Temperature control system of electric heating rice steaming cabinet

Abstract

The invention discloses a temperature control system of an electric heating rice steaming cabinet, which not only can acquire and analyze key characteristic data in a heating process in real time, but also can dynamically adjust PID parameters according to the data by introducing a parameter self-adaption and staged self-adaption control mechanism. The flexible and accurate regulation and control of the heating process of the rice steaming cabinet under different working conditions are realized. Specifically, the system carries out deep analysis on data acquired in an initial heating stage by utilizing time sequence feature coding and a machine learning model, so that an optimal target temperature curve and a PID parameter set corresponding to each stage are generated, and the most suitable control strategy is automatically switched according to the current stage. By means of the mode, the response capacity of the electric heating rice steaming cabinet to environment disturbance and load sudden change can be effectively improved, and intelligent and self-adaptive temperature management under wider working conditions is achieved while the rice quality and energy consumption optimization are guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control, and more specifically, to a temperature control system of an electric rice steaming cabinet. Background Art

[0002] In the modern catering and food processing industries, electric rice steamers, as efficient and convenient heating and cooking equipment, are widely used to prepare large quantities of rice, snacks, and other foods. Since the rice steaming process requires extremely precise temperature control, as excessively high or low temperatures can affect the rice's taste, degree of cookedness, and nutritional content, developing a scientific, rational, and responsive temperature control solution for electric rice steamers is crucial for improving food quality, reducing energy consumption, and enabling intelligent operation.

[0003] Currently, traditional temperature control systems for electric rice steamers mostly use a fixed-parameter PID (proportional-integral-differential) control method. This involves presetting a set of PID parameters and adjusting the heating element power based on the error between the real-time temperature value and the target temperature. While this approach can achieve basic automatic temperature control, due to the frequent changes in load (such as rice volume and water volume) and environmental conditions during actual use, fixed parameters are difficult to adapt to the changing dynamic characteristics of the system under different operating conditions. For example, during the initial heating phase and the heat preservation phase, the system has different requirements for heating rate and temperature control accuracy, and a single PID parameter cannot meet the performance requirements of each stage. Furthermore, because traditional methods lack the ability to adaptively identify load characteristics, they are prone to overshoot, oscillation, or temperature control lag when faced with complex or sudden operating conditions, affecting steaming quality and energy efficiency.

[0004] Therefore, an optimized temperature control system of the electric rice steaming cabinet is expected. Summary of the Invention

[0005] In order to solve the above-mentioned technical problems, the present application is proposed. The embodiment of the present application provides a temperature control system for an electric rice steaming cabinet, which, by introducing parameter adaptation and staged adaptive control mechanisms, can not only collect and analyze key characteristic data of the heating process in real time, but also dynamically adjust the PID parameters based on these data, thereby realizing flexible and precise control of the heating process of the rice steaming cabinet under different working conditions. Specifically, the system uses time series feature coding and machine learning models to conduct an in-depth analysis of the data collected in the initial heating stage, thereby generating an optimal target temperature curve and a set of PID parameters corresponding to each stage, and automatically switching to the most appropriate control strategy according to the current stage. In this way, the response capability of the electric rice steaming cabinet to environmental disturbances and sudden changes in load can be effectively improved, while ensuring the quality of rice and optimizing energy consumption, and realizing intelligent and adaptive temperature management under a wider range of working conditions.

[0006] According to one aspect of the present application, a temperature control system for an electric rice steaming cabinet is provided, comprising: A final target temperature acquisition module is used to obtain the final target temperature set by the user; a temperature control parameter dynamic optimization module, configured to calculate an estimated load characteristic based on a temperature time series of the initial heating stage in response to detecting that the electric rice steamer enters the initial heating stage, and to obtain an optimized target temperature curve and PID parameter sets for different control stages based on the final target temperature and the estimated load characteristic; An adaptive control starting module is used to obtain a current time stamp and a real-time heating temperature value collected by a temperature sensor in response to detecting that the electric rice steaming cabinet enters an adaptive PID control stage; An adaptive control parameter determination module is used to determine the current target temperature and the PID parameters of the control stage corresponding to the current time based on the current time stamp and the optimized target temperature curve; The temperature control module is used to input the temperature error between the real-time heating temperature value and the current target temperature into a PID controller to obtain a heating tube driving signal. The PID controller has PID parameters corresponding to the control stage.

[0007] Compared with the existing technology, the temperature control system of an electric rice steaming cabinet provided by the present application, by introducing parameter adaptation and staged adaptive control mechanisms, can not only collect and analyze key characteristic data of the heating process in real time, but also dynamically adjust the PID parameters based on these data, thereby realizing flexible and precise control of the heating process of the rice steaming cabinet under different working conditions. Specifically, the system uses time series feature coding and machine learning models to conduct an in-depth analysis of the data collected in the initial heating stage, thereby generating the optimal target temperature curve and the PID parameter set corresponding to each stage, and automatically switching to the most appropriate control strategy according to the current stage. In this way, the response ability of the electric rice steaming cabinet to environmental disturbances and sudden changes in load can be effectively improved, while ensuring the quality of rice and energy consumption optimization, and realizing intelligent and adaptive temperature management under a wider range of working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 1 is a block diagram of a temperature control system of an electric rice steaming cabinet according to an embodiment of the present application; Figure 2Schematic diagram of data flow of the temperature control system of the electric rice steaming cabinet according to an embodiment of the present application; Figure 3 This is a block diagram of a temperature control parameter dynamic optimization module in a temperature control system of an electric rice steaming cabinet according to an embodiment of the present application. DETAILED DESCRIPTION

[0010] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0011] As used in this application and the claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not intended to refer to the singular but may include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0012] Although the present application makes various references to certain modules in the system according to embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are illustrative only, and different aspects of the system and method can use different modules.

[0013] Flowcharts are used in this application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the various steps may be processed in reverse order or simultaneously, as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0014] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0015] In the technical solution of the present application, a temperature control system for an electric rice steaming cabinet is proposed. Figure 1 This is a block diagram of a temperature control system of an electric rice steaming cabinet according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the temperature control system of the electric rice steaming cabinet according to the embodiment of the present application. Figure 1 and Figure 2As shown, the electric rice steaming cabinet temperature control system 300 according to an embodiment of the present application includes: a final target temperature acquisition module 310, which is used to obtain the final target temperature set by the user; a temperature control parameter dynamic optimization module 320, which is used to respond to the detection that the electric rice steaming cabinet enters the initial heating stage, calculate the estimated load characteristics based on the temperature time series of the initial heating stage, and obtain the optimized target temperature curve and the PID parameter set of different control stages based on the final target temperature and the estimated load characteristics; an adaptive control starting module 330, which is used to respond to the detection that the electric rice steaming cabinet enters the adaptive PID control stage, obtain the timestamp of the current moment and the real-time heating temperature value collected by the temperature sensor; an adaptive control parameter determination module 340, which is used to determine the current target temperature and the PID parameters of the corresponding control stage at the current moment based on the timestamp of the current moment and the optimized target temperature curve; a temperature control module 350, which is used to input the temperature error between the real-time heating temperature value and the current target temperature into the PID controller to obtain a heating tube driving signal, and the PID controller has PID parameters corresponding to the control stage.

[0016] In particular, the final target temperature acquisition module 310 is used to obtain the final target temperature set by the user. The user-set final target temperature serves as a key input to the entire temperature control system, defining the temperature standard that must be achieved during the rice steaming process. A clear target temperature makes the entire control process controllable and targeted, helping the system quickly adjust to varying loads and environmental conditions. It optimizes the heating intensity by driving the heating tube signal through temperature error feedback, ultimately improving the response speed and accuracy of temperature control.

[0017] In particular, the temperature control parameter dynamic optimization module 320 is used to calculate the estimated load characteristics based on the temperature time series of the initial heating stage in response to detecting that the electric rice steaming cabinet has entered the initial heating stage, and to obtain the optimized target temperature curve and PID parameter sets for different control stages based on the final target temperature and the estimated load characteristics. In a specific example of the present application, Figure 3 As shown, the temperature control parameter dynamic optimization module 320 includes: a real-time temperature acquisition unit 321 in the initial heating stage, which is used to acquire the real-time temperature value of the initial heating stage through a temperature sensor to obtain the temperature time series of the initial heating stage; a time series feature encoding unit 322, which is used to perform time series feature encoding on the temperature time series of the initial heating stage to obtain an estimated load characteristic encoding vector; and a reference information acquisition unit 323, which is used to input the estimated load characteristic encoding vector and the final target temperature into a pre-trained machine learning model to obtain an optimized target temperature curve and a PID parameter set for different control stages.

[0018] Specifically, the initial heating phase real-time temperature acquisition unit 321 is configured to collect real-time temperature values during the initial heating phase using a temperature sensor to generate a temperature time series during the initial heating phase. It should be understood that temperature changes during the initial heating phase reflect the impact of the current load and environmental conditions on the heating process, and this information is crucial for developing subsequent temperature control strategies. Specifically, when the electric rice steamer is started, user-set load parameters such as the amount of rice and water directly affect the dynamic characteristics of temperature changes during the heating process. For example, when the amount of rice is high, the temperature rises slowly during the initial heating phase and exhibits significant thermal inertia; whereas, when the amount of rice is low, the temperature may rise rapidly, with the risk of overshoot. Traditional fixed-parameter PID control cannot detect these differences, resulting in unstable temperature control. Therefore, the system actively collects temperature time series data during the initial heating phase and captures the dynamic characteristics implicit in the load characteristics through time series pattern analysis. This allows the system to not only achieve flexible and precise adaptive PID control that better matches the current operating conditions but also effectively avoids the overshoot, oscillation, and response lag associated with load and environmental variations in traditional fixed-parameter control.

[0019] Specifically, the time series feature encoding unit 322 is configured to perform time series feature encoding on the temperature time series during the initial heating phase to generate an estimated load characteristic encoding vector. In an embodiment of the present application, the temperature time series during the initial heating phase is first sequence-encoded using a forward LSTM model to generate an initial estimated load characteristic encoding vector. It should be understood that the temperature time series during the initial heating phase contains data on the temperature variation over time when the rice steamer initially begins heating. This temperature variation during this phase directly reflects the impact of the current load (e.g., the amount of rice and water) and environmental conditions on the heating process. Sequence encoding using the forward LSTM model enables deep feature extraction of these complex time series signals, thereby generating a representative and discriminative estimated load characteristic encoding vector. The forward LSTM model is a specially designed recurrent neural network that effectively addresses the vanishing or exploding gradient issues that typically occur with traditional recurrent neural networks when learning long time series data. The generated initial estimated load characteristic encoding vector accurately reflects the load state and its influencing characteristics during the initial heating phase, providing a high-quality data foundation for the input of subsequent machine learning models.

[0020] Furthermore, the initial estimated load characteristic encoding vector is optimized for time series fine-grained load assessment to obtain the estimated load characteristic encoding vector. It should be understood that the initial estimated load characteristic encoding vector encodes the temperature time series data of the initial heating stage through the LSTM model. Although it can initially capture the load characteristics, its representation is still subject to information redundancy and local distortion due to the potential complex correlation between sensor noise, nonlinear heat transfer dynamics and load characteristics. Therefore, in the technical solution of the present application, the initial estimated load characteristic encoding vector is optimized for time series fine-grained load assessment to obtain the estimated load characteristic encoding vector. During this process, eigendecomposition is first used to reveal local load characteristic components (such as thermal inertia and heat absorption rate) implicit in the initial encoding. The manifold topology between these components is then modeled. This allows the system to quantify the geometric dependencies between different load characteristics, such as the coupling between the temperature hysteresis caused by increasing rice volume and the thermal conductivity fluctuations caused by changes in water volume. Next, fine-grained correlation optimization based on gated masks and gradient field correction suppresses anomalous correlation interference caused by environmental noise or sudden load changes, while strengthening stable connections between key local features (such as the strong correlation between the gradual temperature rise under high load and the integral control demand). Finally, through dense feedback distillation and self-attention reconstruction, the optimized local load characteristic components are dynamically aggregated into a globally consistent enhanced encoding vector. This allows the load characteristic representation to not only capture the dynamic evolution of the temporal dimension but also reflect the underlying thermodynamic mechanisms under the coupling of multiple factors. The fine-grained optimized estimated load characteristic encoding vector more accurately reflects the real-time thermodynamic state changes of the rice-water mixture during the rice steaming process, rather than relying solely on the statistical patterns of historical temperature curves. This deep optimization enables subsequent machine learning models to more accurately map temperature control strategies that match the current load characteristics, thereby achieving adaptive control of complex working conditions.

[0021] Specifically, the initial estimated load characteristic encoding vector is first subjected to eigendecomposition based on one-dimensional convolutional coding to obtain a set of local implicit eigenvectors of the initial estimated load characteristic. It should be understood that while the initial estimated load characteristic encoding vector generated by the LSTM model can capture the global temporal patterns of temperature changes in the rice steamer during the initial heating phase, its representation as a single high-dimensional vector cannot explicitly represent the local spatiotemporal coupling of the load's dynamic characteristics. For example, during the rice steaming process, physical processes such as rice grain water absorption and expansion, water vapor phase transition, and heat conduction delay do not unfold uniformly along the time axis. Instead, they manifest as dominant local thermodynamic patterns in different time segments (e.g., rapid heat absorption by rice grains in the initial phase leads to a gentle temperature rise, while steam generation in the middle phase causes increased temperature fluctuations). These local patterns may be smoothed or mixed by the LSTM's global encoding, resulting in the subsequent control strategy being unable to accurately distinguish the core load characteristics of different heating phases. Therefore, in order to decompose the composite load characteristics implied in the initial estimated load characteristic coding vector into local dynamic components with physical interpretability, in the technical solution of the present application, the initial estimated load characteristic coding vector is subjected to feature decomposition based on one-dimensional convolutional coding to obtain a set of local implicit feature vectors of the initial estimated load characteristics.

[0022] Specifically, in the rice steamer control scenario, a one-dimensional convolution kernel, sliding along the time dimension, can capture load response patterns within different time windows in a weighted manner: narrower convolution kernels focus on short-term mutations (such as instantaneous temperature fluctuations caused by the opening and closing of steam valves), while wider convolution kernels perceive long-term trends (such as the suppression of sustained temperature rise caused by water absorption by rice grains). By deploying multiple sets of convolution kernels of different scales in parallel, the system can simultaneously extract local load features at multiple time resolutions during the rice steaming process. This decomposition essentially constitutes a structured modeling of the load's dynamic characteristics, providing fine-grained input primitives for subsequent manifold correlation analysis and feature distillation.

[0023] In a specific example of the present application, the initial estimated load characteristic encoding vector is subjected to eigendecomposition based on one-dimensional convolutional coding using the following formula to obtain a set of local implicit eigenvectors of the initial estimated load characteristic; wherein the formula is: in, is the initial estimated load characteristic encoding vector, are the first, second, and third of the initial estimated load characteristic encoding vectors. and eigenvalues, represents one-dimensional convolutional coding, are the first, second, and third in the set of local implicit feature vectors of the initial estimated load characteristics. and An initial estimated local implicit eigenvector of the load characteristics.

[0024] Next, the manifold structure correlation coefficient between any two local implicit eigenvectors of the initial estimated load characteristics in the set of local implicit eigenvectors is calculated to obtain an initial estimated load characteristic manifold structure correlation topology matrix composed of multiple initial estimated load characteristic manifold structure correlation coefficients. It should be understood that while the local implicit eigenvectors obtained through the one-dimensional convolution decomposition initially decouple the load characteristics of different time segments during the rice steaming process (such as heat absorption by rice grains, steam release, and heat dissipation fluctuations), these local features do not exist in isolation in real physical scenarios. Instead, they form a dynamic correlation network through thermodynamic coupling. For example, the heat absorption efficiency of rice grains during the initial heating phase affects the peak heat load of steam generation during the mid-term, while the heat dissipation rate during the holding phase forms a dynamic balance with the steam release intensity. Traditional methods that independently process single features cannot capture these nonlinear correlations across time scales, resulting in blind spots in the control system's response to complex load evolution. Therefore, in the technical solution of the present application, the manifold structure correlation coefficient between any two initial estimated load characteristic local implicit feature vectors in the set of initial estimated load characteristic local implicit feature vectors is calculated to obtain an initial estimated load characteristic manifold structure correlation topology matrix composed of multiple initial estimated load characteristic manifold structure correlation coefficients.

[0025] That is, by constructing the correlation topology of the manifold structure, the cross-stage action mechanism in the dynamic characteristics of the rice steamer load is revealed. Specifically, in the temperature control scenario, each local eigenvector corresponds to the load response pattern within a different time window (such as short-term temperature fluctuations and long-term temperature rise trends), but these patterns are intertwined in the actual physical process through mechanisms such as energy transfer and phase change delay. By calculating the proximity and correlation strength of the eigenvectors in the low-dimensional manifold space, the system can quantify key correlation rules such as "the negative correlation between steam release intensity and heat dissipation efficiency" and "the inhibitory effect of rice grain heat absorption rate on the subsequent temperature rise curve". This correlation topology is essentially a mathematical abstraction of the multi-physical field coupling relationship in the load dynamic system, which provides structured prior knowledge for subsequent feature distillation.

[0026] In a specific example of the present application, the manifold structure correlation coefficient between any two initial estimated load characteristic local implicit feature vectors in the set of initial estimated load characteristic local implicit feature vectors is calculated using the following formula to obtain an initial estimated load characteristic manifold structure correlation topology matrix composed of multiple initial estimated load characteristic manifold structure correlation coefficients; wherein, the formula is: in, represents the one-norm of a vector, Indicates The exponential operation with base , for The transpose of is the first in the set of local implicit feature vectors of the initial estimated load characteristics The initial estimated load characteristic local implicit eigenvector, is the first position of the initial estimated load characteristic manifold structure associated topology matrix eigenvalues.

[0027] Then, the initial estimated load characteristic manifold structure association topology matrix is input into the gated mask function to obtain the initial estimated load characteristic manifold structure fine-grained association mask topology matrix. It should be understood that although the initial estimated load characteristic manifold structure association topology matrix has preliminarily revealed the potential coupling relationship between different local load characteristics (such as the thermodynamic correlation between rice grain water absorption and steam release), in actual dynamic heating scenarios, some associations may be caused by sensor noise, environmental interference, or non-critical secondary effects; in addition, there are dynamic weight differences in key load associations in different heating stages (for example, the initial stage needs to focus on the correlation between the heat absorption rate and the heating curve, while the insulation stage should focus on thermal inertia balance). The unscreened global association topology may confuse the primary and secondary contradictions. Therefore, in the technical solution of the present application, the initial estimated load characteristic manifold structure association topology matrix is input into the gated mask function to obtain the initial estimated load characteristic manifold structure fine-grained association mask topology matrix.

[0028] Specifically, dynamic focusing and noise stripping of load correlations are achieved through a learnable nonlinear gating function. The correlation topology matrix of the initial estimated load characteristic manifold structure, optimized by the gated mask, not only filters out pseudo-correlation noise caused by sensor noise and transient disturbances but, more importantly, highlights the dominant physical laws in the dynamic evolution of the load through a dynamic weight distribution mechanism. This correlation focusing capability enables the reconstructed load characteristic encoding vector to carry information with greater decision-making value, thereby driving the PID parameter generation module to output a control parameter set that strictly matches the current physical state. This effectively avoids parameter misadjustments caused by correlation noise in traditional methods (such as temperature oscillations caused by overcompensation), ultimately achieving smooth temperature tracking and energy consumption optimization for the cooking process under complex working conditions.

[0029] In a specific example of the present application, the initial estimated load characteristic manifold structure association topology matrix is input into the gated mask function using the following formula to obtain the initial estimated load characteristic manifold structure fine-grained association mask topology matrix; wherein, the formula is: in, express function, is the gated mask weight matrix, is the gated mask bias matrix, is the associated topological matrix of the initial estimated load characteristic manifold structure, The fine-grained association mask topology matrix of the manifold structure of the initial estimated load characteristics.

[0030] Furthermore, each eigenvalue in the initial estimated load characteristic manifold structure fine-grained correlation mask topology matrix undergoes geometric correlation optimization based on gradient field tuning to obtain the estimated load characteristic manifold structure fine-grained correlation mask topology matrix. It should be understood that while the generation of the initial estimated load characteristic manifold structure fine-grained correlation mask topology matrix screens the correlation relationships between features through a gating mask, the eigenvalues in the matrix may exhibit local geometric correlation structure inhomogeneity due to the coupling of nonlinear dynamics of the load characteristics (such as differences in heat capacity of rice-water mixtures and fluctuations in heat transfer rates) with environmental noise. For example, a slow temperature rise under high load and a rapid response under low load may result in different correlation strength distributions. The polarization effect of the gating mask may over-enhance certain local correlations, resulting in discretization or distortion of the underlying manifold structure implied in the topology matrix. This nonlinear unsaturation phenomenon compresses the stable solution of the geometric correlation field, making the eigenvalues unable to accurately represent the true coupling relationships of the load characteristics. Therefore, in a preferred example of the present application, each eigenvalue in the initial estimated load characteristic manifold structure fine-grained association mask topology matrix is subjected to estimated load characteristic manifold structure geometric association optimization based on gradient field tuning to obtain the estimated load characteristic manifold structure fine-grained association mask topology matrix.

[0031] Specifically, the introduction of gradient field terms can quantify the nonlinear dependencies between eigenvalues and dynamically adjust each eigenvalue through external field driving terms. For example, when the temperature response curve fluctuates abnormally due to a sudden change in the meter, gradient field tuning can suppress the sudden noise of the local correlation strength. At the same time, mean field tuning balances the mean response of the overall correlation distribution, allowing the topological matrix to maintain the smoothness and consistency of the geometric correlation under the influence of high-order nonlinearities. In this way, the fine-grained correlation mask topological matrix of the initial estimated load characteristic manifold structure can more accurately characterize the potential thermodynamic correlations between load characteristics (such as thermal inertia and the dynamic coupling of energy absorption rate), thereby providing stable structural guidance for subsequent feature-intensive feedback distillation. Ultimately, the optimized fine-grained correlation mask topological matrix of the initial estimated load characteristic manifold structure enhances the robustness of the correlation structure, enabling the system to adaptively identify the deep impact of load changes on temperature control, thereby realizing intelligent temperature regulation under complex working conditions.

[0032] Here, the geometric correlation distribution of the initial estimated load characteristic manifold structure correlation topology matrix on the potential low-dimensional manifold structure will be nonlinear and unsaturated, so that the overall correlation topology distribution of the initial estimated load characteristic manifold structure correlation topology matrix will be compressed due to nonlinear dependence, and this will become more significant due to the correlation polarization enhancement of the gated mask function, affecting the substantial geometric fine-grained correlation structure expression effect of the fine-grained correlation mask topology matrix of the initial estimated load characteristic manifold structure.

[0033] Based on this, for the initial estimated load characteristic manifold structure fine-grained association mask topology matrix Each eigenvalue of , first introduce the gradient field term:

[0034] It is used to correct local nonlinear geometric structure inhomogeneities and thus achieve fine-grained smoothing of geometrically correlated structural fields.

[0035] Then, the gradient field term is used as the external field driving term to perform mean field tuning for each eigenvalue:

[0036] in, is the initial estimated load characteristic manifold structure fine-grained association mask topology matrix The eigenmean of all eigenvalues of , and It is to modulate the scaling weight of the excessive external field drive term. The estimated load characteristic manifold structure is a fine-grained association mask topology matrix of each eigenvalue.

[0037] In this way, under the action of the external field driving term as a high-order gradient, the nonlinear saturation of the geometric correlation distribution under the mean field is reversely promoted (i.e., the strong gradient responsiveness decreases), thereby compensating for the discretization of the sub-geometric correlation structure caused by the correlation polarization enhancement through the mean harmonic response under the mean field, thereby improving the substantial geometric fine-grained correlation structure expression effect of the fine-grained correlation mask topology matrix of the initial estimated load characteristic manifold structure.

[0038] Furthermore, based on the estimated load characteristic manifold structure fine-grained associative mask topology matrix, feature-intensive feedback distillation is performed on each initial estimated load characteristic local implicit feature vector in the set of initial estimated load characteristic local implicit feature vectors to obtain a set of initial estimated load characteristic local implicit distilled feature vectors. It should be understood that although the initial estimated load characteristic local implicit feature vectors preliminarily extract the dynamic characteristics of the load (such as thermal inertia and energy absorption rate) through feature decomposition and associative topology optimization, due to the nonlinear coupling between the load characteristics and the environmental noise (such as heat transfer delay caused by rice volume fluctuations and temperature response curve distortion caused by water volume changes), a single local feature vector may still have information redundancy or local perspective deviation. If these local features are directly used for PID parameter mapping, it may lead to insufficient adaptability of the control strategy to complex working conditions. Therefore, in the technical solution of the present application, based on the estimated load characteristic manifold structure fine-grained associative mask topology matrix, feature-intensive feedback distillation is performed on each initial estimated load characteristic local implicit feature vector in the set of initial estimated load characteristic local implicit feature vectors to obtain a set of estimated load characteristic local implicit distilled feature vectors.

[0039] Specifically, the correlation coefficients in the mask topology matrix define the strength of interactions between different local features (such as the correlation between the temperature rise rate and integral control requirements, or the sensitivity of thermal inertia to differential parameters). The system then uses weighted aggregation to inject information from strongly correlated local features into the current vector. For example, when a local feature (such as a slow temperature rise) is detected to be highly correlated with another feature (such as a sudden change in heat capacity), the distillation process fuses the implicit patterns of these two features to generate enhanced features that reflect both temperature trends and thermodynamic coupling. This dense feedback mechanism eliminates the isolation of each local feature vector, instead forming a coordinated representation through a global correlation network. This allows for a more accurate representation of the multi-dimensional dynamics of load characteristics (such as the nonlinear relationship between metering changes and thermal response). Ultimately, the distilled feature set eliminates noise interference and strengthens the consistency of key patterns, providing high-information-density input for subsequent self-attention reconstruction, enabling the machine learning model to generate temperature control strategies tailored to complex operating conditions.

[0040] In a specific example of the present application, based on the fine-grained associative mask topology matrix of the estimated load characteristic manifold structure, each initial estimated load characteristic local implicit feature vector in the set of initial estimated load characteristic local implicit feature vectors is subjected to feature-intensive feedback distillation to obtain a set of initial estimated load characteristic local implicit distilled feature vectors; wherein, the formula is: in, Indicates point multiplication by position, represents matrix multiplication, express function, is the distillation weight matrix, express The scale, for The corresponding initial estimated load characteristic local implicit distillation feature vector, is the set of local implicit distillation feature vectors of the initial estimated load characteristics, are the first, second, and third in the set of local implicit distillation feature vectors of the initial estimated load characteristics. and The initial estimated load characteristics are locally implicitly distilled feature vectors.

[0041] Subsequently, a self-attention mechanism-based feature reconstruction is performed on the set of local implicit distilled feature vectors of the initial estimated load characteristics to obtain an estimated load characteristic encoding vector. It should be understood that although the local implicit distilled feature vectors of the estimated load characteristics integrate the correlation information between local load characteristics (such as the coupled relationship between thermal inertia, heat absorption rate, and ambient heat dissipation) through feature-intensive feedback distillation, due to the nonlinear complexity of load dynamics (such as the heat transfer delay caused by sudden changes in rice volume and the nonlinear distortion of the temperature response curve caused by water volume fluctuations), the distilled local feature set may still exhibit information dispersion. Therefore, to capture the long-range dependencies between features and integrate the scattered local features into a unified high-order load characteristic representation, the technical solution of this application performs a self-attention mechanism-based feature reconstruction on the set of local implicit distilled feature vectors of the estimated load characteristics to obtain an estimated load characteristic encoding vector. Specifically, the self-attention mechanism automatically identifies key local patterns (such as the thermal inertia-dominant characteristics of the initial heating phase and the temperature stability requirements of the holding phase) by calculating the similarity weights between feature vectors and assigns them higher attention weights. Ultimately, this mechanism enables the system to dynamically optimize control logic based on global load characteristics, achieving a balance between precise temperature regulation and energy efficiency under complex working conditions.

[0042] In a specific example of the present application, the set of local implicit distillation feature vectors of the initial estimated load characteristics is subjected to feature reconstruction based on the self-attention mechanism to obtain an estimated load characteristic encoding vector using the following formula; wherein, the formula is:

[0043] in, 、 and are the learnable weight matrices, is the query feature vector, is the key vector, is a value vector, represents the length of the set of local implicit distillation feature vectors of the initial estimated load characteristics, Represents feature reconstruction of the set of local implicit distillation feature vectors of the initial estimated load characteristics, express function, A vector is encoded for the estimated load characteristic.

[0044] Specifically, the reference information acquisition unit 323 is used to input the estimated load characteristic encoding vector and the final target temperature into a pre-trained machine learning model to obtain an optimized target temperature curve and PID parameter sets for different control stages. The control stages include a rapid heating stage, a precise temperature control stage, and a heat preservation stage. The pre-trained machine learning model is a deep neural network model. Specifically, the estimated load characteristic encoding vector and the final target temperature are input into a deep neural network model trained on a large amount of historical data and multiple operating conditions. This model predicts and optimizes the optimal curve for the heating process and the staged adaptive PID parameters by learning the complex nonlinear mapping relationship between load characteristics and the optimal temperature control strategy. The resulting optimized target temperature curve can more accurately match actual load and environmental changes, and the PID parameters are dynamically adjusted for each control stage (such as initial heating, heating, heat preservation, etc.), avoiding the overshoot, oscillation, and control hysteresis problems associated with traditional fixed parameters. In this way, not only the high responsiveness and accuracy of temperature control are ensured, but also the stability and energy-saving efficiency of the system are improved through phased adaptive control. Ultimately, the electric rice steaming cabinet can achieve intelligent and flexible temperature control under complex working conditions, effectively improving the quality of steamed rice and operating efficiency, and meeting the modern catering industry's demand for high-performance intelligent steaming equipment.

[0045] Specifically, the adaptive control initiation module 330 is configured to, in response to detecting that the electric rice steamer has entered the adaptive PID control phase, obtain the current timestamp and the real-time heating temperature value collected by the temperature sensor. It should be understood that the adaptive PID control phase signifies that the system has completed its initial load characteristic assessment and entered the specific temperature adjustment process. During this phase, accurate timestamp information allows the system to locate the specific moment in the current control process, ensuring precise time-dependent control based on the optimized target temperature curve. The collected real-time temperature value reflects the actual heating status of the rice steamer and serves as the basis for determining temperature errors and adjusting heating intensity. In specific implementations, the system continuously monitors real-time temperature data using temperature sensors and promptly records the corresponding timestamps. This data is input into the control system for comparison with the target temperature to be achieved at the current moment. Combined with the timestamp, the control phase (e.g., heating, insulation, or cooling) is determined. In this way, the system can extract the most appropriate control parameters from the corresponding PID parameter sets for different phases, enabling adaptive switching and dynamic adjustment, thereby improving the response speed and accuracy of temperature control.

[0046] In particular, the adaptive control parameter determination module 340 is configured to determine the current target temperature and the PID parameters for the corresponding control stage based on the current timestamp and the optimized target temperature curve. Specifically, in the embodiment of the present application, the current target temperature and the control stage are first determined based on the current timestamp and the optimized target temperature curve. It should be understood that the heating process of an electric rice steamer is typically divided into multiple stages, such as initial heating, temperature rise, and heat preservation, and each stage has different temperature control targets and parameter requirements. By combining the current timestamp, the system can accurately locate the current time point and, based on the pre-generated optimized target temperature curve, extract the specific target temperature value corresponding to that moment. This temporal correlation ensures that the temperature control strategy accurately responds to the real-time progress of the rice steaming process, rather than relying on static settings. In specific implementations, the system receives real-time time data from the temperature sensor and searches the optimized curve for the corresponding temperature target based on time. It also identifies the current control stage based on time intervals. This process also supports dynamic stage switching, allowing the controller to automatically adjust parameters to adapt to stage characteristics. In this way, an accurate reference benchmark can be provided for subsequent PID control, ensuring that the heating process can closely follow the optimal temperature trajectory and effectively avoid overheating or underheating.

[0047] Furthermore, based on the control stage at the current moment, the PID parameters of the corresponding control stage are extracted from the PID parameter sets of different control stages. It should be understood that the rice steaming process has obvious stage characteristics, such as the initial heating, temperature rise and insulation stages. Each stage has different requirements for temperature control, and the heating speed, stability and response sensitivity involved are different. Traditional fixed PID parameters are difficult to meet the control requirements of these variable stages at the same time. This application calculates and stores the PID parameters optimized for each stage in advance to ensure that the parameter set most suitable for the current stage can be called in a targeted manner during actual operation. In the specific implementation process, the system first determines the control stage by timestamp and optimized target temperature curve, and then extracts the proportional, integral and differential parameters corresponding to the stage from the multi-stage PID parameter set output by the machine learning model, and configures them to the PID controller for real-time adjustment of the driving signal of the heating tube. In this way, the response speed and accuracy of the temperature control process can be improved, overshoot, oscillation or hysteresis caused by parameter mismatch can be prevented, and the temperature change curve can be smooth and stable.

[0048] In particular, the temperature control module 350 is configured to input the temperature error between the real-time heating temperature value and the current target temperature into a PID controller to generate a heating tube drive signal. The PID controller has PID parameters corresponding to the control stage. Specifically, in the embodiment of the present application, the temperature error between the real-time heating temperature value and the current target temperature is first calculated. It should be understood that the temperature error reflects the deviation between the actual heating state of the device and the ideal temperature control target, and is an important basis for the control system to determine whether the current heating effect is achieving the expected effect and whether adjustments are necessary. Specifically, the system uses a temperature sensor to collect the current temperature data of the rice steamer in real time. Combined with the optimized target temperature curve, the target temperature corresponding to the current time point is located using a timestamp. The difference between the two constitutes the temperature error. The purpose of obtaining this temperature error is to provide accurate feedback to the subsequent PID controller, enabling the controller to calculate the corresponding heating tube drive instructions based on the error magnitude and changing trend, thereby achieving dynamic adjustment of the heating power. Using the temperature error, the system can quickly detect temperature deviations and take compensatory measures, effectively avoiding food quality degradation or energy waste due to overheating or underheating. During implementation, the system compares real-time temperature data with the target curve's corresponding temperature value, updating the error in real time to ensure the continuity and timeliness of the temperature control response. This allows the electric rice steamer to dynamically adjust heating intensity based on the error, achieving rapid and stable temperature increases and decreases, ensuring that rice and pastries meet established quality standards throughout the entire cooking process while significantly improving energy efficiency and the system's intelligence.

[0049] Furthermore, the temperature error is input into a PID controller to generate a drive signal for the heating element. The PID controller has PID parameters for the corresponding control stage. It should be understood that the temperature error directly reflects the deviation between the device's current heating temperature and the target temperature and is an important indicator for determining whether the system is achieving the desired temperature control effect. The PID controller comprehensively processes the temperature error through three steps: proportional (P), integral (I), and differential (D). The proportional step adjusts the heating intensity based on the error magnitude; the integral step accumulates past errors to eliminate steady-state deviations; and the differential step predicts the error trend to suppress system overshoot and oscillation. Combined with PID parameters optimized for different control stages (such as initial heating, ramping, and holding), the controller can flexibly adjust the heating signal based on the dynamic characteristics of each stage, ensuring that temperature changes conform to the expected curve and avoiding control failure caused by a single parameter in multi-stage applications. In specific implementation, the system first extracts applicable parameters from a pre-generated PID parameter set based on the current control stage. The real-time temperature error is then input into the PID control algorithm, which calculates and outputs a drive signal for the heating element. This drive signal adjusts the power of the heating element, causing the actual temperature to quickly and stably approach the target temperature. This process achieves closed-loop control from temperature deviation sensing to drive execution, ensuring timely temperature response and stable control. Through phased adaptive PID parameter configuration, the system's adaptability to load changes and environmental disturbances is improved, effectively avoiding the overshoot, oscillation, and hysteresis common in traditional fixed-parameter control.

[0050] As described above, the electric rice steaming cabinet temperature control system 300 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server equipped with an electric rice steaming cabinet temperature control algorithm. In one possible implementation, the electric rice steaming cabinet temperature control system 300 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or hardware module. For example, the electric rice steaming cabinet temperature control system 300 can be a software module in the operating system of the wireless terminal, or it can be an application developed specifically for the wireless terminal. Of course, the electric rice steaming cabinet temperature control system 300 can also be one of the many hardware modules of the wireless terminal.

[0051] Alternatively, in another example, the electric rice steaming cabinet temperature control system 300 and the wireless terminal may also be separate devices, and the electric rice steaming cabinet temperature control system 300 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0052] While various embodiments of the present disclosure have been described above, the above descriptions are illustrative, non-exhaustive, and not intended to be limiting of the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to existing technologies, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A temperature control system for an electric rice steaming cabinet, characterized in that: include: A final target temperature acquisition module is used to obtain the final target temperature set by the user; a temperature control parameter dynamic optimization module, configured to calculate an estimated load characteristic based on a temperature time series of the initial heating stage in response to detecting that the electric rice steamer enters the initial heating stage, and to obtain an optimized target temperature curve and PID parameter sets for different control stages based on the final target temperature and the estimated load characteristic; An adaptive control starting module is used to obtain a current time stamp and a real-time heating temperature value collected by a temperature sensor in response to detecting that the electric rice steaming cabinet enters an adaptive PID control stage; An adaptive control parameter determination module is used to determine the current target temperature and the PID parameters of the control stage corresponding to the current time based on the current time stamp and the optimized target temperature curve; The temperature control module is used to input the temperature error between the real-time heating temperature value and the current target temperature into a PID controller to obtain a heating tube driving signal. The PID controller has PID parameters corresponding to the control stage.

2. The temperature control system of the electric rice steaming cabinet according to claim 1, characterized in that: Temperature control parameter dynamic optimization module, including: The real-time temperature acquisition unit in the initial heating stage is used to acquire the real-time temperature value in the initial heating stage through the temperature sensor to obtain the temperature time series in the initial heating stage; A time series feature encoding unit, configured to perform time series feature encoding on the temperature time series during the initial heating phase to obtain an estimated load characteristic encoding vector; The reference information acquisition unit is used to input the estimated load characteristic encoding vector and the final target temperature into the pre-trained machine learning model to obtain the optimized target temperature curve and the PID parameter set for different control stages.

3. The temperature control system of the electric rice steaming cabinet according to claim 2, characterized in that: Temporal feature encoding unit, including: The sequence encoding subunit is used to perform sequence encoding on the temperature time series in the initial heating stage based on the forward LSTM model to obtain the initial estimated load characteristic encoding vector; The temporal fine-grained load evaluation optimization subunit is used to perform temporal fine-grained load evaluation optimization on the initial estimated load characteristic coding vector to obtain the estimated load characteristic coding vector.

4. The temperature control system of the electric rice steaming cabinet according to claim 3, characterized in that: The timing fine-grained load evaluation optimization subunit includes: a secondary eigendecomposition subunit, configured to perform eigendecomposition based on one-dimensional convolution coding on the initial estimated load characteristic coding vector to obtain a set of local implicit eigenvectors of the initial estimated load characteristic; The secondary sub-unit of temporal feature enhancement is used to perform temporal feature enhancement of the estimated load characteristics on the set of initial estimated load characteristic local implicit feature vectors based on the manifold structure association relationship between any two initial estimated load characteristic local implicit feature vectors in the set of initial estimated load characteristic local implicit feature vectors to obtain an estimated load characteristic encoding vector.

5. The temperature control system of the electric rice steaming cabinet according to claim 4, characterized in that: The secondary sub-unit for temporal feature enhancement includes: A manifold structure fine-grained association encoding three-level subunit is used to calculate the estimated load characteristic manifold structure fine-grained association mask topology matrix based on the manifold structure association relationship between any two initial estimated load characteristic local implicit feature vectors in the set of initial estimated load characteristic local implicit feature vectors; A feature-intensive feedback distillation three-level sub-unit is used to perform feature-intensive feedback distillation on each of the initial estimated load characteristic local implicit feature vectors in the set of initial estimated load characteristic local implicit feature vectors based on the fine-grained associative mask topology matrix of the estimated load characteristic manifold structure to obtain a set of initial estimated load characteristic local implicit distilled feature vectors; The feature reconstruction three-level sub-unit is used to perform feature reconstruction based on the self-attention mechanism on the set of local implicit distillation feature vectors of the initial estimated load characteristics to obtain the estimated load characteristic encoding vector.

6. The temperature control system of the electric rice steaming cabinet according to claim 5, characterized in that: The manifold structured fine-grained associative encoding three-level subunit is used to: Calculating a manifold structure correlation coefficient between any two local implicit eigenvectors of the initial estimated load characteristic in the set of local implicit eigenvectors of the initial estimated load characteristic to obtain an initial estimated load characteristic manifold structure correlation topology matrix composed of a plurality of manifold structure correlation coefficients of the initial estimated load characteristic; Inputting the initial estimated load characteristic manifold structure correlation topology matrix into the gated mask function to obtain the initial estimated load characteristic manifold structure fine-grained correlation mask topology matrix; The geometric correlation optimization of the estimated load characteristic manifold structure based on gradient field tuning is performed on each eigenvalue in the fine-grained correlation mask topology matrix of the initial estimated load characteristic manifold structure to obtain the fine-grained correlation mask topology matrix of the estimated load characteristic manifold structure.

7. The temperature control system of the electric rice steaming cabinet according to claim 1, characterized in that: The control stage includes a rapid temperature rise stage, a precise temperature control stage and a heat preservation stage.

8. The temperature control system of the electric rice steaming cabinet according to claim 1, characterized in that: Adaptive control parameter determination module, used to: Determine the current target temperature and the current control stage based on the current timestamp and the optimized target temperature curve; Based on the control stage at the current moment, the PID parameters of the corresponding control stage are extracted from the PID parameter sets of different control stages.

9. The temperature control system of the electric rice steaming cabinet according to claim 1, characterized in that: Temperature control modules for: Calculate the temperature error between the real-time heating temperature value and the current target temperature; The temperature error is input into a PID controller to obtain a driving signal for the heating tube, and the PID controller has PID parameters corresponding to a control stage.

Images