map is downsampled through a maximum pooling layer to obtain the corresponding spatial feature map; and a The convolution kernel and a global average pooling layer perform convolution operations and spatial dimension reduction on the obtained spatial feature map to obtain an output feature map;

[0036] The fusion layer is used to receive the obtained output feature vector and output feature map, and perform adaptive weighted fusion on them to obtain corresponding output feature information;

[0037] The output layer is used to receive output feature information and based on the fully connected layer and Function, maps the corresponding output feature information into the pre-constructed space and classifies the results; obtains the corresponding dish cooking results;

[0038] Construct a training data set and define As the optimizer continuously optimizes the parameters of the dish quality assessment model during the training process, the corresponding training data set is divided into multiple batches, and is input into the dish quality assessment model one by one based on the time before and after, and the value of the corresponding loss function is recorded. When the value of the batch loss function no longer decreases, the parameters of the dish quality assessment model at this time are saved, and the training of the dish quality assessment model is completed; is a constant.

[0039] Furthermore, the process of labeling includes:

[0040] The constructed coding sequence was divided equally to obtain coded segments, and based on three The convolution kernel performs convolution operation on the corresponding encoding fragment to generate the corresponding request matrix , primary key matrix and the numerical matrix ; Based on it, the feature score corresponding to the corresponding coding segment is obtained ; In the formula, Represents matrix transpose; The vector dimension representing the primary key matrix; Represents a mask operation;

[0041] A score threshold is set. If the feature score is greater than the score threshold, the local features corresponding to the corresponding local fragments are marked.

[0042] Furthermore, the formula for convolution processing is:

[0043] ; In the formula, Refers to the convolution feature map; and Respectively represent the horizontal and vertical coordinate indexes of the pixel points in the corresponding convolution feature map; Represents the layer connected to the input The number index of convolution kernels; and Indicates the corresponding The offset of the spatial horizontal and vertical positions of the convolution kernel on the input dish cooking image and ; represents the input channel index, is the total number of input channels, represents the weight matrix; Indicates The pixel coordinates of the dish cooking image input in the input channel are The pixel value of represents the bias term;

[0044] The formula for downsampling operation is: ; In the formula, Represents the spatial feature map obtained after the downsampling operation is completed. and Respectively represent the indexes of the horizontal and vertical coordinates of the pixel points in the spatial feature map; Represents the first layer connected to the maximum pooling layer indivual Convolution feature map output by the convolution kernel; and Respectively represent the indexes of the horizontal and vertical coordinates of the pixels in the convolution feature map; Indicates finding the vertical spatial position of the corresponding convolution feature map The maximum value of pixels; Indicates finding the horizontal spatial position of the corresponding convolution feature map The maximum value of pixels; and Respectively represent the horizontal and vertical strides of the pooling window in the maximum pooling layer; and Respectively represent the height and width of the pooling window in the maximum pooling layer, Indicates the connection with the maximum pooling layer Index of the number of convolution kernels.

[0045] Furthermore, the process of optimizing the cooking parameters in the cooking process of the target dish based on the obtained quality assessment results includes:

[0046] Obtaining cooking parameters corresponding to the low-quality standard based on the quality assessment result, and comparing the cooking parameters with the pre-set parameter standard to obtain corresponding comparison results, and generating corresponding cooking optimization suggestions based on the comparison results;

[0047] The corresponding quality assessment results and cooking optimization suggestions are fed back to the user in an intuitive manner, such as through the display screen of the cooking equipment or the mobile phone application; the cooking parameter settings during the cooking process of the dish can also be automatically optimized based on user feedback and historical cooking data.

[0048] An intelligent cooking quality assessment and feedback optimization method, comprising:

[0049] Step 1: Based on the pre-built IoT terminal array and terminal, respectively, the cooking parameters and the appearance of the target dish during the cooking process are collected by signal and image, and the corresponding dish cooking information is obtained; the dish cooking information includes the cooking sensor signal and the dish cooking image;

[0050] Step 2: Decompose the cooking sensor signal into Components and perform delayed reconstruction, and obtain the delayed reconstruction The self-clustering coefficient corresponding to the component; and based on it, judge the corresponding Whether the random error of the component meets the requirements, if not, the corresponding The components are adaptively enhanced and reconstructed to obtain corresponding enhanced cooking signals, and the corresponding cooking images of the dishes are simultaneously enhanced to obtain corresponding enhanced cooking images;

[0051] Step 3: Based on the preset cooking standard and in combination with the obtained enhanced cooking signal and enhanced cooking image, the cooking parameters and the appearance of the target dish during the cooking process are evaluated to obtain the corresponding quality evaluation result;

[0052] Step 4: Based on the quality assessment results obtained, the cooking parameters of the target dish are optimized during the cooking process, and cooking optimization suggestions are given.

[0053] Technical effects and advantages of an intelligent cooking quality assessment and feedback optimization system and method of the present invention:

[0054] 1. The present invention forms a closed loop by integrating data collection, processing, evaluation and feedback optimization functions, making the cooking process digital and intelligent; helping users understand the cooking process and optimize operations, promoting the intelligent development of the cooking industry, improving cooking efficiency and quality, and adapting to the needs of modern life;

[0055] 2. By constructing a dish quality evaluation model, it is possible to objectively and accurately evaluate the cooking standards of dishes such as taste, color, and doneness, overcoming the drawbacks of the traditional evaluation method's strong subjectivity and providing a scientific basis for improving cooking quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A schematic diagram of an intelligent cooking quality evaluation and feedback optimization system of the present invention;

[0057] Figure 2 A schematic diagram of an intelligent cooking quality assessment and feedback optimization method of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] Example 1

[0060] See also Figure 1 As shown, this embodiment provides an intelligent cooking quality assessment and feedback optimization system, including:

[0061] The data acquisition module, based on the pre-built IoT terminal array and terminal, respectively performs signal acquisition and image acquisition on the cooking parameters and the appearance of the target dish during the cooking process to obtain the corresponding dish cooking information; the dish cooking information includes the cooking sensor signal and the dish cooking image;

[0062] Data processing module, used to decompose the cooking sensor signal into Components and perform delayed reconstruction, and obtain the delayed reconstruction The self-clustering coefficient corresponding to the component; and based on it, judge the corresponding Whether the random error of the component meets the requirements, if not, the corresponding The components are adaptively enhanced and reconstructed to obtain corresponding enhanced cooking signals, and the corresponding cooking images of the dishes are simultaneously enhanced to obtain corresponding enhanced cooking images;

[0063] The dish evaluation module evaluates the cooking parameters and the appearance of the target dish during the cooking process based on the preset cooking standards and in combination with the obtained enhanced cooking signals and enhanced cooking images to obtain the corresponding quality evaluation results;

[0064] The feedback optimization module optimizes the cooking parameters of the target dish during cooking based on the quality evaluation results obtained, and gives cooking optimization suggestions;

[0065] The modules are connected to each other via wired and / or wireless means to achieve data transmission between modules.

[0066] It should be further explained that, in the specific implementation process, the process of obtaining the corresponding dish cooking information includes:

[0067] The data acquisition module is provided with an acquisition node and an image terminal; the acquisition node refers to a combination of an array of IoT terminals with specific different functions, which is used to collect data on the cooking process of the target dish and obtain the corresponding cooking sensor signal; the cooking sensor signal includes sensor signals corresponding to other cooking parameters such as temperature, weight, humidity and gas; wherein, the acquisition process of the gas sensor signal is as follows: the corresponding IoT terminal responds to the presence of specific gas or chemical substance in the cooking process of the target dish, and when the corresponding IoT terminal recognizes the specific gas or chemical substance, it generates a specific response signal and converts it into an electrical signal, namely, a gas sensor signal;

[0068] The image terminal is used to collect images of the appearance of dishes during the cooking process to obtain corresponding cooking images of the dishes;

[0069] The collected dish cooking images and cooking sensor signals are summarized to obtain corresponding dish cooking information.

[0070] It should be further explained that, in the specific implementation process, the acquisition process of the corresponding enhanced cooking signal and enhanced cooking image includes:

[0071] Perform empirical mode decomposition on the cooking sensor signal in the corresponding dish cooking information to obtain the corresponding residual signal and several Component; the formula for empirical mode decomposition is: ; In the formula, Indicates cooking sensor signal; The decomposition results in indivual Quantity, Represents a time index; represents the residual component, Represents the decomposition result Total number of portions; For the corresponding The number index of the components; it should be further explained that each The conditions that need to be met include The number of extreme points and the number of zero-crossing points in a component cannot differ by more than 1; and at any point, the mean of the upper and lower envelopes formed by the maximum and minimum values ​​must be zero;

[0072] Furthermore, the obtained The components are reconstructed to obtain the corresponding delay components, and the corresponding regression coefficients are constructed based on them. ; In the formula, Indicates the preset parameter threshold, usually a fixed constant; and refer to the parameter values ​​corresponding to any two data points in the corresponding delay component, and ; represent function;

[0073] Obtain the regression coefficients between all data points within the corresponding delay component, and map them into a pre-constructed blank matrix to obtain the corresponding regression matrix; map the matrix elements within the corresponding regression matrix into a pre-constructed complex network to obtain the corresponding network subgraph, where the network subgraph consists of several network nodes and visible edges, and one network node corresponds to one matrix element of the regression matrix; whether there is a visible edge between network nodes depends on whether the two network nodes meet the visibility criterion; the visibility criterion means that for any two points, if the connection line between them is not truncated by any other network points located between these two points, then these two network nodes are "visible" and can be connected into an edge in the network subgraph, and adjacent two nodes must be "visible";

[0074] Perform image analysis on the obtained network subgraph to obtain the self-clustering coefficient corresponding to the corresponding network subgraph ; where is the total number of nodes within the network subgraph; represents the th node corresponding local clustering coefficient; ; where represents the total number of network nodes that have visible edges with the network node ; , and respectively represent the regression coefficients between the network nodes , and ; represents the number index of the network nodes that have visible edges with the point ;

[0075] Set the clustering threshold, compare the obtained self-clustering coefficient with the clustering threshold. If the self-clustering coefficient is less than the clustering threshold, it indicates that the random error of the corresponding component meets the requirements, then no other operations are performed; if the self-clustering coefficient is not less than the clustering threshold, it indicates that the random error of the corresponding component does not meet the requirements, then perform wavelet transform decomposition on the corresponding component to obtain the high-frequency coefficients and low-frequency coefficients at different decomposition levels;

[0076] The formula for wavelet transform decomposition is: ; where , respectively represent the low-pass filter and the high-pass filter; represents the signal displacement; Indicates the number of decomposition levels; and denote the scaling coefficient and wavelet coefficient respectively; Indicates that decomposition is required Quantity; and and All are integers;

[0077] Adaptive threshold processing is performed on the high-frequency coefficients in each decomposition layer, and the high-frequency coefficients and low-frequency coefficients after threshold processing are reconstructed to obtain the corresponding enhancement Quantity;

[0078] The formula for adaptive threshold processing is:

[0079] ; Indicates The high frequency coefficients within the layer decomposition layer, represents symbolic function operation, Indicates a preset high frequency coefficient threshold; Represents the high-frequency coefficient after threshold processing; represents the adjustment coefficient; Represented by natural numbers Exponential function operation with base number;

[0080] The enhanced Quantity and random error meet the requirements The components and the corresponding residual components are recombined to obtain the corresponding enhanced cooking signal;

[0081] The dish cooking image in the corresponding dish cooking information is processed to obtain a corresponding enhanced cooking image, which is then combined with the corresponding enhanced cooking signal to obtain corresponding enhanced cooking information.

[0082] It should be further explained that, in the specific implementation process, the process of obtaining the corresponding quality assessment results includes:

[0083] Input the obtained enhanced cooking information into a pre-built dish quality assessment model to obtain a corresponding model output result, and based on the model output result, obtain a dish cooking result corresponding to the corresponding dish cooking process; the dish cooking result includes predicted cooking standards such as taste, color, and doneness;

[0084] The obtained cooking result of the dish is compared with the preset cooking standard. If the cooking standard in the corresponding cooking result of the dish is less than the preset cooking standard, the corresponding cooking standard is marked as a low-quality standard; if the cooking standard in the corresponding cooking result of the dish is not less than the preset cooking standard, the corresponding cooking standard is marked as a normal standard;

[0085] Then, based on the low quality standard and normal standard marked in the corresponding dish cooking result, the corresponding dish cooking result is scored to obtain a corresponding dish quality score, and a corresponding quality evaluation result is generated based on the score, the quality evaluation result including a single score of each cooking standard and a dish quality score of the overall dish cooking process;

[0086] The construction process of the dish quality assessment model includes:

[0087] The backbone network of the dish quality assessment model is an improved convolutional neural network, which is used to mine and learn the nonlinear mapping relationship between cooking parameters and cooking quality. The basic framework of the improved convolutional neural network is the input layer, feature layer, fusion layer and output layer.

[0088] The input layer is used to receive the input training samples and preprocess them to obtain the corresponding input data. The preprocessing refers to normalizing and scaling the cooking sensor signals and the cooking images in the training samples respectively.

[0089] Feature layers include improvements Coding modules and improvements Modules;

[0090] improve The encoding module consists of a convolution unit and a forward propagation unit. The convolution unit is used to encode the cooking sensor signal in the input data to obtain the corresponding encoding sequence. The encoding operation includes position encoding and time encoding. Then, the corresponding encoding sequence is feature extracted to obtain the corresponding local features. The global temporal dependency in the local features is captured and annotated based on the introduced multi-head causal self-attention mechanism. The annotated local features are input into the forward propagation unit. The forward propagation unit obtains the output feature vector through residual, regularization operation and a forward propagation operation.

[0091] The process of annotation includes: dividing the constructed coding sequence into equal parts to obtain coded segments, and based on three The convolution kernel performs convolution operation on the corresponding encoding fragment to generate the corresponding request matrix , primary key matrix and the numerical matrix ; Then, based on it, the feature score corresponding to the corresponding coding segment is obtained ; In the formula, Represents matrix transpose; The vector dimension representing the primary key matrix; Represents a mask operation that sets the upper triangular elements of a matrix (excluding the diagonal) to ; and by introducing a mask, The multi-head attention mechanism used can fully explore and retain all the relevant relationship information between the elements of the sequence, while suppressing the interference of noise, so that the model can focus on key feature information;

[0092] Set a score threshold. If the feature score is greater than the score threshold, the local features corresponding to the corresponding local fragments will be marked to encourage the model to increase the attention level of the corresponding local area. If the feature score is not greater than the score threshold, no other operations will be performed.

[0093] improve The module is implemented through a The convolution kernel performs convolution processing on the input dish cooking image to obtain the corresponding convolution feature map; the formula for convolution processing is:

[0094] ; In the formula, Refers to the convolution feature map; and Respectively represent the horizontal and vertical coordinate indexes of the pixel points in the corresponding convolution feature map; Represents the layer connected to the input The number index of convolution kernels; and Respectively indicate the corresponding The offset of the spatial horizontal and vertical positions of the convolution kernel on the input dish cooking image and ; represents the input channel index, is the total number of input channels, represents the weight matrix; Indicates The pixel coordinates of the input dish cooking image in the input channel are The pixel value of represents the bias term;

[0095] Then, the convolution feature map is downsampled through a maximum pooling layer to obtain the corresponding spatial feature map; the formula for the downsampling operation is: ; In the formula, Represents the spatial feature map obtained after the downsampling operation is completed. and Respectively represent the indexes of the horizontal and vertical coordinates of the pixel points in the spatial feature map; Represents the first layer connected to the maximum pooling layer indivual Convolution feature map output by the convolution kernel; and Respectively represent the indexes of the horizontal and vertical coordinates of the pixels in the convolution feature map; Indicates finding the vertical spatial position of the corresponding convolution feature map The maximum value of pixels; Indicates finding the horizontal spatial position of the corresponding convolution feature map The maximum pixel value; Note that during the corresponding downsampling operation, it is necessary to first find the maximum pixel value in the vertical space position and then find the maximum pixel value in the horizontal space position; and Respectively represent the horizontal and vertical strides of the pooling window in the maximum pooling layer; and Respectively represent the height and width of the pooling window in the maximum pooling layer; Indicates the connection with the maximum pooling layer The number index of convolution kernels;

[0096] Further, based on the improvement 3 improved modules within the module Extract and learn the visual features of cooking dishes, and then pass through a The convolution kernel and a global average pooling layer perform convolution operations and spatial dimension reduction on the obtained spatial feature map to obtain the output feature map;

[0097] The formula for the convolution operation is: ; In the formula, express Convolution kernel index, Indicates the corresponding The index of the number of channels in the convolution kernel, is the total number of channels; represents the bias term; is the weight matrix; and Respectively refer to the horizontal coordinate and vertical coordinate index of the pixel point in the corresponding spatial feature map; Indicates the corresponding Convolution kernel Spatial feature map of the input within the channel; Represents the spatial feature map after the convolution operation;

[0098] The formula for spatial dimension reduction is: ; In the formula, and Represents the height and width of the spatial feature map after the convolution operation; and are the horizontal and vertical coordinate indices respectively; represents the output feature map;

[0099] The fusion layer is used to receive the obtained output feature vector and output feature map, and perform adaptive weighted fusion on them to obtain the corresponding output feature information; the adaptive weighted fusion process includes: respectively obtaining the corresponding input feature vector and the feature vector corresponding to the output feature map, and processing them based on the fully connected layer set in the fusion layer, and then The function performs feature weight mapping on the input feature vector processed by the fully connected layer and the feature vector corresponding to the output feature map, and performs feature fusion based on the mapped feature weight vector;

[0100] The output layer is used to receive the output feature information and based on the fully connected layer and Function, maps the corresponding output feature information into the pre-constructed space and classifies the results; obtains the corresponding dish cooking results;

[0101] Define the loss function of the dish quality assessment model ; In the formula, Represents the true sample label of the input training sample; Indicates the probability that the prediction result is a certain sample label; Represents the nonlinear combination of training samples and weight parameters and their corresponding features in the input training data set; is the regularization coefficient; represents the weight coefficient; and Respectively represent the index of training samples and the total number of samples;

[0102] Obtaining several sets of cooking information involved in the cooking process of historical dishes, manually annotating them, and constructing a corresponding training data set based on the manually annotated cooking information; the manual annotation includes the taste, color, and doneness of the current dish;

[0103] definition As the optimizer continuously optimizes the parameters of the dish quality assessment model during the training process, the corresponding training data set is divided into multiple batches, and the batches are input into the dish quality assessment model based on the time before and after, and the value of the corresponding loss function is recorded. When the value of the batch loss function no longer decreases, the parameters of the dish quality assessment model at this time are saved, and the training of the dish quality assessment model is completed; is a constant.

[0104] It should be further explained that, in the specific implementation process, the process of optimizing the cooking parameters in the cooking process of the target dish based on the obtained quality assessment results includes:

[0105] Based on the quality assessment result, the cooking parameters corresponding to the low quality standard are obtained, and compared with the pre-set parameter standard to obtain the corresponding comparison result, and the corresponding cooking optimization suggestion is generated based on the comparison result; in one embodiment of the present invention, during the cooking process, if the target dish is not cooked to a sufficient degree of doneness, the temperature and cooking time of the corresponding dish during the cooking process are compared; and based on the comparison result, the corresponding cooking optimization suggestion is generated, such as increasing the cooking time or temperature;

[0106] Furthermore, the corresponding quality assessment results and cooking optimization suggestions are fed back to the user in an intuitive manner, such as through the display screen of the cooking equipment or a mobile phone application. At the same time, the cooking parameter settings can be automatically optimized based on user feedback and historical cooking data to provide more accurate guidance for the next cooking.

[0107] The present invention provides users with a more convenient, efficient and accurate cooking experience through comprehensive and accurate data collection, efficient enhanced processing of cooking information, accurate cooking quality assessment, and intelligent cooking feedback and optimization.

[0108] Example 2

[0109] See also Figure 2 As shown, the part not described in detail in this embodiment is described in Example 1, which provides an intelligent cooking quality evaluation and feedback optimization method, including:

[0110] Step 1: Based on the pre-built IoT terminal array and terminal, respectively, the cooking parameters and the appearance of the target dish during the cooking process are collected by signal and image, and the corresponding dish cooking information is obtained; the dish cooking information includes the cooking sensor signal and the dish cooking image;

[0111] Step 2: Decompose the cooking sensor signal into Components and perform delayed reconstruction, and obtain the delayed reconstruction The self-clustering coefficient corresponding to the component; and based on it, judge the corresponding Whether the random error of the component meets the requirements, if not, the corresponding The components are adaptively enhanced and reconstructed to obtain corresponding enhanced cooking signals, and the corresponding cooking images of the dishes are simultaneously enhanced to obtain corresponding enhanced cooking images;

[0112] Step 3: Based on the preset cooking standard and in combination with the obtained enhanced cooking signal and enhanced cooking image, the cooking parameters and the appearance of the target dish during the cooking process are evaluated to obtain the corresponding quality evaluation result;

[0113] Step 4: Based on the quality assessment results obtained, the cooking parameters of the target dish are optimized during the cooking process, and cooking optimization suggestions are given.

[0114] Example 3

[0115] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the intelligent cooking quality assessment and feedback optimization system and method provided above is implemented.

[0116] Since the electronic device introduced in this embodiment is an electronic device used to implement an intelligent cooking quality assessment and feedback optimization system and method in the embodiment of the present application, based on the intelligent cooking quality assessment and feedback optimization system and method introduced in the embodiment of the present application, the technical personnel of the field can understand the specific implementation of the electronic device of the present embodiment and its various variations, so how the electronic device implements the method in the embodiment of the present application is not described in detail here. As long as the technical personnel of the field implement the electronic device used in the intelligent cooking quality assessment and feedback optimization system and method in the embodiment of the present application, it belongs to the scope of protection of this application.

[0117] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0118] The above are only preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent cooking quality evaluation and feedback optimization system, characterized in that: include: The data acquisition module, based on the pre-built IoT terminal array and terminal, respectively performs signal acquisition and image acquisition on the cooking parameters and the appearance of the target dish during the cooking process to obtain the corresponding dish cooking information; the dish cooking information includes the cooking sensor signal and the dish cooking image; Data processing module, used to decompose the cooking sensor signal into Components and perform delayed reconstruction, and obtain the delayed reconstruction The self-clustering coefficient corresponding to the component; and based on it, judge the corresponding Whether the random error of the component meets the requirements, if not, the corresponding The components are adaptively enhanced and reconstructed to obtain corresponding enhanced cooking signals, and the corresponding cooking images of the dishes are simultaneously enhanced to obtain corresponding enhanced cooking images; The dish evaluation module identifies the cooking parameters and dish appearance of the target dish during the cooking process based on the preset cooking standards and in combination with the obtained enhanced cooking signals and enhanced cooking images, and obtains the corresponding quality evaluation results; The feedback optimization module optimizes the cooking parameters of the target dish during the cooking process based on the quality evaluation results obtained, and gives cooking optimization suggestions.

2. The intelligent cooking quality evaluation and feedback optimization system according to claim 1 is characterized in that: The process of obtaining the corresponding dish cooking information includes: The data acquisition module is provided with an acquisition node and an image terminal; the acquisition node is used to collect data on the cooking process of the target dish and obtain the corresponding cooking sensor signal; The image terminal is used to collect images of the appearance of the dish during the cooking process to obtain corresponding cooking images of the dish; The collected dish cooking images and cooking sensor signals are summarized to obtain corresponding dish cooking information.

3. The intelligent cooking quality evaluation and feedback optimization system according to claim 2 is characterized in that: The corresponding acquisition process of enhanced cooking signals and enhanced cooking images includes: Perform empirical mode decomposition on the corresponding cooking sensor signal to obtain the corresponding residual signal and several Quantity; For the obtained The components are reconstructed with delay to obtain the corresponding delay components, and the corresponding regression coefficients are constructed based on them; the regression coefficients between all data points in the corresponding delay components are obtained, and they are mapped into a pre-constructed blank matrix to obtain the corresponding regression matrix; the matrix elements in the corresponding regression matrix are mapped into the pre-constructed complex network to obtain the corresponding network subgraph; Perform image analysis on the obtained network subgraph to obtain the self-clustering coefficient corresponding to the corresponding network subgraph; Set the clustering threshold, compare the obtained self-clustering coefficient with the clustering threshold, if the self-clustering coefficient is not less than the clustering threshold, The components are decomposed by wavelet transform to obtain high-frequency coefficients and low-frequency coefficients at different decomposition layers; Adaptive threshold processing is performed on the high-frequency coefficients in each decomposition layer, and the high-frequency coefficients and low-frequency coefficients after threshold processing are reconstructed to obtain the corresponding enhancement Quantity; The enhanced Quantity and random error meet the requirements The components and the corresponding residual components are recombined to obtain the corresponding enhanced cooking signal; The dish cooking image in the corresponding dish cooking information is processed to obtain a corresponding enhanced cooking image, which is then combined with the corresponding enhanced cooking signal to obtain corresponding enhanced cooking information.

4. The intelligent cooking quality evaluation and feedback optimization system according to claim 3 is characterized in that: The self-clustering coefficient ; In the formula, is the total number of nodes in the network subgraph; Indicates The local clustering coefficient corresponding to the nodes; ; In the formula, Representation and network nodes The total number of network nodes with visible edges between them; , and Represents network nodes , and The regression coefficient between Representation and point Index of the number of network nodes with visible edges between them; The formula for adaptive threshold processing is: ; Indicates The high frequency coefficients within the layer decomposition layer, represents symbolic function operation, Indicates a preset high frequency coefficient threshold; represents the high frequency coefficient after threshold processing, represents the adjustment coefficient; Represented by natural numbers The exponential function operation with base is.

5. The intelligent cooking quality evaluation and feedback optimization system according to claim 3 is characterized in that: The process of obtaining the corresponding quality assessment results includes: Inputting the obtained enhanced cooking information into a pre-built dish quality assessment model to obtain a corresponding model output result, and based on the model output result, obtaining a dish cooking result corresponding to the corresponding dish cooking process; Comparing the obtained dish cooking results with the pre-set cooking standards, and marking the cooking standards in the corresponding dish cooking results as normal standards and low-quality standards according to the comparison results; Based on the low quality standard and normal standard marked in the corresponding dish cooking result, the corresponding dish cooking result is assigned a score to generate a corresponding quality assessment result.

6. The intelligent cooking quality evaluation and feedback optimization system according to claim 5, characterized in that: The construction process of the dish quality assessment model includes: The backbone network of the dish quality assessment model is an improved convolutional neural network, and the basic framework of the improved convolutional neural network is an input layer, a feature layer, a fusion layer and an output layer; The input layer is used to preprocess the received data to obtain corresponding input data; The feature layer includes improved Coding modules and improvements Modules; The improvements The encoding module is composed of a convolution unit and a forward propagation unit. The convolution unit is used to perform an encoding operation on the cooking sensor signal in the input data to obtain a corresponding encoding sequence; perform feature extraction on the corresponding encoding sequence to obtain corresponding local features and mark them, and input the marked local features into the forward propagation unit to obtain a corresponding output feature vector; The improvements The module is implemented through a The convolution kernel performs convolution processing on the input cooking image of the dish to obtain the corresponding convolution feature map; the convolution feature map is downsampled through a maximum pooling layer to obtain the corresponding spatial feature map, which is then passed through a The convolution kernel and a global average pooling layer perform convolution operations and spatial dimension reduction on the obtained spatial feature map to obtain an output feature map; The fusion layer is used to receive the obtained output feature vector and output feature map, and perform adaptive weighted fusion on them to obtain corresponding output feature information; The output layer is used to receive output feature information and based on the fully connected layer and Function, maps the corresponding output feature information into the pre-constructed space and classifies the results; obtains the corresponding dish cooking results; Construct a training data set and define As the optimizer continuously optimizes the parameters of the dish quality assessment model during the training process, the dish quality assessment model is iteratively trained based on the training data, and the value of the corresponding loss function is recorded. When the value of the batch loss function no longer decreases, the parameters of the dish quality assessment model at this time are saved, and the training of the dish quality assessment model is completed; is a constant.

7. The intelligent cooking quality evaluation and feedback optimization system according to claim 6, characterized in that: The process of labeling includes: The constructed coding sequence was divided equally to obtain coded segments, and based on three The convolution kernel performs convolution operation on the corresponding encoding fragment to generate the corresponding request matrix , primary key matrix and the numerical matrix ; Based on it, the feature score corresponding to the corresponding coding segment is obtained ; In the formula, Represents matrix transpose; The vector dimension representing the primary key matrix; Represents a mask operation; A score threshold is set. If the feature score is greater than the score threshold, the local features corresponding to the corresponding local fragments are marked.

8. The intelligent cooking quality evaluation and feedback optimization system according to claim 6, characterized in that: The formula for convolution processing of the input dish cooking image is: ; In the formula, Refers to the convolution feature map; and Respectively represent the horizontal and vertical coordinate indexes of the pixel points in the corresponding convolution feature map; Represents the layer connected to the input The number index of convolution kernels; and Respectively indicate the corresponding The offset of the spatial horizontal and vertical positions of the convolution kernel on the input dish cooking image and ; represents the input channel index, is the total number of input channels, represents the weight matrix; Indicates The pixel coordinates of the input dish cooking image in the input channel are The pixel value of represents the bias term; The formula for downsampling operation is: ; In the formula, Represents the spatial feature map obtained after the downsampling operation is completed. and Respectively represent the indexes of the horizontal and vertical coordinates of the pixel points in the spatial feature map; Represents the first layer connected to the maximum pooling layer indivual The convolution feature map output by the convolution kernel, and Respectively represent the indexes of the horizontal and vertical coordinates of the pixels in the convolution feature map; Indicates finding the vertical spatial position of the corresponding convolution feature map The maximum value of pixels; Indicates finding the horizontal spatial position of the corresponding convolution feature map The maximum value of pixels; and Respectively represent the horizontal and vertical strides of the pooling window in the maximum pooling layer; and Respectively represent the height and width of the pooling window in the maximum pooling layer, Indicates the connection with the maximum pooling layer Index of the number of convolution kernels.

9. The intelligent cooking quality evaluation and feedback optimization system according to claim 6, characterized in that: The process of optimizing the cooking parameters of the target dish during cooking based on the obtained quality assessment results includes: Obtaining cooking parameters corresponding to the low-quality standard based on the quality assessment result, and comparing the cooking parameters with the pre-set parameter standard to obtain corresponding comparison results, and generating corresponding cooking optimization suggestions based on the comparison results; The corresponding quality assessment results and cooking optimization suggestions are fed back to the user in an intuitive manner and displayed through the display screen of the cooking equipment or the mobile phone application; based on user feedback and historical cooking data, the cooking parameter settings during the cooking process of the dish are automatically optimized.

10. An intelligent cooking quality assessment and feedback optimization method, which is based on the intelligent cooking quality assessment and feedback optimization system according to any one of claims 1 to 9, characterized in that: include: Step 1: Based on the pre-built IoT terminal array and terminal, respectively, the cooking parameters and the appearance of the target dish during the cooking process are collected by signal and image, and the corresponding dish cooking information is obtained; the dish cooking information includes the cooking sensor signal and the dish cooking image; Step 2: Decompose the cooking sensor signal into Components and perform delayed reconstruction, and obtain the delayed reconstruction The self-clustering coefficient corresponding to the component; and based on it, judge the corresponding Whether the random error of the component meets the requirements, if not, the corresponding The components are adaptively enhanced and reconstructed to obtain corresponding enhanced cooking signals, and the corresponding cooking images of the dishes are simultaneously enhanced to obtain corresponding enhanced cooking images; Step 3: Based on the preset cooking standard and in combination with the obtained enhanced cooking signal and enhanced cooking image, the cooking parameters and the appearance of the target dish during the cooking process are evaluated to obtain the corresponding quality evaluation result; Step 4: Based on the quality assessment results obtained, the cooking parameters of the target dish are optimized during the cooking process, and cooking optimization suggestions are given.

Images