Wireless transmission power control method for data acquisition terminal

Abstract

The invention relates to the technical field of wireless communication, in particular to a wireless transmission power control method for a data acquisition terminal, which comprises the following steps of: acquiring historical signal intensity data and dividing windows, calculating fluctuation characteristics and then classifying and building a library, acquiring signals in real time and building windows and judging types, calling corresponding sub-libraries and calculating weights of sampling points. And constructing a weighted difference matrix based on a DTW algorithm, adaptively adjusting a path constraint width to calculate a matching degree, calculating randomness in combination with real-time window fluctuation characteristics, determining the number of fusion windows, screening matching degree window weighted fusion to obtain predicted power, and controlling a transmission module. According to the method, through historical signal intensity window classification database building, real-time window differentiation weight distribution, DTW adaptive matching and dynamic fusion, the wireless transmission power prediction precision is improved, the data transmission reliability and the terminal cruising ability are both considered, and the robustness and adaptability in a complex channel environment are enhanced.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology. In particular, it relates to a method for controlling wireless transmission power in a data acquisition terminal. Background Technology

[0002] Data acquisition terminals are embedded intelligent nodes deployed in physical locations to sense, collect, process, and wirelessly upload environmental or equipment status information. They upload data to the cloud or gateway via wireless links (such as NB-IoT, LoRa, and Wi-Fi) and are widely used in industrial IoT, smart metering, environmental monitoring, and remote operation and maintenance scenarios. In industrial data acquisition scenarios, data acquisition terminals need to wirelessly transmit equipment operating parameters and environmental monitoring data to the gateway; the appropriateness of the wireless transmission power directly determines the transmission reliability and the terminal's battery life.

[0003] Traditional wireless transmission power control methods typically employ fixed power or open-loop power control strategies based on Received Signal Strength Indicator (RSSI). However, actual signal strength is often affected by factors such as the start-up and shutdown of nearby devices, operational mode switching, personnel movement, and multipath effects, exhibiting significant time-varying, random, and fluctuating characteristics. Dynamic Time Warping (DTW) algorithms predict control power by matching the morphological similarity between historical channel sequences and the current sequence. However, existing technologies ignore the differences in the contribution of data at different times within the window to the prediction, and the fixed matching path constraints result in low accuracy in similar sequence matching. Furthermore, the wireless transmission power control of data acquisition terminals relies on real-time signal strength feedback, lacking the ability to adjust in advance, which can easily lead to decreased transmission reliability or insufficient terminal battery life, thus affecting the stability of data transmission. Summary of the Invention

[0004] To address the problems of low matching accuracy caused by traditional wireless transmission power control methods that employ a single sequence matching strategy and fail to consider the differences in data contribution at different times within a window, and the fact that wireless transmission power control relies on real-time signal strength feedback and lacks the ability to adjust in advance, which can easily lead to decreased transmission reliability or insufficient terminal battery life, this invention provides solutions in the following aspects.

[0005] A wireless transmission power control method for a data acquisition terminal includes: acquiring historical signal strength data of the data acquisition terminal; dividing the data into multiple historical windows according to a preset window length; calculating the signal strength change of adjacent sampling points within each historical window; determining the fluctuation trend and fluctuation intensity; and constructing a historical window library based on fluctuation intensity classification; acquiring real-time signal strength data according to the preset window length to form a real-time window; calculating the fluctuation intensity of the real-time window to determine the window type; retrieving the corresponding historical window sub-library from the historical window library based on the determination result; and calculating the weight of each sampling point within the real-time window based on the time decay characteristics; and constructing a real-time window and historical window library based on the DTW algorithm. The weighted difference matrix between historical windows is used to adaptively adjust the path constraint width and search for the optimal path by combining the fluctuation differences between the real-time window and the historical window. Based on the optimal path, the matching degree between the real-time window and each historical window is calculated. The fluctuation characteristics of the real-time window are retrieved to calculate the randomness of the real-time signal strength data and determine the number of fusion windows. Multiple historical windows with matching degrees are selected from the historical window sub-library, and the associated successful transmission power is retrieved for weighted fusion to obtain the wireless transmission power of the real-time window at the next moment. Based on the wireless transmission power at the next moment, the wireless transmission module of the data acquisition terminal is controlled to complete the wireless transmission power control of the data acquisition terminal.

[0006] Optionally, determining the fluctuation trend and fluctuation intensity includes the following steps: Using any historical window as the target window, calculate the change in signal intensity of adjacent sampling points within the target window, construct a change sequence, perform a trend test on the change sequence, and calculate the significance value of the target window to obtain the fluctuation trend of the target window; calculate the mean of the change in signal intensity of adjacent sampling points within the target window as the fluctuation intensity.

[0007] Optionally, the step of building the history window library includes: Using three times the standard deviation of the volatility intensity as the stationarity threshold, if the volatility intensity is less than the stationarity threshold, the corresponding historical window is defined as a stationar window; otherwise, if the volatility intensity is greater than or equal to the stationarity threshold, the corresponding historical window is defined as a volatility window. The stable window is labeled with 0, and the fluctuating window is labeled with 1. The labels of all historical windows are used to build a historical window library. Each historical window corresponds to a historical signal strength sequence, window type label, fluctuation intensity and fluctuation trend. At the same time, each historical window is associated with the wireless transmission power at the time of successful communication.

[0008] Optionally, the step of calculating the weight of each sampling point within the real-time window includes: Based on the fluctuation characteristics of the real-time window, the window type label is confirmed and a preset attenuation coefficient is selected. The normalized position of each sampling point in the real-time window is calculated. The position is multiplied by the attenuation coefficient to obtain the time attenuation factor. The time attenuation factor is mapped through the Sigmoid function to obtain the weight of each sampling point in the real-time window.

[0009] Optionally, the weighted difference matrix can be obtained in the following ways: Calculate the first real-time signal strength sequence corresponding to the real-time window. The sampling point and the historical signal intensity sequence of the nth sampling point The signal strength difference at each sampling point is calculated by squaring the difference to obtain the squared difference. This squared difference is then multiplied by the value at the nth sampling point in the real-time signal strength sequence. The weights of the sampling points are used to obtain the weighted difference matrix of the i-th sampling point. Line number The element values ​​of the column.

[0010] Optionally, the step of adjusting the path constraint width includes: Calculate the difference in fluctuation trend and the difference in fluctuation intensity between the real-time window and the historical window. Perform exponential operations on the difference in fluctuation trend and the difference in fluctuation intensity respectively and add them together to obtain the fluctuation difference index. Multiply the fluctuation difference index, the total number of sampling points in the real-time window and the preset adjustment coefficient respectively and round up to obtain the path constraint width.

[0011] Optionally, the calculation of the randomness of the real-time signal strength includes the following steps: The fluctuation intensity of the real-time window is exponentially mapped using a negative exponential function, and the result of subtracting 1 from the mapping is taken as the fluctuation intensity contribution. The absolute value of the fluctuation trend of the real-time window is exponentially mapped using a negative exponential function to obtain the fluctuation trend contribution. The product of the fluctuation intensity contribution and the fluctuation trend contribution is taken as the randomness of the real-time window.

[0012] Optionally, the method for obtaining the wireless transmission power at the next moment of the real-time window includes: The matching degrees of all fusion windows are summed to obtain the total matching degree of the fusion windows. The ratio between the matching degree of each fusion window and the total matching degree is used as the weight coefficient of each fusion window. The weight coefficient of each fusion window is multiplied by the corresponding associated transmission power to obtain the contribution value of each fusion window to the predicted power. The contribution values ​​of all fusion windows are accumulated and summed to obtain the predicted wireless transmission power of the real-time window at the next moment.

[0013] The present invention has the following effects: 1. This invention constructs a dedicated sub-library by classifying historical signal strength windows according to fluctuation intensity, assigns differentiated weights to real-time window sampling points based on time decay characteristics, and constructs a weighted difference matrix based on the DTW algorithm and adaptively adjusts the path constraint width. This not only solves the problem of traditional single sequence matching ignoring the differences in data contribution within the window, but also adapts to the differences in channel fluctuation characteristics, significantly improving the matching accuracy between real-time and historical sequences, and thus outputting wireless transmission power prediction results that are more in line with the actual channel conditions.

[0014] 2. This invention associates the wireless transmission power of successful communications with historical windows and achieves power prediction in advance based on weighted fusion of highly similar historical windows, avoiding the adjustment lag problem caused by the reliance on real-time signal strength feedback in traditional methods. At the same time, the predicted power not only adapts to the current channel state to ensure data transmission success rate, but also avoids redundant power consumption through precise matching, effectively balancing transmission reliability and terminal battery life.

[0015] 3. This method calculates the degree of randomness by real-time window fluctuation characteristics and dynamically adjusts the number of fusion windows. When the channel fluctuation is complex and highly random, the reference window is expanded to cover diverse fluctuation characteristics, and when the channel is stable, the reference window is simplified to improve computational efficiency. Combined with window type-differentiated configuration of attenuation coefficient and adaptive adjustment of path constraint width, the method can flexibly adapt to dynamic channel changes caused by factors such as interference from nearby devices and personnel movement, and can maintain stable power control performance in various complex wireless communication scenarios. Attached Figure Description

[0016] Figure 1 This is a flowchart of steps S1-S4 in a wireless transmission power control method for a data acquisition terminal according to an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0018] Reference Figure 1 A wireless transmission power control method for a data acquisition terminal includes steps S1-S4, as detailed below: S1: Obtain historical signal strength data from the data acquisition terminal, divide the data into multiple historical windows according to the preset window length, calculate the signal strength change of adjacent sampling points within each historical window, determine the fluctuation trend and fluctuation intensity, and construct a historical window library according to the fluctuation intensity classification.

[0019] A sampling interval is set, for example, 10 seconds / sample. Historical signal strength data is collected using the built-in sensor of the wireless transmission module of the data acquisition terminal. A preset window duration is set, for example, 200 seconds. Each window contains 20 sampling points. The historical signal strength data is divided chronologically to obtain multiple consecutive historical signal strength sequences. Each historical window corresponds to a historical signal strength sequence, and each historical window is associated with a wireless transmission power, which is the actual wireless transmission power used when the first uplink data packet is successfully transmitted after the historical window ends. Min-max normalization is performed on the historical signal strength sequences and associated wireless transmission powers of all historical windows, mapping them to... Interval.

[0020] Using any historical window as the target window, calculate the signal intensity change of adjacent sampling points within the target window, construct a change sequence, and based on the change sequence, use the MK (Mann-Kendall) trend test method to calculate the trend significance corresponding to the target window. The value is used to obtain the fluctuation trend of the target window. Among them, This indicates that the change in signal strength is trending upward (i.e., the fluctuation amplitude within the window is gradually increasing). This indicates that the change in signal strength is decreasing (i.e., the fluctuation amplitude within the window is gradually decreasing). The larger the value, the more significant the fluctuation trend.

[0021] The mean value of the signal strength change among adjacent sampling points within the target window is calculated as the fluctuation intensity. Specifically, the fluctuation intensity satisfies the following relationship: ; In the formula, Indicates the fluctuation intensity of the target window. This indicates the total number of sampling points in the target window. This represents the rate of change in signal quality between adjacent sampling points. Indicates the first Signal quality at each sampling point.

[0022] After obtaining the fluctuation intensity and trend of all historical windows, a stability threshold S0 is set (determined based on the 3σ rule of historical signal intensity data, where σ is the standard deviation of the fluctuation intensity of all historical windows). Fluctuation is then categorized according to intensity: Historical windows with fluctuation intensity less than the stability threshold are defined as stable windows and labeled with 0; historical windows with fluctuation intensity greater than or equal to the stability threshold are defined as fluctuating windows and labeled with 1. A historical window library is constructed based on the label of each historical window. Historical windows with label 0 are stored in the stable window library, and historical windows with label 1 are stored in the fluctuating window library. Each historical window corresponds to a historical signal intensity sequence, a window type label, and a fluctuation intensity... and fluctuation trend Wireless transmission power .

[0023] S2: Obtain real-time signal strength data according to the preset window length to form a real-time window, calculate the fluctuation intensity of the real-time window to determine the window type, and according to the determination result, retrieve the corresponding historical window sub-library from the historical window library, and calculate the weight of each sampling point in the real-time window based on the time decay characteristics.

[0024] It should be noted that the length of the real-time window is the same as the length of the historical window, and the real-time window corresponds to a real-time signal strength sequence.

[0025] A sliding sampling window of the same length as the historical window is used to collect signal strength data from the data acquisition terminal in real time, forming a real-time signal strength sequence. The fluctuation intensity of the real-time window is calculated. If the fluctuation intensity of the real-time window is less than the stability threshold, the real-time window is determined to be a stable window (labeled 0); otherwise, if it is greater than or equal to the stability threshold, the real-time window is determined to be a fluctuating window (labeled 1). Based on the type of the real-time window, a corresponding historical window library is selected, and weights are assigned to each sampling point within the real-time window based on the time decay principle. Sampling points closer to the end of the window are more critical for predicting the next moment, and therefore have higher weights.

[0026] Based on the fluctuation characteristics of the real-time window, determine whether the window is a stable window or a fluctuating window and select a preset attenuation coefficient. Calculate the normalized position of each sampling point in the real-time window, multiply the position by the attenuation coefficient to obtain the time attenuation factor, and map the time attenuation factor using the Sigmoid function to obtain the weight of each sampling point in the real-time window.

[0027] Specifically, the weights satisfy the following relationship: ; In the formula, Indicates the number of times within the real-time window The weight of each sampling point This indicates the total number of sampling points in the real-time window; The label indicating the live window. This indicates that the real-time window is a fluctuating window. This indicates that the real-time window is a stable window; , Indicates the earliest time. The latest time; , These refer to the attenuation coefficients corresponding to fluctuating and stable real-time windows, respectively. , .

[0028] Through the Sigmoid function expression Construct a basic growth model for the weights. The core function of this basic growth model is to ensure that the weight values ​​increase with... The weight increases monotonically with the increase of the sampling point's time index, meaning that the closer the sampling point's time is to the end of the window, the greater its corresponding base weight. This intuitively reflects the relationship between the weight and the sampling point's time index. The positive correlation between them allows later sampling points to be given higher priority in subsequent power prediction.

[0029] Based on this, combined with the type label of the real-time window ( Indicates a stable window. (representing a fluctuating window), allowing for the configuration of differentiated attenuation coefficients for different window types (attenuation coefficients corresponding to stable windows). The fluctuation window corresponds to the attenuation coefficient. This forms a piecewise weight calculation function that adapts to the window characteristics, and strictly limits the attenuation coefficient to satisfy... It can achieve indexing at the same time. The design ensures that the sampling point weight corresponding to the fluctuating window is always higher than that corresponding to the stable window. The core purpose of this design is to enhance the recognizability of the fluctuation features of the fluctuating window in subsequent similar sequence matching: when performing shape matching between real-time windows and historical windows, the sampling point weight of the fluctuating window is higher, and its corresponding channel fluctuation features are given a stronger weight in the difference calculation. This effectively avoids the problem of fluctuation features being masked due to the difference in the magnitude of channel fluctuation amplitude, and ensures that the change pattern of the fluctuating channel can be accurately captured.

[0030] For fluctuation window ( Because signal strength fluctuates drastically, sampling points at the end of the window can more accurately capture the latest changes in the channel, providing a stronger reference value for predicting the wireless transmission power at the next moment. Therefore, a larger attenuation coefficient is used. ,pass right The amplification effect causes the weight to change over time. The increasing trend is steeper, further strengthening the weighting of the last sampling point and highlighting the impact of the latest channel change characteristics on the prediction results; for a stable window ( The signal strength changes gradually and the overall trend is stable. The channel characteristics of each sampling point within the window are all of certain reference value. There is no need to overemphasize the weight of the last sampling point. Therefore, a smaller attenuation coefficient is used. weaken The degree of influence on the weights is determined to ensure that the weight differences among the sampling points within the window are within a small range, thus ensuring a comprehensive consideration of multi-time characteristics under stable channel conditions.

[0031] S3: Construct a weighted difference matrix between the real-time window and the historical window based on the DTW algorithm, adaptively adjust the path constraint width by combining the fluctuation differences between the real-time window and the historical window, and search for the optimal path. Calculate the matching degree between the real-time window and each historical window based on the optimal path.

[0032] The first in the history window library Taking a historical window as an example, calculate the weighted difference matrix between the real-time signal strength sequence and the historical signal strength sequence within the historical window. Specifically, the matrix is Size, where rows correspond to real-time signal strength sequence indices. The column corresponds to the historical signal strength sequence index. .

[0033] Calculate the first in the real-time signal strength sequence The sampling point and the historical signal intensity sequence of the nth sampling point The signal strength difference at each sampling point is squared to amplify the difference and prevent positive and negative values ​​from canceling each other out. The squared difference is then multiplied by the nth value in the real-time signal strength sequence. The weights of the sampling points are used to obtain the weighted difference matrix of the i-th sampling point. Line number The element values ​​of the column.

[0034] Specifically, the elements of the weighted difference matrix satisfy the following relationship: ; In the formula, This represents the real-time signal strength sequence and the first [number] in the historical window library. In the weighted difference matrix of the nth historical signal intensity sequence Line number The element values ​​of the column; Represents the first in the real-time signal strength sequence The weight of each sampling point; Represents the weighted difference matrix in which the first element is... Line number The signal strength difference between columns.

[0035] In the DTW (Dynamic Time Warping) algorithm, the squaring operation can significantly amplify the differences between signal strengths, making the matching process more sensitive to larger differences. Because small changes in signal strength can have a significant impact on the reliability of data transmission, the squaring operation ensures that the difference value is always non-negative, avoiding the situation where positive and negative differences cancel each other out, thus more accurately reflecting the actual difference between the two sequences.

[0036] In a real-time signal strength sequence, the sampling points closer to the end have a higher weight, which amplifies the matching differences in the overall sequence. When searching for the optimal path later, the system tends to match the later points in the real-time signal strength sequence with points in the historical signal strength sequence that have similar shapes.

[0037] In real-world wireless communication scenarios, the signal strength of data acquisition terminals is easily affected by factors such as the start-up and shutdown of nearby equipment, switching of operating conditions, personnel movement, and multipath effects, causing rapid fluctuations. For example, the start-up and shutdown of production equipment in industrial workshops, personnel movement outdoors, and interference from wireless devices in urban areas can all lead to instantaneous changes in signal strength. The later time point (most recent sampling point) of the real-time signal strength sequence carries the latest channel fluctuation information and has higher reference value for predicting the wireless transmission power at the next moment.

[0038] Traditional prediction methods, if overly reliant on early data in the sequence, can easily lead to prediction results lagging behind actual channel changes: when the signal strength drops sharply, the predicted power may be too low due to referencing early stable data, causing transmission failure; when the signal strength improves, the predicted power may be too high due to over-reliance on early data, resulting in wasted power.

[0039] By assigning higher weights to later time points, the prediction model can focus on referencing the latest channel change information in similar sequence matching, more sensitively capturing recent fluctuation trends. This not only improves the accuracy of wireless transmission power prediction and ensures transmission reliability, but also avoids redundant power consumption and extends terminal battery life.

[0040] The path constraint width is dynamically set based on the fluctuation intensity and fluctuation trend of the real-time window. The fluctuation trend difference and fluctuation intensity difference between the real-time window and the historical window are calculated. The fluctuation trend difference and fluctuation intensity difference are exponentially calculated and added to obtain the fluctuation difference index. The fluctuation difference index, the total number of sampling points of the real-time window and the preset adjustment coefficient are multiplied and rounded up to obtain the path constraint width.

[0041] Specifically, the path constraint width satisfies the following relationship: ; In the formula, Indicates the real-time window corresponds to the first The path constraint width for each historical window. The adjustment factor representing the path constraint width can be set to 0.01 to prevent the path constraint width from being too wide. Too large; This indicates the fluctuation intensity of the real-time window compared to the first window in the historical window library. The difference in fluctuation intensity between historical windows is used to accommodate morphological shifts in drastic fluctuations. For example, when the power fluctuation amplitudes between two windows differ significantly, a larger time offset (i.e., a wider path constraint width) is allowed to better match the morphology of the two sequences. This is because sequences with drastic fluctuations may be more prone to temporal shifts, and a wider path constraint width can accommodate such shifts, thereby improving the flexibility and accuracy of the matching. Indicates the real-time window and the first The difference in fluctuation trends between two historical windows is used to accommodate time shifts caused by differences in the trends of fluctuation amplitude changes. When the signal strength change trends between two windows differ significantly, a larger time shift (i.e., an increased path constraint width) is allowed to better match the trend changes of the two sequences. This is because trend differences can lead to significant time shifts between the two sequences, and a larger path constraint width can accommodate such shifts, thereby improving the flexibility and accuracy of the matching. This indicates the total number of sampling points in the real-time window.

[0042] By calculating the constraint path width It can dynamically adjust the search range of the matching path based on the fluctuation characteristics of the real-time window and the historical window, thereby adapting to different fluctuation conditions and improving the accuracy and efficiency of matching. For example, when Poor volatility trend Poor fluctuation intensity adjustment coefficient At that time, the calculation yielded This indicates that the maximum allowed time offset is 3 sampling points when matching the real-time window and the historical window. This means that when constructing the weighted difference matrix, the matching path for each sampling point can only be searched within the range of 3 sampling points before and after the current point, thereby avoiding meaningless matching and improving matching efficiency and accuracy.

[0043] The greater the volatility intensity or the more significant the difference in volatility trends, the wider the path constraint width. Adapting to drastic fluctuations in shape shifts. The time shift caused by the difference in the trend of fluctuation amplitude is adapted. For any element in the weighted difference matrix... Only when the following conditions are met Only then will the point be included in the path search scope.

[0044] Combining the weighted difference matrix and the path constraint width, dynamic programming is used to search for the shortest path in the weighted difference matrix, i.e., the path between the real-time signal strength sequence and the historical window library. The DTW distance of each historical signal intensity sequence is used as the matching degree between the real-time signal intensity sequence corresponding to the real-time window and the historical signal intensity sequence corresponding to each historical window.

[0045] S4: Retrieve the fluctuation characteristics of the real-time window, calculate the randomness of the real-time signal strength, determine the number of fusion windows, select multiple historical windows with matching degrees from the historical window sub-library, and retrieve the associated successful transmission power for weighted fusion to obtain the wireless transmission power of the real-time window at the next moment. Based on the wireless transmission power at the next moment, control the wireless transmission module of the data acquisition terminal to complete the wireless transmission power control of the data acquisition terminal.

[0046] Because the time-varying characteristics of real-time channels vary significantly, the degree of randomness directly determines the number of historical windows required for reference. If the channel state is highly random, it means that its fluctuation pattern is complex and has no fixed trend. Relying on only a small number of historical windows is not enough to fully cover its possible fluctuation characteristics, which can easily lead to biased reference data and thus cause power prediction deviation. If the channel state is not highly random, it means that its fluctuation pattern is relatively stable. Introducing too many historical windows will increase the cost of redundant data processing and reduce the real-time response efficiency of the terminal.

[0047] Therefore, it is necessary to first quantify the randomness of real-time signal strength data, and then dynamically adjust the number of fusion windows based on the quantification results to ensure that power prediction has sufficient and effective references while also taking into account computational efficiency. The specific steps are as follows: The fluctuation intensity of the real-time window is exponentially mapped using a negative exponential function, and the result of subtracting 1 from the mapping is taken as the fluctuation intensity contribution. The absolute value of the fluctuation trend of the real-time window is exponentially mapped using a negative exponential function to obtain the fluctuation trend contribution. The product of the fluctuation intensity contribution and the fluctuation trend contribution is taken as the randomness of the real-time window.

[0048] Specifically, the degree of randomness satisfies the following relationship: ; In the formula, Indicates the degree of randomness of the real-time window; Indicates the intensity of fluctuations in the real-time window; The absolute value representing the fluctuation trend of the real-time window; through Achieve fluctuation intensity The larger the value, the more pronounced the fluctuation in signal strength per unit time. The larger; through Achieving absolute value of fluctuation trend The smaller the value, the less clear the trend in signal strength, indicating that there is no clear pattern of increase or decrease. The larger.

[0049] In the process of wireless transmission power prediction, the number of historical windows selected needs to be adapted to the randomness of real-time signal strength data in order to balance prediction accuracy and terminal operating efficiency. If the randomness of real-time signal strength is high, it indicates that the channel fluctuation characteristics are complex and without fixed patterns, and there may be multiple differentiated fluctuation modes. In this case, if a fixed number of historical windows are used as references, it will not be able to fully cover all kinds of historical fluctuation characteristics that the current channel may correspond to, resulting in one-sided reference data. Consequently, the predicted power derived from this part of the reference data will deviate significantly from the actual channel requirements, affecting the reliability of data transmission. Conversely, if the randomness of real-time signal strength is low, it means that the channel fluctuation pattern is relatively stable, and only a few historical windows with high similarity are needed to provide sufficient reference. In this case, if too many historical windows are selected, a large amount of redundant data with very low correlation to the current channel characteristics will be introduced. This will not only increase the data processing volume and computational complexity of the terminal, but also prolong the response time of power prediction and reduce the terminal's real-time adaptability to dynamic channel changes. Therefore, the number of historical windows used for power fusion is dynamically adjusted based on the randomness. The specific steps are as follows: The maximum number of fusion windows is determined by multiplying the maximum number of fusion windows set by the randomness of the real-time windows and rounding it up. The maximum value between the rounded value and 1 is selected as the number of fusion windows that can be selected for the real-time windows.

[0050] Obtain the matching degree of each historical window corresponding to the real-time window, sort them in ascending order from smallest to largest, select the first preset number of historical windows as the fusion window set for this power prediction based on the number of fusion windows, retrieve the wireless transmission power associated with each fusion window in the set, perform weighted fusion, and determine the predicted wireless transmission power of the real-time window at the next moment.

[0051] The matching degrees of all fusion windows are summed to obtain the total matching degree of the fusion windows. The ratio between the matching degree of each fusion window and the total matching degree is used as the weight coefficient of each fusion window. The weight coefficient of each fusion window is multiplied by the corresponding associated transmission power to obtain the contribution value of each fusion window to the predicted power. The contribution values ​​of all fusion windows are summed to obtain the predicted wireless transmission power of the real-time window at the next moment.

[0052] Specifically, the predicted wireless transmission power at the next moment satisfies the following relationship: ; In the formula, This represents the predicted wireless transmission power for the next moment in the real-time window. Indicates the number of fusion windows; Indicates the first The degree of matching between the merged window and the real-time window; Indicates the first The wireless transmission power associated with each fusion window.

[0053] The predicted wireless transmission power for the next time step within the real-time window is calculated using an inverse normalization formula. Specifically, the inverse normalization formula satisfies the following relationship: ; In the formula, This represents the raw power at the next moment. This represents the predicted wireless transmission power for the next moment in the real-time window. This represents the maximum value of the original power data. This represents the minimum value of the original power data.

[0054] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for controlling wireless transmission power in a data acquisition terminal, characterized in that, include: Acquire historical signal strength data from the data acquisition terminal, divide the historical window into multiple historical windows according to the preset window length, calculate the signal strength change of adjacent sampling points in each historical window, determine the fluctuation trend and fluctuation intensity, and construct a historical window library according to the fluctuation intensity classification. Real-time signal strength data is obtained according to the preset window length to form a real-time window. The fluctuation intensity of the real-time window is calculated to determine the window type. Based on the determination result, the corresponding historical window sub-library in the historical window library is retrieved, and the weight of each sampling point in the real-time window is calculated based on the time decay characteristics. A weighted difference matrix between the real-time window and the historical window is constructed based on the DTW algorithm. The path constraint width is adaptively adjusted by combining the fluctuation difference between the real-time window and the historical window and the optimal path is searched. The matching degree between the real-time window and each historical window is calculated based on the optimal path. The fluctuation characteristics of the real-time window are retrieved, the randomness of the real-time signal strength data is calculated, and the number of fusion windows is determined. Multiple historical windows with matching degrees are selected from the historical window sub-library, and the associated successful transmission power is retrieved for weighted fusion to obtain the wireless transmission power of the real-time window at the next moment. Based on the wireless transmission power at the next moment, the wireless transmission module of the data acquisition terminal is controlled to complete the wireless transmission power control of the data acquisition terminal.

2. The wireless transmission power control method for a data acquisition terminal according to claim 1, characterized in that, Determining the fluctuation trend and fluctuation intensity includes the following steps: Using any historical window as the target window, calculate the change in signal intensity of adjacent sampling points within the target window, construct a change sequence, perform a trend test on the change sequence, and calculate the significance value of the target window to obtain the fluctuation trend of the target window. The mean value of the signal intensity change of adjacent sampling points within the target window is calculated as the fluctuation intensity.

3. The wireless transmission power control method for a data acquisition terminal according to claim 1, characterized in that, The steps for building the history window library include: Using three times the standard deviation of the volatility intensity as the stationarity threshold, if the volatility intensity is less than the stationarity threshold, the corresponding historical window is defined as a stationar window; otherwise, if the volatility intensity is greater than or equal to the stationarity threshold, the corresponding historical window is defined as a volatility window. The stable window is labeled with 0, and the fluctuating window is labeled with 1. The labels of all historical windows are used to build a historical window library. Each historical window corresponds to a historical signal strength sequence, window type label, fluctuation intensity and fluctuation trend. At the same time, each historical window is associated with the wireless transmission power at the time of successful communication.

4. The wireless transmission power control method for a data acquisition terminal according to claim 1, characterized in that, The steps for calculating the weights of each sampling point within the real-time window include: Based on the fluctuation characteristics of the real-time window, the window type label is confirmed and a preset attenuation coefficient is selected. The normalized position of each sampling point in the real-time window is calculated. The position is multiplied by the attenuation coefficient to obtain the time attenuation factor. The time attenuation factor is mapped through the Sigmoid function to obtain the weight of each sampling point in the real-time window.

5. The wireless transmission power control method for a data acquisition terminal according to claim 1, characterized in that, The weighted difference matrix is ​​obtained in the following ways: Calculate the first real-time signal strength sequence corresponding to the real-time window. The sampling point and the historical signal intensity sequence of the nth sampling point The signal strength difference at each sampling point is calculated. The squared difference is then multiplied by the sum of the squared differences at each sampling point in the real-time signal strength sequence. The weights of the sampling points are used to obtain the weighted difference matrix of the i-th sampling point. Line number The element values ​​of the column.

6. The wireless transmission power control method for a data acquisition terminal according to claim 1, characterized in that, The step of adjusting the path constraint width includes: Calculate the difference in fluctuation trend and the difference in fluctuation intensity between the real-time window and the historical window. Perform exponential operations on the difference in fluctuation trend and the difference in fluctuation intensity respectively and add them together to obtain the fluctuation difference index. Multiply the fluctuation difference index, the total number of sampling points in the real-time window and the preset adjustment coefficient respectively and round up to obtain the path constraint width.

7. The wireless transmission power control method for a data acquisition terminal according to claim 1, characterized in that, The calculation of the randomness of real-time signal strength includes the following steps: The fluctuation intensity of the real-time window is exponentially mapped using a negative exponential function, and the result of subtracting 1 from the mapping is taken as the fluctuation intensity contribution. The absolute value of the fluctuation trend of the real-time window is exponentially mapped using a negative exponential function to obtain the fluctuation trend contribution. The product of the contribution of volatility intensity and the contribution of volatility trend is used as the degree of randomness of the real-time window.

8. The wireless transmission power control method for a data acquisition terminal according to claim 1, characterized in that, The methods for obtaining the wireless transmission power at the next moment of the real-time window include: The matching degrees of all fusion windows are summed to obtain the total matching degree of the fusion windows. The ratio between the matching degree of each fusion window and the total matching degree is used as the weight coefficient of each fusion window. The weight coefficient of each fusion window is multiplied by the corresponding associated transmission power to obtain the contribution value of each fusion window to the predicted power. The contribution values ​​of all fusion windows are summed to obtain the predicted wireless transmission power of the real-time window at the next moment.

Images