An IoT Adaptive Networking Method and System for Business Format Energy Analysis

Abstract

This invention relates to the field of business format energy analysis technology, and provides an IoT adaptive networking method and system for business format energy analysis. First, it acquires business format characteristic data and corresponding energy consumption data for different IoT business formats, and establishes a prediction model based on the acquired data. Then, it establishes a networking strategy generation model based on the prediction results of the prediction model to generate energy optimization directions. When a business format changes, the business format characteristic data is input into the prediction model to obtain the subsequent predicted energy consumption data for the corresponding business format. A corresponding networking strategy is generated based on the prediction data, and the IoT performs adaptive networking according to the networking strategy. This invention establishes a prediction model, inputs business format characteristic data into the model for prediction when business format demands change, solves the impact of system response lag, and generates networking strategies based on the prediction results, meeting the energy optimization needs of IoT adaptive networking under various business formats.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industry energy analysis, specifically a method and system for adaptive networking of industry energy analysis based on the Internet of Things. BACKGROUND

[0002] Industry energy analysis is a comprehensive energy evaluation and insight into the Internet of Things scenario under specific business forms. It covers the energy required for the operation of Internet of Things devices, energy consumption during data transmission, and in-depth research on energy supply sources and sustainability in different industries.

[0003] Adaptive networking of the Internet of Things is an advanced networking method that can dynamically adjust network structure and parameters according to environmental changes and business needs. It uses advanced sensor technology to perceive changes in the physical environment, network environment, and business needs, and automatically adjusts parameters to adapt to new conditions based on these perceptions. It can intelligently adjust networking strategies according to different business needs and environmental conditions, thereby reducing energy consumption.

[0004] The essence of industry energy analysis is to conduct in-depth analysis of the energy of Internet of Things devices under different industries, including energy required for device operation, data transmission energy consumption, and sustainability of energy supply. Based on this, adaptive networking of the Internet of Things can dynamically adjust network structure and parameters according to specific industry energy conditions, improving network efficiency and reliability.

[0005] For example, during periods of low business demand, unnecessary device connections and data transmission can be automatically reduced to save energy. At the same time, it can also allocate network resources reasonably according to energy supply conditions to ensure that the network can still operate stably under limited energy conditions. In addition, by sensing environmental changes, adaptive networking can adjust transmission power and other parameters in a timely manner to avoid energy waste and improve energy utilization efficiency. With this networking method, it can better adapt to complex and variable Internet of Things application scenarios, prolong the service life of Internet of Things devices, and reduce operating costs. In practical applications, these not only optimize the performance of Internet of Things systems in different business scenarios, but also reduce energy consumption and improve energy utilization efficiency, providing more stable, efficient, and sustainable network support for Internet of Things applications in various industries, promoting the widespread application and development of Internet of Things technology in different fields.

[0006] However, the existing technology still has the following shortcomings:

[0007] 1. The real-time performance of business energy analysis needs further improvement. Current analysis methods may not be able to accurately capture real-time energy changes in complex and ever-changing business scenarios. Insufficient real-time performance makes it difficult to detect energy changes in a timely manner when setting up IoT networks. This results in a certain lag in the response of adaptive networking, which prevents devices from switching to low-energy mode or shutting down non-critical functions in a timely manner, affecting the normal use of devices. Furthermore, it can lead to IoT devices suddenly losing power or experiencing performance degradation during critical business moments due to inaccurate energy analysis, thereby affecting the normal use of related IoT devices.

[0008] 2. Current business energy analysis often fails to accurately analyze the dynamic energy demand changes of IoT devices under different business scenarios. This makes it difficult for the networking strategy generated based on such analysis to adapt to the fluctuations in energy demand during actual operation of the devices in a timely manner. As a result, the energy optimization is not accurate and efficient enough, and cannot fully meet the energy optimization needs of IoT adaptive networking under various complex business scenarios. Summary of the Invention

[0009] To address the aforementioned technical problems, this invention provides an IoT adaptive networking method and system based on business energy analysis. This method predicts IoT business energy through business energy analysis, generates an IoT adaptive networking strategy based on energy optimization directions, and then performs IoT adaptive networking accordingly, thereby resolving the problems in the prior art.

[0010] An IoT adaptive networking method based on business activity analysis, comprising the following steps:

[0011] S1: Obtain energy information of IoT business formats, and use sensor networks to obtain business format characteristic data of different IoT business formats, as well as energy consumption data of IoT devices in the corresponding business formats. Among them, energy consumption data includes the consumption data of IoT devices themselves and energy transmission consumption data.

[0012] S2: Establish an energy consumption data prediction model. This model can predict energy consumption based on the acquired business characteristic data of different business types and the energy consumption data of IoT devices in the corresponding business types, and generate energy consumption prediction results.

[0013] S3: Establish an IoT adaptive networking strategy generation model. This model can generate an IoT adaptive networking strategy with energy optimization direction based on the energy consumption prediction results of the energy consumption data prediction model.

[0014] S4: Internet of Things adaptive networking optimization, when the format demand changes, input the changed format characteristic data into the energy consumption data prediction model to obtain the subsequent predicted energy consumption data under the corresponding format demand, input the predicted energy consumption data into the Internet of Things adaptive networking strategy generation model to generate the corresponding networking strategy, and then the Internet of Things adjusts the networking strategy according to the newly generated networking strategy to cope with the change of format demand.

[0015] Preferably, in S1, the acquisition of Internet of Things format energy information specifically includes the following steps:

[0016] S11: acquiring format characteristic data of different formats of Internet of Things by using a sensor network;

[0017] S12: acquiring energy consumption data of Internet of Things devices corresponding to different formats of Internet of Things;

[0018] S13: classifying the acquired data into format characteristic data sets according to different formats, each format characteristic data set including corresponding format characteristic data and energy consumption data of Internet of Things devices under the corresponding format.

[0019] Preferably, in S2, the establishment of the energy consumption data prediction model specifically includes the following steps:

[0020] S21: data pre-setting, combining and arranging the collected energy consumption data of Internet of Things devices in time sequence, where t represents a certain moment, each piece of collected energy consumption data of Internet of Things devices corresponds to t to form a set (t), i and j represent different Internet of Things devices, and i x j represents that the Internet of Things device i is connected with the Internet of Things device j; the energy consumption generated by each i x j connection is calculated and predicted;

[0021] S22: model construction, extracting and inputting corresponding energy consumption data according to time sequence:

[0022] S221: data initialization, randomly initializing weight W, bias term b, prediction weight ω and prediction bias β; the initial value C of prediction hidden number h and memory warehouse is initialized as 0 when the data is first calculated;

[0023] S222: information filtering calculation, information filtering determines which information can be deleted from the memory warehouse:

[0024]

[0025] wherein, represents the dot product operation of vectors, f t represents the result of judging whether the input information, X t is to be filtered; h t-1The implicit value for prediction at time t-1 is the information carrier for predicting the business energy value at time t. W and b represent the weight and bias terms, respectively. σ represents the logistic function, and tanh is the hyperbolic tangent function. This is achieved by analyzing the current input information X. t The hidden number h predicted at the previous time step t-1 The processing determines the degree of forgetting of information in the cell state at the previous moment, thereby achieving dynamic adjustment of memory;

[0026] S223: Information update calculation and memory repository update value calculation:

[0027]

[0028] Information updates determine which new information is stored in the memory repository. First, i t Based on the current input X t The hidden number h predicted at the previous time step t-1 Through weight W i The logic function σ determines which of the new input features are important energy information. This provides the data foundation for updating the memory repository values ​​to subsequent updates to the memory repository;

[0029] S224: Update memory repository:

[0030]

[0031] S225: Information Extraction Calculation:

[0032]

[0033] h t =o t ×tanh(C t )

[0034] Among them, C t C is the memory repository at time t. t Think of it as a vector, containing information or elements of many dimensions, o t We use weights, biases, and a memory repository to determine which information is currently needed, i.e., the predicted hidden number h. t Predicting hidden number h t o t With memory warehouse C t The result of multiplying after processing with the hyperbolic tangent function means that o t It will affect the memory repository C t Scaling is applied to each element if o t The smaller the value, the better the prediction of the hidden number h for the corresponding element in the memory repository. tThe smaller the contribution, the easier it is to extract. By judging the size of the contribution, we can determine which information is needed at present. The larger the contribution, the easier it is to be extracted.

[0035] S226: Repeat S221-S224 continuously until all t in set(t) have been calculated;

[0036] S227: Finally, use the predicted implicit number h t Conduct business format energy prediction:

[0037]

[0038] Where E represents the predicted energy consumption data. This refers to the predicted energy consumption data of ij at time t+1, X t ω represents the business characteristic data input at time t; ω and β are the prediction weights and prediction biases used in the prediction calculation.

[0039] S23: Model optimization and adjustment, determine the loss function to measure the difference between the predicted value and the actual value, and select the stochastic gradient descent algorithm to update the model parameters.

[0040] Preferably, in S23, the specific process for model optimization and adjustment is as follows:

[0041] S231: Define the loss function, and choose the mean squared error or mean absolute error loss function to measure the difference between the predicted value and the actual value;

[0042] S232: Select the stochastic gradient descent algorithm as the optimization algorithm;

[0043] S233: Divide the dataset by dividing the actual value data into a training dataset using stratified sampling.

[0044] S234: Model training involves multiple iterations of training on the training dataset. In each iteration, the predicted value of energy consumption data is calculated through forward propagation, the loss is calculated based on the loss function, the gradient is calculated through backpropagation, and the model parameters are updated using stochastic gradient descent.

[0045] S235: Parameter tuning. The validation set is used to tune the model parameters. Grid search or random search methods are used to find the parameter combination that minimizes the model result. The parameters include weights, bias terms, prediction weights, and prediction biases. The data with the smallest loss function value is selected as the model parameters.

[0046] Preferably, in S3, the step of establishing the IoT adaptive networking strategy generation model is as follows:

[0047] S31: Let i be the predicted energy consumption data at time t when i and j are connected, and let the connection method between i and j be k, and let R be... k R is an indicator function, indicating that when i and j are connected using connection method k. k It is 1 if it is true, otherwise it is 0.

[0048] S32: Construct device connection relationships. Based on the functional characteristics and spatial location relationships of the devices, construct a series of device connection relationships and calculate their total energy consumption, considering only the energy consumption between the actually connected devices and the energy consumption of the transmission method used.

[0049] For each device connection combination i×j, the total energy consumption E 总 for:

[0050]

[0051] Where p is the total number of IoT devices in the connection combination, and q is the total number of connection methods k;

[0052] Connection stability constraint: For any k, if the connection method k1 is changed to connection combination k, the stability constraint is determined by the following condition: . Adjustments will only be accepted if the adjusted connection is more stable.

[0053]

[0054] Where μ represents the stability of the connection method, which is determined by the specific connection method;

[0055] S33: Gradually optimize the connection relationships, compare the total energy consumption values ​​of different hypothetical connection combinations, and recalculate the total energy consumption after each combination. If the new connection combination reduces the total energy consumption, accept the adjustment; otherwise, continue to try other adjustments.

[0056] Preferably, in S4, the specific steps of the IoT adaptive networking optimization are as follows:

[0057] S41: When business requirements change, determine the changed business requirements;

[0058] S42: Input the determined business demand into the energy consumption data prediction model in real time, and use the long-term and short-term prediction models to predict the energy consumption data for the subsequent business demand.

[0059] S43: Input the predicted energy consumption data into the IoT adaptive networking strategy generation model to generate the corresponding networking strategy;

[0060] S44: Based on the generated networking strategy, the IoT automatically adjusts the connection relationship between IoT devices, and each IoT device communicates or connects according to the new strategy to achieve adaptive networking.

[0061] An IoT adaptive networking system for business energy analysis includes:

[0062] The business data acquisition module uses a sensor network to acquire business characteristic data under different business formats of the Internet of Things, as well as energy consumption data of IoT devices under the corresponding business formats, including the IoT device's own consumption data and energy transmission consumption data.

[0063] The prediction model building module, which is connected to the business data acquisition module, includes a data preprocessing unit, a model building unit, and a model training unit. The data preprocessing unit cleans and denoises the business characteristic data and energy consumption data under different business conditions provided by the business data acquisition module. The model building unit establishes an energy consumption data prediction model based on the business characteristic data and energy consumption data under different business conditions provided by the business data acquisition module. The model training unit is used to train the built model and adjust the module parameters.

[0064] A network strategy model building module is connected to a prediction model building module. Based on the energy consumption data output by the prediction model building module, an IoT adaptive network strategy with energy optimization direction is generated according to the predicted energy consumption results.

[0065] An adaptive networking module is connected to the prediction model building module and the networking strategy model building module. By monitoring the operating status and data of IoT devices in real time, the module inputs the monitored IoT-related data into the prediction model building module and adjusts the networking of IoT devices according to the networking strategy provided by the networking strategy generation module. The specific execution means include adjusting the connection method to achieve the optimized networking, thereby reducing energy consumption.

[0066] Compared with the prior art, the present invention has the following beneficial effects:

[0067] 1. This invention acquires real-time business characteristic data and equipment energy consumption data under different IoT business formats, and establishes an energy consumption data prediction model. When business demand changes, it can quickly input business characteristic data into the model for prediction, realizing a rapid response to business energy analysis. When business demand changes, it can quickly input business characteristic data into the prediction model to obtain subsequent predicted energy consumption data; this greatly reduces the time lag of system response. Through this timely prediction mechanism, it effectively solves the adverse effects caused by system lag and ensures the normal use of relevant IoT devices.

[0068] 2. This invention establishes an IoT adaptive networking strategy generation model, which generates networking strategies based on accurate energy consumption data prediction results. This model fully considers the business characteristics and energy consumption under different business formats, and can adapt to the fluctuations in energy demand during actual equipment operation in a timely manner. It achieves the accuracy and efficiency of energy optimization, meets the energy optimization needs of IoT adaptive networking under various complex business formats, and solves the problems of inaccurate and inefficient energy optimization in traditional analysis methods. Attached Figure Description

[0069] Figure 1 This is a flowchart of the method steps of the present invention;

[0070] Figure 2 This is a schematic diagram illustrating the specific calculation process in step S22 of the present invention;

[0071] Figure 3 This is a schematic diagram of the specific process of step S23 of the present invention;

[0072] Figure 4 This is a flowchart of the system workflow of the present invention. Detailed Implementation

[0073] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and should not be construed as limiting the scope of the invention.

[0074] This invention provides an IoT adaptive networking method for business energy analysis, the specific steps of which are as follows:

[0075] S1: Obtain energy information of IoT business formats, and use sensor networks to obtain business format characteristic data of different IoT business formats, as well as energy consumption data of IoT devices in the corresponding business formats. Among them, energy consumption data includes the consumption data of IoT devices themselves and energy transmission consumption data.

[0076] S2: Establish an energy consumption data prediction model. This model can predict energy consumption based on the acquired business characteristic data of different business types and the energy consumption data of IoT devices in the corresponding business types, and generate energy consumption prediction results.

[0077] S3: Establish an IoT adaptive networking strategy generation model. This model can generate an IoT adaptive networking strategy with energy optimization direction based on the energy consumption prediction results of the energy consumption data prediction model.

[0078] S4: IoT Adaptive Networking Optimization. When business demand changes, the changed business characteristic data is input into the energy consumption data prediction model to obtain the subsequent predicted energy consumption data under the corresponding business demand. Based on the predicted energy consumption data, the corresponding networking strategy is generated in the IoT adaptive networking strategy generation model. Then, the IoT adjusts itself according to the newly generated networking strategy to cope with the changes in business demand.

[0079] An IoT adaptive networking system for business energy analysis includes:

[0080] The business data acquisition module uses a sensor network to acquire business characteristic data under different business formats of the Internet of Things, as well as energy consumption data of IoT devices under the corresponding business formats, including the IoT device's own consumption data and energy transmission consumption data.

[0081] The prediction model building module, which is connected to the business data acquisition module, includes a data preprocessing unit, a model building unit, and a model training unit. The data preprocessing unit cleans and denoises the business characteristic data and energy consumption data under different business conditions provided by the business data acquisition module. The model building unit establishes an energy consumption data prediction model based on the business characteristic data and energy consumption data under different business conditions provided by the business data acquisition module. The model training unit is used to train the built model and adjust the module parameters.

[0082] A network strategy model building module is connected to a prediction model building module. Based on the energy consumption data output by the prediction model building module, an IoT adaptive network strategy with energy optimization direction is generated according to the predicted energy consumption results.

[0083] An adaptive networking module is connected to the prediction model building module and the networking strategy model building module. By monitoring the operating status and data of IoT devices in real time, the module inputs the monitored IoT-related data into the prediction model building module and adjusts the networking of IoT devices according to the networking strategy provided by the networking strategy generation module. The specific execution means include adjusting the connection method to achieve the optimized networking, thereby reducing energy consumption.

[0084] Example 1:

[0085] like Figures 1-4 As shown in this embodiment, there is an Internet of Things (IoT) system for a smart factory. The production area has various sensors and devices. Different production needs represent different business models. An IoT networking strategy is needed to optimize energy, maintain stable factory operation, and reduce energy consumption.

[0086] S1, acquire energy information for IoT business models, and utilize sensor networks to obtain business model characteristic data under different IoT business models. For example, operational status data of IoT devices under different production conditions;

[0087] Acquire energy consumption data of IoT devices in the corresponding business sectors; energy consumption data includes the energy consumption data of the IoT devices themselves and the energy transmission consumption data.

[0088] The acquired data is classified into business type feature datasets according to different business types. Each business type feature dataset includes corresponding business type feature data and energy consumption data of IoT devices under the corresponding business type.

[0089] The data is preprocessed by using a sliding window-based median filtering algorithm to remove noise and a box plot method to remove outliers.

[0090] S2. Establish a prediction model. Based on the acquired business characteristic data of different business formats and the energy consumption data of IoT devices in the corresponding business formats, establish an energy consumption data prediction model.

[0091] First, data is preset by combining and arranging the collected energy consumption data of IoT devices according to time series. Let t represent a certain moment, and all the collected t constitute a set (t). i and j represent different IoT devices, and i×j means that IoT device i is connected to IoT device j. The energy consumption generated by each i×j connection is predicted and calculated.

[0092] Model construction involves extracting and inputting feature vectors based on time series data.

[0093] The following calculations are performed at different times t: information filtering, information updating, and memory warehouse update values ​​are calculated by inputting feature vectors and the predicted latent number of t; then the memory warehouse is updated and information is extracted to obtain the predicted latent number at the corresponding time; finally, the predicted latent number is used to predict the energy consumption data at t+1.

[0094] Randomly initialize the data with weights W, bias term b, prediction weight ω, and prediction bias β; the prediction latent number h and the initial value C of the memory store are initialized to 0 when the data is first calculated.

[0095] Information filtering calculations determine which information can be deleted from the memory repository:

[0096]

[0097] in, f represents the dot product operation of vectors. t This indicates that for the input information, X t Determine whether the result should be filtered; h t-1The implicit value for prediction at time t-1 is the information carrier for predicting the business energy value at time t. W and b represent the weight and bias terms, respectively. σ represents the logistic function, and tanh is the hyperbolic tangent function. This is achieved by analyzing the current input information X. t The hidden number h predicted at the previous time step t-1 The processing determines the degree of forgetting of information in the cell state at the previous moment, thereby achieving dynamic adjustment of memory;

[0098] Information update calculation and memory repository update value calculation:

[0099]

[0100] Information updates determine which new information is stored in the memory repository. First, i t Based on the current input X t The hidden number h predicted at the previous time step t-1 Through weight W i The logic function σ determines which of the new input features contain the desired energy information. This provides the data foundation for updating the memory repository values ​​to subsequent updates to the memory repository;

[0101] Update memory repository:

[0102]

[0103] Information extraction calculation:

[0104]

[0105] h t =o t ×tanh(C t )

[0106] Among them, C t C is the memory repository at time t. t Think of it as a vector, containing information or elements of many dimensions, o t We use weights, biases, and a memory repository to determine which information is currently needed, i.e., the predicted hidden number h. t Predicting hidden number h t o t With memory warehouse C t The result of multiplying after processing with the hyperbolic tangent function means that o t It will affect the memory repository C t Scaling is applied to each element if o t The smaller the value, the better the prediction of the hidden number h for the corresponding element in the memory repository. t The smaller the contribution, the easier it is to extract. By judging the size of the contribution, we can determine which information is needed at present. The larger the contribution, the easier it is to be extracted.

[0107] Repeat steps S221-S224 until all values ​​of t in set(t) have been calculated.

[0108] Then use the predicted implicit number h t Conduct business format energy prediction:

[0109]

[0110] Where E represents the predicted energy consumption data. This refers to the predicted energy consumption data of ij at time t+1, X t ω represents the business characteristic data input at time t; ω and β are the prediction weights and prediction biases used in the prediction calculation.

[0111] Define a loss function, and choose either mean squared error or mean absolute error loss function to measure the difference between the predicted value and the actual value;

[0112] The stochastic gradient descent algorithm was selected as the optimization algorithm.

[0113] Divide the dataset by dividing the actual value data into a training dataset using stratified sampling.

[0114] Model training involves multiple iterations on the training dataset. In each iteration, the predicted value of energy consumption data is calculated through forward propagation, the loss is calculated based on the loss function, the gradient is calculated through backpropagation, and the model parameters are updated using stochastic gradient descent.

[0115] Parameter tuning: The validation set is used to tune the model parameters. Grid search or random search methods are used to find the parameter combination that minimizes the model result. The parameters include weights, bias terms, prediction weights, and prediction biases. The data with the smallest loss function value is selected as the model parameters.

[0116] By establishing a predictive model, we can predict the energy consumption of different equipment in the future based on historical data and current business characteristics data. This way, when production demand changes, we can input the changed business characteristics data into the predictive model and obtain the predicted data on how energy consumption changes over time under the changed business conditions.

[0117] This provides a basis for adjusting the networking strategy of IoT devices in advance, avoiding energy waste and production instability caused by system lag. For example, when a factory receives a new order and production demands are about to change, the predictive model can predict the trend of energy consumption changes in the equipment in advance, thereby adjusting the networking strategy in a timely manner and ensuring smooth production.

[0118] S3: Establish an IoT adaptive networking strategy generation model. The model can generate an IoT adaptive networking strategy with energy optimization direction based on the energy consumption prediction results of the prediction model.

[0119] set up This represents the predicted energy consumption data at time t when i and j are connected. The connection method between i and j is set to k, and R is set to... k R is an indicator function, indicating that when i and j are connected using connection method k. k It is 1 if it is true, otherwise it is 0.

[0120] Construct device connection relationships. Based on the functional characteristics and spatial location relationships of the devices, construct a series of device connection relationships and calculate their total energy consumption, considering only the energy consumption between the actually connected devices and the energy consumption of the transmission method used.

[0121] For each device connection combination i×j, the total energy consumption E 总 for:

[0122]

[0123] Where p is the total number of IoT devices in the connection combination, and q is the total number of connection methods k;

[0124] Connection stability constraint: For any k, if the connection method k1 is changed to connection combination k, the stability constraint is determined by the following condition: . Adjustments will only be accepted if the adjusted connection is more stable.

[0125]

[0126] Where μ represents the stability of the connection method, which is determined by the specific connection method;

[0127] The connection relationships are gradually optimized, and the total energy consumption values ​​of different hypothetical connection combinations are compared. After each combination, the total energy consumption is recalculated. If the new connection combination reduces the total energy consumption, the adjustment is accepted; otherwise, other adjustments are tried.

[0128] An adaptive networking strategy generation model is established to continuously optimize device connections and find the optimal networking scheme. The total energy consumption of different hypothetical connection combinations is compared, and the total energy consumption is recalculated after each adjustment to ensure that the new connection combination reduces energy consumption. This stepwise optimization method can minimize system energy consumption and improve energy efficiency without affecting production. Simultaneously, the networking strategy generated based on the predictive model results can be dynamically adjusted according to different business needs, ensuring that the system is always in an optimal energy consumption state.

[0129] Example 2:

[0130] like Figures 1-4 As shown in this embodiment, the IoT system of the smart factory in Embodiment 1 needs to adapt to the IoT devices under different business formats in the factory due to market changes and business format requirements, so as to optimize energy, maintain the stable operation of the factory and reduce energy consumption.

[0131] For example, due to market changes, the production scale needs to be reduced, and some large equipment may not be fully utilized for production. Therefore, it is necessary to change the transmission method or adjust the equipment layout to reduce energy loss.

[0132] In Example 1, a complete prediction model has already been established based on historical data, therefore:

[0133] S41: When business needs change, determine the changed business needs. For example, in a smart factory, it may shift from large-scale standardized production to small-batch customized production.

[0134] S42: Input the determined business type demand into the long-term and short-term forecasting models in real time. The long-term and short-term forecasting models predict the energy consumption data for the subsequent business type demand. Input the determined business type characteristic data into the long-term and short-term forecasting models. Based on historical data and current business type demand, the models predict the energy consumption data for the subsequent business type demand. For example, in the small-batch customized production of smart factories, predict the energy consumption of different equipment in different connection methods.

[0135] S43: Generate the corresponding networking strategy based on the predicted energy consumption data;

[0136] S44: Based on the generated networking strategy, the IoT automatically adjusts the connection relationship between IoT devices, and each IoT device communicates or connects according to the new strategy to achieve adaptive networking. For example, in a smart factory, the connection of some devices is turned off or adjusted according to the networking strategy.

[0137] This method enables better collaboration between different IoT devices. For example, on a smart factory production line, when the business model changes and it is predicted that the energy consumption of a certain link's equipment may increase, the working status of the connected equipment can be adjusted, making the energy distribution of the entire production line more reasonable. This not only reduces overall energy consumption but also improves production efficiency.

[0138] The embodiments of the present invention are given for the purposes of illustration and description. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. An industry format energy analysis Internet of Things adaptive networking method, characterized in that, The specific steps are as follows: S1: Obtain the energy information of the Internet of Things industry, use the sensor network to obtain the industry characteristic data under different industries of the Internet of Things, and the energy consumption data of the Internet of Things equipment under the corresponding industry, wherein the energy consumption data includes the self-consumption data of the Internet of Things equipment and the energy transmission consumption data; S2: Establish an energy consumption data prediction model, which can predict the energy consumption according to the obtained industry characteristic data under different industries and the energy consumption data of the Internet of Things equipment under the corresponding industry, and generate an energy consumption prediction result; S3: Establish an Internet of Things adaptive networking strategy generation model, which can generate an energy optimization direction Internet of Things adaptive networking strategy according to the energy consumption prediction result of the energy consumption data prediction model; S4: Internet of Things adaptive networking optimization, when the industry demand changes, input the changed industry characteristic data into the energy consumption data prediction model to obtain the subsequent predicted energy consumption data under the corresponding industry demand, input the predicted energy consumption data into the Internet of Things adaptive networking strategy generation model to generate the corresponding networking strategy, and then the Internet of Things adjusts according to the newly generated networking strategy to cope with the change of industry demand; The specific steps of establishing the energy consumption data prediction model in S2 are as follows: S21: Data pre-setting, combine and arrange the collected energy consumption data of the Internet of Things equipment in time sequence, set t to represent a certain moment, and the t corresponding to each piece of collected energy consumption data of the Internet of Things equipment forms a set T, i and j represent different Internet of Things equipment, and i x j represents the connection between the Internet of Things equipment i and the Internet of Things equipment j; the energy consumption generated by each i x j connection is calculated; S22: Model construction, extract and input the corresponding energy consumption data according to the time sequence; S23: Model optimization and adjustment, determine the loss function to measure the difference between the predicted value and the actual value, and select the stochastic gradient descent algorithm to update the model parameters; The steps of establishing the Internet of Things adaptive networking strategy generation model in S3 are as follows: S31: predicte d energy consumption data at time t when i and j are connected, while the connection mode of i and j is set as k, set is an indicator function, when i and j are connected using connection mode k is 1, otherwise 0; S32: Construct the device connection relationship, construct a series of device connection relationships according to the functional characteristics and spatial position relationship of the device, calculate the total energy consumption, and only consider the energy consumption between the actually connected devices and the energy consumption of the transmission mode used; For each device connection combination i x j, the total energy consumption is: ； Wherein p is the total number of Internet of Things equipment in the connection combination, and q is the total number of connection modes k; Connection stability constraint: for any k, if the connection mode is adjusted to connection combination only if the adjusted connection is more stable, accept the adjustment; S33: Gradually optimize the connection relationship, compare the total energy consumption values of different assumed connection combinations, calculate the total energy consumption after each combination, and if the new connection combination reduces the total energy consumption, accept the adjustment; otherwise, continue to try other adjustments.

2. The industry format energy analysis Internet of Things adaptive networking method of claim 1, wherein: In S1, the specific steps of obtaining the energy information of the Internet of Things industry are as follows: S11: Use the sensor network to obtain the industry characteristic data under different industries of the Internet of Things; S12: Obtain the energy consumption data of the Internet of Things equipment corresponding to different industries of the Internet of Things; S13: The acquired data is classified into industry format characteristic data sets according to different industry formats, and each industry format characteristic data set includes corresponding industry format characteristic data and Internet of Things device energy consumption data under the corresponding industry format.

3. The Internet of Things adaptive networking method for industry format energy analysis according to claim 1, characterized in that: S221: Data initialization, randomly initialize weights W, bias terms b, prediction weights and prediction bias ; prediction hidden number and initial value of memory warehouse initialized to 0 at the beginning of data calculation; S222: Information filtering calculation, information filtering determines which information can be deleted from the memory warehouse: ； wherein, denotes the dot product operation of vectors, denotes the processing of input information to determine whether it needs to be filtered or not; is the predicted hidden number at t-1, which is the information transmission carrier of the predicted format energy value at t, denotes a logical function, which determines the degree of information forgetting in the cell state at the last moment by processing the current input information and the predicted hidden number at the last moment, to realize dynamic adjustment of memory; S223: Information update calculation and memory warehouse update value calculation: ； ； The information update determines which new information is stored into the memory bank. First, Based on the current input And the previous time prediction hidden number Through the weight And the logic function Judge which is the important energy information in the new input feature, The memory bank update value provides data basis for subsequent memory bank update, and tanh is the hyperbolic tangent function. S224: Update the memory warehouse: ； S225: Information extraction calculation: ； ； in, For the memory repository at time t, It can be understood as a vector, containing information or elements of many dimensions. We use weights, biases, and a memory repository to determine which information is currently needed, i.e., to predict hidden numbers. Predicting hidden numbers for With memory warehouse The result of multiplying after processing with the hyperbolic tangent function means Will affect the memory warehouse Scaling each element if The smaller the value, the better the prediction of the hidden number by the corresponding element in the memory repository. The greater the contribution, the smaller the information will be; by judging the size of the contribution, we can determine which information is needed at present. The greater the contribution, the easier it is to be extracted. S226: Repeat S221-S224 continuously until the calculation of all t in the set T is completed; S227: Finally use the predicted hidden number Make a format energy prediction: ； wherein, represents predicted energy consumption data, i.e. the i x j energy consumption data predicted at t+1; and are the prediction weight and the prediction bias used for the prediction calculation.

4. The industry format energy analysis Internet of Things adaptive networking method of claim 1, wherein: In S23, the model optimization adjustment specific process is as follows: S231: Define the loss function, select the mean square error or average absolute error loss function to measure the difference between the predicted value and the actual value; S232: Determine the stochastic gradient descent algorithm as the optimization algorithm; S233: Divide the data set, divide the actual value data into a training data set according to the stratified sampling division method; S234: Model training, perform multiple iterations on the training data set, in each iteration, calculate the energy consumption data predicted value through forward propagation, then calculate the loss according to the loss function, and then calculate the gradient through the back propagation algorithm, and update the model parameters using the stochastic gradient descent algorithm; S235: Parameter adjustment, the validation set is used to adjust the parameters of the model, and the grid search or random search method is used to find the parameter combination that minimizes the model result, the parameters include weight, bias term, prediction weight and prediction bias, and the data with the minimum loss function value is selected as the model parameter.

5. The industry format energy analysis Internet of Things adaptive networking method of claim 1, wherein: In S4, the specific steps of the Internet of Things adaptive networking optimization are as follows: S41: When the industry format demand changes, determine the changed industry format demand; S42: Real-time input the determined industry format demand into the energy consumption data prediction model to predict the subsequent predicted energy consumption data under the corresponding industry format demand through the long-short term prediction model; S43: Input the predicted energy consumption data into the Internet of Things adaptive networking strategy generation model to generate the corresponding networking strategy; S44: According to the generated networking strategy, the Internet of Things automatically adjusts the connection relationship between the Internet of Things devices, and each Internet of Things device communicates or connects according to the new strategy, realizing adaptive networking.

6. An industry energy analysis Internet of Things adaptive networking system applying the method of claim 1, characterized in that, It includes: An industry format data acquisition module that acquires industry format characteristic data under different industry formats of the Internet of Things and energy consumption data of Internet of Things devices under the corresponding industry format through a sensor network, including self-consumption data and energy transmission consumption data of the Internet of Things devices; The prediction model is connected with the industry format data acquisition module, and comprises a data preprocessing unit, a model building unit and a model training unit. The data preprocessing unit cleans and denoises the industry format characteristic data and energy consumption data under different industry formats provided by the industry format data acquisition module. The model building unit establishes an energy consumption data prediction model according to the industry format characteristic data and energy consumption data under different industry formats provided by the industry format data acquisition module. The model training unit is used for training the built model and adjusting module parameters. The networking strategy model building module is connected with the prediction model building module, generates an energy optimization direction Internet of Things adaptive networking strategy according to the predicted energy consumption result based on the energy consumption data prediction model output by the prediction model building module. The adaptive networking module is connected with the prediction model building module and the networking strategy model building module, inputs the monitored Internet of Things related data into the prediction model building module by monitoring the running state and data of the Internet of Things equipment in real time, and adjusts the networking of the Internet of Things equipment according to the networking strategy provided by the networking strategy generation module. The specific execution means comprises adjusting the connection mode to realize the optimized networking, so as to reduce the energy consumption.

Images