Iot data transmission method and system based on iot middle platform

Abstract

The application provides an Internet of Things data transmission method and system based on an Internet of Things middle station, and relates to the technical field of communication.The method comprises the following steps: collecting multi-element perception data based on an Internet of Things perception layer, wherein the Internet of Things perception layer comprises a sensor network composed of multiple sensors; acquiring an adaptive transmission period according to an encoding time sequence matrix corresponding to the multi-element perception data, a data transmission period feature constructed by an Internet of Things middle station, and a bidirectional long short-term memory network model; and dynamically adjusting a data transmission time interval of the Internet of Things perception layer according to the adaptive transmission period.The application is helpful to improve the real-time monitoring capability and response speed of the Internet of Things middle station.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, in particular to an Internet of Things data transmission method and system based on an Internet of Things middle platform. BACKGROUND

[0002] With the rapid development of Internet of Things (IoT) technology, the wide application of various intelligent devices has gradually penetrated into people's daily life and work. Internet of Things devices are connected through wireless networks to realize data exchange and remote control between devices, greatly improving production efficiency and life convenience

[0003] A Chinese patent with publication number CN118869742A discloses a city Internet of Things data processing method and middle platform. The method includes collecting Internet of Things data from different devices and systems, and performing standardized processing on the Internet of Things data. The standardized processing includes data cleaning, conversion, mapping and integration. The Internet of Things data after standardized processing is stored. Based on the standardized processed Internet of Things data, data services are provided for various applications, including data inquiry, data sharing and data visualization. However, the scheme provided by the above-mentioned application relies on centralized supply chain management. The number of sensor node devices is large and the types are diverse. Each sensor node device may include different sensors, actuators, data processing units and communication modules. In this complex distributed environment, it is easy to cause problems such as data transmission delay, network congestion and single point failure. Therefore, it is necessary to provide an Internet of Things data transmission method and system based on an Internet of Things middle platform to improve storage query efficiency. SUMMARY

[0004] Therefore, the present application provides an Internet of Things data transmission method and system based on an Internet of Things middle platform. By using a bidirectional long short-term memory network model combined with an encoded time series matrix and a transmission cycle feature, the appropriate adaptive transmission cycle can be accurately predicted, and the transmission interval is dynamically adjusted so that the sensor network can quickly respond to environmental changes, which helps to improve the real-time monitoring capability and response speed of the Internet of Things middle platform.

[0005] The present application provides an Internet of Things data transmission method based on an Internet of Things middle platform. The method comprises:

[0006] Collecting multi-element perception data based on an Internet of Things perception layer, wherein the Internet of Things perception layer includes a sensor network composed of multiple sensors;

[0007] According to the encoded time series matrix corresponding to the multi-element perception data, the data transmission cycle feature constructed by the Internet of Things middle platform, and the bidirectional long short-term memory network model, an adaptive transmission cycle is obtained;

[0008] The data transmission time interval of the Internet of Things perception layer is dynamically adjusted according to the adaptive transmission period.

[0009] Based on the above technical solutions, preferably, the adaptive transmission period is obtained according to the encoding time sequence matrix corresponding to the multi-element perception data, the data transmission period feature constructed by the Internet of Things middle station, and the bidirectional long short-term memory network model, and specifically includes:

[0010] The encoding time sequence matrix is obtained, and the multi-element time sequence feature is obtained by inputting the encoding time sequence matrix into the bidirectional long short-term memory network model.

[0011] Based on the time period of the two adjacent data transmission processes collected by the Internet of Things middle station, the data transmission period feature is constructed, and the data transmission period feature and the multi-element time sequence feature are input into the bidirectional long short-term memory network model after feature fusion to obtain the optimal adaptive transmission period.

[0012] Based on the above technical solutions, preferably, the encoding time sequence matrix is obtained, and specifically includes:

[0013] The isomorphic Internet of Things sensor set in the Internet of Things perception layer is extracted, and the sequence starting position and initial subsequence length corresponding to the isomorphic Internet of Things sensor set are determined, and the initial encoding time sequence matrix is constructed according to the sequence starting position, the initial subsequence length, and the perception time sequence matrix.

[0014] The initial encoding time sequence matrix is iterated, and the isomorphic Internet of Things sensor set is updated and the processed sensor nodes are removed in each iteration.

[0015] According to the positions of the remaining sensor nodes in the isomorphic Internet of Things sensor set, a plurality of time regularized distance matrices corresponding to the sensor nodes are calculated, and a time regularized distance set is constructed based on the distance metric values of the time regularized distance matrices.

[0016] The time sequence index with the smallest distance metric value is selected from the time regularized distance set, and the time sequence segment corresponding to the time sequence index is merged into the initial encoding time sequence matrix.

[0017] When the isomorphic Internet of Things sensor set is empty, the iteration process is ended and the encoding time sequence matrix is output.

[0018] Further preferably, after the encoding time sequence matrix is obtained, it further includes:

[0019] The time series matrix is input into a multi-head convolutional neural network, a sequence feature vector corresponding to the length of a convolutional filter sequence in the multi-head convolutional neural network is extracted, and the sequence feature vector is fused with an output hidden state vector in the bidirectional long short-term memory network model.

[0020] Further preferably, an expression of a convolution head output function in the multi-head convolutional neural network is:

[0021]

[0022] wherein c h,i represents an output of an i-th feature map of an h-th convolution head, represents an output function of the i-th feature map of the h-th convolution head at a sequence position j, φ represents a moving step of a convolution operation, n1 represents an input time series length covered by a sliding window at each convolution operation, n2 represents a time series length input into the convolution head, represents an attention weight of the i-th feature map of the h-th convolution head at the sequence position j, softmax() represents a normalization function, v T represents a transpose of an attention score weight matrix, tanh() represents a hyperbolic tangent activation function, W represents an input transformation matrix, represents a feature dimension vector contained by each time step in a local time series segment with a length of n1 starting from the sequence position j, U represents a state transformation matrix, h j-1 represents a hidden state vector at the sequence position j, b c represents a bias term corresponding to the attention weight, represents a gating value of the i-th feature map of the h-th convolution head at the sequence position j, σ represents an activation function, [;] represents a vector splicing operation, W g represents a weight matrix corresponding to the gating layer, b g represents a bias term corresponding to the gating layer, represents a residual term of the i-th feature map of the h-th convolution head at the sequence position j, ⊙ represents an element-wise multiplication operation.

[0023] Further preferably, an output feature function expression of the bidirectional long short-term memory network is:

[0024]

[0025] E t = LN(S t + Res(x t ))

[0026] O t = MLP([E t ; F t ])

[0027] wherein, h t denotes the hidden state output function of the bidirectional long short-term memory network at time t, [; ] denotes the vector concatenation operation, denotes the hidden state output vector of the forward long short-term memory network at time t, denotes the hidden state output vector of the backward long short-term memory network at time t, F t denotes the joint feature function, fusion() denotes the operator merging function, H denotes the total number of convolution heads, I denotes the total number of feature maps, S t denotes the feature function after sharing of sensor information, γ denotes the information sharing weight parameter, and ⊙ denotes the element-wise multiplication operation, denotes the output function of the i-th feature map of the h-th convolution head at the sequence position j, E t denotes the feature function after feature enhancement, LN() denotes the layer normalization function, and Res() denotes the residual connection function, O t denotes the output feature function of the bidirectional long short-term memory network, and MLP() denotes the multi-layer perceptron function.

[0028] More preferably, the multi-element perception data collected based on the Internet of Things perception layer further comprises:

[0029] The multi-element perception data is dynamically collected using the Internet of Things perception layer, and the upload data time of each sensor in the Internet of Things perception layer is obtained. The multi-element perception data is received by an edge server, and the multi-element perception data is de-duplicated and time series optimized to obtain a perception time series matrix.

[0030] The perception time series matrix is subjected to event recognition and classification, data aggregation and compression are performed according to the classification results, and the aggregated and compressed perception time series matrix is uploaded to a cloud service layer for decompression storage.

[0031] More preferably, the data transmission time interval of the Internet of Things perception layer is dynamically adjusted according to the adaptive transmission period, and specifically comprises:

[0032] An update instruction including the adaptive transmission period is sent to each sensor node through an Internet of Things middle platform, so that each sensor node dynamically adjusts the data collection interval and the transmission time interval according to the adaptive transmission period.

[0033] In a second aspect of the present application, an Internet of Things data transmission system based on an Internet of Things middle platform is provided, which comprises a data collection module, a period acquisition module and a period adjustment module connected in sequence, wherein,

[0034] The data collection module is used for collecting multi-element perception data based on an Internet of Things perception layer, wherein the Internet of Things perception layer comprises a sensor network composed of multiple sensors.

[0035] The period acquisition module is used for acquiring an adaptive transmission period according to an encoding time sequence matrix corresponding to the multi-element perception data, a data transmission period feature constructed by an Internet of Things middle station, and a bidirectional long short-term memory network model.

[0036] The period adjustment module is used for dynamically adjusting a data transmission time interval of the Internet of Things perception layer according to the adaptive transmission period.

[0037] In a third aspect of the present application, an electronic device is provided, comprising a processor, a memory, a user interface and a network interface, the memory is used for storing instructions, the user interface and the network interface are used for communication with other devices, and the processor is used for executing the instructions stored in the memory.

[0038] The Internet of Things data transmission method and system based on an Internet of Things middle station provided by the present application have the following beneficial effects compared with the prior art:

[0039] (1) The sensor network composed of multiple sensors can comprehensively monitor and collect multi-element perception data in the environment. The multi-element data collected by the perception layer is processed by constructing an encoding time sequence matrix, realizing the ordered integration and standardization of different sensor data. The bidirectional long short-term memory network model is combined with the encoding time sequence matrix and the transmission period feature to accurately predict the adaptive transmission period. According to the adaptive transmission period, the data transmission time interval is dynamically adjusted, unnecessary data transmission is significantly reduced, the network bandwidth occupation is reduced, the energy consumption of the sensor node is saved, the sensor network can quickly respond to environmental changes through dynamic adjustment of the transmission interval, high-frequency data can be obtained at critical moments, and the real-time monitoring capability and response speed of the Internet of Things middle station are improved.

[0040] (2) By extracting the isomorphic Internet of Things sensor set, and determining the corresponding sequence starting position and initial subsequence length, an initial encoding time sequence matrix is constructed, which ensures the ordered integration of different sensor data, helps to reduce data redundancy, avoids data conflict, ensures the accuracy and consistency of the encoding matrix, and in the iteration process, by calculating the time regular distance matrix corresponding to multiple sensor nodes, and constructing a time regular distance set based on the distance metric value, the similarity between different time sequences can be more accurately measured, thereby improving the quality of the encoding matrix, while selecting the time sequence index with the minimum distance metric value from the time regular distance set, and merging the corresponding time sequence segment into the initial encoding time sequence matrix, ensures that the most similar and most representative data segment is integrated first, thereby improving the efficiency and effect of data fusion. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0042] Figure 1 A flowchart of an Internet of Things data transmission method based on an Internet of Things middle station is provided for the present application.

[0043] Figure 2 A framework diagram of a sensor node, a sending node and a collection node is provided for the present application.

[0044] Figure 3 A framework diagram of an Internet of Things data transmission system is provided for the present application.

[0045] Figure 4 A structural diagram of an electronic device is provided for the present application.

[0046] Explanation of reference numerals: 1, Internet of Things data transmission system; 11, data acquisition module; 12, periodic acquisition module; 13, periodic adjustment module; 2, electronic device; 21, processor; 22, communication bus; 23, user interface; 24, network interface; 25, memory. DETAILED DESCRIPTION

[0047] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0048] The present application discloses an Internet of Things data transmission method based on Internet of Things middle platform, referring to Figure 1 The steps of the method include S1-S3.

[0049] Step S1, based on the Internet of Things perception layer, collect multi-element perception data, wherein the Internet of Things perception layer includes a sensor network composed of multiple sensors.

[0050] In the present embodiment, the Internet of Things perception layer includes multiple types of sensors, such as temperature sensors, humidity sensors, light sensors, pressure sensors, gas sensors, motion sensors, etc. These sensors are deployed at different locations according to the specific application scenario requirements to comprehensively monitor environmental parameters. Sensor nodes are distributed in the monitoring area according to the predetermined network topology structure (such as star type, mesh type, tree type, etc.), to ensure the coverage range and redundancy of data collection, and to improve the robustness and reliability of the system.

[0051] Further, each sensor sets a reasonable sampling frequency according to its monitoring object and application requirements. For example, temperature and humidity sensors can sample once a minute, while motion sensors may require a higher sampling frequency to capture rapidly changing motion data. The data collected by each sensor includes continuous analog signals and discrete event data. The data format is unified as digital signals, using standardized data formats (such as JSON, XML, Protobuf, etc.) to facilitate subsequent processing and transmission. The collected raw data is subjected to preliminary filtering and denoising to ensure data accuracy and reliability. For example, a moving average filter is used to eliminate environmental noise interference on temperature sensor data. The sensors are calibrated to correct systematic errors and random errors of the sensors, ensuring data accuracy. For example, gas sensors are calibrated periodically to maintain their measurement accuracy.

[0052] Through the Network Time Protocol (NTP) or a dedicated time synchronization protocol (such as IEEE 1588 Precision Time Protocol), the time synchronization of all sensor nodes is ensured, avoiding data inconsistency caused by clock deviation. Each collected data is stamped with an accurate time stamp, recording the collection time of the data, supporting subsequent data correlation and time series analysis. For example, each temperature data is stamped with a time stamp accurate to milliseconds, to accurately locate in the time series matrix.

[0053] The low-power wireless communication protocol (such as LoRaWAN, Zigbee, Bluetooth Low Energy, etc.) is used to realize the wireless data transmission between the sensor nodes and the edge server, ensuring the stability and low energy consumption of data transmission. In some high-bandwidth or high-reliability scenarios, wired communication protocols (such as Ethernet, RS-485, etc.) can be used to connect sensor nodes and gateway devices. Sensor nodes periodically transmit batch data collected to the edge server, reducing transmission frequency and saving bandwidth. Critical event data (such as abnormally high temperature, dangerous gas leakage, etc.) is immediately sent to the edge server through real-time transmission mechanism to ensure timely response.

[0054] In this step, S11-S12 are also included.

[0055] In step S11, multi-sensor data is dynamically collected using the Internet of Things perception layer, and the upload data time of each sensor in the Internet of Things perception layer is obtained. The multi-sensor data is received by the edge server, and the multi-sensor data is de-duplicated and time series optimized to obtain a perception time series matrix.

[0056] In this step, a message queue system (such as MQTT Broker, Kafka, etc.) is used to receive and buffer data from sensors, and the received data is temporarily stored in the cache (such as memory cache, SSD cache) of the edge server. Using the timestamp information of the data, duplicate or approximately duplicate data is detected and deleted. For example, if the same timestamp produces multiple identical data from a sensor, keep one and delete the rest of the duplicate data. Identify and exclude duplicate data that is completely identical or highly similar, ensuring the uniqueness and accuracy of the data set.

[0057] Further, the de-duplicated multi-sensor data is organized in chronological order to form structured time series data. The data of each sensor is treated as an independent time series dimension and aligned by time. Interpolation algorithms (such as linear interpolation, spline interpolation, etc.) or prediction models (such as moving average, ARIMA, etc.) are used to fill in missing data in the time series. According to the timestamp, the data of different sensors is aligned to a unified time reference to form a unified time interval. For example, all sensor data is aligned to a one-minute time interval. Normalize (such as Min-Max normalization) or standardize (such as Z-score standardization) the time series data to eliminate the dimensional differences of different sensor data. The perception time series matrix is identified and classified for events, and the data is aggregated and compressed according to the classification results, and the aggregated and compressed perception time series matrix is uploaded to the cloud service layer for decompression storage. The optimized time series data is organized into a two-dimensional matrix, where the rows represent time steps and the columns represent data from different sensors.

[0058] In one example, taking the smart agriculture system as an example: the sensor types include soil humidity sensors, temperature sensors, light sensors and gas sensors. The sampling frequency is one sample per minute for soil humidity and temperature, one sample per 10 seconds for the light sensor, and real-time sampling for the gas sensor when an abnormal gas concentration is detected. All sensors are time-synchronized with the edge server through NTP, and the sensor data is in JSON format, with each piece of data containing a sensor ID, data type, value and timestamp. The edge server runs an MQTT Broker, and all sensors publish data to the designated topic through the MQTT protocol, and the edge server subscribes and receives data. The received data is first stored in the Redis cache as temporary storage to improve data processing speed. Using the deduplication function of Redis, based on the sensor ID and timestamp, ensures that there is only one unique data at each time step. Align the soil humidity and temperature data to one per minute, the light data to one per 10 seconds, and fill in the missing data through the interpolation method. Normalize the data of each sensor to the same range to improve the subsequent model processing effect. Organize the soil humidity, temperature, light and gas concentration data once every minute into a 4-column time series matrix, with each row representing the data of a time step. Store the constructed perception time series matrix in the SSD of the edge server, ready to upload to the cloud service layer for further processing.

[0059] Step S2, according to the encoded time series matrix corresponding to the multi-perception data, the data transmission cycle characteristics constructed by the Internet of Things middle station, and the bidirectional long short-term memory network model, an adaptive transmission cycle is obtained.

[0060] In this step, S21-S22 are also included.

[0061] Step S21, obtain the encoded time series matrix, input the encoded time series matrix into the bidirectional long short-term memory network model, and obtain the multi-time series features.

[0062] Further, step S21 further includes S211-S215.

[0063] Step S211, extract the isomorphic Internet of Things sensor set in the Internet of Things perception layer, and determine the sequence starting position and initial subsequence length corresponding to the isomorphic Internet of Things sensor set, construct an initial encoded time series matrix according to the sequence starting position, initial subsequence length and perception time series matrix.

[0064] In this step, the sensor features used to determine isomorphism are determined, including at least sensor type (such as temperature, humidity, pressure), sampling rate, data precision, and communication protocol. A feature vector is generated for each sensor, and the similarity between sensors is calculated using appropriate similarity or distance measurement methods (such as cosine similarity, Euclidean distance). Apply clustering algorithms (such as K-means, hierarchical clustering) or threshold-based classification methods to divide sensors with high similarity into the same isomorphic set to obtain multiple isomorphic sensor sets. The sensors in each set have high similarity in the defined feature dimensions.

[0065] The earliest time point in all isomorphic sensor data is selected as the starting position of the sequence to ensure data synchronization. If there is data missing or abnormal, the first time point with complete data record is selected as the starting position. According to the periodic characteristics of sensor data (such as daily cycle, weekly cycle), the subsequence length covering a complete cycle is selected to capture the trend and pattern of data. In the absence of obvious periodicity, a fixed sliding window length can be set, such as 30 minutes, 1 hour, etc. The specific length is determined according to application requirements and data characteristics.

[0066] According to the determined sequence starting position and initial subsequence length, the corresponding data segments are extracted from the perception time series matrix. This involves cutting the specified time period of data from each isomorphic sensor's data stream.

[0067] The above data processing missing values and outliers to ensure data integrity and consistency. Ensure that the data of all sensors are aligned on the time axis to avoid data inconsistency caused by time offset. The extracted and preprocessed data is arranged according to the sensor set and time sequence to form a new matrix. The rows of the matrix usually represent different isomorphic sensors, and the columns represent time steps. According to specific requirements, data encoding processing may be required, such as normalization, standardization or feature transformation.

[0068] In one example, when storing the signals collected by the sensors, not only the values of non-zero elements are stored, but also the position information corresponding to the sensors is stored. In the local sampling stage, each sensor node samples the signal and determines whether it is a sending node or a resting node with a certain probability. In the data uploading stage, the remaining nodes do not send data, and the sending nodes send the sampled data to the sink point one by one. The sink point collects all the data of the sending nodes, amplifies and forwards it to the cloud service layer, and the cloud service layer reconstructs the signal transmission process in the observation area through optimization algorithms.

[0069] As Figure 2As shown, assuming there are M nodes in the IoT perception layer, m nodes are selected from the IoT perception layer. Each time a new node is set, the new node is connected to the selected m nodes. The number of IoT perception layer nodes generated after the tth moment is t+M, and the number of edges in the IoT perception layer is mt.

[0070] The IoT perception layer clustering coefficient can be expressed as:

[0071]

[0072] Step S212, iteration is performed on the initial encoding time sequence matrix, and the isomorphic IoT sensor set is updated and the processed sensor nodes are removed in each iteration.

[0073] In this step, the initial encoding time sequence matrix constructed in step S211 is used as the starting point of iteration. The isomorphic sensor set initially contains all sensor nodes identified as isomorphic. The processed sensor node list is initially empty and is used to record the processed sensor nodes during iteration.

[0074] A sensor node is selected from the current isomorphic sensor set. The selection strategy can be based on specific requirements, such as polling selection, priority of high priority sensors, etc. The processed data segment is integrated into the initial encoding time sequence matrix. This step ensures that the matrix structure remains consistent and reflects the latest processing results. The currently processed sensor node is removed from the isomorphic sensor set and added to the processed sensor node list. This step avoids the same sensor being repeatedly processed in subsequent iterations. According to the latest processing state, the isomorphic sensor set is re-evaluated and updated, including re-matching features, re-computing sensor similarity, or adjusting isomorphic relationships, to ensure the accuracy and relevance of the isomorphic sensor set.

[0075] Step S213, according to the positions of the remaining sensor nodes in the isomorphic IoT sensor set, a plurality of time regularized distance matrices corresponding to the sensor nodes are calculated, and a time regularized distance set is constructed based on the distance metric values of the time regularized distance matrices.

[0076] In this step, the remaining sensor nodes and their corresponding positions are extracted from the isomorphic sensor set, and the time sequence data of each sensor node in the perception time sequence matrix is obtained. The comparison sensor pair is selected according to the requirements, and the distance between the sensor pairs is calculated. There are usually two ways (calculating the distance between each pair of sensors in the set or comparing the central sensor pair, i.e. selecting one or more central sensors and calculating the distance with other sensors one by one).

[0077] For each pair of selected sensor nodes, the time series data of the two sensor nodes are obtained, the DTW distance between the two time series is calculated, the nonlinear alignment of the time axis is allowed to capture similar patterns, and the calculated DTW distance is stored in the corresponding position matrix. For each pair of sensor nodes, a distance value is generated, and finally a distance matrix is formed. Each element (p, q) in the matrix represents the DTW distance between sensor node p and sensor node q.

[0078] From the generated multiple time warping distance matrices, distance metric values are extracted, and the distance values in all matrices are summarized into a unified set. The statistical characteristics of the distance values are calculated, such as the average distance, the minimum distance, the maximum distance, etc. According to the specific application requirements, meaningful distance metric values are selected into the time warping distance set, such as the minimum distance for identifying the most similar sensor pair; threshold screening, i.e. selecting sensor pairs with distance values below a certain threshold, indicating that they have high similarity; ranking selection to select the top N minimum distance values. The filtered key distance values form the final time warping distance set.

[0079] Step S214, from the time warping distance set, select the time series index with the minimum distance metric value, and merge the time series segment corresponding to the time series index into the initial encoding time series matrix.

[0080] In this step, all distance metric values stored in the time warping distance set are accessed, each distance metric value corresponds to a specific time series index, indicating the similarity of the sequence to the current encoding matrix. The time series index corresponding to the minimum distance metric value is recorded, which identifies the most similar time series segment. According to the selected time series index, the time series segment corresponding to it is extracted from the perception time series matrix. The extracted time series segment is added to the current initial encoding time series matrix, and if there is related data in the encoding matrix, it is ensured that the merging of the new segment will not introduce data conflicts or redundancy. The processed sensor nodes are removed from the homogeneous Internet of Things sensor set to avoid repeated processing in subsequent iterations. Prepare for the next round of iteration to continue selecting the next minimum distance metric value of the time series segment for merging.

[0081] Step S215, when the homogeneous Internet of Things sensor set is empty, end the iteration process and output the encoding time series matrix.

[0082] In this step, the initial encoding time series matrix is constructed by extracting the isomorphic Internet of Things sensor set and determining its corresponding sequence starting position and initial subsequence length, ensuring the ordered integration of different sensor data. This helps to reduce data redundancy, avoid data conflicts, and ensure the accuracy and consistency of the encoding matrix. In the iterative process, the isomorphic Internet of Things sensor set is constantly updated and the processed sensor nodes are removed, ensuring that each iteration can efficiently process the remaining sensor data. This dynamic updating mechanism improves the running efficiency of the algorithm, reduces unnecessary computational overhead, and improves the overall processing speed. By calculating the time regularized distance matrix corresponding to multiple sensor nodes and constructing a time regularized distance set based on the distance metric value, the similarity between different time series can be more accurately measured. This precise similarity analysis helps to effectively fuse subsequent data and improve the quality of the encoding matrix. From the time regularized distance set, the time series index with the smallest distance metric value is selected, and the corresponding time series segment is merged into the initial encoding time series matrix, ensuring that the most similar and representative data segments are integrated first. This process improves the efficiency and effectiveness of data fusion, making the encoding matrix more representative and practical. When the isomorphic Internet of Things sensor set is empty, the iteration process is automatically ended and the encoding time series matrix is output, avoiding unnecessary additional calculations and improving the automation level and running efficiency of the method. At the same time, the output encoding time series matrix is efficient and complete, suitable for subsequent data analysis and application.

[0083] In step S22, based on the time period of the two adjacent data transmission processes collected by the Internet of Things middle station, the data transmission period feature is constructed, and the data transmission period feature and the multivariate time series feature are fused to input the bidirectional long short-term memory network model to obtain the optimal adaptive transmission period.

[0084] In this step, the timestamp of each transmission is extracted from the data transmission record, and the time difference between two adjacent transmissions is calculated as the transmission period feature. The constructed data transmission period feature and the extracted multivariate time series feature are merged to form a comprehensive feature vector. The fusion method can use simple feature concatenation, or more complex feature fusion methods such as weighted fusion, principal component analysis (PCA) dimensionality reduction, etc.

[0085] The data set is divided into training set, validation set and test set according to a certain proportion. The proportion is usually 70% training set, 15% validation set and 15% test set. According to the continuity of time series, the sliding window method is used to slice the long sequence into multiple sub-sequences to adapt to the input requirements of the bidirectional long short-term memory network model.

[0086] The bidirectional long short-term memory network model comprises an input layer, a bidirectional LSTM layer, a full connection layer and an output layer, wherein the input layer is used for receiving the fused feature vector, the shape is (time step, feature dimension), the bidirectional LSTM layer comprises a forward LSTM and a backward LSTM two parts, and is used for capturing context information of a sequence, the full connection layer maps the output of the LSTM layer to a target variable space, that is, a predicted transmission period, and the output layer is used for outputting a predicted transmission period value.

[0087] Further, the encoded time sequence matrix is input into the multi-head convolutional neural network, a sequence feature vector corresponding to a convolutional filter sequence length in the multi-head convolutional neural network is extracted, and the sequence feature vector is fused with a hidden state output vector in the bidirectional long short-term memory network model.

[0088] In the embodiment, since the multi-head convolution allows the model to use different convolutional filters on different "heads", features are extracted from multiple angles, and the expression ability of the model is enhanced. The encoded time sequence matrix is received, and the shape is (sample number, time step, feature number, channel number). Usually, the channel number is 1, indicating a single time sequence feature. The convolutional filter has a plurality of convolution kernels (filters), each filter has a different sequence length (kernel size), the output feature map of the multi-head convolution is flattened into a one-dimensional vector as a subsequent feature vector.

[0089] In this step, the expression of the convolution head output function in the multi-head convolutional neural network is:

[0090]

[0091] wherein c h,i represents the output of the i-th feature map of the h-th convolution head, represents the output function of the i-th feature map of the h-th convolution head at the sequence position j, φ represents the moving step of the convolution operation, n1 represents the length of the input time sequence covered by the sliding window at each convolution operation, and n2 represents the length of the time sequence input into the convolution head, represents the attention weight of the i-th feature map of the h-th convolution head at the sequence position j, softmax() represents a normalization function, and v T represents the transpose of the attention score weight matrix, tanh() represents a hyperbolic tangent activation function, W represents an input transformation matrix, represents a feature dimension vector contained by each time step in a local time sequence segment with a length of n1 starting from the sequence position j, U represents a state transformation matrix, and h j-1 represents the hidden state vector at the sequence position j, b c represents a bias term corresponding to the attention weight, denotes the gating value of the i-th feature map of the h-th convolution head at the sequence position j, σ denotes an activation function, [; ] denotes a vector concatenation operation, W g denotes the weight matrix corresponding to the gating layer, b g denotes the bias term corresponding to the gating layer, denotes the residual term of the i-th feature map of the h-th convolution head at the sequence position j, ⊙ denotes an element-wise multiplication operation.

[0092] Further, the feature extraction function of the bidirectional long short-term memory network is represented as:

[0093]

[0094] wherein, denotes the hidden state output vector of the forward long short-term memory network at time t, LSTM f denotes the forward LSTM unit, denotes the hidden state output vector of the backward long short-term memory network at time t, LSTM f denotes the backward LSTM unit.

[0095]

[0096] E t = LN(S t + Res(x t ))

[0097] The output feature function expression of the bidirectional long short-term memory network is:

[0098] O t = MLP([E t ; F t ])

[0099] wherein, h t denotes the hidden state output function of the bidirectional long short-term memory network at time t, [; ] denotes a vector concatenation operation, denotes the hidden state output vector of the forward long short-term memory network at time t, denotes the hidden state output vector of the backward long short-term memory network at time t, F t denotes the joint feature function, fusion() denotes an operator merging function, H denotes the total number of convolution heads, I denotes the total number of feature maps, S t denotes the feature function after sharing of sensor information, γ denotes an information sharing weight parameter, ⊙ denotes an element-wise multiplication operation, denotes the output function of the i-th feature map of the h-th convolution head at the sequence position j, E t denotes the feature function after feature enhancement, LN() denotes a layer normalization function, Res() denotes a residual connection function, Ot where denotes the output feature function of the bidirectional long short-term memory network, and MLP() denotes a multi-layer perceptron function.

[0100] At step S3, the data transmission time interval of the perception layer of the Internet of Things is dynamically adjusted according to the adaptive transmission period.

[0101] In this embodiment, the Internet of Things middle platform sends an update instruction to each sensor node, and the update instruction includes an adaptive transmission period, so that each sensor node dynamically adjusts the data collection interval and transmission time interval according to the adaptive transmission period.

[0102] In one example, a smart agricultural system uses Internet of Things (IoT) technology to monitor and manage farmland environments. The smart agricultural system includes an Internet of Things middle platform and multiple sensor nodes distributed in different farmland areas. These sensors are used to monitor key parameters such as soil moisture, temperature, and light intensity in real time. According to environmental changes and data analysis results, the Internet of Things middle platform needs to dynamically adjust the data collection and transmission intervals of each sensor node to optimize resource usage and improve data real-time performance.

[0103] The instruction adopts a standardized data format, such as JSON or XML, to ensure compatibility and parsability between the middle platform and sensor nodes. The core content of the instruction includes the adaptive transmission period, i.e., the new data collection and transmission time interval. In addition, other configuration parameters such as data collection accuracy and transmission priority can also be included as needed. For example, the update instruction may contain a command type (such as "UPDATE_TRANSMISSION_INTERVAL") and a payload section that specifies the new transmission interval time (e.g., 600 seconds, i.e., 10 minutes).

[0104] Further, MQTT (Message Queue Telemetry Transport protocol) is selected as the communication protocol because of its lightweight and publish / subscribe mode, which is very suitable for resource-constrained sensor nodes. MQTT supports topic filtering, making it easy to selectively send instructions. In addition, other optional protocols include CoAP (Constrained Application Protocol) and HTTP / HTTPS, but according to the resources and network environment of the sensor nodes, MQTT is usually the better choice.

[0105] The Internet of Things middle platform generates new adaptive transmission period parameters based on the collected data and pre-set adaptive algorithms (such as using a bidirectional long short-term memory network model for prediction). The middle platform encapsulates these parameters in a pre-defined instruction format and sends them to the target sensor nodes through the selected communication protocol (such as MQTT). For example, if the system detects that the soil moisture in a certain area does not change much, the middle platform may decide to adjust the transmission interval from every 5 minutes to every 10 minutes to reduce data transmission frequency and save bandwidth and power consumption.

[0106] At the same time, identity verification and authorization mechanisms (such as username / password or TLS certificates) are used to ensure that only authorized intermediaries can send instructions to sensor nodes. At the same time, data encryption protocols (such as TLS / SSL) are used to protect data in transit, preventing unauthorized parties from intercepting or tampering with instructions. Sensor nodes have the ability to receive and process update instructions. Specifically, sensor nodes will continuously listen to instruction topics from intermediaries, and once they receive an update instruction, they will parse the instruction content, extract new transmission interval parameters, and update internal configurations to adjust the time interval of data collection and transmission. For example, when a sensor node receives an instruction to update the transmission interval to 600 seconds, it will immediately adjust its internal timer to reduce the frequency of data collection and transmission to once every 10 minutes.

[0107] After successfully receiving and applying the update instruction, the sensor node will send an acknowledgment message to the intermediary, indicating that the instruction has been correctly processed. If the instruction processing fails, the node will also send a failure confirmation message with the corresponding error information. For example, after successfully updating the transmission interval, the sensor node will send a confirmation message containing the "success" status and the new transmission interval time to the intermediary. If an error occurs during the update process, the node will send a confirmation message containing the "failure" status and error details.

[0108] Further, the Internet of Things intermediary will record the time of each instruction sending, target node and confirmation status. If there is a sending failure or no confirmation is received, the intermediary can resend the instruction or mark the problem node for subsequent processing. Sensor nodes should also record logs of receiving and executing instructions, including instruction content, reception time, and execution status. Based on the feedback from sensor nodes and the system running situation, the Internet of Things intermediary can dynamically adjust the transmission period strategy. For example, if the intermediary receives frequent feedback from sensor nodes in a certain area that transmission has failed, it may adjust the transmission interval or optimize network configuration. In addition, the intermediary can periodically evaluate the effectiveness of the adaptive transmission period, and continuously optimize the transmission strategy and algorithm model according to actual data and application requirements, to achieve more efficient resource utilization and more real-time data monitoring.

[0109] By dynamically adjusting the data transmission interval according to actual needs, unnecessary frequent transmission is avoided, effectively reducing the occupation of network bandwidth, improving the overall network efficiency, and the adaptive transmission period can be adjusted in real time according to network conditions, preventing data congestion during network load peaks, ensuring smoothness and stability of data transmission. Dynamic adjustment of transmission interval means that sensor devices can enter a low-power state when they do not need to transmit data frequently, significantly reducing battery consumption and prolonging the service life of the device. Through synchronization control and priority mechanism, timely transmission of important data is ensured, reducing data loss due to data conflicts or network fluctuations, and improving the reliability of data transmission. Time series matrix optimization of multi-sensor data improves the degree of data structuring, facilitating subsequent event identification and classification, and improving data processing accuracy. Using a bidirectional long short-term memory network (Bi-LSTM) model, the optimal transmission period is predicted based on historical data and real-time network conditions, enabling the system to adapt to environmental changes and demand changes autonomously. The support of multiple sensor types and adaptive adjustment mechanism make the system highly flexible and able to adapt to different application scenarios and changes in demand.

[0110] Based on the above method, an Internet of Things data transmission system based on Internet of Things middle platform, referring to Figure 3 , the Internet of Things data transmission system 1 includes a data acquisition module 11, a cycle acquisition module 12, and a cycle adjustment module 13, wherein,

[0111] The data acquisition module 11 is used to acquire multi-sensor data based on the Internet of Things perception layer, wherein the Internet of Things perception layer includes a sensor network composed of multiple sensors;

[0112] The cycle acquisition module 12 is used to acquire an adaptive transmission period according to the encoded time series matrix corresponding to the multi-sensor data, the data transmission cycle characteristics constructed by the Internet of Things middle platform, and the bidirectional long short-term memory network model;

[0113] The cycle adjustment module 13 is used to dynamically adjust the data transmission time interval of the Internet of Things perception layer according to the adaptive transmission period.

[0114] In one example, the cycle acquisition module 12 is used to acquire an encoded time series matrix, input the encoded time series matrix into a bidirectional long short-term memory network model, acquire multi-time series features, construct data transmission cycle characteristics based on the time period of two adjacent data transmission processes collected by the Internet of Things middle platform, and input the data transmission cycle characteristics and multi-time series features into the bidirectional long short-term memory network model after feature fusion to obtain the optimal adaptive transmission period.

[0115] In one example, the periodicity acquisition module 12 is configured to extract a set of homogeneous IoT sensors in the IoT perception layer, determine a sequence start position and an initial subsequence length corresponding to the set of homogeneous IoT sensors, construct an initial encoded time series matrix according to the sequence start position, the initial subsequence length, and the perception time series matrix, iterate the initial encoded time series matrix, and update the set of homogeneous IoT sensors and remove processed sensor nodes in each iteration, calculate a plurality of time warping distance matrices corresponding to the sensor nodes according to the remaining sensor node positions in the set of homogeneous IoT sensors, construct a time warping distance set based on distance metric values of the time warping distance matrices, select a time series index with a minimum distance metric value from the time warping distance set, merge a time series segment corresponding to the time series index into the initial encoded time series matrix, and end the iteration process and output an encoded time series matrix when the set of homogeneous IoT sensors is empty.

[0116] In one example, the periodicity acquisition module 12 is configured to input the encoded time series matrix into a multi-head convolutional neural network, extract a sequence feature vector corresponding to a convolutional filter sequence length in the multi-head convolutional neural network, and perform feature fusion between the sequence feature vector and a hidden state output vector in a bidirectional long short-term memory network model.

[0117] In one example, the expression of the convolution head output function in the multi-head convolutional neural network is as follows:

[0118]

[0119] wherein c h,i represents the output of the i-th feature map of the h-th convolution head, represents the output function of the i-th feature map of the h-th convolution head at the sequence position j, φ represents the moving step of the convolution operation, n1 represents the length of the input time series covered by the sliding window at each convolution operation, and n2 represents the length of the time series input into the convolution head, represents the attention weight of the i-th feature map of the h-th convolution head at the sequence position j, and softmax() represents a normalization function, T represents the transpose of the attention score weight matrix, tanh() represents a hyperbolic tangent activation function, and W represents an input transformation matrix, represents the feature dimension vector contained in each time step in the local time series segment with a length of n1 starting from the sequence position j, U represents a state transformation matrix, and h j-1 represents the hidden state vector at the sequence position j, and b c represents the bias term corresponding to the attention weight, denotes the gating value of the i-th feature map of the h-th convolution head at the sequence position j, σ denotes an activation function, [; ] denotes a vector concatenation operation, W g denotes the weight matrix corresponding to the gating layer, b g denotes the bias term corresponding to the gating layer, denotes the residual term of the i-th feature map of the h-th convolution head at the sequence position j, ⊙ denotes an element-wise multiplication operation.

[0120] In one example, the output feature function of the bidirectional long short-term memory network is expressed as:

[0121]

[0122] E t = LN(S t + Res(x t ))

[0123] O t = MLP([E t ; F t ])

[0124] where h t denotes the hidden state output function of the bidirectional long short-term memory network at time t, [; ] denotes a vector concatenation operation, denotes the hidden state output vector of the forward long short-term memory network at time t, denotes the hidden state output vector of the backward long short-term memory network at time t, F t denotes the joint feature function, fusion() denotes an operator merging function, H denotes the total number of convolution heads, I denotes the total number of feature maps, S t denotes the feature function after sharing sensor information, γ denotes an information sharing weight parameter, ⊙ denotes an element-wise multiplication operation, denotes the output function of the i-th feature map of the h-th convolution head at the sequence position j, E t denotes the feature function after feature enhancement, LN() denotes a layer normalization function, Res() denotes a residual connection function, O t denotes the output feature function of the bidirectional long short-term memory network, MLP() denotes a multi-layer perceptron function.

[0125] In one example, the data collection module 11 is configured to dynamically collect multi-sensor data using the Internet of Things perception layer, obtain the upload data time of each sensor in the Internet of Things perception layer, receive the multi-sensor data through the edge server, and perform deduplication and time series optimization on the multi-sensor data to obtain a perception time series matrix; perform event identification and classification on the perception time series matrix, aggregate and compress the perception time series matrix according to the classification result, and upload the aggregated and compressed perception time series matrix to the cloud service layer for decompression storage.

[0126] In one example, the periodic adjustment module 13 is configured to send an update instruction to each sensor node through the Internet of Things middle platform, the update instruction including an adaptive transmission period, so that each sensor node dynamically adjusts the data collection interval and the transmission time interval according to the adaptive transmission period.

[0127] Please refer to Figure 4 , the embodiment of the present application provides a structural schematic diagram of an electronic device. As shown in Figure 4 , the electronic device 2 can include at least one processor 21, at least one network interface 24, a user interface 23, a memory 25, and at least one communication bus 22.

[0128] The communication bus 22 is configured to realize the connection and communication among the components.

[0129] The user interface 23 can include a display screen (Display), a camera (Camera), and can also include a standard wired interface and a wireless interface.

[0130] The network interface 24 can optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0131] The processor 21 can include one or more processing cores. The processor 21 connects various parts within the server through various interfaces and lines, performs various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 25, and calling data stored in the memory 25. Alternatively, the processor 21 can be implemented in at least one of a hardware form of a digital signal processing (DSP), a field-programmable gate array (FPGA), and a programmable logic array (PLA). The processor 21 can be integrated with a combination of one or more of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes operating systems, user interfaces, and application programs; the GPU is responsible for rendering and drawing the content required to be displayed on the display screen; and the modem is used for processing wireless communication. It can be understood that the above-mentioned modem can also not be integrated into the processor 21, but can be realized by a separate chip.

[0132] The memory 25 can include a random access memory (RAM) and a read-only memory (ROM). Alternatively, the memory 25 includes a non-transitory computer-readable storage medium. The memory 25 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 25 can include a program storage area and a data storage area, wherein the program storage area can store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area can store data involved in the above-mentioned various method embodiments, etc. The memory 25 can also be at least one storage device located away from the aforementioned processor 21. As shown, the memory 25 as a computer storage medium can include an operating system, a network communication module, a user interface module, and an application program based on the Internet of Things data transmission method of the Internet of Things middle station. Figure 4 As shown, the memory 25 as a computer storage medium can include an operating system, a network communication module, a user interface module, and an application program based on the Internet of Things data transmission method of the Internet of Things middle station.

[0133] In Figure 4In the electronic device 2 shown, the user interface 23 is mainly used to provide an interface for the user to input, and obtain data input by the user; and the processor 21 can be used to invoke an application program stored in the memory 25 and based on the Internet of Things data transmission method of the Internet of Things middle station, which, when executed by one or more processors, causes the electronic device to perform one or more methods as described in the above embodiments.

[0134] A computer-readable storage medium stores instructions that, when executed by one or more processors, cause a computer to perform one or more methods as described in the above embodiments.

[0135] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all described as a series of action combinations, but those skilled in the art should know that the present application is not limited to the order of the actions described, because according to the present application, certain steps can be performed in other order or at the same time. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0136] In the above embodiments, the description of each embodiment is focused on, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0137] In several embodiments provided in the present application, it should be understood that the disclosed device can be implemented by other means. For example, the device embodiments described above are only schematic, and the division of units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed each other can be indirect coupling or communication connection through some service interface, device or unit, which can be electrical or other forms.

[0138] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0139] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The above integrated unit can be realized in the form of hardware or in the form of software functional unit.

[0140] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable memory. Based on such understanding, the technical solutions of the present application essentially or say the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a memory and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the embodiments of the present application. The aforementioned memory includes: a U disk, a mobile hard disk, a magnetic disk or an optical disk and various program code storage media.

[0141] The above is only an exemplary embodiment of the present disclosure, and cannot limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure. Other embodiments of the present disclosure will be readily apparent to those skilled in the art upon considering the specification and practicing the true principles of the disclosure. The present application is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or conventional techniques in the art that are not described in the present disclosure.

[0142] The above is only a preferred embodiment of the present application and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. An IoT data transmission method based on an IoT middleware platform, characterized in that, The method includes: The Internet of Things (IoT) sensing layer collects multi-sensor data, wherein the IoT sensing layer includes a sensor network composed of multiple sensors. Based on the encoded time series matrix corresponding to the multi-sensor data, the data transmission cycle characteristics constructed by the IoT platform, and the bidirectional long short-term memory network model, the adaptive transmission cycle is obtained. The step of obtaining the adaptive transmission period based on the encoded time series matrix corresponding to the multi-dimensional sensing data, the data transmission cycle characteristics constructed by the IoT platform, and the bidirectional long short-term memory network model specifically includes: Obtain the encoded time series matrix, and input the encoded time series matrix into a bidirectional long short-term memory network model to obtain multivariate time series features; Based on the time period of two adjacent data transmission processes collected by the IoT platform, a data transmission period feature is constructed. The data transmission period feature and the multivariate time series feature are then fused and input into the bidirectional long short-term memory network model to obtain the optimal adaptive transmission period. The acquisition of the encoded time series matrix specifically includes: Extract the homogeneous IoT sensor set from the IoT perception layer, and determine the sequence start position and initial subsequence length corresponding to the homogeneous IoT sensor set. Based on the sequence start position, the initial subsequence length, and the perception time series matrix, construct an initial encoding time series matrix. The initial encoded time series matrix is ​​iterated, and the homogeneous IoT sensor set is updated and the processed sensor nodes are removed in each iteration; Based on the locations of the remaining sensor nodes in the isomorphic IoT sensor set, calculate multiple time-warped distance matrices corresponding to the sensor nodes, and construct a time-warped distance set based on the distance metric values ​​of the time-warped distance matrices. Select the time series index with the minimum distance metric value from the time warp distance set, and merge the time series segment corresponding to the time series index into the initial encoded time series matrix; When the homogeneous IoT sensor set is empty, the iteration process ends and the encoded time series matrix is ​​output. The data transmission time interval of the IoT sensing layer is dynamically adjusted according to the adaptive transmission cycle.

2. The IoT data transmission method based on an IoT middleware platform as described in claim 1, characterized in that, After obtaining the encoded time series matrix, the process further includes: The encoded time series matrix is ​​input into a multi-head convolutional neural network, and the sequence feature vector corresponding to the sequence length of the convolutional filter in the multi-head convolutional neural network is extracted. The sequence feature vector is then fused with the hidden state output vector in the bidirectional long short-term memory network model.

3. The IoT data transmission method based on an IoT middleware platform as described in claim 2, characterized in that, The expression for the output function of the convolution head in the multi-head convolutional neural network is as follows: Among them, c h,i This represents the output of the i-th feature map of the h-th convolutional head. Let represent the output function of the i-th feature map of the h-th convolutional head at sequence position j, φ represent the stride of the convolution operation, n1 represent the length of the input time series covered by the sliding window during each convolution operation, and n2 represent the length of the input convolutional head's time series. V represents the attention weight of the i-th feature map of the h-th convolutional head at sequence position j, where softmax() represents the normalization function. T The expression represents the transpose of the attention score weight matrix, tanh() represents the hyperbolic tangent activation function, and W represents the input transformation matrix. Let U represent the feature dimension vector contained in each time step of a local time series segment starting from position j and of length n1, and let h represent the state transition matrix. j-1 Let b represent the hidden state vector at position j in the sequence. c This represents the bias term corresponding to the attention weights. Let σ represent the gating value of the i-th feature map of the h-th convolutional head at sequence position j, σ represent the activation function, and [;] represent the vector concatenation operation. g b represents the weight matrix corresponding to the gated layer. g This represents the bias term corresponding to the gating layer. Let represent the residual term of the i-th feature map of the h-th convolutional head at sequence position j, and ⊙ represent the element-wise multiplication operation.

4. The IoT data transmission method based on an IoT middleware platform as described in claim 1, characterized in that, The output feature function expression of the bidirectional long short-term memory network is: E t =LN(S t +Res(x t )) O t <MLP([E t F t ]) Among them, h t Let represent the hidden state output function of the bidirectional long short-term memory network at time t, and [;] denote the vector concatenation operation. Let represent the hidden state output vector of the feedforward long short-term memory network at time t. F represents the hidden state output vector of the feedforward long short-term memory network at time t. t The fusion() operator represents the joint feature function, H represents the total number of convolution heads, I represents the total number of feature maps, and S represents the total number of feature maps. t This represents the characteristic function after sensor information sharing, where γ represents the information sharing weight parameter, and ⊙ represents the element-wise multiplication operation. E represents the output function of the i-th feature map of the h-th convolutional head at sequence position j. t LN() represents the feature function after feature enhancement, Res() represents the layer normalization function, and O represents the residual connection function. t Let MLP() represent the output feature function of the bidirectional long short-term memory network.

5. The IoT data transmission method based on an IoT middleware platform as described in claim 1, characterized in that, The collection of multi-sensor data based on the Internet of Things sensing layer also includes: The IoT sensing layer is used to dynamically collect multi-sensor data and obtain the upload time of each sensor in the IoT sensing layer. The multi-sensor data is received through the edge server, and the multi-sensor data is deduplicated and optimized for time series to obtain a sensing time series matrix. The perceived time series matrix is ​​subjected to event identification and classification. Based on the classification results, the data is aggregated and compressed, and the aggregated and compressed perceived time series matrix is ​​uploaded to the cloud service layer for decompression and storage.

6. The IoT data transmission method based on an IoT middleware platform as described in claim 5, characterized in that, The step of dynamically adjusting the data transmission time interval of the IoT sensing layer according to the adaptive transmission period specifically includes: The IoT platform sends update instructions to each sensor node, and the update instructions include the adaptive transmission period, so that each sensor node can dynamically adjust the data acquisition interval and transmission time interval according to the adaptive transmission period.

7. An IoT data transmission system based on an IoT middleware platform, characterized in that, The IoT data transmission system (1) includes a data acquisition module (11), a period acquisition module (12), and a period adjustment module (13) connected in sequence. The data acquisition module (11) is used to acquire multi-sensor data based on the Internet of Things (IoT) sensing layer, wherein the IoT sensing layer includes a sensor network composed of multiple sensors. The cycle acquisition module (12) is used to acquire an adaptive transmission cycle based on the encoded time series matrix corresponding to the multi-dimensional sensing data, the data transmission cycle characteristics constructed by the Internet of Things platform, and the bidirectional long short-term memory network model. The step of obtaining the adaptive transmission period based on the encoded time series matrix corresponding to the multi-dimensional sensing data, the data transmission cycle characteristics constructed by the IoT platform, and the bidirectional long short-term memory network model specifically includes: Obtain the encoded time series matrix, and input the encoded time series matrix into a bidirectional long short-term memory network model to obtain multivariate time series features; Based on the time period of two adjacent data transmission processes collected by the IoT platform, a data transmission period feature is constructed. The data transmission period feature and the multivariate time series feature are then fused and input into the bidirectional long short-term memory network model to obtain the optimal adaptive transmission period. The acquisition of the encoded time series matrix specifically includes: Extract the homogeneous IoT sensor set from the IoT perception layer, and determine the sequence start position and initial subsequence length corresponding to the homogeneous IoT sensor set. Based on the sequence start position, the initial subsequence length, and the perception time series matrix, construct an initial encoding time series matrix. The initial encoded time series matrix is ​​iterated, and the homogeneous IoT sensor set is updated and the processed sensor nodes are removed in each iteration; Based on the locations of the remaining sensor nodes in the isomorphic IoT sensor set, calculate multiple time-warped distance matrices corresponding to the sensor nodes, and construct a time-warped distance set based on the distance metric values ​​of the time-warped distance matrices. Select the time series index with the minimum distance metric value from the time warp distance set, and merge the time series segment corresponding to the time series index into the initial encoded time series matrix; When the homogeneous IoT sensor set is empty, the iteration process ends and the encoded time series matrix is ​​output. The period adjustment module (13) is used to dynamically adjust the data transmission time interval of the Internet of Things sensing layer according to the adaptive transmission period.

8. An electronic device, characterized in that, The device includes a processor (21), a memory (25), a user interface (23), and a network interface (24). The memory (25) is used to store instructions. The user interface (23) and the network interface (24) are used to communicate with other devices. The processor (21) is used to execute the instructions stored in the memory (25) to cause the electronic device (2) to perform the method as described in any one of claims 1-6.

Images