A method and system for optimizing data transmission quality for smart meters

Abstract

The present application relates to the field of data processing, more particularly, the present application relates to a kind of data transmission quality optimization method and system for smart meter, method includes: obtaining each meter historical sensitivity sequence and key update record, constructs transmission sensitive vector;To meter cluster and give sensitive grade;Transmission cycle is divided into multiple time slot and establishes corresponding transmission scheduling group, according to sensitive grade, meter is distributed to each group to form initial scheduling scheme;Network and computing resource state is collected when current transmission cycle starts, computing resource stress;Resource stress is compared with double threshold, according to the comparison result, with the sum of transmission sensitivity of group meter as load, according to load height, dynamically adjust the corresponding order of each transmission scheduling group and transmission time slot and issue transmission instruction.The present application realizes meter group peak-shaving transmission by the double-layer scheduling mechanism of the combination of historical behavior classification and real-time resource perception, significantly improves concentrator processing efficiency.

Description

Technical Field

[0001] This invention relates to the field of data processing. More specifically, this invention relates to a method and system for optimizing data transmission quality in smart meters. Background Technology

[0002] As key metering devices in the power system, smart meters need to upload data to the concentrator regularly. With the surge in the number of meters, concurrent data uploads from multiple meters can easily lead to network congestion, peak instantaneous load on the concentrator, increased transmission latency, and increased packet loss rate, seriously affecting the real-time performance and reliability of the data.

[0003] In the prior art, for example, the patent authorization document CN120934759B provides a method and apparatus for encrypted transmission of electricity meter data based on edge computing. This method for encrypted transmission of electricity meter data dynamically decides whether to update the encryption key by performing sensitivity analysis on a single electricity meter, thereby reducing unnecessary encryption and decryption overhead while ensuring security.

[0004] However, existing technologies only optimize the encryption strategy for a single meter. When a large number of meters transmit data simultaneously, the concentrator will still face challenges in terms of computing resources and network bandwidth competition, leading to a decline in overall transmission quality. Summary of the Invention

[0005] To address the technical problem that existing technologies only optimize encryption strategies for single meters, and that concentrators still face challenges in computing resource constraints and network bandwidth competition when a large number of meters transmit data simultaneously, leading to a decline in overall transmission quality, this invention provides solutions in the following aspects.

[0006] In a first aspect, a method for optimizing data transmission quality for smart meters includes:

[0007] The system acquires a sensitivity sequence composed of the sensitivity calculated from each meter in the meter group during each transmission within a historical period, as well as a key update record calculated based on the sensitivity exceeding a preset threshold; the meters report data to the concentrator according to a preset transmission period.

[0008] Based on the sensitivity sequence and key update record, a transmission sensitivity vector is constructed for the corresponding electricity meter. Using all transmission sensitivity vectors as samples, the electricity meter group is clustered to obtain multiple electricity meter clusters, and a sensitivity level is assigned to each electricity meter cluster.

[0009] A transmission cycle is divided into multiple transmission time slots, and a transmission scheduling group is established for each transmission time slot. The meter group is allocated to multiple transmission scheduling groups according to the sensitivity level of each meter, forming an initial correspondence between each transmission scheduling group and the transmission time slot.

[0010] At the start of the current transmission cycle, collect the current network status parameters and the concentrator calculates the resource status parameters, and calculate the resource pressure at the start of the current transmission cycle based on all collected status parameters.

[0011] The resource pressure is compared with a preset first pressure threshold and a second pressure threshold; based on the comparison result, the sum of the transmission sensitivity of all meters in the transmission scheduling group is used as the load, and the corresponding order of the multiple transmission scheduling groups and transmission slots is rearranged according to the load level; wherein, the transmission sensitivity is calculated from the components in the transmission sensitivity vector;

[0012] At the beginning of each transmission time slot, a transmission instruction is sent to all meters in the transmission scheduling group corresponding to that transmission time slot.

[0013] Preferably, the transmission sensitivity vector includes a normalized fluctuation coefficient and a key update frequency;

[0014] Specifically, the coefficient of variation of the sensitivity sequence is calculated and normalized to obtain the normalized fluctuation coefficient; the ratio of the number of times the sensitivity exceeds the preset threshold within the historical period to the total number of transmissions within the historical period is calculated.

[0015] The transmission sensitivity is the product of the normalized fluctuation coefficient and the key update frequency.

[0016] Preferably, assigning a sensitivity level to each meter cluster includes:

[0017] Using the transmission sensitivity vector of each meter as a sample, the K-means clustering algorithm is used to divide the meter group into several meter clusters;

[0018] Calculate the cluster sensitivity index for each meter cluster, where the cluster sensitivity index is the arithmetic mean of the transmission sensitivity of all meters in the meter cluster.

[0019] The sensitivity levels of each meter cluster are sequentially marked according to the cluster sensitivity index in ascending order.

[0020] Preferably, the step of allocating the meter group to multiple transmission scheduling groups according to the sensitivity level of each meter includes:

[0021] Electricity meters with a sensitivity level greater than the preset threshold are classified as high-sensitivity meters. Each high-sensitivity meter is then assigned to all transmission scheduling groups in a round-robin fashion according to its sensitivity level from high to low.

[0022] Electricity meters with a sensitivity level less than or equal to a preset threshold are classified as low-sensitivity meters. A greedy load balancing algorithm is used to sequentially assign each low-sensitivity meter to the transmission scheduling group with the lowest initial load.

[0023] The initial load of the transmission scheduling group is the sum of the transmission sensitivities of the meters already allocated within the group.

[0024] Preferably, the process of using the greedy equilibrium algorithm includes:

[0025] Create an empty member list for each transmission scheduling group and initialize the load of each transmission scheduling group to zero;

[0026] Select any low-sensitivity meter to be assigned as the target meter, and calculate the temporary load of the transmission scheduling group after the target meter is temporarily added to each transmission scheduling group. Construct a load set by combining the initial load of all other transmission scheduling groups with the temporary load of the transmission scheduling group to which the target meter is added. Then, iterate through and assign the target meter to all transmission scheduling groups to obtain multiple load sets. Calculate the sample variance of each load set, select the transmission scheduling group to which the target meter belongs in the load set with the smallest sample variance as the target group, add the corresponding low-sensitivity meter to the empty member list constructed by the target group, and update the temporary load of the target group.

[0027] The algorithm terminates once all low-sensitivity meters have been allocated.

[0028] Preferably, the calculation process for the resource pressure includes:

[0029] Extract the bandwidth utilization and packet loss rate from the network status parameters and the CPU utilization and memory usage rate from the computing resource status parameters; calculate the product of bandwidth utilization and packet loss rate, and the product of CPU utilization and memory usage rate for the current transmission cycle;

[0030] The average of the product of bandwidth utilization and packet loss rate of all transmission slots in the previous transmission cycle of the current transmission cycle is used as the first average. The average of the product of CPU utilization and memory utilization of all transmission slots in the previous transmission cycle of the current transmission cycle is used as the second average. The first average is divided by the sum of the first average and the second average to obtain the network state weight. The difference between 1 and the network state weight is used as the resource state weight.

[0031] The first component is obtained by multiplying the network state weight by the product of the bandwidth utilization and packet loss rate of the current transmission cycle; the second component is obtained by multiplying the resource state weight by the product of the CPU utilization and memory occupancy of the current transmission cycle; the resource pressure is obtained by adding the first component and the second component.

[0032] Preferably, the comparison results include:

[0033] The first pressure threshold is less than the second pressure threshold;

[0034] When the resource pressure is less than or equal to the first pressure threshold, the transmission scheduling group with the highest temporary load is adjusted to the first transmission slot, and the remaining transmission scheduling groups are allocated subsequent transmission slots in order of temporary load from high to low.

[0035] When the resource pressure is greater than the first pressure threshold and less than the second pressure threshold, the initial correspondence is maintained;

[0036] When the resource pressure is greater than or equal to the second pressure threshold, the transmission scheduling group with the lowest current load is adjusted to the first transmission slot, and the remaining transmission scheduling groups are assigned to the subsequent transmission slots in order of load from low to high.

[0037] In a second aspect, a data transmission quality optimization system for smart meters includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the data transmission quality optimization method for smart meters described in any one of the claims is implemented.

[0038] The beneficial effects of this invention are:

[0039] 1. This invention overcomes the limitations of existing technologies that only optimize encryption strategies for single electricity meters. It constructs a three-layer progressive scheduling framework from a system-level perspective: "historical behavior grading - static load planning - real-time resource feedback." By extracting the fluctuation characteristics of the historical sensitivity sequence of electricity meters and the key update frequency, a transmission sensitivity vector representing transmission risk is generated. Combined with clustering algorithms, objective risk grading of the meter group is achieved, solving the technical challenge of large differences in the behavior patterns of massive numbers of electricity meters and the difficulty in unified scheduling. This provides a refined decision-making basis for differentiated peak-shifting transmission.

[0040] 2. This invention divides the transmission cycle into time slots and establishes transmission scheduling groups. Based on the sensitivity level, it uses a round-robin allocation and greedy load balancing algorithm to generate an initial scheduling scheme, so that highly sensitive meters are evenly distributed and low-sensitive meters are evenly filled. This achieves static optimal load balancing between groups, avoids competition between the concentrator's instantaneous calculation peak and network bandwidth from the source, and significantly reduces transmission delay and packet loss risk.

[0041] 3. This invention introduces a dual-threshold resource pressure sensing mechanism, which integrates network bandwidth utilization, packet loss rate, and concentrator CPU and memory usage in real time to construct resource pressure indicators. Based on the comparison between the pressure indicators and preset thresholds, the execution sequence of each transmission scheduling group is dynamically adjusted, enabling the scheduling strategy to adapt to fluctuating resource environments: when resources are abundant, high-load groups are prioritized to shorten the overall cycle; when resources are scarce, low-load groups are prioritized to quickly reduce pressure. This mechanism ensures system stability while avoiding frequent strategy oscillations caused by minor fluctuations in indicators, significantly improving the robustness and environmental adaptability of the scheduling system. Attached Figure Description

[0042] Figure 1 This is a flowchart of steps S1-S3 in a data transmission quality optimization method for smart meters according to an embodiment of the present invention.

[0043] Figure 2 This is a schematic diagram of a data transmission quality optimization system for smart meters according to an embodiment of the present invention. Detailed Implementation

[0044] 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.

[0045] The specific application scenarios targeted by this invention are as follows:

[0046] This invention is built on the existing single-meter encrypted transmission framework disclosed in the prior art. The core of the prior art is: for a single smart meter, a sensitivity is calculated based on the degree of deviation between its historical data and the data to be transmitted, and the encryption key is dynamically determined based on whether the sensitivity exceeds a preset threshold, thereby optimizing the key management overhead while ensuring security.

[0047] This invention addresses the scenario where several smart meters (hereinafter referred to as "meter group") within a certain area need to periodically (e.g., hourly) upload data to the same concentrator (deployed on an edge computing node). In addition to processing data, the concentrator also has the capability to perform clustering calculations and transmission scheduling.

[0048] Reference Figure 1 A method for optimizing data transmission quality in smart meters includes steps S1-S3, as detailed below:

[0049] S1: Obtain the meter transmission sensitivity vector based on the meter's historical sensitivity and key update records.

[0050] To overcome the shortcomings of existing technologies that only focus on individual meter encryption and cannot assess system-level transmission risks, this invention aims to establish a feature system that can objectively and quantitatively characterize the potential impact of historical meter transmission behavior on concentrator load. This requires extracting two key dimensions of information from the historical data of each meter: the volatility of its data transmission and the frequency with which it triggers encryption computation overhead. This information will serve as the data foundation for subsequent scientific classification and intelligent scheduling.

[0051] A concentrator is known to manage a group of smart meters. Based on historical data, such as the past 30 days, the concentrator collects and calculates the following characteristics for each meter:

[0052] Obtain each segment of data to be transmitted as defined in the prior art. sensitivity (Its physical meaning is: a quantitative value of the degree of deviation between the data sequence to be transmitted and historical data from the same period. For details on its calculation method, please refer to the existing technology.)

[0053] The sensitivity of each transmission of the electricity meter in the historical period is sorted in chronological order to construct a sensitivity sequence. The mean and standard deviation of the sensitivity sequence are calculated respectively. The coefficient of variation of the sensitivity sequence is calculated based on the mean and standard deviation of the sensitivity sequence.

[0054] Calculating this coefficient of variation eliminates the influence of the absolute scale of the data, objectively reflecting the relative volatility of the sensitivity sequence. This facilitates horizontal comparisons between different meters. A meter with a high coefficient of variation indicates that its historical transmission behavior is variable and unpredictable. In future transmissions, it is more likely to unexpectedly trigger high-overhead encryption key update operations due to sudden changes in sensitivity, thus posing a potential, intermittent, high-load threat to the concentrator. This step transforms qualitative behavioral instability into a comparable, quantitative risk indicator.

[0055] In addition, the total number of transmissions of the electricity meter in the historical period, i.e. the total number of sensitivities, and the number of times the sensitivities exceeded a preset threshold (e.g., 0.6), are counted. Then the key update frequency is the ratio of the number of times the sensitivities exceeded the preset threshold to the total number of sensitivities.

[0056] By calculating the key update frequency, the key update behavior is quantified into a frequency value between 0 and 1, which intuitively reflects the magnitude of the historical encryption overhead of the electricity meter. The higher the key update frequency, the stronger the electricity meter's tendency to occupy computing resources in the past. The higher the key update frequency, the more it means that the electricity meter has occupied more computing resources in the past. More importantly, it reveals a certain inherent characteristic of the electricity consumption behavior or environment measured by the electricity meter (such as highly fluctuating loads or frequently started and stopped equipment), making it statistically likely that it will continue to have high encryption computing requirements in future transmissions.

[0057] Furthermore, for each meter in the meter group, the coefficient of variation of its sensitivity sequence is normalized (e.g., using the maximum-minimum normalization method) to obtain the normalized fluctuation coefficient. Then, the transmission behavior characteristics of each meter can be represented by a two-dimensional transmission sensitivity vector:

[0058]

[0059] In the formula, For the first The transmission behavior characteristics of an individual electricity meter For the first Normalized fluctuation coefficient of individual electricity meters For the first Key update frequency for each electricity meter.

[0060] The aforementioned transmission sensitivity vector comprehensively characterizes the potential impact pattern of the meter on the system load in historical transmission from two dimensions: volatility and encryption overhead, providing accurate input data for the next step of group analysis.

[0061] S2: Calculate the transmission sensitivity of the electricity meter based on the meter's transmission sensitivity vector and perform clustering to classify the sensitivity level.

[0062] The above-mentioned S1 has generated a transmission sensitivity vector describing the behavioral characteristics of each electricity meter. In order to implement differentiated scheduling strategies, it is necessary to fuse the two independent features in the transmission sensitivity vector to obtain an index that can comprehensively and singularly characterize the overall transmission risk level of the electricity meters, thereby ranking the electricity meters. At the same time, facing tens of thousands of electricity users, it is necessary to use unsupervised machine learning algorithms to automatically discover the potential risk group structure from the data itself, so as to achieve objective and dynamic classification.

[0063] Transmission sensitivity vectors of each meter Normalized fluctuation coefficient in and key update frequency Multiplying them together gives the transmission sensitivity of each meter.

[0064] The aforementioned transmission sensitivity uses multiplication rather than simple addition. Through the multiplicative relationship, a nonlinear amplification effect is created. Only when the meter has both a high key update frequency and a high relative fluctuation intensity will its sensitivity index be high, thus enabling the identification of high-risk meters that frequently trigger encryption and exhibit unstable behavior.

[0065] Furthermore, using the transmission sensitivity vectors of all the meters obtained in S1 as input, the K-means clustering algorithm is used to divide the meter group. The algorithm initializes the cluster centers using the K-means++ method, and iteratively calculates the similarity using the Euclidean distance between the transmission sensitivity vectors of the meters as a similarity measure until the cluster allocation result is stable or the preset maximum number of iterations is reached, for example, 100 times.

[0066] Among them, the meter cluster K is within a reasonable range (e.g., using the existing profile coefficient method) through existing technology. , The optimal value is dynamically determined for the total number of meters contained in the meter group.

[0067] The K-means clustering algorithm automatically and objectively identifies groups with different transmission behavior patterns within a group of electricity meters, avoiding subjective classification bias. Because the input vectors have been normalized, the algorithm can effectively classify data based on comprehensive information from two dimensions.

[0068] After the above clustering is completed, for the generated K meter clusters, the cluster sensitivity index of each cluster is calculated, which is the arithmetic mean of the transmission sensitivity of all meters in the cluster.

[0069] Furthermore, all meter clusters are sorted in ascending order of their cluster sensitivity indicators, and then labeled with sensitivity levels according to the sorting order. All meters within a meter cluster inherit the sensitivity level of the meter cluster.

[0070] Using the above-mentioned risk rating system, the electricity meter group is clearly divided into categories ranging from low risk (Level 1) to high risk (Level 2). This process assigns a clear sensitivity level label to each meter, establishing a continuous spectrum of meters, thus enabling refined risk classification of the meter group and providing a direct basis for subsequent differentiated scheduling.

[0071] In summary, the above-mentioned S2 transformation from individual risk quantification to group intelligent classification accurately captures the dual high-risk patterns that pose the greatest threat to system load in meter behavior by designing a transmission sensitivity with nonlinear characteristics. Then, through an autonomous clustering process, it objectively divides the meter group into sensitivity levels with different average risk levels.

[0072] S3: Generate and dynamically adjust peak-shifting transmission scheduling strategies based on meter sensitivity levels and real-time resource status.

[0073] The preliminary operations have completed the risk identification and classification of electricity meter transmission. The next step is to translate this into action, i.e., generate specific transmission control commands. However, an excellent scheduling strategy must possess two capabilities: first, the ability to perform preventative, static load balancing planning based on historical risk (level); and second, the ability to sense the real-time system status and make dynamic, emergency adjustments. This requires designing a two-tiered decision-making mechanism that integrates rule-based prevention and feedback-based adaptation.

[0074] First, the transmission period for data sequence transmission is set according to existing technology. ( =1 hour) means that the meter transmits a data sequence once per hour, and the transmission period is evenly divided into 1 hour. Number of transmission time slots The settings can be dynamically configured based on the total number of meters in the meter group and the network capacity, for example: , This represents the total number of meters contained in the meter group. This is the theoretical carrying capacity estimated based on the network bandwidth of a single time slot and the average data volume of a single electricity meter.

[0075] Furthermore, create A transmission scheduling group, set during initialization. The transmission scheduling groups correspond to the 1st, 2nd, 3rd, ..., Each transmission slot is selected, and then allocation begins. The specific allocation operation is as follows:

[0076] Distribute highly sensitive meters evenly: All highly sensitive meters (such as those with a sensitivity level greater than the threshold) The system iterates through each meter cluster in descending order of sensitivity level. For each meter cluster, a round-robin allocation method is used to sequentially assign all high-sensitivity meters to... Within each transmission scheduling group, ensure that meters from the same highly sensitive cluster are distributed as evenly as possible across different groups.

[0077] Low-sensitivity meter equalization: Replace the remaining low-sensitivity meters (i.e., those with a sensitivity level less than or equal to...) The load balancing algorithm is used for allocation. The operation of this greedy load balancing algorithm is as follows:

[0078] Initialization: Create an empty member list for each transmission scheduling group, set the sum of the transmission sensitivities of the allocated meters in each transmission scheduling group to 0, and call the sum of the transmission sensitivities of the allocated meters in the transmission scheduling group the initial load.

[0079] Iterative allocation: For each low-sensitivity meter to be matched, perform the following operations:

[0080] First, obtain the transmission sensitivity of each low-sensitivity meter.

[0081] Next, a low-sensitivity meter is selected as the target meter for allocation. Once the target meter is assigned to a transmission scheduling group, the sum of the temporary load (initial load) after the target meter is added and the transmission sensitivity of the target meter is calculated. Then, the initial loads of all other transmission scheduling groups and the temporary loads of the transmission scheduling group containing the target meter are combined to form a load set. Finally, the target meter is iterated through and allocated to all transmission scheduling groups to obtain the desired result. A load set.

[0082] Furthermore, the sample variance of each load set is calculated, and the transmission scheduling group corresponding to the target meter of the load set with the smallest sample variance is selected as the target group. If multiple sample variances are the same and the smallest, the transmission scheduling group with the smallest temporary load is selected first.

[0083] After obtaining the target group, the corresponding low-sensitivity meters can be added to the empty member list of the target group and the temporary load of the target group can be updated.

[0084] Finally, once all low-sensitivity meters have been allocated, the algorithm terminates, at which point the empty member list and final load of each transmission scheduling group are determined.

[0085] Through the above Each transmission scheduling group generates an initial time slot allocation plan, which preliminarily determines the order of transmission time slots corresponding to each transmission scheduling group. Here, it is assumed that the first transmission scheduling group corresponds to the first transmission time slot, and so on, thus forming the initial transmission scheduling scheme. The greedy load balancing algorithm aims to make the final load of all transmission scheduling groups as balanced as possible.

[0086] In summary, the preceding steps have completed risk classification based on historical behavior, and generated a highly balanced static transmission scheduling group accordingly. However, considering that network bandwidth and edge computing node resource status are dynamically fluctuating, unpredictable instantaneous pressure may arise due to background traffic, hardware failures, sudden tasks, etc. If the static scheduling plan is mechanically executed, and high-load groups are still transmitted sequentially when resources are scarce, it is very easy to trigger a chain reaction such as concentrator overload and increased network packet loss, leading to a deterioration in transmission quality. Therefore, based on the above static planning, a real-time resource status perception and feedback control mechanism is introduced to enable the scheduling strategy to have environmental adaptability. The specific operation is as follows:

[0087] First, at the start of the current transmission cycle, the concentrator collects network status parameters (current bandwidth utilization, packet loss rate) and its own computing resource status parameters (CPU utilization, memory usage) in real time. Based on the collected status parameters, it calculates the resource pressure at the start of the current transmission cycle, which can be expressed as:

[0088]

[0089] In the formula, This represents the resource pressure at the start of the current transmission cycle. As the first component, For the second component, This represents the real-time bandwidth utilization at the start of the current transmission cycle. This represents the real-time packet loss rate at the start of the current transmission cycle. This represents the real-time CPU utilization at the start of the current transmission cycle. This represents the real-time memory usage at the start of the current transmission cycle. For network state weights, To calculate resource state weights.

[0090] Specifically, the average of the product of bandwidth utilization and packet loss rate of all transmission slots in the previous transmission cycle of the current transmission cycle is calculated as the first average. The average of the product of CPU utilization and memory occupancy of all transmission slots in the previous transmission cycle of the current transmission cycle is calculated as the second average. The first average is divided by the ratio of the sum of the first average and the second average as the aforementioned network state weight.

[0091] In the above resource pressure calculation, high utilization accompanied by low packet loss indicates a good network condition, but high utilization accompanied by high packet loss is a clear signal of network congestion. Therefore, through calculation... This amplifies the correlation, meaning that network pressure only increases significantly when both are high.

[0092] Similarly, through calculation This can more effectively characterize the overall strain on computing resources. A high single indicator may only be a temporary fluctuation, but two indicators being high simultaneously usually means that computing resources are truly strained.

[0093] After obtaining the resource pressure at the start of the current transmission cycle, further dynamic adjustments are made to the initial transmission scheduling scheme described above. Each transmission scheduling group executes the transmission slot order within the current transmission cycle and issues instructions:

[0094] First, to achieve refined perception and differentiated control of system resource status, two resource pressure thresholds are further introduced. (marked as the first pressure threshold) and (Marked as the second pressure threshold) (satisfied) These two resource pressure thresholds can be determined by analyzing historical data. For example, the quantiles can be taken as the 30th and 70th percentiles respectively.

[0095] when This indicates sufficient resource redundancy, with the core optimization goal being to improve efficiency and reduce latency. The transmission scheduling group with the highest temporary load is moved to the first transmission slot for transmission, and the remaining transmission scheduling groups are allocated subsequent transmission slots in descending order of temporary load.

[0096] when This indicates a balance between resource supply and demand, with the core optimization goal being stable execution and maintaining performance. Transmission is performed according to the initial time slot allocation plan described above.

[0097] when This indicates that resources are facing a bottleneck, and the core optimization goal is to quickly reduce voltage and prevent overload. The transmission scheduling group with the lowest temporary load is moved to the first transmission slot for transmission, and the remaining transmission scheduling groups are allocated subsequent transmission slots in order of temporary load from low to high.

[0098] Once the transmission time slot allocation order is determined, the system simultaneously sends transmission instructions to all meters in the transmission scheduling group assigned to that transmission time slot at the start of the corresponding transmission time slot, controlling them to complete data upload within that transmission time slot.

[0099] By adopting the aforementioned dual thresholds, firstly, precise matching of state and policy is achieved, enabling the system to take the most appropriate action in different scenarios; secondly, a stable buffer zone is formed between high and low thresholds through hysteresis design, effectively avoiding frequent policy switching (oscillation) caused by small fluctuations in pressure indicators, and greatly enhancing the robustness and stability of the scheduling system.

[0100] In summary, this invention optimizes the data transmission quality of smart meters through three logically rigorous and sequential steps: First, a standardized transmission sensitivity vector is constructed for each meter based on historical sensitivity and key update records, transforming behavioral patterns into quantifiable features; second, a clustering algorithm is used to classify the meter group by risk based on this transmission sensitivity vector, marking the sensitivity level of each meter; finally, an initial transmission scheduling group for load balancing is generated based on the sensitivity level, and combined with real-time calculated resource pressure indicators, a dual-threshold decision mechanism is used to dynamically adjust the execution sequence of each transmission scheduling group, outputting adaptive peak-shifting scheduling instructions, thereby achieving an overall improvement in load balancing, congestion avoidance, and transmission reliability at the system level.

[0101] This invention also provides a data transmission quality optimization system for smart meters. For example... Figure 2 As shown, the system includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement the data transmission quality optimization method for a smart meter according to the first aspect of the present invention.

[0102] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0103] 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 optimizing data transmission quality in smart meters, characterized in that, include: The system acquires a sensitivity sequence composed of the sensitivity calculated for each transmission of each meter in the meter group within a historical period, and a key update frequency calculated based on the sensitivity exceeding a preset threshold. The key update frequency is the ratio of the number of times the sensitivity exceeds the preset threshold to the total number of sensitivities. The meters report data to the concentrator according to a preset transmission cycle. Based on the sensitivity sequence and key update frequency, a transmission sensitivity vector is constructed for the corresponding electricity meter. Using all transmission sensitivity vectors as samples, the electricity meter group is clustered to obtain multiple electricity meter clusters, and a sensitivity level is assigned to each electricity meter cluster. The transmission sensitivity vector includes a normalized fluctuation coefficient and a key update frequency; Specifically, the coefficient of variation of the sensitivity sequence is calculated and normalized to obtain the normalized fluctuation coefficient; the ratio of the number of times the sensitivity exceeds the preset threshold within the historical period to the total number of transmissions within the historical period is calculated. Transmission sensitivity is the product of the normalized fluctuation coefficient and the key update frequency. Assigning a sensitivity level to each meter cluster includes: Using the transmission sensitivity vector of each meter as a sample, the K-means clustering algorithm is used to divide the meter group into several meter clusters; Calculate the cluster sensitivity index for each meter cluster, where the cluster sensitivity index is the arithmetic mean of the transmission sensitivity of all meters in the meter cluster. The sensitivity levels of each meter cluster are sequentially marked according to the cluster sensitivity index in ascending order; A transmission cycle is divided into multiple transmission time slots, and a transmission scheduling group is established for each transmission time slot. The meter group is allocated to multiple transmission scheduling groups according to the sensitivity level of each meter, forming an initial correspondence between each transmission scheduling group and the transmission time slot. At the start of the current transmission cycle, collect the current network status parameters and the concentrator calculates the resource status parameters, and calculate the resource pressure at the start of the current transmission cycle based on all collected status parameters. The resource pressure is compared with a preset first pressure threshold and a second pressure threshold; based on the comparison result, the sum of the transmission sensitivities of all meters in the transmission scheduling group is used as the load, and the corresponding order of the multiple transmission scheduling groups and transmission slots is rearranged according to the load level; wherein, the transmission sensitivity is calculated from the components in the transmission sensitivity vector; At the beginning of each transmission time slot, a transmission instruction is sent to all meters in the transmission scheduling group corresponding to that transmission time slot.

2. The method for optimizing data transmission quality in smart meters according to claim 1, characterized in that, The method of allocating the electricity meter group to multiple transmission scheduling groups according to the sensitivity level of each meter includes: Electricity meters with a sensitivity level greater than the preset threshold are classified as high-sensitivity meters. Each high-sensitivity meter is then assigned to all transmission scheduling groups in a round-robin fashion according to its sensitivity level from high to low. Electricity meters with a sensitivity level less than or equal to a preset threshold are classified as low-sensitivity meters. A greedy load balancing algorithm is used to sequentially assign each low-sensitivity meter to the transmission scheduling group with the lowest initial load. The initial load of the transmission scheduling group is the sum of the transmission sensitivities of the meters already allocated within the group.

3. The method for optimizing data transmission quality in smart meters according to claim 2, characterized in that, The process of using the greedy equilibrium algorithm includes: Create an empty member list for each transmission scheduling group and initialize the load of each transmission scheduling group to zero; Select any low-sensitivity meter to be assigned as the target meter, and calculate the temporary load of each transmission scheduling group after temporarily adding the target meter to each group. Construct a load set by combining the initial load of all other transmission scheduling groups with the temporary load of the group containing the target meter. Then, iterate through all transmission scheduling groups to assign the target meter, resulting in multiple load sets. Calculate the sample variance of each load set, and select the transmission scheduling group containing the target meter corresponding to the load set with the smallest sample variance as the target group. Then, assign the corresponding low-sensitivity meter... Add the target group to the empty member list and update the temporary load of the target group; The algorithm terminates once all low-sensitivity meters have been allocated.

4. The method for optimizing data transmission quality in smart meters according to claim 1, characterized in that, The calculation process for the resource pressure includes: Extract the bandwidth utilization and packet loss rate from the network status parameters and the CPU utilization and memory usage rate from the computing resource status parameters; calculate the product of bandwidth utilization and packet loss rate, and the product of CPU utilization and memory usage rate for the current transmission cycle; The average of the product of bandwidth utilization and packet loss rate of all transmission slots in the previous transmission cycle of the current transmission cycle is used as the first average. The average of the product of CPU utilization and memory utilization of all transmission slots in the previous transmission cycle of the current transmission cycle is used as the second average. The first average is divided by the sum of the first average and the second average to obtain the network state weight. The difference between 1 and the network state weight is used as the resource state weight. The first component is obtained by multiplying the network state weight by the product of the bandwidth utilization and packet loss rate of the current transmission cycle; the second component is obtained by multiplying the resource state weight by the product of the CPU utilization and memory occupancy of the current transmission cycle; the resource pressure is obtained by adding the first component and the second component.

5. A method for optimizing data transmission quality in smart meters according to claim 1 or 4, characterized in that, The comparison results include: The first pressure threshold is less than the second pressure threshold; When the resource pressure is less than or equal to the first pressure threshold, the transmission scheduling group with the highest temporary load is adjusted to the first transmission slot, and the remaining transmission scheduling groups are allocated subsequent transmission slots in order of temporary load from high to low. When the resource pressure is greater than the first pressure threshold and less than the second pressure threshold, the initial correspondence is maintained; When the resource pressure is greater than or equal to the second pressure threshold, the transmission scheduling group with the lowest current load is adjusted to the first transmission slot, and the remaining transmission scheduling groups are assigned to the subsequent transmission slots in order of load from low to high.

6. A data transmission quality optimization system for smart meters, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the data transmission quality optimization method for a smart meter according to any one of claims 1-5.

Images