Internet of Things Control Method and System

Abstract

The present application discloses an Internet of Things control method and system. It uses a temperature sensor to continuously monitor and collect the temperature data inside the refrigeration thermostat, and introduces data processing and analysis algorithms based on artificial intelligence and deep learning at the backend to perform time series analysis on this temperature data. Then, it uses the contribution degree of the pseudo-anchoring center to filter the local time series features of the temperature inside the refrigeration thermostat. At the same time, it aggregates and infers the local time series information of the temperature inside the thermostat through the spatio-temporal transfer information of the local temperature time series, thereby predicting the real-time temperature data. Subsequently, the predicted real-time temperature data is transmitted to the actuator module, and the actuator module performs corresponding actions based on this real-time temperature data and control signals to achieve intelligent control of the Internet of Things. In this way, it is possible to reduce the influence of irrelevant noise interference in the temperature time series through feature filtering, and at the same time use the spatio-temporal transfer method of local temperature time series semantics to aggregate and infer the time series information of the temperature inside the thermostat, so as to more accurately predict the real-time temperature data inside the thermostat. This intelligent prediction method improves the generalization ability of the system, enabling the Internet of Things control system to better adapt to new situations and execute control actions more timely and effectively.

Description

Technical Field

[0001] This application relates to the field of intelligent control, and more specifically, to an Internet of Things control method and system. Background Art

[0002] A refrigerating constant temperature bath is a device widely used in various research fields such as chemistry, biomedicine, and materials science. It is mainly used to provide a stable temperature environment for various experiments. The automatic control system of the refrigerating constant temperature bath is a system that realizes the automatic regulation and control of the refrigerating constant temperature bath, keeping the temperature in the refrigerating constant temperature bath constant within a predetermined range. However, there are some deficiencies in the traditional control system of the refrigerating constant temperature bath. For example, when collecting temperature data, it is easily affected by noise and other interference factors, resulting in inaccurate temperature measurement. Moreover, the control system cannot promptly detect the change of the actual temperature, leading to temperature fluctuations in the constant temperature bath. In addition, when the temperature changes dynamically, it is difficult for the system to quickly adjust the strategy according to the temperature change, and so on.

[0003] In view of the above technical problems, Chinese Patent CN117784849A proposes an automatic control system for a refrigerating constant temperature bath based on artificial intelligence. It can collect temperature data regularly, convert the residual signal between the predicted temperature and the measured temperature into a time series, perform EMD decomposition, extract the intrinsic mode function components, reduce noise interference, and improve the accuracy and stability of the data. Moreover, through the temperature prediction module, a radial basis function neural network constructed by using a Gaussian kernel function is used for temperature prediction. By adjusting the connection weights to minimize the error between the actual temperature and the predicted value, the accuracy of temperature prediction is improved. Then, the system module optimizes the neural network parameters, optimizes the system performance by selecting actions and adding noise, state transition storage, calculating the loss function and policy gradient, etc., and uses a soft update mechanism to update the target network parameters to enhance the adaptability of the system to temperature changes. In this way, it can adjust decisions in a timely manner according to temperature changes and improve the system stability.

[0004] However, in the above-mentioned automatic control system of the refrigerating thermostat, during the prediction process of temperature data, it uses a Gaussian kernel function to construct a radial basis function neural network for temperature prediction. However, the radial basis function neural network performs poorly in dealing with complex non-linear relationships and it is difficult to obtain accurate temperature prediction values. That is to say, although the radial basis function neural network can handle non-linear relationships to a certain extent, its generalization ability may be limited when dealing with complex and variable temperature data. This means that when facing temperature changes in different environments, the model may not be able to adapt well to new situations, resulting in a decrease in the accuracy of prediction. In addition, the above solution reduces noise interference through EMD decomposition, but this method may not be sufficient to completely eliminate all types of noise. Especially in industrial or laboratory environments, various interference factors may cause large fluctuations in temperature data, thereby affecting the prediction accuracy of temperature and resulting in subsequent control deviations.

[0005] Therefore, an optimized Internet of Things control system is desired. Summary of the Invention

[0006] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide an Internet of Things control method and system. It real-time monitors and collects the temperature data inside the refrigerating thermostat through a temperature sensor, and introduces data processing and analysis algorithms based on artificial intelligence and deep learning at the backend to perform time series analysis on the temperature data. Then, it uses the contribution degree of the pseudo-anchoring center to filter the local time series features of the temperature inside the refrigerating thermostat. At the same time, it aggregates and infers the local time series information of the temperature inside the thermostat through the spatio-temporal transfer information of the temperature local time series, so as to predict the real-time temperature data. Then, the predicted real-time temperature data is transmitted to the actuator module, and the actuator module performs corresponding actions based on the real-time temperature data and the control signal to achieve intelligent control of the Internet of Things. In this way, it can reduce the influence of irrelevant noise interference on the temperature time series through feature filtering, and at the same time use the spatio-temporal transfer method of the temperature local time series semantics to aggregate and infer the time series information of the temperature inside the thermostat, so as to more accurately predict the real-time temperature data inside the thermostat. This intelligent prediction method improves the generalization ability of the system, enables the Internet of Things control system to better adapt to new situations, and thus can execute control actions more timely and effectively.

[0007] According to one aspect of the present application, there is provided an Internet of Things control system, which includes:

[0008] A temperature acquisition and prediction module, configured to obtain the time queue of the temperature data inside the refrigerating thermostat collected by the temperature sensor, and perform time series information transfer aggregation analysis and prediction on the time queue of the temperature data inside the refrigerating thermostat to obtain real-time temperature data;

[0009] A user interface setting module for setting a user interface for mode selection, parameter adjustment, and fault prompting;

[0010] An actuator module for receiving the real-time temperature data and the control signal sent by the user interface, and performing a control action based on the real-time temperature data and the control signal;

[0011] A system monitoring module for remotely monitoring the Internet of Things control system and warning of faults.

[0012] According to another aspect of the present application, there is provided an Internet of Things control method, which includes:

[0013] Obtaining a time queue of the temperature data inside the refrigerated thermostat collected by a temperature sensor, and performing time series information transfer aggregation analysis and prediction on the time queue of the temperature data inside the refrigerated thermostat to obtain real-time temperature data;

[0014] Setting a user interface for mode selection, parameter adjustment, and fault prompting;

[0015] Receiving the real-time temperature data and the control signal sent by the user interface, and performing a control action based on the real-time temperature data and the control signal;

[0016] Remotely monitoring the Internet of Things control system and warning of faults.

[0017] Compared with the prior art, an Internet of Things control method and system provided by the present application monitors and collects the temperature data inside the refrigerated thermostat in real time through a temperature sensor, and introduces data processing and analysis algorithms based on artificial intelligence and deep learning at the backend to perform time series analysis on the temperature data, and then uses the contribution degree of the pseudo-anchoring center to filter the local time series features of the temperature inside the refrigerated thermostat. At the same time, the local time series information of the temperature inside the thermostat is aggregated and inferred through the spatio-temporal transfer information of the local temperature time series, so as to predict the real-time temperature data. Then, the predicted real-time temperature data is transmitted to the actuator module, so that the actuator module performs corresponding actions based on the real-time temperature data and the control signal, realizing intelligent control of the Internet of Things. In this way, the influence of irrelevant noise interference of the temperature time series can be reduced by means of feature filtering, and at the same time, the spatio-temporal transfer method of the local temperature time series semantics is used to aggregate and infer the time series information of the temperature inside the thermostat, so as to more accurately predict the real-time temperature data inside the thermostat. This intelligent prediction method improves the generalization ability of the system, enables the Internet of Things control system to better adapt to new situations, and thus performs control actions more timely and effectively. Description of the Drawings

[0018] The embodiments of the present application will be described in more detail with reference to the accompanying drawings. The above and other objects, features, and advantages of the present application will become more apparent. The drawings are used 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 to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0019] Figure 1 It is a block diagram of an Internet of Things control system according to an embodiment of the present application;

[0020] Figure 2 It is a schematic diagram of data flow of an Internet of Things control system according to an embodiment of the present application;

[0021] Figure 3 It is a block diagram of a temperature acquisition and prediction module in an Internet of Things control system according to an embodiment of the present application;

[0022] Figure 4 It is a block diagram of a temperature local temporal information transfer and aggregation representation unit in an Internet of Things control system according to an embodiment of the present application;

[0023] Figure 5 It is a flowchart of an Internet of Things control method according to an embodiment of the present application. Detailed Embodiments

[0024] Next, exemplary embodiments according to the present application will be described in detail with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0025] As shown in the present application and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0026] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0027] In this application, flowcharts are used to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the operations above or below do not necessarily have to be performed precisely in order. Instead, as needed, various steps can be performed in reverse order or simultaneously. Also, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0028] Next, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the exemplary embodiments described herein.

[0029] In the technical solution of this application, an Internet of Things control system is proposed. Figure 1 It is a block diagram of the Internet of Things control system according to an embodiment of this application. Figure 2 It is a schematic diagram of the data flow of the Internet of Things control system according to an embodiment of this application. As Figure 1 and Figure 2 shown, the Internet of Things control system 300 according to the embodiment of this application includes: a temperature acquisition and prediction module 310, configured to obtain a time queue of temperature data inside the refrigeration thermostat collected by a temperature sensor, and perform time series information transfer aggregation analysis and prediction on the time queue of temperature data inside the refrigeration thermostat to obtain real-time temperature data; a user interface setting module 320, configured to set a user interface for mode selection, parameter adjustment, and fault prompt; an actuator module 330, configured to receive the real-time temperature data and a control signal sent by the user interface, and perform a control action based on the real-time temperature data and the control signal; a system monitoring module 340, configured to remotely monitor the Internet of Things control system and give an early warning of faults.

[0030] Specifically, the temperature acquisition and prediction module 310 is configured to obtain a time queue of temperature data inside the refrigeration thermostat collected by a temperature sensor, and perform time series information transfer aggregation analysis and prediction on the time queue of temperature data inside the refrigeration thermostat to obtain real-time temperature data. In a specific example of this application, as Figure 3As shown, the temperature acquisition and prediction module 310 includes: a temperature time series data encoding unit 311, configured to perform standardization processing on the time queue of the temperature data inside the refrigerated thermostat and then perform temperature sequence encoding to obtain a sequence of local time series correlation features of the temperature inside the thermostat; a temperature local time series information transfer and aggregation representation unit 312, configured to perform information transfer and aggregation representation on the sequence of local time series correlation features of the temperature inside the thermostat to obtain a temperature time series information propagation and aggregation representation inside the thermostat; a real-time temperature data prediction unit 313, configured to determine the real-time temperature data based on the temperature time series information propagation and aggregation representation inside the thermostat.

[0031] Specifically, the temperature time series data encoding unit 311 is used to perform standardization processing on the time queue of the temperature data inside the refrigerated thermostat and then perform temperature sequence encoding to obtain a sequence of local time series correlation features of the temperature inside the thermostat. In a specific example of the present application, first, the time queue of the temperature data inside the thermostat is standardized to obtain the time queue of the standardized temperature data inside the thermostat. It should be understood that since the temperature data collected by the temperature sensor may have different magnitudes and distribution situations, without standardization processing, larger values and more dense data may dominate the learning process of the model, while ignoring those features with smaller values and sparser data distributions but equally important. Therefore, in order to ensure data consistency in subsequent processing steps and reduce the influence caused by different data magnitudes or distributions, in the technical solution of the present application, the time queue of the temperature data inside the thermostat is standardized to obtain the time queue of the standardized temperature data inside the thermostat. Next, the time queue of the standardized temperature data inside the thermostat is input into the temperature sequence encoder based on the 1D-CNN model to obtain a sequence of local time series correlation feature vectors of the temperature inside the thermostat; by inputting the time queue of the standardized temperature data inside the thermostat into the temperature sequence encoder based on the 1D-CNN model for feature encoding, the local time series feature information of the standardized temperature data inside the thermostat in the time dimension is extracted, thereby obtaining a sequence of local time series correlation feature vectors of the temperature inside the thermostat. Subsequently, the sequence of local time series correlation feature vectors of the temperature inside the thermostat is subjected to feature filtering processing based on the central contribution degree to obtain a filtered sequence of local time series correlation feature vectors of the temperature inside the thermostat as the sequence of local time series correlation features of the temperature inside the thermostat. Here, considering that each local time series correlation feature vector in the sequence of local time series correlation feature vectors of the temperature inside the thermostat respectively contains local time series correlation feature information regarding each local time period of the temperature inside the thermostat in the time dimension, and these local time series features of the temperature are not all equally important for subsequent real-time temperature data prediction tasks and control execution tasks. At the same time, there are some irrelevant interference and noise information in these different local time series features of the temperature. Especially in industrial or laboratory environments, various interference factors may cause large fluctuations in temperature data, thereby affecting the prediction accuracy of the temperature and resulting in subsequent control deviations.Based on this, in order to select the most representative and influential temperature local time series features from among the numerous temperature local time series correlation feature vectors inside the thermostat, thereby improving the prediction performance and generalization ability of the model, in the technical solution of this application, the sequence of the temperature local time series correlation feature vectors inside the thermostat is further subjected to feature filtering processing based on central contribution degree to obtain the sequence of the filtered temperature local time series correlation feature vectors inside the thermostat as the sequence of the temperature local time series correlation features inside the thermostat.

[0032] Among them, the process of performing clustering analysis on the sequence of the temperature local time series correlation feature vectors inside the thermostat to obtain the temperature local time series clustering center representation vector of the thermostat includes: calculating the position-wise mean between each of the temperature local time series correlation feature vectors in the sequence of the temperature local time series correlation feature vectors inside the thermostat to obtain the temperature local time series clustering center representation vector of the thermostat.

[0033] More specifically, taking the temperature local time series clustering center representation vector as the pseudo-anchoring center, the process of calculating the semantic contribution degree representation matrix of each of the temperature local time series correlation feature vectors in the sequence of the temperature local time series correlation feature vectors inside the thermostat with respect to the pseudo-anchoring center to obtain the sequence of the temperature local time series contribution degree representation matrices includes: calculating the position-wise division between the transposed vector of the temperature local time series clustering center representation vector and the two-norm of the temperature local time series clustering center representation vector to obtain the transformed temperature local time series clustering center vector; after calculating the vector multiplication between each of the temperature local time series correlation feature vectors and the transformed temperature local time series clustering center vector respectively, using the Softmax function to perform soft maximum normalization processing on the obtained sequence of correlation matrices to obtain the sequence of the temperature local time series contribution degree representation matrices.

[0034] More specifically, the process of performing feature filtering processing based on central contribution degree on the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat to obtain the sequence of filtered local temporal correlation feature vectors of the internal temperature of the thermostat as the sequence of local temporal correlation features of the internal temperature of the thermostat includes: First, perform clustering analysis on the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat to obtain the local temporal clustering center representation vector of the internal temperature of the thermostat; Through clustering analysis, the local temporal clustering center representation vector of the internal temperature of the thermostat can be extracted as a pseudo-anchoring center, which helps to capture the internal structure and distribution of temperature temporal data. The determination of the pseudo-anchoring center provides a benchmark for the importance evaluation of feature vectors, enabling the model to focus on those features that are most meaningful for temperature prediction. Next, calculate the semantic contribution degree of each local temporal correlation feature vector in the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat relative to the local temporal clustering center representation vector of the internal temperature of the thermostat to obtain a sequence of masked sparse local temporal contribution degree representation matrices of the internal temperature of the thermostat; Calculating the semantic contribution degree quantifies the contribution and importance of each local temporal correlation feature of the internal temperature of the thermostat to the overall semantics, so as to ignore those features that are unhelpful or even harmful to temperature prediction in subsequent processing. Furthermore, using each masked sparse local temporal contribution degree representation matrix in the sequence of masked sparse local temporal contribution degree representation matrices of the internal temperature of the thermostat as a modulation matrix, calculate the matrix product between it and each local temporal correlation feature vector of the internal temperature of the thermostat to obtain the sequence of filtered local temporal correlation feature vectors of the internal temperature of the thermostat. Specifically, first use the local temporal clustering center representation vector of the internal temperature of the thermostat as a pseudo-anchoring center, calculate the semantic contribution degree representation matrix of each local temporal correlation feature vector in the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat relative to the pseudo-anchoring center to obtain a sequence of local temporal contribution degree representation matrices of the internal temperature of the thermostat; Then perform dropout processing on each local temporal contribution degree representation matrix in the sequence of local temporal contribution degree representation matrices of the internal temperature of the thermostat to obtain the sequence of masked sparse local temporal contribution degree representation matrices of the internal temperature of the thermostat. That is, use the dropout mechanism to sparsify the local temporal contribution degree representation matrix of the internal temperature of the thermostat, generating a sequence of masked sparse local temporal contribution degree representation matrices of the internal temperature of the thermostat, to enhance the generalization ability of the model and reduce the number of parameters. The contribution degree pseudo-masked sparse processing strengthens the model's attention to key temperature temporal features by randomly "turning off" some unimportant temperature local temporal features, while reducing the complexity of the model and reducing the risk of overfitting.Finally, the matrix representing the local temporal contribution degree of the internal temperature of the mask sparse thermostat is used as the modulation matrix, and matrix multiplication operation is performed with the local temporal correlation feature vectors of the internal temperature of each thermostat to achieve feature filtering. That is, by weighting the original feature vectors with the modulation matrix, the unimportant or noisy temperature temporal features are filtered out. After feature filtering, a sequence of filtered local temporal correlation feature vectors of the internal temperature of the thermostat is obtained, which contains the important local temporal semantics and features that have been evaluated and screened. These features are used for subsequent prediction and can more accurately reflect the actual change trend of the internal temperature of the thermostat, thereby improving the accuracy of temperature prediction.

[0035] In summary, in the above embodiment, the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat is subjected to feature filtering processing based on the central contribution degree to obtain a sequence of filtered local temporal correlation feature vectors of the internal temperature of the thermostat as the sequence of local temporal correlation features of the internal temperature of the thermostat, including: performing feature filtering processing based on the central contribution degree on the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat with the following feature filtering formula to obtain the sequence of filtered local temporal correlation feature vectors of the internal temperature of the thermostat as the sequence of local temporal correlation features of the internal temperature of the thermostat; wherein, the feature filtering formula is:

[0036] X = {x1, x2,..., x n}

[0037]

[0038] {x i ’} = {WF i * x i}

[0039] wherein, X is the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat, x1, x2, x n are the 1st, 2nd, and nth local temporal correlation feature vectors of the internal temperature of the thermostat in the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat respectively, x i is the ith local temporal correlation feature vector of the internal temperature of the thermostat in the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat, N is the number of feature vectors in the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat, x c is the vector representing the local temporal clustering center of the internal temperature of the thermostat, ||·||2 represents the two-norm of the vector, Softmax(·) is the Softmax function, dropout is the random inactivation process, WF iis the masked sparse contribution degree representation matrix of the internal temperature local temporal correlation feature vector corresponding to the i-th thermostat, and x i ’ is the i-th filtered internal temperature local temporal correlation feature vector in the sequence of the filtered internal temperature local temporal correlation feature vectors of the thermostat.

[0040] Specifically, the temperature local temporal information transfer and aggregation representation unit 312 is used to perform information transfer and aggregation representation on the sequence of the internal temperature local temporal correlation features of the thermostat to obtain the internal temperature temporal information propagation and aggregation representation of the thermostat. It should be understood that since the sequence of the filtered internal temperature local temporal correlation feature vectors of the thermostat respectively contains the internal temperature local temporal correlation features in the thermostat after feature filtering and purification, there is a temporal transfer characteristic between these temperature local temporal features after feature filtering. This characteristic not only exists in the temporal correlation attenuation in time but also exists in the span information of the features of the two in space. Based on this, in order to better capture and utilize the dynamic changes of temperature data in time and space, thereby improving the prediction accuracy and response speed of the system, in the technical solution of this application, the sequence of the internal temperature local temporal correlation features of the thermostat is further subjected to information transfer and aggregation representation to obtain the internal temperature temporal information propagation and aggregation representation of the thermostat. In a specific example of this application, as Figure 4 shown, the temperature local temporal information transfer and aggregation representation unit 312 includes: an internal temperature local temporal correlation feature extraction sub-unit 3121, which is used to extract the sequence of the current internal temperature local temporal correlation feature and the historical internal temperature local temporal correlation feature from the sequence of the internal temperature local temporal correlation features of the thermostat; an internal temperature dynamic energy spatio-temporal transfer weight calculation sub-unit 3122, which is used to calculate the sequence of the internal temperature dynamic energy spatio-temporal transfer weights of the sequence of the historical internal temperature local temporal correlation features relative to the current internal temperature local temporal correlation feature; a node feature transfer and fusion sub-unit 3123, which is used to perform node feature transfer and fusion on the sequence of the internal temperature local temporal correlation features based on the sequence of the internal temperature dynamic energy spatio-temporal transfer weights to obtain the internal temperature temporal information propagation and aggregation representation.

[0041] More specifically, the internal temperature local time-series correlation feature extraction subunit 3121 of the thermostat is configured to extract a sequence of the current internal temperature local time-series correlation feature and the historical internal temperature local time-series correlation feature from the sequence of the internal temperature local time-series correlation features of the thermostat. In the technical solution of the present application, the currently filtered internal temperature local time-series correlation feature vector is extracted from the sequence of the filtered internal temperature local time-series correlation feature vectors of the thermostat as the current internal temperature local time-series correlation feature of the thermostat, and the other filtered internal temperature local time-series correlation feature vectors in the sequence of the filtered internal temperature local time-series correlation feature vectors of the thermostat are defined as the historical filtered internal temperature local time-series correlation feature vectors to obtain a sequence of the historical filtered internal temperature local time-series correlation feature vectors as the sequence of the historical internal temperature local time-series correlation features.

[0042] More specifically, the internal temperature dynamic energy spatio-temporal transfer weight calculation sub-unit 3122 of the thermostat is used to calculate the sequence of the internal temperature dynamic energy spatio-temporal transfer weights of the sequence of the historical internal temperature local temporal correlation features of the thermostat relative to the current internal temperature local temporal correlation features of the thermostat. In a specific example of the present application, first, calculate the internal temperature local temporal static energy values of each of the historical filtered internal temperature local temporal correlation feature vectors in the sequence of the historical filtered internal temperature local temporal correlation feature vectors of the thermostat; then, based on the spatial span coefficient between each of the historical filtered internal temperature local temporal correlation feature vectors and the current filtered internal temperature local temporal correlation feature vector, determine the internal temperature local temporal energy spatial decay factor; then, based on the time span coefficient between each of the historical filtered internal temperature local temporal correlation feature vectors and the current filtered internal temperature local temporal correlation feature vector, determine the internal temperature local temporal energy time decay factor; further, based on the internal temperature local temporal energy spatial decay factor and the internal temperature local temporal energy time decay factor, perform spatio-temporal modulation on the internal temperature local temporal static energy values of each of the historical filtered internal temperature local temporal correlation feature vectors to obtain the internal temperature local temporal dynamic energy spatio-temporal transfer values of each of the historical filtered internal temperature local temporal correlation feature vectors relative to the current filtered internal temperature local temporal correlation feature vector; subsequently, input the internal temperature local temporal dynamic energy spatio-temporal transfer values of each of the historical filtered internal temperature local temporal correlation feature vectors into the energy transfer gating module based on the masking function to obtain the sequence of the internal temperature local temporal dynamic energy spatio-temporal transfer weights. Here, by calculating the static energy values of each of the historical filtered internal temperature local temporal correlation feature vectors and combining the spatial span coefficient and the time span coefficient of the current filtered internal temperature local temporal correlation feature vector, historical information and the current state can be integrated to form an aggregated representation vector. This vector contains the dynamic change information of the temperature local temporal features in terms of time and spatial spans, which helps the model to more comprehensively understand the trend of temperature change. In particular, the static energy value is related to the maximum eigenvalue, mean value and variance of each of the historical filtered internal temperature local temporal correlation feature vectors. The static energy value provides a quantitative measure of the inherent property of the feature vector in space, provides a benchmark for the spatio-temporal decay of potential energy, and helps the model to understand the static importance of each node in the spatial dimension and the time dimension.The spatial span coefficient and the temporal span coefficient respectively reflect the spatial distance and the temporal distance between nodes. The static energy values of the local temporal correlation feature vectors of the internal temperature of the thermostat after each history filtering are modulated by using the energy spatial decay factor and the energy temporal decay factor, so as to combine the spatial decay factor and the temporal decay factor through spatio-temporal modulation to adjust the static energy values, in order to reflect the dynamic changes of node features in space-time, enabling the model to adjust the weights and intensities of information transmission according to the spatial and temporal distances between nodes, and enhancing the sensitivity of the model to the locality and timeliness of temperature time series. In this way, the dynamic changes and information transmission effects between different nodes of the temperature local time series feature data in the full time domain can be captured more accurately. Furthermore, through the energy transfer gating module, important spatio-temporal transfer information regarding the temperature local time series can be screened out, and these information can be strengthened through the masking function. This mechanism can help the model identify which historical temperature local time series information is crucial for the prediction of the current state, thereby improving the accuracy and effectiveness of the temperature local time series feature information transmission. By adaptively adjusting the information flow, the model can pay more attention to the features that contribute to the task.

[0043] Among them, the process of calculating the internal temperature local temporal static energy value of each internal temperature local temporal correlation feature vector of the thermostat after history filtering in the sequence of the internal temperature local temporal correlation feature vectors of the thermostat after history filtering includes: respectively calculating the mean and the standard deviation of the internal temperature local temporal correlation feature vectors of the thermostat after history filtering to obtain the internal temperature local temporal feature mean and the internal temperature local temporal standard deviation after history filtering; calculating the fourth power of the position-wise difference between the internal temperature local temporal correlation feature vectors of the thermostat after history filtering and the internal temperature local temporal feature mean after history filtering, and then calculating the expected value of the fourth power modulated internal temperature local temporal offset vector after history filtering to obtain the internal temperature local temporal feature offset expectation factor after filtering; calculating the division between the internal temperature local temporal feature offset expectation factor after filtering and the fourth power of the internal temperature local temporal standard deviation after history filtering, and then adding it to the maximum eigenvalue in the internal temperature local temporal correlation feature vectors of the thermostat after history filtering to obtain the internal temperature local temporal static energy value.

[0044] More specifically, the process of determining the local temporal energy spatial decay factor of the internal temperature of the thermostat based on the spatial span coefficient between each of the locally temporally correlated feature vectors of the historical filtered internal temperature of the thermostat and the locally temporally correlated feature vector of the currently filtered internal temperature of the thermostat includes: calculating the spatial distance span between the locally temporally correlated feature vector of the historical filtered internal temperature of the thermostat and the locally temporally correlated feature vector of the currently filtered internal temperature of the thermostat to obtain the locally temporally spatial span coefficient of the historical filtered internal temperature of the thermostat; calculating the division between the local temporal static energy value of the internal temperature of the thermostat corresponding to the locally temporally correlated feature vector of the historical filtered internal temperature of the thermostat and the locally temporally spatial span coefficient of the historical filtered internal temperature of the thermostat to obtain the local temporal energy spatial decay factor of the internal temperature of the thermostat.

[0045] More specifically, the process of determining the local temporal energy temporal decay factor of the internal temperature of the thermostat based on the time span coefficient between each of the locally temporally correlated feature vectors of the historical filtered internal temperature of the thermostat and the locally temporally correlated feature vector of the currently filtered internal temperature of the thermostat includes: after calculating the subtraction between the timestamp of the locally temporally correlated feature vector of the currently filtered internal temperature of the thermostat and the timestamp of the locally temporally correlated feature vector of the historical filtered internal temperature of the thermostat, rounding down the obtained timestamp value to obtain the locally temporally time span coefficient of the filtered internal temperature of the thermostat; dividing the locally temporally time span coefficient of the filtered internal temperature of the thermostat by the inverse scaling parameter of the control time decay period, and taking the obtained time span decay representation coefficient as the exponential power, calculating the natural exponential function value with the natural constant e as the base to obtain the time span decay class support representation coefficient; calculating the division between the local temporal static energy value of the internal temperature of the thermostat and the time span decay class support representation coefficient to obtain the local temporal energy temporal decay factor of the internal temperature of the thermostat.

[0046] More specifically, the process of inputting the local temporal dynamic energy spatio-temporal transfer values of each of the locally temporally correlated feature vectors of the historical filtered internal temperature of the thermostat into the energy transfer gating module based on the masking function to obtain a sequence of local temporal dynamic energy spatio-temporal transfer weights includes: in the energy transfer gating module based on the masking function, in response to the local temporal dynamic energy spatio-temporal transfer value being greater than the threshold hyperparameter, normalizing the local temporal dynamic energy spatio-temporal transfer value greater than the threshold hyperparameter through the sigmoid function to obtain the local temporal dynamic energy spatio-temporal transfer weight.

[0047] More specifically, the node feature transfer and fusion subunit 3123 is used to perform node feature transfer and fusion on the sequence of local temporal correlation features of the internal temperature of the thermostat based on the sequence of spatio-temporal transfer weights of the dynamic energy of the internal temperature of the thermostat, so as to obtain the spatio-temporal propagation aggregation representation of the internal temperature of the thermostat. In a specific example of the present application, first, based on the sequence of spatio-temporal transfer weights of the local temporal dynamic energy of the internal temperature of the thermostat, the weighted sum of the sequence of local temporal correlation feature vectors of the internal temperature of the thermostat after historical filtering is calculated to obtain the local temporal node feature transfer aggregation modulation vector of the internal temperature of the thermostat after historical filtering; that is, by weighted summing the spatio-temporal transfer values of the historical node features, the information from different historical nodes is integrated, providing rich context information of the local temporal features of the temperature for the model, which is crucial for understanding the dynamics and development trends of temperature changes and helps improve the model's processing ability for complex temperature temporal information transfer and prediction tasks. Furthermore, the element-wise sum of the local temporal node feature transfer aggregation modulation vector of the internal temperature of the thermostat after historical filtering and the local temporal correlation feature vector of the internal temperature of the thermostat after current filtering is calculated to obtain the spatio-temporal significant propagation aggregation representation vector of the internal temperature of the thermostat as the spatio-temporal propagation aggregation representation of the internal temperature of the thermostat. Here, the element-wise sum of the local temporal node feature transfer aggregation modulation vector of the internal temperature of the thermostat after historical filtering and the local temporal correlation feature vector of the internal temperature of the thermostat after current filtering is used to generate a spatio-temporal propagation aggregation representation of the internal temperature of the thermostat that combines the local temporal historical information and the current state of the temperature, providing rich features for subsequent tasks such as temperature prediction. Such a representation not only contains the local temporal feature information of the temperature at the current moment but also integrates the temperature change trend over a past period of time, enabling the model to more accurately infer and predict future temperature changes and enhancing the model's processing ability for complex tasks.

[0048] In summary, in the above embodiment, the sequence of local temporal correlation features of the internal temperature of the thermostat is subjected to information transfer aggregation representation to obtain the spatio-temporal propagation aggregation representation of the internal temperature of the thermostat, including: the sequence of local temporal correlation features of the internal temperature of the thermostat is subjected to information transfer aggregation representation according to the following feature message propagation formula to obtain the spatio-temporal propagation aggregation representation of the internal temperature of the thermostat; wherein, the feature message propagation formula is:

[0049] X’ = {x1’, x2’,..., x i ’,..., x p ’}

[0050]

[0051] Among them, X’ is the sequence of the filtered local temporal correlation feature vectors of the internal temperature of the thermostat, and x1’, x2’, x i ’, x p ’ are the 1st, 2nd, ith, and pth filtered local temporal correlation feature vectors of the internal temperature of the thermostat in the sequence of the filtered local temporal correlation feature vectors of the internal temperature of the thermostat. x i ’(j) is the eigenvalue at the jth position in the ith filtered local temporal correlation feature vector of the internal temperature of the thermostat. μ i and σ i are the feature mean and standard deviation of the ith filtered local temporal correlation feature vector of the internal temperature of the thermostat respectively. E is to calculate the expected value, and max(·) represents taking the maximum value in the vector. is the local temporal static energy value of the internal temperature of the thermostat. Count(x i ’ → x p ’) represents the spatial span coefficient between x i ’ and x p ’. is the energy space decay factor between x i ’ and x p ’. t p and t i represent the timestamps of x p ’ and x i ’ respectively. represents the floor operation. L is the inverse scaling parameter for controlling the time decay period, and exp(·) is the natural exponential function value. is the energy time decay factor between x i and x p ’. α and β are trainable hyperparameters. x (i→p) is the dynamic energy spatio-temporal transfer value of x i ’ relative to x p ’. mask(·) is the masking operation. p is the number of vectors in the sequence of the filtered local temporal correlation feature vectors of the internal temperature of the thermostat. θ is the threshold hyperparameter, and sigmoid(·) is the sigmoid function. x p+1 ’ is the spatio-temporal significant propagation aggregation representation vector of the internal temperature of the thermostat.

[0052] It is worth mentioning that in other specific examples of the present application, the sequence of the internal temperature local time series correlation features of the thermostat can also be aggregated and represented by other means to obtain the internal temperature time series information propagation and aggregation representation of the thermostat. For example: input the sequence of the internal temperature local time series correlation features of the thermostat; select a suitable model, such as a recurrent neural network like LSTM (Long Short-Term Memory) or GRU (Gated Recurrent Unit), or a model based on the attention mechanism like Transformer; use the extracted features to train the model so that it can learn the pattern of temperature data changing over time; define how to transfer information from one time point to another in the model, such as through the connection between recurrent layers; perform aggregation operations at different levels of the model, such as average pooling, max pooling, etc., to synthesize information at different time points; to obtain the internal temperature time series information propagation and aggregation representation of the thermostat.

[0053] Specifically, the real-time temperature data prediction unit 313 is used to determine the real-time temperature data based on the internal temperature time series information propagation and aggregation representation of the thermostat. In a specific example of the present application, the internal temperature spatio-temporal significant propagation and aggregation representation vector of the thermostat is input into a decoder-based temperature predictor to obtain the decoded value of the real-time temperature data. That is, the spatio-temporal information transfer and aggregation representation features of the temperature inside the thermostat are used for decoding regression to predict the real-time temperature data inside the thermostat. In this way, the interference of irrelevant noise in the temperature time series can be reduced, and at the same time, the real-time temperature data inside the thermostat can be predicted more accurately. This intelligent prediction method improves the generalization ability of the system, enabling the Internet of Things control system to better adapt to new situations and execute control actions more timely and effectively.

[0054] In a preferred example, inputting the internal temperature spatio-temporal significant propagation and aggregation representation vector of the thermostat into a decoder-based temperature predictor to obtain the decoded value of the real-time temperature data includes:

[0055] Determine the maximum eigenvalue of the internal temperature spatio-temporal significant propagation and aggregation of the internal temperature spatio-temporal significant propagation and aggregation representation vector of the thermostat, the minimum eigenvalue of the internal temperature spatio-temporal significant propagation and aggregation, and calculate the average value of the internal temperature spatio-temporal significant propagation and aggregation and the standard deviation of the internal temperature spatio-temporal significant propagation and aggregation of the feature set of the internal temperature spatio-temporal significant propagation and aggregation representation vector of the thermostat;

[0056] Calculate the quotient of the average value of the internal temperature spatio-temporal significant propagation and aggregation and the standard deviation of the internal temperature spatio-temporal significant propagation and aggregation to obtain the standardized value of the internal temperature spatio-temporal significant propagation and aggregation statistics;

[0057] Calculate the reciprocal of each eigenvalue of the spatio-temporally significantly propagated aggregation representation vector of the internal temperature of the thermostat, and after dot-multiplying with the difference between the spatio-temporally significantly propagated aggregation maximum eigenvalue and the spatio-temporally significantly propagated aggregation minimum eigenvalue of the internal temperature of the thermostat, perform a dot-subtraction with the spatio-temporally significantly propagated aggregation statistical normalization value of the internal temperature of the thermostat to obtain the spatio-temporally significantly propagated aggregation distribution approximation vector of the internal temperature of the thermostat;

[0058] Calculate the exponential function with the natural constant as the base and each eigenvalue of the spatio-temporally significantly propagated aggregation distribution approximation vector of the internal temperature of the thermostat as the exponent to obtain the spatio-temporally significantly propagated aggregation distribution class approximation vector of the internal temperature of the thermostat;

[0059] Add the spatio-temporally significantly propagated aggregation distribution class approximation vector of the internal temperature of the thermostat to the spatio-temporally significantly propagated aggregation statistical normalization value of the internal temperature of the thermostat, and calculate the base-two logarithm of the absolute value of each eigenvalue of the added vector to obtain the optimized spatio-temporally significantly propagated aggregation representation vector of the internal temperature of the thermostat; and

[0060] Input the optimized spatio-temporally significantly propagated aggregation representation vector of the internal temperature of the thermostat into a decoder-based temperature predictor to obtain the decoded value of the real-time temperature data.

[0061] Among them, the optimized representation of the spatio-temporally significantly propagated aggregation representation vector of the internal temperature of the thermostat, denoted as V, is:

[0062]

[0063]

[0064] δ = v max -v min

[0065] Among them, V is the spatio-temporally significantly propagated aggregation representation vector of the internal temperature of the thermostat, V ⊙-1 is the reciprocal of each eigenvalue of the spatio-temporally significantly propagated aggregation representation vector of the internal temperature of the thermostat, v max and v min are the spatio-temporally significantly propagated aggregation maximum eigenvalue and the spatio-temporally significantly propagated aggregation minimum eigenvalue of the internal temperature of the thermostat of the spatio-temporally significantly propagated aggregation representation vector of the internal temperature of the thermostat, and δ is the difference between the spatio-temporally significantly propagated aggregation maximum eigenvalue and the spatio-temporally significantly propagated aggregation minimum eigenvalue of the internal temperature of the thermostat, μ is the spatio-temporally significantly propagated aggregation mean, σ is the spatio-temporally significantly propagated aggregation standard deviation, η is the spatio-temporally significantly propagated aggregation statistical normalization value, V e is the spatio-temporally significantly propagated aggregation distribution class approximation vector of the internal temperature of the thermostat, + is dot addition, ⊙ is dot multiplication, - is dot subtraction, exp() is exponential operation, log is logarithm with base 2, and V′ is the spatio-temporally significant propagation aggregation representation vector of the internal temperature of the optimized thermostat.

[0066] Here, in the preferred example, since the sequences of the filtered local temporal correlation feature vectors of the internal temperature of the thermostat respectively represent the filtered query semantic features of the local temporal correlation of the internal temperature of the thermostat in each local time domain based on the semantic query of the contribution degree of the pseudo-anchoring center, when input into the feature message propagation network based on the spatio-temporal transfer characteristics of node potential energy, the node semantic filtering query differences will have different semantic propagation weights based on the spatio-temporal transfer characteristics of potential energy, making the spatio-temporally significant propagation aggregation representation vector of the internal temperature of the thermostat also have a diverse set expression distribution of temporal aggregation characteristics. Therefore, it is desired to improve the balance between the regression mapping accuracy and integrity when the spatio-temporally significant propagation aggregation representation vector of the internal temperature of the thermostat is input into the temperature predictor based on the decoder for decoding regression, thereby improving the accuracy of the decoded value of the obtained real-time temperature data.

[0067] Therefore, by randomly statistically normalizing the diverse feature sets of the spatio-temporally significant propagation aggregation representation vector of the internal temperature of the thermostat, an approximation of the standardized continuous probability density distribution of the response hypothesis test of the confidence space constructed based on the overall eigenvalue of the spatio-temporally significant propagation aggregation representation vector of the internal temperature of the thermostat with respect to each eigenvalue of the spatio-temporally significant propagation aggregation representation vector is performed, thereby establishing the target reachability from the diverse feature distribution of the spatio-temporally significant propagation aggregation representation vector of the internal temperature of the thermostat to the unified regression target, so as to realize the balance executability between the mapping accuracy and mapping integrity in the decoding regression process based on the diverse feature distribution of the spatio-temporally significant propagation aggregation representation vector of the internal temperature of the thermostat, and improve the accuracy of the decoded value of the real-time temperature data obtained by inputting the spatio-temporally significant propagation aggregation representation vector of the internal temperature of the thermostat into the temperature predictor based on the decoder. In this way, the real-time temperature data inside the thermostat can be predicted more accurately, the generalization ability of the system is improved, and the Internet of Things control system can better adapt to new situations, so as to execute control actions more timely and effectively.

[0068] It is worth mentioning that in other specific examples of this application, the time queue of the temperature data inside the refrigeration thermostat collected by the temperature sensor can also be obtained through other means, and the time queue of the temperature data inside the refrigeration thermostat can be subjected to time series information transfer aggregation analysis and prediction to obtain real-time temperature data. For example: input the time queue of the temperature data inside the refrigeration thermostat; select a model suitable for time series data analysis, such as LSTM (Long Short-Term Memory Network), GRU (Gated Recurrent Unit), Transformer, etc.; use the extracted features and label data to train the model to ensure that the model can learn the pattern of temperature data changing over time; define an information transfer mechanism in the model, such as realizing information transfer through the connection between recursive layers; perform aggregation operations at different levels of the model, such as average pooling, max pooling, etc., to synthesize information at different time points; continuously collect real-time temperature data from the temperature sensor and add it to the time queue; perform the same preprocessing steps on the real-time collected data to ensure the consistency and accuracy of the data; extract features from the real-time data to form the same feature representation as the training data; input the features of the real-time data into the trained model, and output the temperature prediction value at a future time point to obtain the real-time temperature data.

[0069] Specifically, the user interface setting module 320 is used to set the user interface for mode selection, parameter adjustment, and fault prompt. The user interface setting module plays a crucial role in the refrigeration thermostat control system. It not only provides an entry for users to interact with the system but also enhances the usability and user experience of the system. In one example, the user interface setting module allows users to select different working modes and adjust various control parameters, such as set temperature, temperature fluctuation range, heating / cooling rate, etc. In addition, by monitoring the system status in real time, when a fault or abnormality is detected, it promptly issues a warning or prompt to the user. In this way, the operation experience of users and the reliability of the system are significantly improved, ensuring that the refrigeration thermostat operates efficiently and stably in various application scenarios.

[0070] Specifically, the actuator module 330 is configured to receive the real-time temperature data and the control signal sent by the user interface, and perform a control action based on the real-time temperature data and the control signal. In one example, the actuator module performs a corresponding control action according to the real-time temperature data and the control signal set by the user. For example: when the real-time temperature is lower than the set temperature, the actuator module increases the heating speed by adjusting the current or voltage of the heater to increase the power of the heater; when the real-time temperature is higher than the set temperature, the actuator module increases the cooling speed by adjusting the fan speed or refrigerant flow rate of the cooler to increase the power of the cooler; when a temperature anomaly or equipment failure is detected, the heater is turned off through the actuator module, the cooler is started, a fault alarm signal is issued, and the fault information is recorded. Through the effective operation of the actuator module, it can ensure that the refrigeration thermostat operates stably and efficiently under various working conditions, meeting the user's temperature control requirements.

[0071] Specifically, the system monitoring module 340 is configured to remotely monitor the Internet of Things control system and give early warnings of faults. That is, through the Internet or local area network, users can remotely monitor the system at any location. In one example, a threshold is set to monitor the system status in real time. When the parameters exceed the threshold, it indicates that an anomaly or potential fault is detected, and an early warning is immediately issued. In this way, the flexibility and efficiency of management can be improved.

[0072] As described above, the Internet of Things control system 300 according to the embodiments of the present application can be implemented in various wireless terminals, such as a server with Internet of Things control algorithms. In a possible implementation manner, the Internet of Things control system 300 according to the embodiments of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the Internet of Things control system 300 can be a software module in the operating system of the wireless terminal, or can be an application program developed for the wireless terminal; of course, the Internet of Things control system 300 can also be one of the many hardware modules of the wireless terminal.

[0073] Alternatively, in another example, the Internet of Things control system 300 and the wireless terminal can also be separate devices, and the Internet of Things control system 300 can be connected to the wireless terminal through a wired and / or wireless network and transmit interaction information according to a predefined data format.

[0074] Furthermore, an Internet of Things control method is also provided.

[0075] Figure 5 is a flowchart of the Internet of Things control method according to the embodiments of the present application. As Figure 5As shown, the Internet of Things control method according to an embodiment of the present application includes the steps of: S1, obtaining a time queue of temperature data inside the refrigeration constant temperature bath collected by a temperature sensor, and performing time series information transfer aggregation analysis and prediction on the time queue of temperature data inside the refrigeration constant temperature bath to obtain real-time temperature data; S2, setting a user interface for mode selection, parameter adjustment, and fault prompt; S3, receiving the real-time temperature data and a control signal sent by the user interface, and performing a control action based on the real-time temperature data and the control signal; S4, remotely monitoring the Internet of Things control system and giving an early warning of faults.

[0076] In summary, the Internet of Things control method according to an embodiment of the present application is clarified. It uses a temperature sensor to monitor and collect the temperature data inside the refrigeration constant temperature bath in real time, and introduces data processing and analysis algorithms based on artificial intelligence and deep learning at the backend to perform time series analysis on the temperature data. Then, it uses the contribution degree of the pseudo-anchoring center to filter the local time series features of the temperature inside the refrigeration constant temperature bath. At the same time, it aggregates and infers the local time series information of the temperature inside the constant temperature bath through the spatio-temporal transfer information of the local temperature time series, so as to predict the real-time temperature data. Then, by transmitting the predicted real-time temperature data to the actuator module, the actuator module performs corresponding actions based on the real-time temperature data and the control signal, realizing the intelligent control of the Internet of Things. In this way, it is possible to reduce the influence of irrelevant noise interference on the temperature time series through feature filtering, and at the same time use the spatio-temporal transfer method of local temperature time series semantics to aggregate and infer the time series information of the temperature inside the constant temperature bath, so as to more accurately predict the real-time temperature data inside the constant temperature bath. This intelligent prediction method improves the generalization ability of the system, enabling the Internet of Things control system to better adapt to new situations, so as to perform control actions more timely and effectively.

[0077] The embodiments of the present disclosure have been described above. The above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to technologies in the market, or to enable other ordinary skill in the art in the technical field to understand the embodiments disclosed herein.

Claims

1. An Internet of Things control system, characterized in that, Including: A temperature acquisition and prediction module, configured to obtain a time queue of temperature data inside the refrigerated thermostat collected by a temperature sensor, and perform time series information transfer aggregation analysis and prediction on the time queue of the temperature data inside the refrigerated thermostat to obtain real-time temperature data; A user interface setting module, configured to set a user interface for mode selection, parameter adjustment, and fault prompt; An actuator module, configured to receive the real-time temperature data and a control signal sent by the user interface, and perform a control action based on the real-time temperature data and the control signal; A system monitoring module, configured to remotely monitor the Internet of Things control system and give early warnings for faults; Wherein, the temperature acquisition and prediction module includes: a temperature time series data encoding unit, configured to perform normalization processing on the time queue of the temperature data inside the refrigerated thermostat and then perform temperature sequence encoding to obtain a sequence of local time series correlation features of the temperature inside the thermostat; a temperature local time series information transfer aggregation representation unit, configured to perform information transfer aggregation representation on the sequence of the local time series correlation features of the temperature inside the thermostat to obtain a temperature time series information propagation aggregation representation inside the thermostat; a real-time temperature data prediction unit, configured to determine the real-time temperature data based on the temperature time series information propagation aggregation representation inside the thermostat; Wherein, the temperature local time series information transfer aggregation representation unit includes: A local time series correlation feature extraction subunit of the temperature inside the thermostat, configured to extract a sequence of current local time series correlation features of the temperature inside the thermostat and a sequence of historical local time series correlation features of the temperature inside the thermostat from the sequence of the local time series correlation features of the temperature inside the thermostat; A dynamic energy spatio-temporal transfer weight calculation subunit of the temperature inside the thermostat, configured to calculate a sequence of dynamic energy spatio-temporal transfer weights of the temperature inside the thermostat of the sequence of the historical local time series correlation features of the temperature inside the thermostat relative to the current local time series correlation features of the temperature inside the thermostat; A node feature transfer and fusion subunit, configured to perform node feature transfer and fusion on the sequence of the local time series correlation features of the temperature inside the thermostat based on the sequence of the dynamic energy spatio-temporal transfer weights of the temperature inside the thermostat to obtain the temperature time series information propagation aggregation representation inside the thermostat.

2. The Internet of Things control system according to claim 1, characterized in that The temperature time series data encoding unit includes: A normalization subunit, configured to normalize the time queue of the temperature data inside the thermostat to obtain a time queue of normalized temperature data inside the thermostat; A temperature sequence encoding subunit, configured to input the time queue of the normalized temperature data inside the thermostat into a temperature sequence encoder based on a 1D-CNN model to obtain a sequence of local time series correlation feature vectors of the temperature inside the thermostat; A feature filtering subunit, configured to perform feature filtering processing based on central contribution degree on the sequence of the local time series correlation feature vectors of the temperature inside the thermostat to obtain a filtered sequence of the local time series correlation feature vectors of the temperature inside the thermostat as the sequence of the local time series correlation features of the temperature inside the thermostat.

3. The Internet of Things control system according to claim 2, characterized in that, The feature filtering subunit includes: The clustering analysis secondary subunit is used to perform clustering analysis on the sequence of the local temporal correlation feature vectors of the internal temperature of the thermostat to obtain the local temporal clustering center representation vector of the internal temperature of the thermostat; The semantic contribution degree calculation secondary subunit is used to calculate the semantic contribution degree of each local temporal correlation feature vector of the internal temperature of the thermostat in the sequence of the local temporal correlation feature vectors of the internal temperature of the thermostat relative to the local temporal clustering center representation vector of the internal temperature of the thermostat to obtain a sequence of masked sparse local temporal contribution degree representation matrices of the internal temperature of the thermostat; The feature modulation secondary subunit is used to use each masked sparse local temporal contribution degree representation matrix in the sequence of the masked sparse local temporal contribution degree representation matrices as a modulation matrix, and calculate the matrix product between it and each local temporal correlation feature vector of the internal temperature of the thermostat respectively to obtain the sequence of the filtered local temporal correlation feature vectors of the internal temperature of the thermostat.

4. The Internet of Things control system according to claim 3, characterized in that, The semantic contribution degree calculation secondary subunit is used for: Taking the local temporal clustering center representation vector of the internal temperature of the thermostat as a pseudo-anchoring center, calculating the semantic contribution degree representation matrix of each local temporal correlation feature vector of the internal temperature of the thermostat in the sequence of the local temporal correlation feature vectors of the internal temperature of the thermostat relative to the pseudo-anchoring center to obtain a sequence of local temporal contribution degree representation matrices of the internal temperature of the thermostat; Performing dropout processing on each local temporal contribution degree representation matrix in the sequence of the local temporal contribution degree representation matrices of the internal temperature of the thermostat to obtain the sequence of the masked sparse local temporal contribution degree representation matrices of the internal temperature of the thermostat.

5. The Internet of Things control system according to claim 4, wherein, The local temporal correlation feature extraction subunit of the internal temperature of the thermostat is used for: extracting the currently filtered local temporal correlation feature vector of the internal temperature of the thermostat from the sequence of the filtered local temporal correlation feature vectors of the internal temperature of the thermostat as the current local temporal correlation feature of the internal temperature of the thermostat, and defining the other filtered local temporal correlation feature vectors in the sequence of the filtered local temporal correlation feature vectors of the internal temperature of the thermostat as the historical filtered local temporal correlation feature vectors of the internal temperature of the thermostat to obtain a sequence of historical filtered local temporal correlation feature vectors of the internal temperature of the thermostat as the sequence of the historical local temporal correlation features of the internal temperature of the thermostat.

6. The Internet of Things control system according to claim 5, characterized in that, The local temporal dynamic energy spatio-temporal transfer weight calculation subunit of the internal temperature of the thermostat is used for: Calculating the local temporal static energy value of each historical filtered local temporal correlation feature vector of the internal temperature of the thermostat in the sequence of the historical filtered local temporal correlation feature vectors of the internal temperature of the thermostat; Based on the spatial span coefficient between each historical filtered local temporal correlation feature vector of the internal temperature of the thermostat and the currently filtered local temporal correlation feature vector of the internal temperature of the thermostat, determining the local temporal energy spatial attenuation factor; Determine the local temporal energy time decay factor of the internal temperature of the thermostat based on the time span coefficient between the local temporal correlation feature vectors of the internal temperature of the thermostat after each historical filtering and the local temporal correlation feature vector of the internal temperature of the thermostat after the current filtering; Perform spatio-temporal modulation on the local temporal static energy values of the internal temperature of the thermostat for the local temporal correlation feature vectors of the internal temperature of the thermostat after each historical filtering based on the local temporal energy spatial decay factor of the internal temperature of the thermostat and the local temporal energy time decay factor of the internal temperature of the thermostat to obtain the local temporal dynamic energy spatio-temporal transfer values of the local temporal correlation feature vectors of the internal temperature of the thermostat after each historical filtering with respect to the local temporal correlation feature vector of the internal temperature of the thermostat after the current filtering; Input the local temporal dynamic energy spatio-temporal transfer values of the local temporal correlation feature vectors of the internal temperature of the thermostat after each historical filtering into an energy transfer gating module based on a masking function to obtain a sequence of the local temporal dynamic energy spatio-temporal transfer weights.

7. The Internet of Things control system according to claim 6, characterized in that, The node feature transfer and fusion sub-unit is used for: Calculate the weighted sum of the sequence of the local temporal correlation feature vectors of the internal temperature of the thermostat after historical filtering based on the sequence of the local temporal dynamic energy spatio-temporal transfer weights to obtain a local temporal node feature transfer aggregation modulation vector of the internal temperature of the thermostat after historical filtering; Calculate the element-wise sum of the local temporal node feature transfer aggregation modulation vector of the internal temperature of the thermostat after historical filtering and the local temporal correlation feature vector of the internal temperature of the thermostat after the current filtering to obtain a spatio-temporal significant propagation aggregation representation vector of the internal temperature of the thermostat as the spatio-temporal information propagation aggregation representation of the internal temperature of the thermostat.

8. The Internet of Things control system according to claim 7, wherein The real-time temperature data prediction unit is used for: Input the spatio-temporal significant propagation aggregation representation vector of the internal temperature of the thermostat into a temperature predictor based on a decoder to obtain a decoded value of the real-time temperature data.

9. An Internet of Things control method, using the Internet of Things control system according to claim 1, characterized in that, Including: Obtain a time queue of the temperature data inside the refrigerating thermostat collected by a temperature sensor, and perform spatio-temporal information transfer aggregation analysis and prediction on the time queue of the temperature data inside the refrigerating thermostat to obtain real-time temperature data; Set up a user interface for mode selection, parameter adjustment, and fault prompt; Receive the real-time temperature data and the control signal sent by the user interface, and perform control actions based on the real-time temperature data and the control signal; Remotely monitor the Internet of Things control system and give early warnings about faults.

Images