Self-adaptive dormancy scheduling method for wireless sensing node powered by solar energy

Abstract

The invention relates to the technical field of wireless sensing nodes, and discloses a self-adaptive sleep scheduling method for a wireless sensing node powered by solar energy. According to the method, energy collection data and environmental parameters of a solar cell panel are monitored in real time, an energy prediction model is constructed to generate an energy supply curve, a dynamic sleep time threshold value is calculated according to the energy supply curve and task load priority weight, a multi-stage sleep mode is generated through a fuzzy decision rule, and time-sharing sleep scheduling of communication, sensors and calculation modules is achieved. Meanwhile, a sleep mode can be switched by detecting the emergency degree change of a task queue, an emergency scheduling strategy is started and a standby power supply is activated when energy is insufficient, and data can be recorded to update model parameters to realize self-optimization iteration. According to the method, the energy utilization efficiency and the working stability of the wireless sensing node are improved, and the problems of unstable solar energy supply and unreasonable task scheduling are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless sensor nodes, and specifically to an adaptive sleep scheduling method for wireless sensor nodes powered by solar energy. Background Art

[0002] In the era of the booming development of the Internet of Things today, wireless sensor nodes, as key devices for data collection and transmission, have continuously expanded their application scenarios and are widely used in fields such as environmental monitoring, smart homes, and industrial automation. However, these nodes usually rely on battery power supply, and the battery capacity is limited. Frequent battery replacement is not only costly but also impossible in some inaccessible areas, seriously restricting the long-term stable operation of wireless sensor nodes.

[0003] Solar energy, as a clean and renewable energy source, provides a new solution for the power supply of wireless sensor nodes. Using solar panels to collect energy can theoretically enable nodes to get rid of the dependence on traditional batteries and achieve long-term autonomous operation. However, solar energy supply has instability, and its energy collection efficiency is affected by various environmental factors such as light intensity and weather conditions. During the day with sufficient sunlight, more energy can be collected; while at night or on rainy and cloudy days, the energy collection amount will be greatly reduced or even zero. This instability makes it a major problem how to effectively manage the energy of wireless sensor nodes powered by solar energy.

[0004] Traditional wireless sensor node sleep scheduling methods are mostly fixed-mode and do not fully consider the dynamic changes of energy collection and the differences in task loads. For example, some methods simply let the node enter the sleep or wake-up state at preset time intervals. When the energy is sufficient, it may cause waste of resources; when the energy is insufficient, it cannot guarantee the execution of key tasks, resulting in data loss or monitoring interruption. At the same time, due to the lack of scheduling in combination with environmental parameters and energy prediction, the node may consume too much energy at inappropriate times and cannot adapt to the complex and changeable working environment.

[0005] In terms of task scheduling, there are also deficiencies in the existing technologies. Wireless sensor nodes usually need to process various tasks with different priorities and energy consumptions, such as real-time data collection, data transmission, and complex computing tasks. However, traditional scheduling methods often adopt a unified scheduling strategy and do not reasonably allocate energy according to the priority weights of tasks, resulting in important tasks not being processed in time, while some low-priority tasks occupy too many resources, reducing the overall working efficiency of the node.

[0006] In addition, with the increasingly complex application scenarios of wireless sensor nodes, the requirements for their energy utilization efficiency and working stability are getting higher and higher. For example, in field environmental monitoring, the nodes need to operate stably for a long time to obtain continuous environmental data; in industrial production monitoring, the nodes need to transmit data in a timely and accurate manner to ensure production safety. Therefore, it is urgent to develop a method for adaptive sleep scheduling based on solar energy harvesting, environmental parameters, and task load, which is of great significance for improving the performance of wireless sensor nodes and expanding their application scope. Summary of the Invention

[0007] The purpose of the present invention is to provide an adaptive sleep scheduling method for wireless sensor nodes powered by solar energy to solve the problems proposed in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solution: An adaptive sleep scheduling method for wireless sensor nodes powered by solar energy, the method includes:

[0009] Real-time monitoring of the energy harvesting data of the solar panel, the energy harvesting data includes the current voltage, current, and remaining energy storage capacity;

[0010] Collecting the light intensity, temperature, and humidity parameters of the environment where the wireless sensor node is located;

[0011] Based on the energy harvesting data and environmental parameters, constructing an energy prediction model to generate an energy supply curve within a preset future time;

[0012] According to the energy supply curve and the priority weight of the current task load of the node, calculating the dynamic sleep time threshold;

[0013] Based on the preset fuzzy decision rule, dividing the interval of the dynamic sleep time threshold to generate a multi-level sleep mode;

[0014] According to the multi-level sleep mode, performing time-sharing sleep scheduling on the communication module, sensor module, and computing module of the wireless sensor node.

[0015] Preferably, the steps for constructing the energy prediction model include:

[0016] Performing time series decomposition on the historical energy harvesting data to extract the trend term, periodic term, and residual term;

[0017] Using the adaptive moving average algorithm to smooth and correct the trend term, and extracting the frequency domain characteristics of the periodic term based on wavelet transform;

[0018] Fusing the corrected trend term, frequency domain characteristics, and real-time environmental parameters to construct a multivariate regression prediction equation;

[0019] Optimize the coefficient matrix of the regression equation by the gradient descent method to generate the energy supply curve.

[0020] Preferably, the calculation steps of the dynamic sleep time threshold include:

[0021] Divide the task levels according to the priority weights of the task loads and assign corresponding energy consumption coefficients;

[0022] Calculate the energy surplus or deficit per unit time based on the slope change rate of the energy supply curve;

[0023] Combine the energy consumption coefficient and the energy surplus / deficit value, and use the dynamic programming algorithm to solve the optimal sleep time threshold.

[0024] Preferably, the setting steps of the fuzzy decision rule include:

[0025] Define the membership functions corresponding to the sleep time threshold intervals, including the extremely low power consumption mode, the balanced mode, and the high performance mode;

[0026] According to the stability index of the energy supply curve and the task queue length, set the trigger conditions of the fuzzy rule base, and use the centroid method for defuzzification to discretize the continuous threshold into a multi-level sleep mode.

[0027] Preferably, the steps of the time-sharing sleep scheduling further include:

[0028] Adopt a heartbeat packet interval adaptive adjustment strategy for the communication module, and dynamically extend or shorten the heartbeat period according to the sleep mode;

[0029] Adopt an event-driven wake-up mechanism for the sensor module, and activate the sampling function only when the preset environmental parameter change threshold is triggered;

[0030] Adopt a task sharding scheduling algorithm for the computing module, and disassemble the high-load tasks into multiple low-power subtasks and execute them in batches.

[0031] Preferably, the steps of the time series decomposition include:

[0032] Use the empirical mode decomposition algorithm to decompose the historical energy data into multiple intrinsic mode functions;

[0033] Calculate the instantaneous frequency of each mode function through the Hilbert transform, and screen out the dominant periodic components;

[0034] Reconstruct the dominant periodic components and the trend term into the input feature vector of the energy prediction model.

[0035] Preferably, the method further includes:

[0036] Real-time detect the change in the urgency of the task queue and trigger the adaptive switching instruction of the sleep mode;

[0037] When the energy supply curve is below the critical threshold, enable the emergency scheduling strategy and activate the low-power backup power supply;

[0038] Record the historical scheduling data and update the parameter weights of the energy prediction model to achieve self-optimizing iteration;

[0039] The enabling conditions of the emergency scheduling strategy include:

[0040] When the remaining energy storage capacity is below the first preset threshold and the energy supply curve continues to decline, force a switch to the extremely low-power mode;

[0041] When the ambient light intensity is below the second preset threshold for more than the set duration, start the parallel power supply mechanism of the backup power supply;

[0042] Close the non-essential peripheral interfaces and limit the maximum working current below the safety threshold.

[0043] Preferably, the execution steps of the dynamic programming algorithm include:

[0044] Establish a state transition equation with time slices as the stage variable and the remaining energy as the state variable;

[0045] Based on the Bellman optimality principle, recursively solve the optimal sleep decision sequence for each stage;

[0046] Introduce a relaxation factor to balance the computational complexity and the solution accuracy, and generate a real-time feasible scheduling scheme.

[0047] Preferably, the steps of the self-optimizing iteration include:

[0048] Perform residual analysis on the actual energy consumption and the predicted value in the historical scheduling data, calculate the model deviation index, dynamically adjust the weight coefficients of the regression equation based on the deviation index, and update the model parameters using the incremental learning strategy;

[0049] Perform Gaussian kernel width adaptive adjustment on the membership functions in the fuzzy decision rule base to optimize the rule matching accuracy.

[0050] Preferably, the activation logic of the low-power backup power supply includes:

[0051] When the main power supply voltage drops to the undervoltage protection point, switch to the supercapacitor for temporary power supply;

[0052] During the period when the backup power supply is enabled, use pulse width modulation technology to dynamically adjust the duty cycle of the supply voltage;

[0053] Real-time monitor the remaining capacity of the backup power supply and perform a progressive switching strategy after the main power supply is restored.

[0054] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0055] From the perspective of energy management, by real-time monitoring the energy collection data of solar panels, including the current voltage, current, and remaining energy storage capacity, and combining the collected environmental light intensity, temperature, and humidity parameters to construct an energy prediction model, an energy supply curve for a preset future time is generated. This measure enables the node to predict the energy supply situation in advance, changing the blindness of energy management in traditional methods. For example, when it is predicted that the light intensity will be insufficient in the upcoming period, the node adjusts the sleep strategy in advance to reduce unnecessary energy consumption and ensure that basic functions can still be maintained during the energy trough period. At the same time, the dynamic sleep time threshold is calculated based on the energy supply curve, making the determination of the node's sleep time more scientific and reasonable, and avoiding energy waste or shortage caused by a fixed sleep strategy. When the light is sufficient during the day, the node can appropriately shorten the sleep time to complete more tasks in a timely manner; while at night or in bad weather, the sleep time is extended to reduce energy consumption, thus significantly improving the energy utilization efficiency.

[0056] In terms of task scheduling, the task levels are divided according to the priority weights of the task loads and the energy consumption coefficients are assigned. The dynamic programming algorithm is used to solve the optimal sleep time threshold in combination with the energy surplus or deficit value. This method ensures that high-priority tasks can obtain energy resources first and be processed in a timely manner, avoiding the loss or delay of important data. In the industrial production monitoring scenario, for the key data collection tasks related to production safety, due to their high priority, the system will give priority to allocating energy to execute them quickly to ensure the safety and stability of the production process. Low-priority tasks are only executed when there is sufficient energy, effectively balancing the relationship between task execution and energy consumption and improving the overall working efficiency of the node.

[0057] The introduction of fuzzy decision rules further optimizes the sleep scheduling. By defining the membership functions corresponding to different sleep time threshold intervals, such as the very low power consumption mode, the balanced mode, and the high-performance mode, and setting the trigger conditions of the fuzzy rule base according to the stability index of the energy supply curve and the task queue length, the centroid method is used for defuzzification to generate multi-level sleep modes. This enables the node to quickly and accurately select the appropriate sleep mode according to the complex and changeable working conditions. When the energy supply is stable and the tasks are few, it automatically switches to the very low power consumption mode to minimize energy consumption; when the task volume increases and the energy supply is sufficient, it switches to the balanced mode or the high-performance mode to ensure the smooth completion of tasks.

[0058] The time-sharing sleep scheduling strategy optimizes the communication, sensor, and computing modules respectively. The communication module adopts a heartbeat packet interval adaptive adjustment strategy, dynamically extending or shortening the heartbeat cycle according to the sleep mode, reducing unnecessary communication energy consumption. During periods with low data transmission requirements, the heartbeat cycle is extended, the working frequency of the communication module is reduced, and power consumption is decreased. The event-driven wake-up mechanism of the sensor module activates the sampling function only when the preset environmental parameter change threshold is triggered, avoiding energy waste caused by frequent sampling. The task sharding scheduling algorithm of the computing module disassembles high-load tasks into multiple low-power subtasks and executes them in batches, reducing the energy consumption of the computing module when processing complex tasks.

[0059] The present invention also has the ability of self-optimizing iteration. By recording historical scheduling data and updating the parameter weights of the energy prediction model, and adaptively adjusting the membership functions in the fuzzy decision rule base, the system can continuously adapt to changes in the environment and tasks, and continuously improve performance. The emergency scheduling strategy and the low-power backup power activation logic enhance the survival ability of the node in extreme situations. When the energy supply curve is below the critical threshold, the emergency scheduling strategy is enabled, non-essential peripheral interfaces are turned off, the maximum working current is restricted, and the backup power is activated to ensure that the node can still maintain key functions in a harsh environment, guaranteeing the reliability and stability of the system. Brief Description of the Drawings

[0060] Figure 1 is the working principle diagram of the adaptive sleep scheduling method for the solar-powered wireless sensor node described in the present invention;

[0061] Figure 2 is the construction diagram of the energy prediction model;

[0062] Figure 3 is the calculation diagram of the dynamic sleep time threshold;

[0063] Figure 4 is the flow chart related to emergency scheduling and self-optimization. Detailed Embodiments

[0064] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0065] Please refer to Figures 1 - 4, the present invention provides an adaptive sleep scheduling method for solar-powered wireless sensor nodes, aiming to solve the problems of energy management and task scheduling of solar-powered wireless sensor nodes, and improve the energy utilization efficiency and working stability of the nodes. The specific implementation manners of the present invention are elaborated in detail below. The overall implementation scheme is as follows:

[0066] Real-time monitoring of energy harvesting data: Special voltage sensors, current sensors, and remaining energy storage capacity detection circuits are used to monitor the energy harvesting data of solar panels in real time. These sensors convert the collected analog signals into digital signals and transmit them to the main control chip of the wireless sensor node. Among them, the current voltage, current, and remaining energy storage capacity data can reflect the working state and energy storage situation of the solar panel in real time.

[0067] Collection of environmental parameters: Light intensity sensors, temperature sensors, and humidity sensors are arranged around the wireless sensor node. These sensors continuously collect the light intensity, temperature, and humidity parameters of the surrounding environment and transmit the data to the main control chip. These environmental parameters are of great significance for constructing the subsequent energy prediction model.

[0068] Construct an energy prediction model and generate an energy supply curve: An energy prediction model is constructed by combining the data obtained in Step 1 and Step 2. By analyzing the relationship between historical energy harvesting data and real-time environmental parameters, this model predicts the energy supply situation within a preset future time, and then generates an energy supply curve to provide a basis for subsequent sleep scheduling.

[0069] Calculate the dynamic sleep time threshold: Based on the generated energy supply curve and the priority weight of the current task load of the node, the dynamic sleep time threshold is obtained through a specific calculation method. This threshold will be dynamically adjusted according to the changes in energy supply and task load to ensure that the node reduces energy consumption as much as possible while meeting the task requirements.

[0070] Generate multi-level sleep modes: Based on the preset fuzzy decision rules, the dynamic sleep time threshold is divided into intervals to generate multi-level sleep modes. These sleep modes correspond to different power consumption levels and performance performances to adapt to different working scenarios.

[0071] Time-sharing sleep scheduling: According to the generated multi-level sleep modes, time-sharing sleep scheduling is performed on the communication module, sensor module, and computing module of the wireless sensor node. In this way, the working and sleep times of each module are reasonably controlled to further optimize the energy consumption of the node.

[0072] The technical solution of the present invention is further described in detail below through 5 embodiments.

[0073] Embodiment 1:

[0074] This embodiment elaborates in detail the construction process of the energy prediction model.

[0075] First, perform time series decomposition on historical energy collection data. Use the Empirical Mode Decomposition (EMD) algorithm to decompose the historical energy data into multiple Intrinsic Mode Functions (IMFs). Assume the historical energy data is , and after EMD decomposition, a series of intrinsic mode functions , . are obtained. Calculate the instantaneous frequency of each mode function through Hilbert transform, and screen out the dominant periodic components. Let be the screened dominant periodic component. Then reconstruct the dominant periodic component and the trend term into the input feature vector of the energy prediction model.

[0076] Use the adaptive moving average algorithm to smooth and correct the trend term. The adaptive moving average algorithm can dynamically adjust the size and weight of the moving window according to the change of data, making the trend term smoother and more accurate. Extract the frequency domain features of the periodic term based on wavelet transform. Wavelet transform can decompose the signal at different scales, so as to extract the frequency domain features of the periodic term.

[0077] Integrate the corrected trend term, frequency domain features and real-time environmental parameters to construct a multivariate regression prediction equation. Let the prediction equation be , where represents the predicted energy value, represents the th input variable (including the corrected trend term, frequency domain features and real-time environmental parameters, etc.), are the corresponding coefficients, is the constant term, is the number of input variables.

[0078] Optimize the coefficient matrix of the regression equation through the gradient descent method. The gradient descent method is an iterative optimization algorithm. By continuously adjusting the values of the coefficient matrix, the error between the predicted value and the actual value is minimized. Finally, generate an energy supply curve, which can accurately predict the energy supply situation within a preset future time.

[0079] In practical applications, for example, in an environmental monitoring project, collect the energy collection data and environmental parameter data of the past month, and use the above method to construct an energy prediction model. After multiple tests and optimizations, the prediction error of the energy supply within the next day by this model is controlled within 10%, providing reliable data support for subsequent sleep scheduling.

[0080] Embodiment 2:

[0081] This embodiment focuses on explaining the calculation steps of the dynamic sleep time threshold.

[0082] Divide the task levels according to the priority weights of the task loads and assign corresponding energy consumption coefficients. Assume that the tasks are divided into three levels: high, medium, and low, corresponding to the energy consumption coefficients , , , respectively, and . The priority weights of the tasks can be determined according to factors such as the importance and urgency of the tasks.

[0083] Calculate the energy surplus or deficit per unit time based on the slope change rate of the energy supply curve. Let the energy supply curve be , and within the time interval , the slope change rate of the energy supply curve is , where is the first derivative of the energy supply curve. According to the slope change rate , the energy surplus or deficit per unit time can be calculated. If , it indicates an energy surplus, ; if , it indicates an energy deficit, , is the unit time interval.

[0084] Combine the energy consumption coefficient and the energy surplus / deficit value, and use the dynamic programming algorithm to solve for the optimal sleep time threshold. Establish a state transition equation with the time slice as the stage variable and the remaining energy as the state variable. Let the time slice be , and the remaining energy be , and the state transition equation is , where represents the remaining energy at the th time slice, represents the execution time of the th task in the current time slice, represents the number of tasks executed within the current time slice.

[0085] Recursively solve the optimal sleep decision sequence for each stage based on the Bellman optimality principle. The Bellman optimality principle states that an optimal strategy has the property that regardless of the initial state and initial decision, for the state resulting from the initial decision, the remaining decisions must form an optimal strategy. By continuously iterating and calculating, find the optimal sleep time threshold to minimize the energy consumption of the node while meeting the task requirements.

[0086] Introduce a relaxation factor to balance the computational complexity and solution accuracy, and generate a real-time feasible scheduling scheme. The relaxation factor can relax the requirement for the optimal solution to a certain extent, thereby reducing the computational amount and improving the real-time performance of the algorithm. In practical applications, for example, in a smart home monitoring system, according to the priorities and energy consumption coefficients of different tasks, combined with the changes in the energy supply curve, the dynamic sleep time threshold is calculated through the dynamic programming algorithm. Through actual tests, this method can effectively balance the energy consumption of nodes and the task execution efficiency, enabling the nodes to extend the working time while ensuring the normal progress of the monitoring tasks.

[0087] Embodiment 3:

[0088] This embodiment details the setting steps of the fuzzy decision rule.

[0089] Define the membership functions corresponding to the sleep time threshold intervals, including the very low power consumption mode, the balanced mode, and the high-performance mode. Taking the sleep time threshold as an example, let the membership function of the very low power consumption mode be , the membership function of the balanced mode be , and the membership function of the high-performance mode be . These membership functions can adopt common function forms such as Gaussian functions and triangular functions. For example, the membership function in the form of a Gaussian function is

[0090]

[0091] where is the input variable (i.e., the sleep time threshold ), is the central value of the function, is the standard deviation, and the membership functions of different modes are determined by adjusting the and values to determine their respective ranges and shapes.

[0092] Set the trigger conditions of the fuzzy rule base according to the stability index of the energy supply curve and the task queue length. The stability index of the energy supply curve can be obtained by calculating the variance of the curve, etc. Let the stability index be , and the task queue length be . For example, when is less than a certain threshold and is less than another threshold, the very low power consumption mode is triggered; when and are within a certain range, the balanced mode is triggered; when is greater than a certain threshold or is greater than another threshold, the high-performance mode is triggered.

[0093] The centroid method is used for defuzzification, and the continuous threshold is discretized into multiple sleep modes. The centroid method is a commonly used defuzzification method, and its calculation formula is

[0094]

[0095] where is the finally determined discretized sleep time threshold,[[]] is the value of the discretized sleep time threshold,[[]] is the corresponding membership degree,[[]] is the number of discretized points. In this way, the continuous sleep time threshold is converted into a specific multi-level sleep mode, so as to perform more accurate sleep scheduling for wireless sensor nodes.

[0096] In actual application scenarios, such as in an agricultural environment monitoring project, by long-term monitoring of the energy supply curve and task queue length data, the parameters of the membership function and the triggering conditions of the fuzzy rule base are adjusted according to actual needs. Through on-site testing, this fuzzy decision rule setting method can accurately select the appropriate sleep mode according to different working conditions, effectively improving the energy utilization efficiency of wireless sensor nodes. For example, when the light is insufficient at night, the energy supply is relatively stable and the task queue is short, the node can quickly switch to the ultra-low power consumption mode to reduce energy consumption; while when the light is sufficient during the day and the task demand is large, it automatically switches to the balanced mode or high-performance mode to ensure the smooth completion of the monitoring task.

[0097] Example 4:

[0098] This embodiment details the specific steps and related strategies of time-sharing sleep scheduling.

[0099] An adaptive adjustment strategy for the heartbeat packet interval is adopted for the communication module, and the heartbeat period is dynamically extended or shortened according to the sleep mode. In the ultra-low power consumption mode, since the task load of the node is low and the requirement for communication real-time performance is not high, the heartbeat period can be greatly extended. Let the heartbeat period in the ultra-low power consumption mode be , in this mode, in order to minimize the energy consumption of the communication module, the heartbeat period may be set to several minutes or even longer. In the balanced mode, the heartbeat period is adjusted according to the actual situation, usually shorter than the heartbeat period in the ultra-low power consumption mode but longer than that in the high-performance mode to balance communication requirements and energy consumption. In the high-performance mode, in order to ensure the timely transmission of data, the heartbeat period is set shorter, generally between a few seconds and dozens of seconds. Through this strategy of adaptively adjusting the heartbeat packet interval, it is possible to effectively reduce the energy consumption of the communication module while meeting communication requirements.

[0100] An event-driven wake-up mechanism is adopted for the sensor module, and the sampling function is activated only when the preset environmental parameter change threshold is triggered. For example, for a light intensity sensor, a light intensity change threshold is preset. . When the change amount of the environmental light intensity exceeds , the sensor module is woken up and starts sampling. Assume the current light intensity is . After a period of time, the light intensity becomes . If , the wake-up mechanism of the sensor module is triggered. For the temperature sensor and the humidity sensor, a similar principle is adopted, and the temperature change threshold and the humidity change threshold are set respectively. Through this event-driven wake-up mechanism, the frequent sampling of the sensor module when the environmental parameters do not change significantly is avoided, thereby reducing the energy consumption of the sensor module.

[0101] A task sharding scheduling algorithm is adopted for the computing module, and the high-load task is disassembled into multiple low-power subtasks and executed in batches. Assume there is a high-load computing task , whose computing amount is large, and direct execution will consume a large amount of energy. This task is decomposed into subtasks . According to the characteristics of the task and the performance of the computing module, the computing amount of each subtask is reasonably allocated. During the execution process, these subtasks are executed in batches in a certain order. After executing a subtask, the computing module can enter a short sleep state and then execute the next subtask. This can effectively reduce the energy consumption of the computing module when executing high-load tasks and ensure the smooth completion of the task at the same time.

[0102] In an actual industrial monitoring scenario, the wireless sensor node needs to monitor the operating status of the device in real time, including multiple parameters such as temperature, pressure, and vibration. By adopting the above-mentioned time-sharing sleep scheduling strategy, when the device is running stably and the data changes little, the communication module can automatically extend the heartbeat period and reduce unnecessary communication energy consumption; the sensor module is only woken up for sampling when the parameters change significantly, avoiding the increase in energy consumption caused by frequent sampling; the computing module executes the complex data analysis tasks in slices, reducing the overall energy consumption while ensuring the accuracy of data analysis. After actual operation tests, after adopting these strategies, the energy consumption of the wireless sensor node is reduced by 30%-40% compared with that before optimization, greatly extending the working life of the node.

[0103] Example 5:

[0104] This example mainly describes other related functions and strategies in the present invention, including task queue urgency detection, emergency scheduling strategy, self-optimizing iteration, and low-power backup power activation logic.

[0105] Real-time detect the change in the urgency level of the task queue and trigger the adaptive switching instruction for the sleep mode. The urgency level of the task queue can be determined according to factors such as the priority of the task and the waiting time of the task. For example, for tasks with a higher priority and a longer waiting time, increase the urgency level of the task queue. When it is detected that the urgency level of the task queue changes, trigger the adaptive switching instruction for the sleep mode according to the current energy supply situation and the sleep mode. If the current is in the extremely low power consumption mode, but the urgency level of the task queue suddenly increases and the energy supply can meet the requirements of a higher performance mode, then switch to the balanced mode or the high performance mode to process the urgent tasks as soon as possible.

[0106] When the energy supply curve is below the critical threshold, enable the emergency scheduling strategy and activate the low power consumption backup power supply. The enabling conditions for the emergency scheduling strategy include: when the remaining energy storage capacity is below the first preset threshold and the energy supply curve continues to decline, force a switch to the extremely low power consumption mode. This is to minimize the energy consumption of the node to the greatest extent in the case of severe energy shortage and ensure that the node can continue to operate. When the ambient light intensity is below the second preset threshold for more than the set duration start the parallel power supply mechanism of the backup power supply. Because too low light intensity will cause a significant decrease in the energy collection efficiency of the solar panel, starting the backup power supply at this time can ensure the normal operation of the node. At the same time, turn off the non-essential peripheral interfaces and limit the maximum working current to the safety threshold or below to further reduce the energy consumption.

[0107] Record the historical scheduling data and update the parameter weights of the energy prediction model to achieve self-optimizing iteration. Perform residual analysis on the actual energy consumption and the predicted value in the historical scheduling data to calculate the model deviation index. Let the actual energy consumption be , the predicted energy consumption be , the model deviation index can be obtained by calculating the mean square error (MSE), , where is the number of data samples. Dynamically adjust the weight coefficients of the regression equation based on the deviation index and update the model parameters using the incremental learning strategy. The incremental learning strategy can gradually update the model parameters when new data arrives without having to retrain the entire model, improving the update efficiency of the model. Perform Gaussian kernel width adaptive adjustment on the membership function in the fuzzy decision rule base to optimize the rule matching accuracy. By continuously adjusting the Gaussian kernel width, the membership function can better adapt to the actual situation and improve the accuracy of fuzzy decision-making.

[0108] The activation logic of the low power consumption backup power supply includes: when the main power supply voltage drops to the undervoltage protection point When the situation occurs, switch to the temporary power supply of the supercapacitor. The supercapacitor has the characteristics of fast charge and discharge speed and can quickly provide power support when the main power supply voltage is insufficient. During the period when the backup power supply is enabled, the pulse width modulation technology (PWM) is used to dynamically adjust the duty cycle of the supply voltage. Let the duty cycle of PWM be , by adjusting value, the output voltage and power of the backup power supply can be controlled to meet the different energy consumption requirements of the node. The remaining capacity of the backup power supply is monitored in real time, and a progressive switching strategy is executed after the main power supply is restored. The progressive switching strategy can avoid problems such as voltage fluctuations caused by suddenly switching back to the main power supply when the main power supply is restored, and ensure the stability of the node operation.

[0109] In a meteorological monitoring project in a remote area, due to the complex environment, the solar power supply is unstable. By implementing the above functions and strategies, when the energy supply is insufficient, the node can timely switch to the emergency mode, enable the backup power supply and adjust the working state, ensuring the continuous monitoring and transmission of meteorological data. At the same time, through self-optimizing iteration, the energy prediction model and the fuzzy decision rule base are continuously optimized, making the energy management of the node more reasonable and greatly improving the working reliability in harsh environments. After long-term operation monitoring, the failure rate of the node is reduced by more than 50% compared with the situation without adopting these strategies, effectively ensuring the smooth progress of meteorological monitoring work.

[0110] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0111] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A self - adaptive sleep scheduling method for solar - powered wireless sensor nodes, characterized in that, The method includes: real-time monitoring of the energy collection data of the solar panel, where the energy collection data includes the current voltage, current, and remaining energy storage capacity; collecting the light intensity, temperature, and humidity parameters of the environment where the wireless sensor node is located; constructing an energy prediction model based on the energy collection data and environmental parameters to generate an energy supply curve for a preset future time period; calculating a dynamic sleep time threshold according to the energy supply curve and the priority weight of the current task load of the node; performing interval division on the dynamic sleep time threshold based on a preset fuzzy decision rule to generate a multi-level sleep mode; and performing time-sharing sleep scheduling on the communication module, sensor module, and computing module of the wireless sensor node according to the multi-level sleep mode.

2. The adaptive sleep scheduling method for wireless sensor nodes according to claim 1, wherein The steps for constructing the energy prediction model include: performing time series decomposition on historical energy collection data to extract the trend term, periodic term, and residual term; using an adaptive moving average algorithm to smooth and correct the trend term, and extracting the frequency domain features of the periodic term based on wavelet transform; fusing the corrected trend term, frequency domain features, and real-time environmental parameters to construct a multivariable regression prediction equation; and optimizing the coefficient matrix of the regression equation by the gradient descent method to generate an energy supply curve.

3. The adaptive sleep scheduling method for wireless sensor nodes according to claim 1, characterized in that The steps for calculating the dynamic sleep time threshold include: dividing the task levels according to the priority weight of the task load and assigning corresponding energy consumption coefficients; calculating the energy surplus or deficit per unit time based on the slope change rate of the energy supply curve; and combining the energy consumption coefficient and the energy surplus / deficit value to solve for the optimal sleep time threshold using the dynamic programming algorithm.

4. The adaptive sleep scheduling method for wireless sensor nodes according to claim 1, characterized in that, The steps for setting the fuzzy decision rule include: defining membership functions corresponding to the sleep time threshold intervals, including the extremely low power consumption mode, balanced mode, and high-performance mode; setting the trigger conditions of the fuzzy rule base according to the stability index of the energy supply curve and the task queue length, and using the centroid method for defuzzification to discretize the continuous threshold into a multi-level sleep mode.

5. The adaptive sleep scheduling method for wireless sensor nodes according to claim 1, characterized in that, The steps for time-sharing sleep scheduling further include: adopting a heartbeat packet interval adaptive adjustment strategy for the communication module to dynamically extend or shorten the heartbeat period according to the sleep mode; adopting an event-driven wake-up mechanism for the sensor module to activate the sampling function only when a preset environmental parameter change threshold is triggered; and adopting a task sharding scheduling algorithm for the computing module to break down high-load tasks into multiple low-power subtasks and execute them in batches.

6. The adaptive sleep scheduling method for wireless sensor nodes according to claim 2, wherein The steps for time series decomposition include: using the empirical mode decomposition algorithm to decompose historical energy data into multiple intrinsic mode functions; calculating the instantaneous frequency of each mode function through Hilbert transform and screening out the dominant periodic components; and reconstructing the dominant periodic components and the trend term into the input feature vector of the energy prediction model.

7. The adaptive sleep scheduling method for wireless sensor nodes according to claim 1, characterized in that The method further includes: detecting the change in the urgency of the task queue in real time and triggering an adaptive switching instruction for the sleep mode; when the energy supply curve is lower than the critical threshold, enabling an emergency scheduling strategy and activating a low-power backup power supply; recording historical scheduling data and updating the parameter weights of the energy prediction model to achieve self-optimizing iteration; the enabling conditions of the emergency scheduling strategy include: when the remaining energy storage capacity is lower than the first preset threshold and the energy supply curve continues to decline, forcibly switching to an extremely low-power mode; when the ambient light intensity is lower than the second preset threshold for more than a set duration, starting a parallel power supply mechanism for the backup power supply; closing unnecessary peripheral interfaces and limiting the maximum working current to below the safety threshold.

8. The adaptive sleep scheduling method for wireless sensor nodes according to claim 3, wherein The execution steps of the dynamic programming algorithm include: establishing a state transition equation with time slices as stage variables and remaining energy as state variables; recursively solving the optimal sleep decision sequence for each stage based on the Bellman optimality principle; introducing a relaxation factor to balance the computational complexity and solution accuracy to generate a real-time feasible scheduling plan.

9. The adaptive sleep scheduling method for wireless sensor nodes according to claim 7, wherein The steps of the self-optimizing iteration include: performing residual analysis on the actual energy consumption and predicted value in the historical scheduling data, calculating the model deviation index, dynamically adjusting the weight coefficients of the regression equation based on the deviation index, and updating the model parameters using an incremental learning strategy; adaptively adjusting the Gaussian kernel width of the membership function in the fuzzy decision rule base to optimize the rule matching accuracy.

10. The adaptive sleep scheduling method for wireless sensor nodes according to claim 7, wherein, The activation logic of the low-power backup power supply includes: when the main power supply voltage drops to the undervoltage protection point, switching to the supercapacitor for temporary power supply; during the period when the backup power supply is enabled, dynamically adjusting the duty cycle of the supply voltage using pulse width modulation technology; real-time monitoring the remaining capacity of the backup power supply and performing a progressive switching strategy after the main power supply is restored.

Images