Wireless sensor network node charging state monitoring and energy saving regulation method

Abstract

The application discloses a wireless sensor network node charging state monitoring and energy-saving regulation method, relates to the technical field of wireless sensors, and is based on pre-acquired energy source types, battery data and acoustic spectrum data of the wireless sensor network node, calculates a frequency self-adaptive charging state index, obtains a predicted energy value and a fluctuation variance, analyzes node working requirements to generate a working requirement vector, carries out working mode regulation through a reinforcement learning model containing an energy fluctuation penalty term, outputs optimized working mode parameters, optimizes communication parameters to generate communication optimization parameters, carries out node-network collaborative topology optimization through local regulation cost calculation, outputs topology optimization parameters, implements regulation and generates a regulation effect report, solves the limitation problems existing in the current wireless sensor network node charging state monitoring and energy-saving regulation process, and guarantees the reliability and authenticity of the analysis results.

Description

Technical Field

[0001] This application relates to the field of wireless sensor technology, specifically to a method for monitoring the charging status and energy-saving control of wireless sensor network nodes. Background Technology

[0002] The problem of energy harvesting instability in current wireless sensor network nodes has not been fully solved. Therefore, it is very important to achieve dynamic energy-saving regulation by integrating environmental frequency adaptive interference models, fluctuation-aware reinforcement learning, and node-network bidirectional optimization mechanisms.

[0003] Existing technologies, such as the invention application patent with announcement number CN111988791A, disclose a method for improving the computing power of wireless charging network nodes based on fog computing. This method includes the following steps: inputting the data required for computation; determining the energy harvested by the wireless charging network node, the computational energy consumption of the wireless charging network node, and the energy consumption of unloading tasks to the fog server based on the input computational data; establishing an optimization problem model regarding the energy beamforming vector of the power transmitter, the time allocation of the wireless charging network node, the computation frequency, and the power allocation; solving and outputting the maximum system computation rate. This invention, however, considers the energy causal relationship between the wireless charging network node, the power transmitter, and multiple fog servers, as well as the system computational power of the wireless charging network node and multiple fog servers. It jointly optimizes the energy beamforming vector of the power transmitter, the time allocation of the wireless charging network node, the computation frequency, and the power allocation to determine the objective of maximizing the system computation rate.

[0004] Regarding the above solutions, the inventors of this application have found that the above technology has at least the following technical problems: 1. The current environmental interference identification accuracy is low, and the general charging status monitoring model cannot adapt to dynamic environmental noise, resulting in a high misjudgment rate. Especially in the vibration energy harvesting scenario, the physical characteristics of the energy source are not bound, and it is impossible to distinguish between effective energy and noise interference, which frequently triggers invalid mode switching and increases energy consumption.

[0005] 2. Current traditional reinforcement learning control methods only focus on static energy consumption and task latency, ignoring the volatility of intermittent energy collection. Sudden energy depletion leads to a high node downtime rate, making it difficult to guarantee task continuity. Node-network collaboration is insufficient, and topology optimization algorithms rely on a single energy index to select the best node without considering the node's local working state and task priority. High-load nodes are prone to interrupting critical tasks due to resource contention, which shortens the lifespan of wireless sensor network nodes. Summary of the Invention

[0006] To address the aforementioned technical shortcomings, the purpose of this application is to provide a method for monitoring the charging status and controlling energy efficiency of wireless sensor network nodes.

[0007] To solve the above-mentioned technical problems, this application adopts the following technical solution: This application provides a method for monitoring the charging status and energy-saving control of wireless sensor network nodes, including: S1, calculating the frequency adaptive charging status index based on the energy source type, battery data and acoustic spectrum data of the pre-acquired wireless sensor network node, and then obtaining the predicted energy value and fluctuation variance.

[0008] S2. Based on the pre-acquired workload data, analyze the node workload requirements to generate a workload requirement vector.

[0009] S3. Based on the working demand vector, the frequency adaptive charging state index, the predicted energy value and fluctuation variance, the working mode is adjusted through a reinforcement learning model, and optimized working mode parameters are output. Communication optimization parameters are generated based on the optimized working mode parameters.

[0010] S4. Based on the predicted energy value, optimized working mode parameters, link quality data, and task priority, perform node-network collaborative topology optimization through local control cost calculation, and output topology optimization parameters.

[0011] S5. Based on communication optimization parameters, optimized working mode parameters, and topology optimization parameters, implement regulation and generate a regulation effect report.

[0012] Preferably, the battery data includes the remaining battery percentage and voltage fluctuation rate.

[0013] Preferably, the calculation of the frequency adaptive charging state index based on the pre-acquired energy harvesting data, battery data, and acoustic spectrum data of the wireless sensor network nodes includes: analyzing the acoustic spectrum data to obtain the frequency with the maximum amplitude as the main frequency; extracting the resonant frequency according to the energy source type, and then using the ratio of the resonant frequency to the main frequency with a preset scaling factor to obtain the interference factor.

[0014] The signal-to-noise ratio is integrated by calculating exponential decay, and the interference factor is combined with the signal-to-noise ratio. The result of multiplying the ratio of the interference factor to the signal-to-noise ratio by 0.2 is recorded as the noise effect.

[0015] Extract the remaining charge percentage and voltage fluctuation rate from the battery data. Multiply the noise impact by the noise impact weight factor, the remaining charge percentage by the remaining charge percentage weight factor, and the voltage fluctuation rate by the voltage fluctuation rate weight factor. Then sum the results of each multiplication to obtain the frequency adaptive charging state index.

[0016] Preferably, obtaining the predicted energy value and fluctuation variance includes: retrieving the adaptive charging state index of each frequency from the historical records, inputting it into the energy value prediction model, multiplying the result of the mixed weight coefficient with the state value and the predicted value by weighted summation to obtain the predicted energy value, summing the squares of the predicted energy value minus the actual energy values, and dividing by the total number of actual energy values ​​to obtain the fluctuation variance.

[0017] Preferably, the workload data is a task queue, which includes the number of high-priority tasks, the number of medium-priority tasks, and the number of low-priority tasks.

[0018] Preferably, the step of analyzing node workload requirements based on pre-acquired workload data to generate a workload requirement vector includes: defining the workload requirement vector as a three-dimensional vector based on the pre-acquired workload data, including the weighted sum of requirements for high-priority tasks, the weighted sum of requirements for medium-priority tasks, and the weighted sum of requirements for low-priority tasks; the weighted sum of requirements for high-priority tasks is the product of the number of high-priority tasks and the high-priority weight coefficient, the weighted sum of requirements for medium-priority tasks is the product of the number of medium-priority tasks and the medium-priority weight coefficient, and the weighted sum of requirements for low-priority tasks is the product of the number of low-priority tasks and the low-priority weight coefficient, so as to generate a workload requirement vector and derive task priorities.

[0019] Preferably, the step of regulating the working mode through a reinforcement learning model and outputting optimized working mode parameters includes: integrating the input working demand vector, frequency adaptive charging state index, predicted energy value, and fluctuation variance through a reinforcement learning model; the reinforcement learning model selects actions based on the state space and dynamically generates optimized working mode parameters; wherein the state space is defined as a four-dimensional vector, action space constraints are applied, and working modes are set, including low-power mode, medium-efficiency mode, and high-performance mode; and a key constraint is set to forcibly exclude the high-performance mode when the fluctuation variance is greater than 0.1 multiplied by the predicted average energy value; the reinforcement learning model is optimized through a reward function, which includes energy saving value, task delay duration, and fluctuation variance, specifically a weighted fusion of the product of energy saving value and the weight factor corresponding to energy saving value, the product of task delay duration and the weight factor corresponding to task delay duration, and the product of fluctuation variance and the weight factor corresponding to fluctuation variance.

[0020] Preferably, the step of generating communication optimization parameters based on optimized operating mode parameters includes: taking the optimized operating mode parameters as input, dynamically adjusting the modulation depth based on the optimized operating mode parameters, and using the product of the optimized operating mode parameters divided by a scaling exponent of 0.6 and the initial adjustment depth as the modulation depth; simultaneously, multiplying the ratio of the optimized operating mode parameters to 0.6 by 2 by the initial bit error rate value to obtain the bit error rate threshold; the communication optimization parameters are defined as a two-dimensional vector, and the modulation depth and bit error rate threshold are used as two items in the communication optimization parameters, thereby generating the communication optimization parameters.

[0021] Preferably, the node-network cooperative topology optimization through local control cost calculation, outputting topology optimization parameters, includes: taking predicted energy value, optimized working mode parameters, link quality data, and task priority as input data, and generating topology optimization parameters through local control cost calculation and optimal node selection; wherein the absolute value of the node's optimized working mode parameter multiplied by the task priority weight coefficient is used as the local control cost; the optimal node is selected by maximizing the energy efficiency ratio, and the node corresponding to the maximum ratio of the predicted energy value to the product of the local control cost and the link quality data is selected as the optimal node, and the optimal node is denoted as... ; and then according to the formula Obtain topology optimization parameters ,in This represents the associated path weight corresponding to the best node.

[0022] Preferably, the step of implementing control and generating a control effect report based on communication optimization parameters, optimized operating mode parameters, and topology optimization parameters includes: inputting communication optimization parameters, optimized operating mode parameters, topology optimization parameters, optimal nodes, and associated path weights; implementing control measures such as: adjusting the modulation scheme based on communication optimization parameters to achieve communication optimization; switching node operating modes based on optimized operating mode parameter values; selecting the optimal node and optimizing paths based on topology optimization parameters to complete network topology reconstruction; and simultaneously monitoring control effect data, including energy consumption reduction, latency optimization rate, and usage cycle gain, and generating a time-series performance curve as the control effect report based on the control effect data.

[0023] The beneficial effects of this application are as follows: 1. The wireless sensor network node charging status monitoring and energy-saving control method provided in this application calculates the frequency adaptive charging status index based on the pre-acquired energy source type, battery data, and acoustic spectrum data of the wireless sensor network node, and obtains the predicted energy value and fluctuation variance. It analyzes the node's working requirements to generate a working requirement vector, performs working mode control through a reinforcement learning model that includes an energy fluctuation penalty term, outputs optimized working mode parameters, optimizes communication parameters to generate communication optimization parameters, performs node-network cooperative topology optimization through local control cost calculation, outputs topology optimization parameters, implements control, and generates a control effect report. This solves the limitations of current wireless sensor network node charging status monitoring and energy-saving control processes, and ensures the reliability and authenticity of the analysis results.

[0024] 2. In this application, end-to-end training specifically involves joint optimization of LSTM and Markov parts. The mixed weight coefficients are used as trainable parameters to balance the contributions of both. The input data comes from historical records and current features, and the output is used for downstream risk control to ensure data consistency. The training is based on gradient descent-driven LSTM and statistically driven Markov initialization, which is consistent with time series prediction.

[0025] 3. The process of generating the work requirement vector in this application is highly independent, relying only on the input workload data and not involving the output of other steps, ensuring modular processing; the output work requirement vector value is input to the work mode control module for dynamically adjusting resource allocation, such as allocating computing resources or bandwidth based on the work requirement vector value; for example, high-priority tasks have high weighted demand sum values, triggering a high resource allocation mode; task priority data is also used for historical data analysis or optimization; complex task queues are converted into numerical vectors through simple weighted quantization, and the weight coefficient settings are consistent with resource allocation rules, ensuring logical coherence.

[0026] 4. The optimized working mode parameters in this application are reused in the historical analysis module to optimize long-term strategies. The volatility variance is directly embedded in the reward function as a risk suppression term, reducing the downtime rate by 23%. The dynamic constraint of the action space fits the energy volatility risk. Among them, the working demand vector provides task requirements, the frequency adaptive charging status index provides historical characteristics, and the predicted energy value and volatility variance provide energy prediction and risk. The four work together to ensure the comprehensiveness of decision-making. Risk perception is performed based on volatility variance, and a forced degradation mode is implemented to achieve fault prevention.

[0027] 5. Topology optimization parameters are applied to the deployment module to guide network topology reconstruction, such as cluster head node routing allocation; local control costs are integrated with working modes and task priorities to avoid isolated optimization; the best node is selected by maximizing energy efficiency, while link quality data is used directly in the selection to prevent low-stability nodes from becoming the best nodes; predicted energy values, optimized working mode parameters, and task priorities are used to ensure data consistency. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a flowchart illustrating the steps involved in implementing the method described in this application.

[0030] Figure 2 This is a diagram showing the signal connections for a wireless sensor.

[0031] Figure 3 This is a node diagram of a wireless sensor network. Detailed Implementation

[0032] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0033] Please see Figure 1 As shown, this application provides a method for monitoring the charging status and energy-saving control of wireless sensor network nodes, including: S1, calculating a frequency-adaptive charging status index based on the energy source type, battery data and acoustic spectrum data of the pre-acquired wireless sensor network node, and then obtaining the predicted energy value and fluctuation variance.

[0034] It should be noted that wireless sensor network nodes are primarily charged through energy harvesting, with rechargeable batteries serving as a backup; energy sources include light, vibration, and wind.

[0035] In one specific instance, the battery data includes the remaining percentage of charge and voltage fluctuation rate.

[0036] In a specific example, the calculation of the frequency-adaptive charging state index based on the energy harvesting data, battery data, and acoustic spectrum data of the pre-acquired wireless sensor network nodes includes: analyzing the acoustic spectrum data to obtain the frequency with the maximum amplitude as the main frequency; extracting the resonant frequency according to the energy source type, and then using the ratio of the resonant frequency to the main frequency with a preset scaling factor to obtain the interference factor.

[0037] The signal-to-noise ratio is integrated by calculating exponential decay, and the interference factor is combined with the signal-to-noise ratio. The result of multiplying the ratio of the interference factor to the signal-to-noise ratio by 0.2 is recorded as the noise effect.

[0038] Extract the remaining charge percentage and voltage fluctuation rate from the battery data. Multiply the noise impact by the noise impact weight factor, the remaining charge percentage by the remaining charge percentage weight factor, and the voltage fluctuation rate by the voltage fluctuation rate weight factor. Then sum the results of each multiplication to obtain the frequency adaptive charging state index.

[0039] It should be noted that the frequency adaptive charging status indicator is used to improve the accuracy of environmental interference identification.

[0040] It should be noted that the resonant frequency is preset according to the energy type, such as the resonant frequency of the solar panel being set to... This indicates that the solar panel has no mechanical resonance, and the resonant frequency of the vibration energy collector is set to... , indicating that the vibration energy harvester is at its mechanical resonant frequency; the noise frequency and energy source resonant characteristics are correlated in the interference factor to quantify the physical impact of environmental interference.

[0041] It should be noted that the analysis of the acoustic spectrum data to determine the frequency with the largest amplitude as the dominant frequency involves performing a 512-point Fast Fourier Transform (FFT) on the acoustic spectrum data. The FFT is an efficient computational method used to convert time-domain signals into frequency-domain signals and identify the main frequency components. The original measurement values ​​of the acoustic spectrum data at each acquisition time point are multiplied by a rotation factor of 512, and the results are superimposed to output the amplitude frequency. The frequency with the largest amplitude is then used as the dominant frequency, which is used to quantify the main interference sources. The constant 512 represents the number of sampling points in the FFT and determines the frequency resolution.

[0042] It should be noted that a higher signal-to-noise ratio (SNR) indicates higher signal quality; when the interference factor is combined with the SNR, the noise impact increases exponentially when the interference factor increases or the SNR decreases.

[0043] It should be noted that the weighting factors for noise impact, remaining power percentage, and voltage fluctuation rate are obtained through factor analysis. First, the information on noise impact, remaining power percentage, and voltage fluctuation rate is condensed, and then the variance explained after rotation is obtained. The weights are obtained by dividing the cumulative variance explained rate.

[0044] It should be noted that factor analysis is a well-known technique. It is a multivariate statistical analysis method that starts by studying the internal dependencies of variables and reduces some variables with complex relationships to a few comprehensive factors. Information condensation is expressed as the calculation of the median. The variance explained rate is the amount of information extracted by the factors, and the variance explained rate is equal to the ratio of the eigenvalues ​​to the total number of analysis items. The rotated variance explained rate is expressed as the variance explained by the factors after maximum variance rotation.

[0045] In a specific example, the process of obtaining the predicted energy value and the fluctuation variance includes: retrieving the adaptive charging state index for each frequency from the historical records, inputting it into the energy value prediction model, multiplying the mixed weight coefficient with the state value and the predicted value by a weighted sum to obtain the predicted energy value, summing the squares of the predicted energy value minus the squares of each actual energy value, and dividing by the total number of actual energy values ​​to obtain the fluctuation variance.

[0046] It should be noted that volatility variance quantifies the risk of energy fluctuations and addresses the problem of intermittent energy harvesting.

[0047] It should be noted that the energy value prediction model is a combination of LSTM and Markov state transition models. First, data preparation is performed: the frequency-adaptive charging state index sequence from historical records is divided into training and test sets. The input is the frequency-adaptive charging state index sequence, and the output is the corresponding future energy value. Next, model initialization is performed: the LSTM part randomly initializes the weight matrix (e.g., input gate, forget gate, output gate weights) and bias vector, using a normal distribution or random method; the Markov part statistically estimates the initial state transition probability based on historical data (e.g., by counting state transition frequencies), and state classification depends on the discretization of the feature vector of the frequency-adaptive charging state index; the mixed weight coefficients are initialized to preset values ​​(e.g., 0.5) as trainable parameters. Finally, training iterations and optimization are performed: batches of data (e.g., mini-batch) are randomly sampled from the training set, the frequency-adaptive charging state index sequence is input into the LSTM network, and the LSTM prediction value is output; the feature vector of the current frequency-adaptive charging state index is input into the Markov model, and the state value is output; the predicted energy value and fluctuation variance are calculated. The Adam optimizer is used to calculate the loss gradient (with a learning rate of 0.001), and backpropagation is used to update the LSTM parameters (weights and biases) and the hybrid weight coefficients. The training loop is repeated while monitoring the training set loss. The process stops when the loss converges and changes less than a threshold. Performance metrics (such as mean squared error) are calculated using the test set, and hyperparameters (such as hybrid weight coefficients and the number of LSTM layers) are fine-tuned based on the results. The model parameters are dynamically calibrated using historical data to enhance generalization ability.

[0048] It should be noted that the actual energy values ​​are taken from historical records.

[0049] It should be noted that the mixed weight coefficient is a preset hyperparameter, ranging from [0,1], used to adjust the contribution ratio of the LSTM and Markov parts. It is optimized through historical data, such as by minimizing the mean square error, to ensure that the model balances the long-term trend and the fluctuation of the current state.

[0050] In this application, end-to-end training specifically involves joint optimization of LSTM and Markov components. Hybrid weight coefficients are used as trainable parameters to balance the contributions of both. Input data comes from historical records and current features, and the output is used for downstream risk control to ensure data consistency. The training is based on gradient descent-driven LSTM and statistically driven Markov initialization, which is consistent with time series prediction.

[0051] S2. Based on the pre-acquired workload data, analyze the node workload requirements to generate a workload requirement vector.

[0052] In a specific instance, the workload data is a task queue, which includes the number of high-priority tasks, the number of medium-priority tasks, and the number of low-priority tasks.

[0053] It should be noted that high-priority tasks are those that need to be processed immediately; medium-priority tasks are those that are routine and allow for moderate delays; and low-priority tasks are those that are background tasks and allow for delays.

[0054] In a specific example, the step of analyzing node workload requirements based on pre-acquired workload data to generate a workload requirement vector includes: defining the workload requirement vector as a three-dimensional vector based on the pre-acquired workload data, including the weighted sum of requirements for high-priority tasks, the weighted sum of requirements for medium-priority tasks, and the weighted sum of requirements for low-priority tasks; the weighted sum of requirements for high-priority tasks is the product of the number of high-priority tasks and the high-priority weight coefficient, the weighted sum of requirements for medium-priority tasks is the product of the number of medium-priority tasks and the medium-priority weight coefficient, and the weighted sum of requirements for low-priority tasks is the product of the number of low-priority tasks and the low-priority weight coefficient, to generate the workload requirement vector and derive task priorities.

[0055] It should be noted that the process of generating the work requirement vector is as follows: Priority rules are set, the task queue is quantified, and the weighted sum of requirements for each priority category is calculated; the work requirement vector is defined as a three-dimensional vector with the following expression: ,in This is represented as the weighted sum of requirements for high-priority tasks. This is represented as the weighted sum of requirements for medium-priority tasks. This is expressed as a weighted sum of requirements for low-priority tasks; the calculation formula is based on task counts and fixed weights. , , The final work requirement vector expression is: ,in This represents the number of high-priority tasks. This represents the number of medium-priority tasks. This represents the number of low-priority tasks; Represented as a high-priority weight coefficient, Represented as a medium-priority weighting coefficient, This is represented as a low-priority weight coefficient.

[0056] It should be noted that the above example of generating the work requirement vector is as follows: assuming there is a node with a work requirement task queue containing 3 high-priority tasks, 2 medium-priority tasks, and 1 low-priority task, the work requirement vector expression is: The generated work requirement vector value is .

[0057] The process of generating the work requirement vector in this application is highly independent, relying only on the input workload data and not involving the output of other steps, ensuring modular processing. The output work requirement vector value is input to the work mode control module for dynamically adjusting resource allocation, such as allocating computing resources or bandwidth based on the work requirement vector value. For example, high-priority tasks have high weighted demand sum values, triggering a high resource allocation mode. Task priority data is also used for historical data analysis or optimization. Complex task queues are converted into numerical vectors through simple weighted quantization, and the weight coefficients are set in accordance with resource allocation rules to ensure logical consistency.

[0058] S3. Based on the working demand vector, the frequency adaptive charging state index, the predicted energy value and fluctuation variance, the working mode is adjusted through a reinforcement learning model, and optimized working mode parameters are output. Communication optimization parameters are generated based on the optimized working mode parameters.

[0059] In a specific example, the step of regulating the working mode through a reinforcement learning model and outputting optimized working mode parameters includes: integrating the input working demand vector, frequency adaptive charging state index, predicted energy value, and fluctuation variance through the reinforcement learning model; the reinforcement learning model selects actions based on the state space and dynamically generates optimized working mode parameters; wherein the state space is defined as a four-dimensional vector, and action space constraints are applied to set working modes, including low-power mode, medium-efficiency mode, and high-performance mode; and a key constraint is set to forcibly exclude the high-performance mode when the fluctuation variance is greater than 0.1 multiplied by the predicted average energy value; the reinforcement learning model is driven to optimize through a reward function, which includes energy saving value, task delay duration, and fluctuation variance, specifically a weighted fusion of the product of energy saving value and the weight factor corresponding to energy saving value, the product of task delay duration and the weight factor corresponding to task delay duration, and the product of fluctuation variance and the weight factor corresponding to fluctuation variance.

[0060] It should be noted that reinforcement learning models select actions based on the state space and dynamically generate optimal working mode parameters. The specific process is as follows: First, the state space is defined as a four-dimensional vector, with the specific expression being... ,in Represented as a work requirement vector, This is represented as a frequency-adaptive charging state indicator. Represented as predicted energy value, Represented as variance; the workload requirement vector quantifies task priority requirements, such as the proportion of high-priority tasks. A workload mode is selected, including low-power mode, medium-efficiency mode, and high-performance mode. The low-power mode has a mapping value of 0.3, which reduces the sampling frequency and implements energy-saving strategies; the medium-efficiency mode has a mapping value of 0.6, which balances performance and energy consumption; and the high-performance mode has a mapping value of 0.9, which maximizes computing power. Key constraints are set when... At this time, the high-performance mode is forcibly excluded, leaving only 0.3 or 0.6 in the motion space. This is represented as the mean of the predicted energy value, avoiding the risk of downtime under high fluctuations. Model optimization is driven by a reward function, which is... ,in Expressed as an energy saving value, it is the ratio of actual energy consumption to baseline energy consumption; This represents the task delay time, which is the ratio of the actual completion time to the expected time. , and These represent the weighting factors corresponding to energy savings, task delay duration, and variance, respectively. The final reinforcement learning model selects actions based on the state space and dynamically generates optimized working mode parameters.

[0061] It should be noted that the weighting factor corresponding to the energy saving value amplifies the energy-saving benefits and prioritizes energy saving; the weighting factor corresponding to the task delay duration amplifies the task delay cost and provides secondary control over delay; and the weighting factor corresponding to the volatility variance is the volatility penalty coefficient, which strengthens the suppression of volatility risk and avoids risk.

[0062] The optimized working mode parameters in this application are reused in the historical analysis module to optimize long-term strategies. The volatility variance is directly embedded in the reward function as a risk suppression term, reducing the downtime rate by 23%. The dynamic constraint of the action space is adapted to the energy volatility risk. Among them, the working demand vector provides task requirements, the frequency adaptive charging status index provides historical characteristics, and the predicted energy value and volatility variance provide energy prediction and risk. The four work together to ensure comprehensive decision-making. Risk perception is performed based on volatility variance, and a forced degradation mode is implemented to achieve fault prevention.

[0063] In a specific example, the step of generating communication optimization parameters based on optimized operating mode parameters includes: taking the optimized operating mode parameters as input, dynamically adjusting the modulation depth based on the optimized operating mode parameters, and using the product of the optimized operating mode parameters divided by a scaling exponent of 0.6 and the initial adjustment depth as the modulation depth; simultaneously, multiplying the ratio of the optimized operating mode parameters to 0.6 by 2 by the initial bit error rate value to obtain the bit error rate threshold; the communication optimization parameters are defined as a two-dimensional vector, and the modulation depth and bit error rate threshold are used as two items in the communication optimization parameters, thereby generating the communication optimization parameters.

[0064] It should be noted that the communication optimization parameters are defined as a two-dimensional vector, and the modulation depth and bit error rate threshold are used as two items in the communication optimization parameters to generate the communication optimization parameters. The use of the modulation depth and bit error rate threshold as two items in the communication optimization parameters enables dual driving of low power mode and high performance mode, and dynamic adjustment to meet real-time requirements.

[0065] It should be noted that the initial adjustment depth is defined by the communication protocol. For example, the initial adjustment depth of Quadrature Phase Shift Keying (QPSK) is 0.5. The scaling index is preset to 1.0, which controls the adjustment range.

[0066] It should be noted that the modulation depth is dynamically adjusted based on the optimized operating mode parameters to ensure that the optimized operating mode parameters and the modulation depth are positively correlated, thereby optimizing the balance between energy consumption and performance.

[0067] It should be noted that the logic for adjusting the modulation depth is as follows: In low-power mode, when the optimized operating mode parameter is equal to 0.3, the modulation depth is reduced to reduce energy consumption, such as reducing the modulation depth to 0.25, sacrificing the transmission rate; in high-performance mode, when the optimized operating mode parameter is equal to 0.9, the modulation depth is increased to enhance data throughput, such as setting the modulation depth to 0.75; in medium-efficiency mode, when the optimized operating mode parameter is equal to 0.6, the baseline value is maintained, such as setting the modulation depth to 0.5.

[0068] It should be noted that the initial bit error rate value is defined by the communication protocol, such as... .

[0069] It should be noted that the bit error rate threshold is negatively correlated with the optimized working mode parameters to ensure that resource allocation meets the mode requirements.

[0070] It should be noted that the logic for adjusting the bit error rate threshold is as follows: In low-power mode, when the optimized operating mode parameter is equal to 0.3, the bit error rate threshold is increased to 1.5 times the initial bit error rate value, tolerating a higher bit error rate to save energy; in high-performance mode, when the optimized operating mode parameter is equal to 0.9, the bit error rate threshold is decreased to 0.5 times the initial bit error rate value, strictly limiting bit errors to enhance reliability; in medium-efficiency mode, the optimized operating mode parameter is equal to 0.6, maintaining the initial bit error rate value.

[0071] S4. Based on the predicted energy value, optimized working mode parameters, link quality data and task priority, perform node-network cooperative topology optimization through local control cost calculation, and output topology optimization parameters.

[0072] It should be noted that the link quality data refers to the link quality from the node to the network center.

[0073] It should be noted that link quality is obtained by deploying probes on one side of the link.

[0074] It should be noted that the node-network collaborative topology optimization parameters are applied to the deployment module to guide network topology reconstruction, such as optimal node routing allocation.

[0075] In a specific example, the node-network cooperative topology optimization through local control cost calculation, outputting topology optimization parameters, includes: taking predicted energy value, optimized operating mode parameters, link quality data, and task priority as input data, and generating topology optimization parameters through local control cost calculation and optimal node selection; wherein the absolute value of the node's optimized operating mode parameter multiplied by the task priority weight coefficient is used as the local control cost; the optimal node is selected by maximizing the energy efficiency ratio, and the node corresponding to the maximum ratio of the predicted energy value to the product of the local control cost and the link quality data is selected as the optimal node, and the optimal node is denoted as... ; and then according to the formula Obtain topology optimization parameters ,in This represents the associated path weight corresponding to the best node.

[0076] It should be noted that the calculation logic of the local control cost is as follows: quantify the node resource control cost. For example, when the optimized working mode parameter is equal to 0.9 in high-performance mode and a high-priority task is executed, the local control cost is 0.9 × 1.0 = 0.9; when the optimized working mode parameter is equal to 0.3 in low-power mode and a low-priority task is executed, the local control cost is 0.3 × 0.5 = 0.15.

[0077] It should be noted that the calculation logic for selecting the optimal node by maximizing the energy efficiency ratio is as follows: prioritize nodes with higher predicted energy values ​​in the numerator; prioritize nodes with lower local control costs and higher link quality data values ​​in the denominator; and maximize the energy efficiency ratio overall to ensure that the optimal node has high energy, low cost, and excellent link quality. For example, if node A has a predicted energy value of 80, a local control cost of 0.2, and a link quality data value of 0.9, then node A has an energy efficiency ratio of 444; if node B has a predicted energy value of 100, a local control cost of 0.9, and a link quality data value of 0.5, then node B has an energy efficiency ratio of 222. Since 444 is greater than 222, node A is selected as the optimal node.

[0078] It should be noted that the associated path weight matrix is ​​generated by integrating link quality data and predicted energy values. The specific method is as follows: according to the calculation formula... The path weights of the nodes are derived, where Minimize the difference in energy magnitude and highlight the relative advantage.

[0079] The topology optimization parameters in this application are applied to the deployment module to guide network topology reconstruction, such as optimal node routing allocation; local control costs are integrated with working modes and task priorities to avoid isolated optimization; the best node is selected by maximizing energy efficiency, while link quality data is used directly in the selection to prevent low-stability nodes from becoming the best nodes; and data consistency is ensured by using predicted energy values, optimized working mode parameters, and task priorities.

[0080] S5. Based on communication optimization parameters, optimized working mode parameters, and topology optimization parameters, implement regulation and generate a regulation effect report.

[0081] In a specific example, the process of implementing control and generating a control effect report based on communication optimization parameters, optimized operating mode parameters, and topology optimization parameters includes: inputting communication optimization parameters, optimized operating mode parameters, topology optimization parameters, optimal node and associated path weights; implementing control measures such as: adjusting the modulation scheme based on communication optimization parameters to achieve communication optimization; switching node operating modes based on optimized operating mode parameter values; selecting the optimal node and optimizing paths based on topology optimization parameters to complete network topology reconstruction; and simultaneously monitoring control effect data, including energy consumption reduction, latency optimization rate, and usage cycle gain, and generating a time-series performance curve as the control effect report based on the control effect data.

[0082] It should be noted that system parameters can be dynamically configured, such as through API service calls.

[0083] It should be noted that the ratio of standard energy consumption minus actual energy consumption to standard energy consumption is used as the energy consumption reduction rate; the ratio of standard latency minus actual latency to standard latency is used as the latency optimization rate; and the ratio of the current predicted usage period minus the previous predicted usage period to the previous predicted usage period is used as the usage period gain. Here, actual energy consumption refers to the energy consumption data of the nodes collected in real time by intelligent sensors; actual latency is the task execution time; and the current predicted usage period is the node lifespan. A stream processing framework is used to clean and aggregate the control effect data, generating a time-series performance curve. The control effect data is fed back to S1, S2, S3, and S4 to drive iterative optimization of the model.

[0084] This application calculates frequency-adaptive charging state indices based on pre-acquired energy source types, battery data, and acoustic spectrum data of wireless sensor network nodes, and derives predicted energy values ​​and fluctuation variances. It analyzes node operating requirements to generate an operating requirement vector, uses a reinforcement learning model including an energy fluctuation penalty term to regulate operating modes, outputs optimized operating mode parameters, optimizes communication parameters to generate communication optimization parameters, performs node-network cooperative topology optimization through local regulation cost calculation, outputs topology optimization parameters, implements regulation, and generates a regulation effect report. This addresses the limitations of current wireless sensor network node charging state monitoring and energy-saving regulation processes, ensuring the reliability and authenticity of the analysis results.

[0085] The above content is merely an example and illustration of the concept of this application. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the inventive concept or exceed the scope defined in this specification, they should all fall within the protection scope of this application.

Claims

1. A method for monitoring the charging status and controlling energy saving of wireless sensor network nodes, characterized in that, include: S1. Based on the energy source type, battery data and acoustic spectrum data of the pre-acquired wireless sensor network nodes, calculate the frequency adaptive charging state index, and then obtain the predicted energy value and fluctuation variance. The frequency-adaptive state-of-charge index is calculated based on pre-acquired energy harvesting data, battery data, and acoustic spectrum data from wireless sensor network nodes, including: Based on the analysis of acoustic spectrum data, the frequency with the maximum amplitude is obtained as the main frequency; the resonant frequency is extracted according to the energy source type, and the ratio of the resonant frequency to the main frequency is compared with the preset scaling factor to obtain the interference factor. The signal-to-noise ratio is integrated by calculating exponential decay, and the interference factor is combined with the signal-to-noise ratio. The result of multiplying the ratio of the interference factor to the signal-to-noise ratio by 0.2 is recorded as the noise effect. Extract the remaining power percentage and voltage fluctuation rate from the battery data. Multiply the noise impact with the noise impact weight factor, the remaining power percentage with the remaining power percentage weight factor, and the voltage fluctuation rate with the voltage fluctuation rate weight factor, and sum the results of each multiplication to obtain the frequency adaptive charging state index. The process of deriving the predicted energy value and fluctuation variance includes: Retrieve adaptive charging state indicators for each frequency from the historical records, input them into the energy value prediction model, multiply the result of the mixed weight coefficient with the state value and the predicted value, and sum the weighted sum to obtain the predicted energy value. Sum the squares of the predicted energy value minus the actual energy values ​​and divide by the total number of actual energy values ​​to obtain the fluctuation variance. S2. Based on the pre-acquired workload data, analyze the node workload requirements to generate a workload requirement vector; S3. Based on the working demand vector, the frequency adaptive charging state index, the predicted energy value and fluctuation variance, the working mode is adjusted through a reinforcement learning model, and optimized working mode parameters are output. Communication optimization parameters are generated based on the optimized working mode parameters. S4. Based on the predicted energy value, optimized working mode parameters, link quality data and task priority, perform node-network cooperative topology optimization through local control cost calculation, and output topology optimization parameters. S5. Based on communication optimization parameters, optimized working mode parameters, and topology optimization parameters, implement regulation and generate a regulation effect report.

2. The method for monitoring the charging status and controlling energy saving of wireless sensor network nodes according to claim 1, characterized in that, The battery data includes the remaining percentage of charge and voltage fluctuation rate.

3. The method for monitoring the charging status and controlling energy saving of wireless sensor network nodes according to claim 2, characterized in that, The step of analyzing node workload requirements based on pre-acquired workload data to generate a workload requirement vector includes: Based on the pre-acquired workload data, the work requirement vector is defined as a three-dimensional vector, which includes the weighted sum of requirements for high-priority tasks, the weighted sum of requirements for medium-priority tasks, and the weighted sum of requirements for low-priority tasks. The weighted sum of requirements for high-priority tasks is the product of the number of high-priority tasks and the high-priority weight coefficient; the weighted sum of requirements for medium-priority tasks is the product of the number of medium-priority tasks and the medium-priority weight coefficient; and the weighted sum of requirements for low-priority tasks is the product of the number of low-priority tasks and the low-priority weight coefficient. This process generates the work requirement vector and derives the task priorities.

4. The method for monitoring the charging status and controlling energy saving of wireless sensor network nodes according to claim 3, characterized in that, The process of regulating the working mode through a reinforcement learning model and outputting optimized working mode parameters includes: By integrating the input working demand vector, frequency adaptive charging state index, predicted energy value, and fluctuation variance through a reinforcement learning model, the reinforcement learning model selects actions based on the state space and dynamically generates optimized working mode parameters. The state space is defined as a four-dimensional vector, and action space constraints are applied to set working modes, including low-power mode, medium-efficiency mode, and high-performance mode. A key constraint is set to forcibly exclude the high-performance mode when the fluctuation variance is greater than 0.1 times the result of the predicted average energy value. The reinforcement learning model is optimized through a reward function, which includes energy saving value, task delay duration, and fluctuation variance. Specifically, it is a weighted fusion of the product of energy saving value and its corresponding weight factor, the product of task delay duration and its corresponding weight factor, and the product of fluctuation variance and its corresponding weight factor.

5. The method for monitoring the charging status and controlling energy saving of wireless sensor network nodes according to claim 4, characterized in that, The process of generating communication optimization parameters based on optimized working mode parameters includes: The optimized operating mode parameters are used as inputs, and the modulation depth is dynamically adjusted based on the optimized operating mode parameters. The modulation depth is obtained by multiplying the optimized operating mode parameters by the scaling exponent of 0.6 and the initial adjustment depth. At the same time, the bit error rate threshold is obtained by multiplying the ratio of the optimized operating mode parameters to 0.6 by 2 and the initial bit error rate value. The communication optimization parameters are defined as a two-dimensional vector, and the modulation depth and bit error rate threshold are used as two items in the communication optimization parameters to generate the communication optimization parameters.

6. The method for monitoring the charging status and controlling energy saving of wireless sensor network nodes according to claim 5, characterized in that, The node-network cooperative topology optimization through local control cost calculation, outputting topology optimization parameters, includes: Using predicted energy values, optimized operating mode parameters, link quality data, and task priorities as input data, topology optimization parameters are generated through local control cost calculation and optimal node selection. The local control cost is calculated by multiplying the absolute value of the node's optimized operating mode parameter by the task priority weight coefficient. The optimal node is selected by maximizing the energy efficiency ratio; the node corresponding to the maximum ratio of the predicted energy value to the product of the local control cost and the link quality data is designated as the optimal node, and this optimal node is denoted as […]. ; and then according to the formula Obtain topology optimization parameters ,in This represents the associated path weight corresponding to the best node.

7. The method for monitoring the charging status and controlling energy saving of wireless sensor network nodes according to claim 6, characterized in that, The process of implementing control based on communication optimization parameters, optimized operating mode parameters, and topology optimization parameters, and generating a control effect report, includes: Input communication optimization parameters, optimized operating mode parameters, topology optimization parameters, optimal node and associated path weights; implement control measures such as: adjusting the modulation scheme based on communication optimization parameters to achieve communication optimization; switching node operating modes based on optimized operating mode parameter values; selecting the optimal node and optimizing paths based on topology optimization parameters to complete network topology reconstruction; simultaneously monitor control effect data, including energy consumption reduction, latency optimization rate and usage cycle gain, and generate time-series performance curves as control effect reports based on the control effect data.

Images