Highway load prediction regulation and control method and system based on low-power-consumption Internet of Things

Abstract

The invention provides an expressway load prediction regulation and control method and system based on a low-power-consumption internet of things, and relates to the technical field of expressway load prediction, and the method comprises the steps: obtaining multi-modal data; constructing a multi-modal data structure by using the preprocessed multi-modal data, inputting the multi-modal data structure into a multi-modal cooperative load prediction model, extracting local multi-scale features by using a parallel multi-scale TCN network, extracting global features by using a Transform encoder and a multi-head self-attention mechanism, and fusing the local multi-scale features and the global features to obtain a multi-modal cooperative load prediction model; obtaining a future load prediction value; and constructing a highway load regulation and control model based on the load predicted value and the multi-modal data, constructing a target function by taking the lowest operation cost, the lowest carbon emission and the highest energy efficiency as targets, and solving the target function by adopting a differential evolution algorithm to obtain an optimal scheduling strategy of the highway load regulation and control model. And performing adaptive regulation and control on the expressway power grid by using the optimal scheduling strategy.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of highway load prediction, in particular to a highway load prediction and regulation method and system based on low-power Internet of Things. BACKGROUND

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] With the increase of highway traffic volume, the energy demand of facilities along the line is becoming increasingly complex, especially on devices such as traffic signals, lighting systems, toll stations, and electric vehicle charging piles. The traditional energy management mode is difficult to effectively respond to the time-varying and burstiness of these loads, often leading to energy waste and uneven load problems. Existing load control systems rely on centralized scheduling, which has problems such as response lag, low scheduling efficiency, and energy waste, and does not fully consider complex dynamic environmental factors (such as traffic flow, weather changes, etc.).

[0004] The emergence of Internet of Things technology, especially Low-Power Wide-Area Network (LPWAN) technology, provides a new solution for intelligent control of highway load side. Low-power Internet of Things devices have the advantages of long-distance communication, low power consumption, and real-time data collection, and can be deployed with sensors along the highway to realize real-time monitoring of various facilities. Through these devices, information such as device operating status, energy consumption data, and environmental changes can be collected to provide real-time data support for subsequent load scheduling.

[0005] However, existing Internet of Things applications are mostly focused on data collection and monitoring, lacking efficient intelligent scheduling and real-time response mechanisms, making it difficult to respond to highway load fluctuations and emergencies. Currently, load-side monitoring systems along highways generally use Internet of Things technology for data collection and remote monitoring. Common methods include: (1) LoRaWAN or NB-IoT-based wireless sensor network, deploying low-power sensor nodes on load devices (such as lighting, signal control, toll stations), and uploading operating parameters to the central platform through low-rate wireless links. The advantage is that it is easy to set up, low in power consumption, and has a wide coverage. (2) Fiber or Ethernet-based wired monitoring system, deploying monitoring devices at key nodes and transmitting data back to the dispatch center in real time through wired links. The advantage is stable communication and high bandwidth. (3) Hybrid communication mode, core nodes use wired direct connection, and edge nodes use wireless relay to core nodes and then back.

[0006] However, the above existing solutions still have the following problems: (1) Insufficient data fusion capability: Most systems only collect single-type data (such as power load), lacking comprehensive processing capability for multi-source information such as traffic flow and weather conditions.

[0007] (2) Lack of real-time performance: Low-power wireless communication has limited bandwidth, which cannot support high-frequency and large-scale data reporting, resulting in data delay and information lag.

[0008] (3) Low network robustness: Most wireless networks use a single central gateway, which is vulnerable to single-point failures and lacks self-healing capabilities.

[0009] (4) Lack of control functions: Existing Internet of Things systems are mainly used for state monitoring and abnormal alarm, and lack dynamic control mechanisms based on prediction and optimization.

[0010] (5) Poor environmental adaptability: In extreme weather, high traffic, and other conditions, network performance and data collection accuracy can easily decline. SUMMARY

[0011] To solve the above problems, the present disclosure proposes a highway load prediction and control method and system based on low-power Internet of Things, which combines the load prediction model of low-power Internet of Things with the load scheduling algorithm, through multi-modal data collaborative load prediction and multi-objective optimization of load scheduling, realizes real-time prediction, dynamic scheduling and optimization control of highway load, to improve energy utilization efficiency, reduce energy waste, and ensure the safety and stability of system operation.

[0012] According to some embodiments, the present disclosure adopts the following technical solutions: The highway load prediction and control method based on low-power Internet of Things comprises: Obtaining multi-modal data of electricity, environment and traffic of the highway and preprocessing; Fusing the preprocessed multi-modal data in multiple dimensions to construct a multi-modal data structure, inputting the multi-modal data structure into a multi-modal collaborative load prediction model, extracting local multi-scale features using a parallel multi-scale TCN network, extracting global features using a Transformer encoder and a multi-head self-attention mechanism, fusing the local multi-scale features and the global features, and predicting the output using the fused features to obtain the future load prediction value; Based on the load prediction value and the multi-modal data, a highway load control model is constructed, a target function is constructed with the lowest operating cost, the lowest carbon emission and the highest energy efficiency as the target, a differential evolution algorithm is used to solve the target function, the optimal scheduling strategy of the highway load control model is obtained, and the optimal scheduling strategy is used to adaptively control the highway power grid.

[0013] According to some embodiments, the present disclosure adopts the following technical solutions: The highway load prediction and control system based on low-power Internet of Things comprises: The data acquisition module is used to acquire multimodal data on power, environment, and traffic of the highway and to perform preprocessing. The load forecasting module is used to fuse preprocessed multimodal data in multiple dimensions to construct a multimodal data structure. The multimodal data structure is then input into the multimodal collaborative load forecasting model. A parallel multi-scale TCN network is used to extract local multi-scale features, and a Transformer encoder and a multi-head self-attention mechanism are used to extract global features. The local multi-scale features are fused with the global features, and the fused features are used to predict the output to obtain the future load forecast value. The scheduling optimization and adaptive control module is used to construct a highway load control model based on load forecasts and multimodal data. It constructs an objective function with the goals of minimizing operating costs, carbon emissions, and energy efficiency, and solves the objective function using a differential evolution algorithm to obtain the optimal scheduling strategy for the highway load control model. The optimal scheduling strategy is then used to adaptively control the highway power grid.

[0014] According to some embodiments, the present disclosure adopts the following technical solutions: A computer program product includes a computer program that, when executed by a processor, implements the aforementioned highway load prediction and control method based on low-power Internet of Things.

[0015] According to some embodiments, the present disclosure adopts the following technical solutions: A non-transitory computer-readable storage medium is provided for storing computer instructions, which, when executed by a processor, implement the aforementioned highway load prediction and control method based on low-power Internet of Things.

[0016] According to some embodiments, the present disclosure adopts the following technical solutions: An electronic device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the highway load prediction and control method based on low-power Internet of Things.

[0017] Compared with the prior art, the beneficial effects of this disclosure are as follows: The highway load prediction and regulation method based on low-power Internet of Things of the present disclosure can provide accurate data support for intelligent control by deploying low-power Internet of Things sensors to collect real-time highway power consumption load, running state and surrounding environmental information (such as traffic flow, weather, etc.). The collected data is analyzed using a multi-modal collaborative load prediction model to predict the load trend in real time. Based on load demand, energy supply and external environmental factors, energy resources are intelligently dispatched, power supply schemes are optimized, energy load is balanced, and energy waste is reduced. The optimal scheduling strategy is generated based on the differential evolution algorithm, and the optimal scheduling strategy is used to control the power grid, adjust the energy storage system and load distribution strategy, and achieve load balancing, energy efficiency optimization and maximum utilization of renewable energy.

[0018] The highway load prediction and regulation method based on low-power Internet of Things of the present disclosure constructs a multi-modal collaborative load prediction model, extracts local multi-scale features using a parallel multi-scale TCN network, extracts global features using a Transformer encoder and a multi-head self-attention mechanism, fuses local multi-scale features and global features, and uses fused features to predict output to obtain future load prediction values. Real-time response to load fluctuations, energy efficiency optimization and renewable energy utilization rate improvement are achieved by combining scheduling strategies.

[0019] The highway load prediction and regulation method based on low-power Internet of Things of the present disclosure constructs a highway load regulation model based on load prediction values and multi-modal data, constructs an objective function with the lowest operating cost, the lowest carbon emissions and the highest energy efficiency as the target, and solves the objective function using a differential evolution algorithm. Through this optimization process, the method can achieve dynamic balance in cost, emissions and energy efficiency, and maintain stable convergence between global search and local optimization. Finally, the optimal highway load regulation scheme is output to achieve the coordinated improvement of economy, greenness and efficiency.

[0020] The highway load prediction and regulation method based on low-power Internet of Things of the present disclosure constructs an adaptive regulation mechanism through real-time monitoring and data analysis, which can adjust the load scheduling strategy according to real-time monitoring data and historical load prediction conditions. In the control process, a combination of fuzzy control and PID control is used to further improve scheduling accuracy and system stability. The fuzzy controller dynamically adjusts the charging and discharging strategy of the energy storage device through fuzzy rule reasoning combined with real-time load fluctuation, reduces energy waste and improves grid stability. PID control is used for fine-grained adjustment to achieve rapid response and load adjustment, ensuring smooth and safe operation of the power grid. The present disclosure can quickly respond to sudden load changes or equipment failures, automatically adjust the power supply strategy, and ensure the safe and stable operation of the system. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the disclosure, illustrate exemplary embodiments of the disclosure, and to explain the disclosure without limiting the disclosure thereto.

[0022] Figure 1 A flow chart of a highway load prediction and regulation method based on a low-power Internet of Things according to an embodiment of the disclosure. DETAILED DESCRIPTION

[0023] The disclosure will be further described below with reference to the drawings and embodiments.

[0024] It should be noted that the following detailed description is illustrative only, and is intended to provide further description of the disclosure. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the disclosure belongs.

[0025] It should be noted that the terms used herein are only intended to describe specific embodiments, and are not intended to limit the exemplary embodiments according to the disclosure. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and it should also be understood that when the terms "comprise" and / or "include" are used in the specification, there is a presence of a feature, step, operation, device, component, and / or combinations thereof.

[0026] Embodiment 1 In an embodiment of the disclosure, a highway load prediction and regulation method based on a low-power Internet of Things is provided, comprising the following steps: step one: obtaining power, environmental, and traffic multi-modal data of the highway and pre-processing the data; Step two: multi-dimensionally fusing the pre-processed multi-modal data, constructing a multi-modal data structure, inputting the multi-modal data structure into a multi-modal collaborative load prediction model, extracting local multi-scale features using a parallel multi-scale TCN network, extracting global features using a Transformer encoder and a multi-head self-attention mechanism, fusing the local multi-scale features and the global features, predicting an output using the fused features, and obtaining a future load prediction value; Step three: based on the load prediction value and the multi-modal data, constructing a highway load regulation model, constructing an objective function with the lowest operating cost, the lowest carbon emission, and the highest energy efficiency as the target, solving the objective function using a differential evolution algorithm, obtaining an optimal scheduling strategy of the highway load regulation model, and adaptively regulating the highway power grid using the optimal scheduling strategy.

[0027] As an embodiment, the low-power Internet of Things-based highway load prediction and regulation method of the present disclosure realizes real-time monitoring, dynamic scheduling and optimization control of highway load by combining low-power Internet of Things technology and intelligent scheduling algorithm, so as to improve energy utilization efficiency, reduce energy waste, and ensure the safety and stability of system operation. The specific implementation process is as follows: Step 1: Obtain the multi-modal data of power, environment and traffic of the highway and perform preprocessing; Specifically, real-time monitoring of electrical parameters such as voltage, current and power is performed through voltage and current sensors and other power load sensors, environmental factors such as temperature, humidity, wind speed and air pressure are collected through environmental monitoring sensors (temperature and humidity sensors, wind speed sensors, etc.), and data such as vehicle flow, density and road load are collected through traffic flow sensors.

[0028] The sampling frequency of all sensors is adjusted by period and load characteristics, and a dynamic sampling strategy is adopted to reduce data redundancy and energy consumption.

[0029] Further, the collected raw data is denoised, time-synchronized and data-interpolated to ensure the integrity and accuracy of the data. A Kalman filter is used to smooth the sensor data and eliminate transient fluctuations and measurement errors.

[0030] Step 2: Multi-dimensional fusion of preprocessed multi-modal data to construct a multi-modal data structure; Specifically, the preprocessed multi-modal data includes data of three modalities, namely power dimension, environment dimension and traffic dimension; The specific data of the power dimension includes: three-phase voltage / current (V a ,V b ,V c ), frequency f t , harmonic content H t , total active / reactive power P t ,Q t .

[0031] The specific data of the environment dimension includes: temperature and humidity (T t ,H t ), wind speed / direction (WS t ,WD t ), solar irradiance G t , weather category (coded).

[0032] The specific data of the traffic dimension includes: lane flow N ti , speed S ti , density D ti , traffic event code (such as congestion, accident).

[0033] Then the unified multi-modal data structure is:

[0034] Further, the multi-modal data structure is subjected to data enhancement. First, outlier detection is performed, and the interquartile range method IQR is used to remove outliers. Then, wavelet denoising is performed, and after smoothing the data, time difference features are constructed. The specific process is as follows: (1) Calculate the quartile For each feature dimension of the data sample, sort it from small to large, and calculate the first quartile Q1 (25% quartile) and the third quartile Q3 (75% quartile).

[0035] (2) Calculate the interquartile range

[0036] (3) Determine the outlier judgment threshold Define the lower and upper bounds:

[0037] Where k is the expansion coefficient, usually k = 1.5, and in the case of large data fluctuations, it can be adjusted to k = 3 to reduce false positives.

[0038] (4) Outlier removal If a sample feature value xi satisfies:

[0039] It is determined to be an outlier and is removed or replaced with the median.

[0040] (5) The specific process of wavelet denoising is as follows: After removing outliers, the discrete wavelet transform (DWT) is used to decompose, threshold filter and reconstruct the time series signal, eliminating high-frequency noise components; then, the data is smoothed by sliding average or exponential smoothing to suppress local fluctuations.

[0041]

[0042] (6) Time difference feature construction The first and second order time difference features are constructed for the smoothed time series data:

[0043] It is used to assist the model in capturing the mutation point.

[0044] Step 3: input the multi-modal data structure into the multi-modal collaborative load prediction model, extract local multi-scale features using a parallel multi-scale TCN network, extract global features using a Transformer encoder and a multi-head self-attention mechanism, fuse the local multi-scale features and the global features, use the fused features to predict the output, and obtain the future load prediction value; Specifically, the multi-modal collaborative load prediction model is a TCN-Transformer collaborative prediction network that combines the local convolution advantages of TCN and the global dependency modeling capabilities of Transformer. The processing process is as follows: Input data sequence First, Local multi-scale features are extracted using a parallel multi-scale TCN network. Each TCN branch has a different dilation rate, and 1-D convolution is used for time modeling. The local multi-scale features are obtained by concatenation, as follows: Each TCN branch has a different dilation rate d l Time modeling is performed using 1-D convolution to obtain local multi-scale features:

[0045] Concatenate the multi-scale convolution results:

[0046] where h l is the output feature vector of the lth branch at time step t; W l is the weight parameter of the lth branch at the convolution kernel position, used to weight the input features; X t-k+1:t is the feature vector of the input sequence at time t-k+1:t, where d l is the dilation rate (dilation rate) used for jump sampling input, which expands the convolution kernel receptive field without increasing the parameter amount; B l is the convolution bias term of the lth branch. H TCN is the concatenated local multi-scale feature matrix, which is one of the inputs to the subsequent Transformer; Concat[•] is the concatenation of the outputs of multiple branches in the feature dimension.

[0047] Further, in order to capture the global dependency relationship, the local multi-scale features are further input into the Transformer encoder, and the Transformer encoder is used for position encoding to preserve the sequence order information. First, input the embedding layer:

[0048] where, PEtEncoding positions, We Trainable embedding matrix.

[0049] Further, local feature fusion is performed using a multi-head self-attention mechanism:

[0050] Further, residual and layer normalization are performed to obtain global features:

[0051] Finally, the local multi-scale features and global features are fused, and the fused features are used to predict the output to obtain the future load prediction value:

[0052] Where σ is a linear or ReLU activation function, and the output is a future load or power value prediction.

[0053] Step 4: Based on the load prediction value and multi-modal data, a highway load regulation model is constructed, and a target function is constructed with the goals of minimizing operating cost, minimizing carbon emissions, and maximizing energy efficiency. The differential evolution algorithm is used to solve the target function to obtain the optimal scheduling strategy of the highway load regulation model. Specifically, a target function is constructed with the goals of minimizing operating cost, minimizing carbon emissions, and maximizing energy efficiency:

[0054] Where the operating cost includes: (1) Generation cost function: Considering that the unit generation cost changes with the load, a piecewise linear model is used:

[0055] Where: c1 < c2, indicating that the electricity price is low at low load and high at high load; L th is the load threshold, and I(•) is the indicator function.

[0056] (2) Start-stop cost: Considering the start-stop cost of adjustable loads (such as charging loads, lighting loads),

[0057] Where: ut∈{0,1} is whether the load is enabled at time t; γ is the start-stop loss cost factor.

[0058] (3) Demand response incentive model: If load shifting is achieved to avoid peak, consider the demand response incentive revenue,

[0059] Where δ is the unit load transfer subsidy; ΔL t This refers to the amount of load transferred during peak hours.

[0060] The final operating cost objective function is:

[0061] Furthermore, the carbon emission function is:

[0062] Energy efficiency objective function:

[0063] The objective function of this disclosure is a multi-objective optimization problem, including:

[0064] Where x is the scheduling decision vector (energy storage charging and discharging, load allocation, etc.); p t L is the electricity price for time period t; t The load value for time period t; ϵ t R is the carbon emission factor per unit load for time period t; t C represents the actual load response during time period t. t This represents the theoretically controllable capacity for time period t.

[0065] To simultaneously optimize three objectives, this method employs a decomposition-based multi-objective differential evolution (MOEA / D-DE) solution framework, decomposing the multi-objective problem into several weighted single-objective sub-problems, including: The weighted Chebyshev method is used to decompose the multi-objective optimization problem into weighted single-objective optimization subproblems. For the th i Each subproblem is a single-objective optimization problem, with a corresponding weight vector defined. The single-objective optimization function for each subproblem is also defined, including: Decompose the multi-objective optimization problem into several weighted single-objective sub-problems: For the i Each sub-problem is defined with a corresponding weight vector:

[0066] satisfy:

[0067] The weighted Chebyshev method is used for decomposition, defined as follows: Current ideal point:

[0068] wherein:

[0069] Define the single-objective optimization function of decomposition for each sub-problem:

[0070] Further, the differential evolution algorithm is used to locally search and globally optimize the single-objective optimization sub-problem, and update the optimal solution of the sub-problem. The specific solving process is as follows: In the present disclosure, the individual (solution vector) of differential evolution optimization is defined as:

[0071] respectively represent the start-stop state of the adjustable load at time t , the charge-discharge power of the energy storage system at time t , the reduction amount of demand response at time t , the electricity purchase amount and the renewable energy output amount.

[0072] For each current individual x i , randomly select three different individuals x r1 , x r2 , x r3 from the neighborhood or global population to generate a differential mutation vector:

[0073] wherein F is a scaling factor, and r1, r2, r3 are different random individuals.

[0074] Further, the variable boundary is constrained and projected:

[0075] to ensure that the charge-discharge power, demand response, etc. do not exceed the device capacity and adjustable range.

[0076] Further, cross operation (Binomial): generate a trial individual by binomial crossover:

[0077] wherein CR is the crossover probability, usually 0.8-0.9; J rand is to ensure that at least one dimension comes from the mutation vector.

[0078] Further, select and update the neighborhood solution set, compare the fitness of the mutant and the individuals in its neighborhood under the Tchebycheff function, including: Calculate the target value of the current individual:

[0079] Calculate the target value of the newly generated trial individual:

[0080] If:

[0081] Replace u i with x i . Select the better solution to update the population, iterate until the convergence condition is met or the maximum number of iterations is reached, and obtain the optimal solution.

[0082] Step 5: Use the optimal scheduling strategy to adaptively regulate the expressway power grid; Specifically, based on the optimal scheduling strategy generated by the above optimization algorithm, the scheduling module controls the power grid, adjusts the energy storage system and load distribution strategy, realizes load balancing, energy efficiency optimization and maximization of renewable energy utilization. The specific control process is based on the dynamic adjustment of the operation mode of the power grid based on the load prediction and optimization scheduling results of the system.

[0083] Further, the present disclosure sets up an adaptive feedback mechanism to adjust the load scheduling strategy according to real-time operation data and historical load prediction. In the control process, a combination of fuzzy control and PID control is adopted, the fuzzy controller adjusts the charging and discharging strategy of the energy storage device dynamically through fuzzy rule reasoning combined with real-time load fluctuation; PID control is used for fine-grained adjustment to realize rapid response and load adjustment. Including: (1) Fuzzy control 1) Control target: Dynamically adjust the charging and discharging power of the energy storage system to reduce load fluctuations and peak clipping, and improve system stability.

[0084] 2) Input variables: Load deviation:

[0085] Wherein, represents the predicted load, represents the actual load.

[0086] Load deviation change rate:

[0087] 3) Output energy storage charging and discharging adjustment amount .

[0088] 4) Fuzzy membership function: Three fuzzy sets are defined:

[0089] 5) Fuzzy demodulation (weighted average method):

[0090] Wherein, μ i The membership degree of the first i rule; P i The charge and discharge adjustment amount corresponding to the first i rule.

[0091] Update the energy storage power:

[0092] Further, the PID controller is used for fine-grained and rapid adjustment of energy storage and adjustable load, to ensure system frequency and voltage stability, specifically including: (1) PID control object The load deviation is also taken as the control input.

[0093] (2) PID control formula:

[0094]

[0095] When the actual value is realized, it is discretized as:

[0096] When the fuzzy control gives a wide range of adjustment, the PID control is used for fine adjustment of the energy storage charge and discharge power:

[0097] The system continues to circulate, and the control amount is refreshed every Δt to realize dynamic self-adaptive load adjustment.

[0098] Embodiment 2 In an embodiment of the present disclosure, a highway load prediction and control system based on low-power Internet of Things is provided, comprising: A data acquisition module is configured to acquire multi-modal data of power, environment and traffic of the highway, and perform preprocessing. ​The load prediction module is configured to perform multi-dimensional fusion on the preprocessed multi-modal data, construct a multi-modal data structure, input the multi-modal data structure into a multi-modal collaborative load prediction model, extract local multi-scale features by using a parallel multi-scale TCN network, extract global features by using a Transformer encoder and a multi-head self-attention mechanism, fuse the local multi-scale features and the global features, and predict an output by using the fused features to obtain a future load prediction value. The scheduling optimization and adaptive regulation and control module is configured to construct a highway load regulation and control model based on the load prediction value and the multi-modal data, construct an objective function with the lowest operation cost, the lowest carbon emission and the highest energy efficiency as targets, solve the objective function by using a differential evolution algorithm, obtain an optimal scheduling strategy of the highway load regulation and control model, and perform adaptive regulation and control on the highway power grid by using the optimal scheduling strategy.

[0099] Embodiment 3 In an embodiment of the present disclosure, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the highway load prediction and regulation method based on a low-power Internet of Things.

[0100] Embodiment 4 In an embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, which is configured to store computer instructions, which, when executed by a processor, implement the highway load prediction and regulation method based on a low-power Internet of Things.

[0101] Embodiment 5 In an embodiment of the present disclosure, an electronic device is provided, comprising a processor, a memory and a computer program; wherein the processor is connected with the memory, and the computer program is stored in the memory; when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device implements the highway load prediction and regulation method based on a low-power Internet of Things.

[0102] The present disclosure is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present disclosure. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device that implements the functions described in the flowcharts and / or block diagrams. Figure 1 one flow or multiple flows and / or blocks Figure 1means for performing the function specified by the block or blocks.

[0103] These computer program instructions can also be loaded into computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate a computer implemented process, so that the instructions executed on the computer or other programmable data processing devices provide a process for implementing the flowchart Figure 1 flowchart or flowcharts and / or block Figure 1 steps of a function specified by a block or blocks.

[0104] Although the specific embodiments of the present disclosure are described above with reference to the drawings, the description is not a limitation on the scope of protection of the present disclosure, and those skilled in the art should understand that various modifications or changes made on the basis of the technical solutions of the present disclosure without creative labor are still within the scope of protection of the present disclosure.

Claims

1. A method for predictive regulation of highway load based on low-power Internet of Things, characterized in that, include: Acquire multimodal data on power, environment, and traffic of highways, and perform preprocessing. The preprocessed multimodal data is fused in multiple dimensions to construct a multimodal data structure. The multimodal data structure is then input into the multimodal collaborative load prediction model. A parallel multi-scale TCN network is used to extract local multi-scale features, and a Transformer encoder and a multi-head self-attention mechanism are used to extract global features. The local multi-scale features are fused with the global features, and the fused features are used to predict the output to obtain the future load prediction value. Based on load forecasts and multimodal data, a highway load regulation model is constructed. The objective function is constructed with the goals of minimizing operating costs, minimizing carbon emissions, and maximizing energy efficiency. The differential evolution algorithm is used to solve the objective function to obtain the optimal scheduling strategy of the highway load regulation model. The optimal scheduling strategy is then used to adaptively regulate the highway power grid.

2. The highway load prediction and control method based on low-power Internet of Things as described in claim 1, characterized in that, The power data in the multimodal data includes three-phase voltage, three-phase current, frequency, harmonic content, and active and reactive power of the power grid; the environmental data includes temperature and humidity, wind speed and direction, solar irradiance, and weather type codes; the traffic data includes lane flow, speed, density, and traffic event codes. The preprocessing process includes denoising, time synchronization, and data interpolation of the multimodal data, and smoothing the data using a Kalman filter. The preprocessed multimodal data is then fused to obtain the multimodal data structure.

3. The highway load prediction and control method based on low-power Internet of Things as described in claim 1, characterized in that, To augment the multimodal data structure, outlier detection is performed first, and outliers are removed using the interquartile range method. Then, wavelet denoising is performed, and the data is smoothed before time difference features are constructed.

4. The highway load prediction and control method based on low-power Internet of Things as described in claim 1, characterized in that, The augmented multimodal data structure is input into the multimodal collaborative load prediction model, which is a TCN-Transformer collaborative prediction network. A parallel multi-scale TCN network is used to extract local multi-scale features. Each TCN branch has a different expansion rate. Temporal modeling is performed separately through 1-D convolution, and the local multi-scale features are concatenated to obtain the local multi-scale features. A Transformer encoder is used for position encoding, and a multi-head self-attention mechanism is used to fuse local features to obtain global features. The prediction output is fused using an activation function to obtain the future load prediction value.

5. The highway load prediction and control method based on low-power Internet of Things as described in claim 1, characterized in that, The objective function is constructed with the goals of minimizing operating costs, minimizing carbon emissions, and maximizing energy efficiency. The operating costs include power generation costs and start-up / shutdown costs. A weighted Chebyshev method is used to decompose the multi-objective optimization problem into weighted single-objective optimization subproblems. For the ... i Each single-objective optimization subproblem is given a corresponding weight vector, and a decomposed single-objective optimization function is defined for each subproblem. The differential evolution algorithm is used to perform local search and global optimization on the single-objective optimization subproblems, and the optimal solution of the subproblems is updated.

6. The highway load prediction and control method based on low-power Internet of Things as described in claim 1, characterized in that, Adaptive regulation of the highway power grid using optimal scheduling strategies includes: adjusting the load scheduling strategy based on real-time operating data and historical load forecasts through an adaptive feedback mechanism; and using a combination of fuzzy control and PID control during the control process. The fuzzy controller uses fuzzy rule reasoning and real-time load fluctuations to dynamically adjust the charging and discharging strategies of energy storage devices. PID control is used for fine-grained adjustment to achieve rapid response and load adjustment.

7. A highway load forecasting and control system based on low-power Internet of Things, characterized in that, include: The data acquisition module is used to acquire multimodal data on power, environment, and traffic of the highway and to perform preprocessing. The load forecasting module is used to fuse preprocessed multimodal data in multiple dimensions to construct a multimodal data structure. The multimodal data structure is then input into the multimodal collaborative load forecasting model. A parallel multi-scale TCN network is used to extract local multi-scale features, and a Transformer encoder and a multi-head self-attention mechanism are used to extract global features. The local multi-scale features are fused with the global features, and the fused features are used to predict the output to obtain the future load forecast value. The scheduling optimization and adaptive control module is used to construct a highway load control model based on load forecasts and multimodal data. It constructs an objective function with the goals of minimizing operating costs, carbon emissions, and energy efficiency, and solves the objective function using a differential evolution algorithm to obtain the optimal scheduling strategy for the highway load control model. The optimal scheduling strategy is then used to adaptively control the highway power grid.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the highway load prediction and control method based on any one of claims 1-6.

9. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium is used to store computer instructions, which, when executed by a processor, implement the highway load prediction and control method based on low-power Internet of Things as described in any one of claims 1-6.

10. An electronic device, characterized in that, include: The device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to perform the highway load prediction and control method based on any one of claims 1-6.

Images